Reviewer #1 (Public Review):

By identifying a loss of function mutant of IQCH in infertile patient, Ruan et al. shows that IQCH is essential for spermiogenesis by generating a knockout mouse model of IQCH. Similar to infertile patient with mutant of IQCH, Iqch knockout mice are characterized by a cracked flagellar axoneme and abnormal mitochondrial structure. Mechanistically, IQCH regulates the expression of RNA-binding proteins (especially HNRPAB), which are indispensable for spermatogenesis.

Although this manuscript contains a potentially interesting piece of work that delineates a mechanism of IQCH that associates with spermatogenesis, this reviewer feels that a number of issues require clarification and re-evaluation for a better understanding of the role of IQCH in spermatogenesis.

Line 251 - 253, "To elucidate the molecular mechanism by which IQCH regulates male fertility, we performed liquid chromatography tandem mass spectrometry (LC‒MS/MS) analysis using mouse sperm lysates and detected 288 interactors of IQCH (Figure 5-source data 1)."

The reviewer had already raised significant concerns regarding the text above, noting that "LC‒MS/MS analysis using mouse sperm lysates" would not identify interactors of IQCH. However, this issue was not addressed in the revised manuscript. In the Methods section detailing LC-MS/MS, the authors stated that it was conducted on "eluates obtained from IP". However, there was no explanation provided on how IP for LC-MS/MS was performed. Additionally, it was unclear whether LC-MS or LC-MS/MS was utilized. The primary concern is that if LC‒MS/MS was conducted for the IP of IQCH, IQCH itself should have been detected in the results; however, as indicated by Figure 5-source data 1, IQCH was not listed.

Reviewer #3 (Public Review):

In this study, Ruan et al. investigate the role of the IQCH gene in spermatogenesis, focusing on its interaction with calmodulin and its regulation of RNA-binding proteins. The authors examined sperm from a male infertility patient with an inherited IQCH mutation as well as Iqch CRISPR knockout mice. The authors found that both human and mouse sperm exhibited structural and morphogenetic defects in multiple structures, leading to reduced fertility in Ichq-knockout male mice. Molecular analyses such as mass spectrometry and immunoprecipitation indicated that RNA-binding proteins are likely targets of IQCH, with the authors focusing on the RNA-binding protein HNRPAB as a critical regulator of testicular mRNAs. The authors used in vitro cell culture models to demonstrate an interaction between IQCH and calmodulin, in addition to showing that this interaction via the IQ motif of IQCH is required for IQCH's function in promoting HNRPAB expression. In sum, the authors concluded that IQCH promotes male fertility by binding to calmodulin and controlling HNRPAB expression to regulate the expression of essential mRNAs for spermatogenesis. These findings provide new insight into molecular mechanisms underlying spermatogenesis and how important factors for sperm morphogenesis and function are regulated.

The strengths of the study include the use of mouse and human samples, which demonstrate a likely relevance of the mouse model to humans; the use of multiple biochemical techniques to address the molecular mechanisms involved; the development of a new CRISPR mouse model; ample controls; and clearly displayed results. Assays are done rigorously and in a quantitative manner. Overall, the claims made by the authors in this manuscript are well-supported by the data provided.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Strengths:

The study was designed as a 6-month follow-up, with repeated behavioral and EEG measurements through disease development, providing valuable and interesting findings on AD progression and the effect of early-life choline supplantation. Moreover, the behavioral data that suggest an adverse effect of low choline in WT mice are interesting and important beyond the context of AD.

Thank you for identifying several strengths.

Weaknesses:

(1) The multiple headings and subheadings, focusing on the experimental method rather than the narrative, reduce the readability.

We have reduced the number of headings.

(2) Quantification of NeuN and FosB in WT littermates is needed to demonstrate rescue of neuronal death and hyperexcitability by high choline supplementation and also to gain further insights into the adverse effect of low choline on the performance of WT mice in the behavioral test.

We agree and have added WT data for the NeuN and ΔFosB analyses. These data are included in the text and figures. For NeuN, the Figure is Figure 6. For ΔFosB it is Figure 7. In brief, the high choline diet restored NeuN and ΔFosB to the levels of WT mice.

Below is Figure 6 and its legend to show the revised presentation of data for NeuN. Afterwards is the revised figure showing data for ΔFosB. After that are the sections of the Results that have been revised.

Author response image 1.

Choline supplementation improved NeuN immunoreactivity (ir) in hilar cells in Tg2576 animals. A. Representative images of NeuN-ir staining in the anterior DG of Tg2576 animals. (1) A section from a Tg2576 mouse fed the low choline diet. The area surrounded by a box is expanded below. Red arrows point to NeuN-ir hilar cells. Mol=molecular layer, GCL=granule cell layer, HIL=hilus. Calibration for the top row, 100 µm; for the bottom row, 50 µm. (2) A section from a Tg2576 mouse fed the intermediate diet. Same calibrations as for 1. (3) A section from a Tg2576 mouse fed the high choline diet. Same calibrations as for 1. B. Quantification methods. Representative images demonstrate the thresholding criteria used to quantify NeuN-ir. (1) A NeuN-stained section. The area surrounded by the white box is expanded in the inset (arrow) to show 3 hilar cells. The 2 NeuN-ir cells above threshold are marked by blue arrows. The 1 NeuN-ir cell below threshold is marked by a green arrow. (2) After converting the image to grayscale, the cells above threshold were designated as red. The inset shows that the two cells that were marked by blue arrows are red while the cell below threshold is not. (3) An example of the threshold menu from ImageJ showing the way the threshold was set. Sliders (red circles) were used to move the threshold to the left or right of the histogram of intensity values. The final position of the slider (red arrow) was positioned at the onset of the steep rise of the histogram. C. NeuN-ir in Tg2576 and WT mice. Tg2576 mice had either the low, intermediate, or high choline diet in early life. WT mice were fed the standard diet (intermediate choline). (1) Tg2576 mice treated with the high choline diet had significantly more hilar NeuN-ir cells in the anterior DG compared to Tg2576 mice that had been fed the low choline or intermediate diet. The values for Tg2576 mice that received the high choline diet were not significantly different from WT mice, suggesting that the high choline diet restored NeuN-ir. (2) There was no effect of diet or genotype in the posterior DG, probably because the low choline and intermediate diet did not appear to lower hilar NeuN-ir.

Author response image 2.

Choline supplementation reduced ∆FosB expression in dorsal GCs of Tg2576 mice. A. Representative images of ∆FosB staining in GCL of Tg2576 animals from each treatment group. (1) A section from a low choline-treated mouse shows robust ∆FosB-ir in the GCL. Calibration, 100 µm. Sections from intermediate (2) and high choline (3)-treated mice. Same calibration as 1. B. Quantification methods. Representative images demonstrating the thresholding criteria established to quantify ∆FosB. (1) A ∆FosB -stained section shows strongly-stained cells (white arrows). (2) A strict thresholding criteria was used to make only the darkest stained cells red. C. Use of the strict threshold to quantify ∆FosB-ir. (1) Anterior DG. Tg2576 mice treated with the choline supplemented diet had significantly less ∆FosB-ir compared to the Tg2576 mice fed the low or intermediate diets. Tg2576 mice fed the high choline diet were not significantly different from WT mice, suggesting a rescue of ∆FosB-ir. (2) There were no significant differences in ∆FosB-ir in posterior sections. D. Methods are shown using a threshold that was less strict. (1) Some of the stained cells that were included are not as dark as those used for the strict threshold (white arrows). (2) All cells above the less conservative threshold are shown in red. E. Use of the less strict threshold to quantify ∆FosB-ir. (1) Anterior DG. Tg2576 mice that were fed the high choline diet had less ΔFosB-ir pixels than the mice that were fed the other diets. There were no differences from WT mice, suggesting restoration of ∆FosB-ir by choline enrichment in early life. (2) Posterior DG. There were no significant differences between Tg2576 mice fed the 3 diets or WT mice.

Results, Section C1, starting on Line 691:

“To ask if the improvement in NeuN after MCS in Tg256 restored NeuN to WT levels we used WT mice. For this analysis we used a one-way ANOVA with 4 groups: Low choline Tg2576, Intermediate Tg2576, High choline Tg2576, and Intermediate WT (Figure 5C). Tukey-Kramer multiple comparisons tests were used as the post hoc tests. The WT mice were fed the intermediate diet because it is the standard mouse chow, and this group was intended to reflect normal mice. The results showed a significant group difference for anterior DG (F(3,25)=9.20; p=0.0003; Figure 5C1) but not posterior DG (F(3,28)=0.867; p=0.450; Figure 5C2). Regarding the anterior DG, there were more NeuN-ir cells in high choline-treated mice than both low choline (p=0.046) and intermediate choline-treated Tg2576 mice (p=0.003). WT mice had more NeuN-ir cells than Tg2576 mice fed the low (p=0.011) or intermediate diet (p=0.003). Tg2576 mice that were fed the high choline diet were not significantly different from WT (p=0.827).”

Results, Section C2, starting on Line 722:

“There was strong expression of ∆FosB in Tg2576 GCs in mice fed the low choline diet (Figure 7A1). The high choline diet and intermediate diet appeared to show less GCL ΔFosB-ir (Figure 7A2-3). A two-way ANOVA was conducted with the experimental group (Tg2576 low choline diet, Tg2576 intermediate choline diet, Tg2576 high choline diet, WT intermediate choline diet) and location (anterior or posterior) as main factors. There was a significant effect of group (F(3,32)=13.80, p=<0.0001) and location (F(1,32)=8.69, p=0.006). Tukey-Kramer post-hoc tests showed that Tg2576 mice fed the low choline diet had significantly greater ΔFosB-ir than Tg2576 mice fed the high choline diet (p=0.0005) and WT mice (p=0.0007). Tg2576 mice fed the low and intermediate diets were not significantly different (p=0.275). Tg2576 mice fed the high choline diet were not significantly different from WT (p>0.999). There were no differences between groups for the posterior DG (all p>0.05).”

“∆FosB quantification was repeated with a lower threshold to define ∆FosB-ir GCs (see Methods) and results were the same (Figure 7D). Two-way ANOVA showed a significant effect of group (F(3,32)=14.28, p< 0.0001) and location (F(1,32)=7.07, p=0.0122) for anterior DG but not posterior DG (Figure 7D). For anterior sections, Tukey-Kramer post hoc tests showed that low choline mice had greater ΔFosB-ir than high choline mice (p=0.0024) and WT mice (p=0.005) but not Tg2576 mice fed the intermediate diet (p=0.275); Figure 7D1). Mice fed the high choline diet were not significantly different from WT (p=0.993; Figure 7D1). These data suggest that high choline in the diet early in life can reduce neuronal activity of GCs in offspring later in life. In addition, low choline has an opposite effect, suggesting low choline in early life has adverse effects.”

(3) Quantification of the discrimination ratio of the novel object and novel location tests can facilitate the comparison between the different genotypes and diets.

We have added the discrimination index for novel object location to the paper. The data are in a new figure: Figure 3. In brief, the results for discrimination index are the same as the results done originally, based on the analysis of percent of time exploring the novel object.

Below is the new Figure and legend, followed by the new text in the Results.

Author response image 3.

Novel object location results based on the discrimination index. A. Results are shown for the 3 months-old WT and Tg2576 mice based on the discrimination index. (1) Mice fed the low choline diet showed object location memory only in WT. (2) Mice fed the intermediate diet showed object location memory only in WT. (3) Mice fed the high choline diet showed memory both for WT and Tg2576 mice. Therefore, the high choline diet improved memory in Tg2576 mice. B. The results for the 6 months-old mice are shown. (1-2) There was no significant memory demonstrated by mice that were fed either the low or intermediate choline diet. (3) Mice fed a diet enriched in choline showed memory whether they were WT or Tg2576 mice. Therefore, choline enrichment improved memory in all mice.

Results, Section B1, starting on line 536:

“The discrimination indices are shown in Figure 3 and results led to the same conclusions as the analyses in Figure 2. For the 3 months-old mice (Figure 3A), the low choline group did not show the ability to perform the task for WT or Tg2576 mice. Thus, a two-way ANOVA showed no effect of genotype (F(1,74)=0.027, p=0.870) or task phase (F(1,74)=1.41, p=0.239). For the intermediate diet-treated mice, there was no effect of genotype (F(1,50)=0.3.52, p=0.067) but there was an effect of task phase (F(1,50)=8.33, p=0.006). WT mice showed a greater discrimination index during testing relative to training (p=0.019) but Tg2576 mice did not (p=0.664). Therefore, Tg2576 mice fed the intermediate diet were impaired. In contrast, high choline-treated mice performed well. There was a main effect of task phase (F(1,68)=39.61, p=<0.001) with WT (p<0.0001) and Tg2576 mice (p=0.0002) showing preference for the moved object in the test phase. Interestingly, there was a main effect of genotype (F(1,68)=4.50, p=0.038) because the discrimination index for WT training was significantly different from Tg2576 testing (p<0.0001) and Tg2576 training was significantly different from WT testing (p=0.0003).”

“The discrimination indices of 6 months-old mice led to the same conclusions as the results in Figure 2. There was no evidence of discrimination in low choline-treated mice by two-way ANOVA (no effect of genotype, (F(1,42)=3.25, p=0.079; no effect of task phase, F(1,42)=0.278, p=0.601). The same was true of mice fed the intermediate diet (genotype, F(1,12)=1.44, p=0.253; task phase, F(1,12)=2.64, p=0.130). However, both WT and Tg2576 mice performed well after being fed the high choline diet (effect of task phase, (F(1,52)=58.75, p=0.0001, but not genotype (F(1,52)=1.197, p=0.279). Tukey-Kramer post-hoc tests showed that both WT (p<0.0001) and Tg2576 mice that had received the high choline diet (p=0.0005) had elevated discrimination indices for the test session.”

(4) The longitudinal analyses enable the performance of multi-level correlations between the discrimination ratio in NOR and NOL, NeuN and Fos levels, multiple EEG parameters, and premature death. Such analysis can potentially identify biomarkers associated with AD progression. These can be interesting in different choline supplementation, but also in the standard choline diet.

We agree and added correlations to the paper in a new figure (Figure 9). Below is Figure 9 and its legend. Afterwards is the new Results section.

Author response image 4.

Correlations between IIS, Behavior, and hilar NeuN-ir. A. IIS frequency over 24 hrs is plotted against the preference for the novel object in the test phase of NOL. A greater preference is reflected by a greater percentage of time exploring the novel object. (1) The mice fed the high choline diet (red) showed greater preference for the novel object when IIS were low. These data suggest IIS impaired object location memory in the high choline-treated mice. The low choline-treated mice had very weak preference and very few IIS, potentially explaining the lack of correlation in these mice. (2) There were no significant correlations for IIS and NOR. However, there were only 4 mice for the high choline group, which is a limitation. B. IIS frequency over 24 hrs is plotted against the number of dorsal hilar cells expressing NeuN. The dorsal hilus was used because there was no effect of diet on the posterior hilus. (1) Hilar NeuN-ir is plotted against the preference for the novel object in the test phase of NOL. There were no significant correlations. (2) Hilar NeuN-ir was greater for mice that had better performance in NOR, both for the low choline (blue) and high choline (red) groups. These data support the idea that hilar cells contribute to object recognition (Kesner et al. 2015; Botterill et al. 2021; GoodSmith et al. 2022).

Results, Section F, starting on Line 801:

“F. Correlations between IIS and other measurements

As shown in Figure 9A, IIS were correlated to behavioral performance in some conditions. For these correlations, only mice that were fed the low and high choline diets were included because mice that were fed the intermediate diet did not have sufficient EEG recordings in the same mouse where behavior was studied. IIS frequency over 24 hrs was plotted against the preference for the novel object in the test phase (Figure 9A). For NOL, IIS were significantly less frequent when behavior was the best, but only for the high choline-treated mice (Pearson’s r, p=0.022). In the low choline group, behavioral performance was poor regardless of IIS frequency (Pearson’s r, p=0.933; Figure 9A1). For NOR, there were no significant correlations (low choliNe, p=0.202; high choline, p=0.680) but few mice were tested in the high choline-treated mice (Figure 9B2).

We also tested whether there were correlations between dorsal hilar NeuN-ir cell numbers and IIS frequency. In Figure 9B, IIS frequency over 24 hrs was plotted against the number of dorsal hilar cells expressing NeuN. The dorsal hilus was used because there was no effect of diet on the posterior hilus. For NOL, there was no significant correlation (low choline, p=0.273; high choline, p=0.159; Figure 9B1). However, for NOR, there were more NeuN-ir hilar cells when the behavioral performance was strongest (low choline, p=0.024; high choline, p=0.016; Figure 9B2). These data support prior studies showing that hilar cells, especially mossy cells (the majority of hilar neurons), contribute to object recognition (Botterill et al. 2021; GoodSmith et al. 2022).”

We also noted that all mice were not possible to include because they died or other reasons, such a a loss of the headset (Results, Section A, Lines 463-464): Some mice were not possible to include in all assays either because they died before reaching 6 months or for other reasons.

Reviewer #2 (Public Review):

Strengths:

The strength of the group was the ability to monitor the incidence of interictal spikes (IIS) over the course of 1.2-6 months in the Tg2576 Alzheimer's disease model, combined with meaningful behavioral and histological measures. The authors were able to demonstrate MCS had protective effects in Tg2576 mice, which was particularly convincing in the hippocampal novel object location task.

We thank the Reviewer for identifying several strengths.

Weaknesses:

Although choline deficiency was associated with impaired learning and elevated FosB expression, consistent with increased hyperexcitability, IIS was reduced with both low and high choline diets. Although not necessarily a weakness, it complicates the interpretation and requires further evaluation.

We agree and we revised the paper to address the evaluations that were suggested.

Reviewer #1 (Recommendations For The Authors):

(1) A reference directing to genotyping of Tg2576 mice is missing.

We apologize for the oversight and added that the mice were genotyped by the New York University Mouse Genotyping core facility.

Methods, Section A, Lines 210-211: “Genotypes were determined by the New York University Mouse Genotyping Core facility using a protocol to detect APP695.”

(2) Which software was used to track the mice in the behavioral tests?

We manually reviewed videos. This has been clarified in the revised manuscript. Methods, Section B4, Lines 268-270: Videos of the training and testing sessions were analyzed manually. A subset of data was analyzed by two independent blinded investigators and they were in agreement.

(3) Unexpectedly, a low choline diet in AD mice was associated with reduced frequency of interictal spikes yet increased mortality and spontaneous seizures. The authors attribute this to postictal suppression.

We did not intend to suggest that postictal depression was the only cause. It was a suggestion for one of many potential explanations why seizures would influence IIS frequency. For postictal depression, we suggested that postictal depression could transiently reduce IIS. We have clarified the text so this is clear (Discussion, starting on Line 960):

If mice were unhealthy, IIS might have been reduced due to impaired excitatory synaptic function. Another reason for reduced IIS is that the mice that had the low choline diet had seizures which interrupted REM sleep. Thus, seizures in Tg2576 mice typically started in sleep. Less REM sleep would reduce IIS because IIS occur primarily in REM. Also, seizures in the Tg2576 mice were followed by a depression of the EEG (postictal depression; Supplemental Figure 3) that would transiently reduce IIS. A different, radical explanation is that the intermediate diet promoted IIS rather than low choline reducing IIS. Instead of choline, a constituent of the intermediate diet may have promoted IIS.

However, reduced spike frequency is already evident at 5 weeks of age, a time point with a low occurrence of premature death. A more comprehensive analysis of EEG background activity may provide additional information if the epileptic activity is indeed reduced at this age.

We did not intend to suggest that premature death caused reduced spike frequency. We have clarified the paper accordingly. We agree that a more in-depth EEG analysis would be useful but is beyond the scope of the study.

(4) Supplementary Fig. 3 depicts far more spikes / 24 h compared to Fig. 7B (at least 100 spikes/24h in Supplementary Fig. 3 and less than 10 spikes/24h in Fig. 7B).

We would like to clarify that before and after a seizure the spike frequency is unusually high. Therefore, there are far more spikes than prior figures.

We clarified this issue by adding to the Supplemental Figure more data. The additional data are from mice without a seizure, showing their spikes are low in frequency.

All recordings lasted several days. We included the data from mice with a seizure on one of the days and mice without any seizures. For mice with a seizure, we graphed IIS frequency for the day before, the day of the seizure, and the day after. For mice without a seizure, IIS frequency is plotted for 3 consecutive days. When there was a seizure, the day before and after showed high numbers of spikes. When there was no seizure on any of the 3 days, spikes were infrequent on all days.

The revised figure and legend are shown below. It is Supplemental Figure 4 in the revised submission.

Author response image 5.

IIS frequency before and after seizures. A. Representative EEG traces recorded from electrodes implanted in the skull over the left frontal cortex, right occipital cortex, left hippocampus (Hippo) and right hippocampus during a spontaneous seizure in a 5 months-old Tg2576 mouse. Arrows point to the start (green arrow) and end of the seizure (red arrow), and postictal depression (blue arrow). B. IIS frequency was quantified from continuous video-EEG for mice that had a spontaneous seizure during the recording period and mice that did not. IIS frequency is plotted for 3 consecutive days, starting with the day before the seizure (designated as day 1), and ending with the day after the seizure (day 3). A two-way RMANOVA was conducted with the day and group (mice with or without a seizure) as main factors. There was a significant effect of day (F(2,4)=46.95, p=0.002) and group (seizure vs no seizure; F(1,2)=46.01, p=0.021) and an interaction of factors (F(2,4)=46.68, p=0.002)..Tukey-Kramer post-hoc tests showed that mice with a seizure had significantly greater IIS frequencies than mice without a seizure for every day (day 1, p=0.0005; day 2, p=0.0001; day 3, p=0.0014). For mice with a seizure, IIS frequency was higher on the day of the seizure than the day before (p=0.037) or after (p=0.010). For mice without a seizure, there were no significant differences in IIS frequency for day 1, 2, or 3. These data are similar to prior work showing that from one day to the next mice without seizures have similar IIS frequencies (Kam et al., 2016).

In the text, the revised section is in the Results, Section C, starting on Line 772:

“At 5-6 months, IIS frequencies were not significantly different in the mice fed the different diets (all p>0.05), probably because IIS frequency becomes increasingly variable with age (Kam et al. 2016). One source of variability is seizures, because there was a sharp increase in IIS during the day before and after a seizure (Supplemental Figure 4). Another reason that the diets failed to show differences was that the IIS frequency generally declined at 5-6 months. This can be appreciated in Figure 8B and Supplemental Figure 6B. These data are consistent with prior studies of Tg2576 mice where IIS increased from 1 to 3 months but then waxed and waned afterwards (Kam et al., 2016).”

(5) The data indicating the protective effect of high choline supplementation are valuable, yet some of the claims are not completely supported by the data, mainly as the analysis of littermate WT mice is not complete.

We added WT data to show that the high choline diet restored cell loss and ΔFosB expression to WT levels. These data strengthen the argument that the high choline diet was valuable. See the response to Reviewer #1, Public Review Point #2.

• Line 591: "The results suggest that choline enrichment protected hilar neurons from NeuN loss in Tg2576 mice." A comparison to NeuN expression in WT mice is needed to make this statement.

These data have been added. See the response to Reviewer #1, Public Review Point #2.

• Line 623: "These data suggest that high choline in the diet early in life can reduce hyperexcitability of GCs in offspring later in life. In addition, low choline has an opposite effect, again suggesting this maternal diet has adverse effects." Also here, FosB quantification in WT mice is needed.

These data have been added. See the response to Reviewer #1, Public Review Point #2.

(7) Was the effect of choline associated with reduced tauopathy or A levels?

The mice have no detectable hyperphosphorylated tau. The mice do have intracellular A before 6 months. This is especially the case in hilar neurons, but GCs have little (Criscuolo et al., eNeuro, 2023). However, in neurons that have reduced NeuN, we found previously that antibodies generally do not work well. We think it is because the neurons become pyknotic (Duffy et al., 2015), a condition associated with oxidative stress which causes antigens like NeuN to change conformation due to phosphorylation. Therefore, we did not conduct a comparison of hilar neurons across the different diets.

(8) Since the mice were tested at 3 months and 6 months, it would be interesting to see the behavioral difference per mouse and the correlation with EEG recording and immunohistological analyses.

We agree that would be valuable and this has been added to the paper. Please see response to Reviewer #1, Public Review Point #4.

Reviewer #2 (Recommendations For The Authors):

There were several areas that could be further improved, particularly in the areas of data analysis (particularly with images and supplemental figures), figure presentation, and mechanistic speculation.

Major points:

(1) It is understandable that, for the sake of labor and expense, WT mice were not implanted with EEG electrodes, particularly since previous work showed that WT mice have no IIS (Kam et al. 2016). However, from a standpoint of full factorial experimental design, there are several flaws - purists would argue are fatal flaws. First, the lack of WT groups creates underpowered and imbalanced groups, constraining statistical comparisons and likely reducing the significance of the results. Also, it is an assumption that diet does not influence IIS in WT mice. Secondly, with a within-subject experimental design (as described in Fig. 1A), 6-month-old mice are not naïve if they have previously been tested at 3 months. Such an experimental design may reduce effect size compared to non-naïve mice. These caveats should be included in the Discussion. It is likely that these caveats reduce effect size and that the actual statistical significance, were the experimental design perfect, would be higher overall.

We agree and have added these points to the Limitations section of the Discussion. Starting on Line 1050: In addition, groups were not exactly matched. Although WT mice do not have IIS, a WT group for each of the Tg2576 groups would have been useful. Instead, we included WT mice for the behavioral tasks and some of the anatomical assays. Related to this point is that several mice died during the long-term EEG monitoring of IIS.

(2) Since behavior, EEG, NeuN and FosB experiments seem to be done on every Tg2576 animal, it seems that there are missed opportunities to correlate behavior/EEG and histology on a per-mouse basis. For example, rather than speculate in the discussion, why not (for example) directly examine relationships between IIS/24 hours and FosB expression?

We addressed this point above in responding to Reviewer #1, Public Review Point #4.

(3) Methods of image quantification should be improved. Background subtraction should be considered in the analysis workflow (see Fig. 5C and Fig. 6C background). It would be helpful to have a Methods figure illustrating intermediate processing steps for both NeuN and FosB expression.

We added more information to improve the methods of quantification. We did use a background subtraction approach where ImageJ provides a histogram of intensity values, and it determines when there is a sharp rise in staining relative to background. That point is where we set threshold. We think it is a procedure that has the least subjectivity.

We added these methods to the Methods section and expanded the first figure about image quantification, Figure 6B. That figure and legend are shown above in response to Reviewer #1, Point #2.

This is the revised section of the Methods, Section C3, starting on Line 345:

“Photomicrographs were acquired using ImagePro Plus V7.0 (Media Cybernetics) and a digital camera (Model RET 2000R-F-CLR-12, Q-Imaging). NeuN and ∆FosB staining were quantified from micrographs using ImageJ (V1.44, National Institutes of Health). All images were first converted to grayscale and in each section, the hilus was traced, defined by zone 4 of Amaral (1978). A threshold was then calculated to identify the NeuN-stained cell bodies but not background. Then NeuN-stained cell bodies in the hilus were quantified manually. Note that the threshold was defined in ImageJ using the distribution of intensities in the micrograph. A threshold was then set using a slider in the histogram provided by Image J. The slider was pushed from the low level of staining (similar to background) to the location where staining intensity made a sharp rise, reflecting stained cells. Cells with labeling that was above threshold were counted.”

(4) This reviewer is surprised that the authors do not speculate more about ACh-related mechanisms. For example, choline deficiency would likely reduce Ach release, which could have the same effect on IIS as muscarinic antagonism (Kam et al. 2016), and could potentially explain the paradoxical effects of a low choline diet on reducing IIS. Some additional mechanistic speculation would be helpful in the Discussion.

We thank the Reviewer for noting this so we could add it to the Discussion. We had not because we were concerned about space limitations.

The Discussion has a new section starting on Line 1009:

“Choline and cholinergic neurons

There are many suggestions for the mechanisms that allow MCS to improve health of the offspring. One hypothesis that we are interested in is that MCS improves outcomes by reducing IIS. Reducing IIS would potentially reduce hyperactivity, which is significant because hyperactivity can increase release of A. IIS would also be likely to disrupt sleep since it represents aberrant synchronous activity over widespread brain regions. The disruption to sleep could impair memory consolidation, since it is a notable function of sleep (Graves et al. 2001; Poe et al. 2010). Sleep disruption also has other negative consequences such as impairing normal clearance of A (Nedergaard and Goldman 2020). In patients, IIS and similar events, IEDs, are correlated with memory impairment (Vossel et al. 2016).

How would choline supplementation in early life reduce IIS of the offspring? It may do so by making BFCNs more resilient. That is significant because BFCN abnormalities appear to cause IIS. Thus, the cholinergic antagonist atropine reduced IIS in vivo in Tg2576 mice. Selective silencing of BFCNs reduced IIS also. Atropine also reduced elevated synaptic activity of GCs in young Tg2576 mice in vitro. These studies are consistent with the idea that early in AD there is elevated cholinergic activity (DeKosky et al. 2002; Ikonomovic et al. 2003; Kelley et al. 2014; Mufson et al. 2015; Kelley et al. 2016), while later in life there is degeneration. Indeed, the chronic overactivity could cause the degeneration.

Why would MCS make BFCNs resilient? There are several possibilities that have been explored, based on genes upregulated by MCS. One attractive hypothesis is that neurotrophic support for BFCNs is retained after MCS but in aging and AD it declines (Gautier et al. 2023). The neurotrophins, notably nerve growth factor (NGF) and brain-derived neurotrophic factor (BDNF) support the health of BFCNs (Mufson et al. 2003; Niewiadomska et al. 2011).”

Minor points:

(1) The vendor is Dyets Inc., not Dyets.

Thank you. This correction has been made.

(2) Anesthesia chamber not specified (make, model, company).

We have added this information to the Methods, Section D1, starting on Line 375: The animals were anesthetized by isoflurane inhalation (3% isoflurane. 2% oxygen for induction) in a rectangular transparent plexiglas chamber (18 cm long x 10 cm wide x 8 cm high) made in-house.

(3) It is not clear whether software was used for the detection of behavior. Was position tracking software used or did blind observers individually score metrics?

We have added the information to the paper. Please see the response to Reviewer #1, Recommendations for Authors, Point #2.

(4) It is not clear why rat cages and not a true Open Field Maze were used for NOL and NOR.

We used mouse cages because in our experience that is what is ideal to detect impairments in Tg2576 mice at young ages. We think it is why we have been so successful in identifying NOL impairments in young mice. Before our work, most investigators thought behavior only became impaired later. We would like to add that, in our experience, an Open Field Maze is not the most common cage that is used.

(5) Figure 1A is not mentioned.

It had been mentioned in the Introduction. Figure B-D was the first Figure mentioned in the Results so that is why it might have been missed. We now have added it to the first section of the Results, Line 457, so it is easier to find.

6) Although Fig 7 results are somewhat complicated compared to Fig. 5 and 6 results, EEG comes chronologically earlier than NeuN and FosB expression experiments.

We have kept the order as is because as the Reviewer said, the EEG is complex. For readability, we have kept the EEG results last.

(7) Though the statistical analysis involved parametric and nonparametric tests, It is not clear which normality tests were used.

We have added the name of the normality tests in the Methods, Section E, Line 443: Tests for normality (Shapiro-Wilk) and homogeneity of variance (Bartlett’s test) were used to determine if parametric statistics could be used. We also added after this sentence clarification: When data were not normal, non-parametric data were used. When there was significant heteroscedasticity of variance, data were log transformed. If log transformation did not resolve the heteroscedasticity, non-parametric statistics were used. Because we added correlations and analysis of survival curves, we also added the following (starting on Line 451): For correlations, Pearson’s r was calculated. To compare survival curves, a Log rank (Mantel-Cox) test was performed.

Figures:

(1) In Fig. 1A, Anatomy should be placed above the line.

We changed the figure so that the word “Anatomy” is now aligned, and the arrow that was angled is no longer needed.

In Fig. 1C and 1D, the objects seem to be moved into the cage, not the mice. This schematic does not accurately reflect the Fig. 1C and 1D figure legend text.

Thank you for the excellent point. The figure has been revised. We also updated it to show the objects more accurately.

Please correct the punctuation in the Fig. 1D legend.

Thank you for mentioning the errors. We corrected the legend.

For ease of understanding, Fig. 1C and 1D should have training and testing labeled in the figure.

Thank you for the suggestion. We have revised the figure as suggested.

Author response image 6.

(2) In Figure 2, error bars for population stats (bar graphs) are not obvious or missing. Same for Figure 3.

We added two supplemental figures to show error bars, because adding the error bars to the existing figures made the symbols, colors, connecting lines and error bars hard to distinguish. For novel object location (Fig. 2) the error bars are shown in Supp. Fig. 2. For novel object recognition, the error bars are shown in Supplemental Fig. 3.

(3) The authors should consider a Methods figure for quantification of NeuN and deltaFOSB (expansions of Fig. 5C and Fig. 6C).

Please see Reviewer #1, Public Review Point #2.

(4) In Figure 5, A should be omitted and mentioned in the Methods/figure legend. B should be enlarged. C should be inset, zoomed-in images of the hilus, with an accompanying analysis image showing a clear reduction in NeuN intensity in low choline conditions compared to intermediate and high choline conditions. In D, X axes could delineate conditions (figure legend and color unnecessary). Figure 5C should be moved to a Methods figure.

We thank the review for the excellent suggestions. We removed A as suggested. We expanded B and included insets. We used different images to show a more obvious reduction of cells for the low choline group. We expanded the Methods schematics. The revised figure is Figure 6 and shown above in response to Reviewer 1, Public Review Point #2.

(5) In Figure 6, A should be eliminated and mentioned in the Methods/figure legend. B should be greatly expanded with higher and lower thresholds shown on subsequent panels (3x3 design).

We removed A as suggested. We expanded B as suggested. The higher and lower thresholds are shown in C. The revised figure is Figure 7 and shown above in response to Reviewer 1, Public Review Point #2.

(6) In Figure 7, A2 should be expanded vertically. A3 should be expanded both vertically and horizontally. B 1 and 2 should be increased, particularly B1 where it is difficult to see symbols. Perhaps colored symbols offset/staggered per group so that the spread per group is clearer.

We added a panel (A4) to show an expansion of A2 and A3. However, we did not see that a vertical expansion would add information so we opted not to add that. We expanded B1 as suggested but opted not to expand B2 because we did not think it would enhance clarity. The revised figure is below.

Author response image 7.

(7) Supplemental Figure 1 could possibly be combined with Figure 1 (use rounded corner rat cage schematic for continuity).

We opted not to combine figures because it would make one extremely large figure. As a result, the parts of the figure would be small and difficult to see.

(8) Supplemental Figure 2 - there does not seem to be any statistical analysis associated with A mentioned in the Results text.

We added the statistical information. It is now Supplemental Figure 4:

Author response image 8.

Mortality was high in mice treated with the low choline diet. A. Survival curves are shown for mice fed the low choline diet and mice fed the high choline diet. The mice fed the high choline diet had a significantly less severe survival curve. B. Left: A photo of a mouse after sudden unexplained death. The mouse was found in a posture consistent with death during a convulsive seizure. The area surrounded by the red box is expanded below to show the outstretched hindlimb (red arrow). Right: A photo of a mouse that did not die suddenly. The area surrounded by the box is expanded below to show that the hindlimb is not outstretched.

The revised text is in the Results, Section E, starting on Line 793:

“The reason that low choline-treated mice appeared to die in a seizure was that they were found in a specific posture in their cage which occurs when a severe seizure leads to death (Supplemental Figure 5). They were found in a prone posture with extended, rigid limbs (Supplemental Figure 5). Regardless of how the mice died, there was greater mortality in the low choline group compared to mice that had been fed the high choline diet (Log-rank (Mantel-Cox) test, Chi square 5.36, df 1, p=0.021; Supplemental Figure 5A).”

Also, why isn't intermediate choline also shown?

We do not have the data from the animals. Records of death were not kept, regrettably.

Perhaps labeling of male/female could also be done as part of this graph.

We agree this would be very interesting but do not have all sex information.

B is not very convincing, though it is understandable once one reads about posture.

We have clarified the text and figure, as well as the legend. They are above.

Are there additional animals that were seen to be in a specific posture?

There are many examples, and we added them to hopefully make it more convincing.

We also added posture in WT mice when there is a death to show how different it is.

Is there any relationship between seizures detected via EEG, as shown in Supplemental Figure 3, and death?

Several mice died during a convulsive seizure, which is the type of seizure that is shown in the Supplemental Figure.

(9) Supplemental Figure 3 seems to display an isolated case in which EEG-detected seizures correlate with increased IIEs. It is not clear whether there are additional documented cases of seizures that could be assembled into a meaningful population graph. If this data does not exist or is too much work to include in this manuscript, perhaps it can be saved for a future paper.

We have added other cases and revised the graph. This is now Supplemental Figure 4 and is shown above in response to Reviewer #1, Recommendation for Authors Point #4.

Frontal is misspelled.

We checked and our copy is not showing a misspelling. However, we are very grateful to the Reviewer for catching many errors and reading the manuscript carefully.

(10) Supplemental Figure 4 seems incomplete in that it does not include EEG data from months 4, 5, and 6 (see Fig. 7B).

We have added data for these ages to the Supplemental Figure (currently Supplemental Figure 6) as part B. In part A, which had been the original figure, only 1.2, 2, and 3 months-old mice were shown because there were insufficient numbers of each sex at other ages. However, by pooling 1.2 and 2 months (Supplemental Figure 6B1), 3 and 4 months (B2) and 5 and 6 months (B3) we could do the analysis of sex. The results are the same – we detected no sex differences.

Author response image 9.

IIS frequency was similar for each sex. A. IIS frequency was compared for females and males at 1.2 months (1), 2 months (2), and 3 months (3). Two-way ANOVA was used to analyze the effects of sex and diet. Female and male Tg2576 mice were not significantly different. B. Mice were pooled at 1.2 and 2 months (1), 3 and 4 months (2) and 5 and 6 months (3). Two-way ANOVA analyzed the effects of sex and diet. There were significant effects of diet for (1) and (2) but not (3). There were no effects of sex at any age.

(1) There were significant effects of diet (F(2,47)=46.21, p<0.0001) but not sex (F(1,47)=0.106, p=0.746). Female and male mice fed the low choline diet or high choline diet were significantly different from female and male mice fed the intermediate diet (all p<0.05, asterisk).

(2) There were significant effects of diet (F(2,32)=10.82, p=0.0003) but not sex (F(1,32)=1.05, p=0.313). Both female and male mice of the low choline group were significantly different from male mice fed the intermediate diet (both p<0.05, asterisk) but no other pairwise comparisons were significant.

(3) There were no significant differences (diet, F(2,23)=1.21, p=0.317); sex, F(1,23)=0.844, p=0.368).

The data are discussed the Results, Section G, tarting on Line 843:

In Supplemental Figure 6B we grouped mice at 1-2 months, 3-4 months and 5-6 months so that there were sufficient females and males to compare each diet. A two-way ANOVA with diet and sex as factors showed a significant effect of diet (F(2,47)=46.21; p<0.0001) at 1-2 months of age, but not sex (F1,47)=0.11, p=0.758). Post-hoc comparisons showed that the low choline group had fewer IIS than the intermediate group, and the same was true for the high choline-treated mice. Thus, female mice fed the low choline diet differed from the females (p<0.0001) and males (p<0.0001) fed the intermediate diet. Male mice that had received the low choline diet different from females (p<0.0001) and males (p<0.0001) fed the intermediate diet. Female mice fed the high choline diet different from females (p=0.002) and males (p<0.0001) fed the intermediate diet, and males fed the high choline diet difference from females (p<0.0001) and males (p<0.0001) fed the intermediate diet.

For the 3-4 months-old mice there was also a significant effect of diet (F(2,32)=10.82, p=0.0003) but not sex (F(1,32)=1.05, p=0.313). Post-hoc tests showed that low choline females were different from males fed the intermediate diet (p=0.007), and low choline males were also significantly different from males that had received the intermediate diet (p=0.006). There were no significant effects of diet (F(2,23)=1.21, p=0.317) or sex (F(1,23)=0.84, p=0.368) at 5-6 months of age.

eLife assessment

In this fundamental work, the authors demonstrated that maternal choline supplementation improved spatial memory, reduced hyperexcitability, and restored NeuN expression in a familial Alzheimer's disease mouse model. Interestingly, choline deficiency increased mortality, while paradoxically reduced hyperexcitability. Using behavior, electrophysiological, and histological measures, the authors present convincing evidence supporting the significant role of maternal choline supplementation in protecting hippocampal functions vulnerable to Alzheimer's disease.

Reviewer #1 (Public Review):

Summary:

Chartampila et al. describe the effect of early-life choline supplementation on cognitive functions and epileptic activity in a mouse model of Alzheimer's disease. The cognitive abilities were assessed by the novel object recognition test and the novel object location test, performed in the same cohort of mice at 3 months and 6 months of age. Neuronal loss was tested using NeuN immunoreactivity, and neuronal hyperexcitability was examined using FosB and video-EEG recordings, providing multi-level correlations between these different parameters.

Strengths:

The study was designed as a 6-month follow-up, with repeated behavioral and EEG measurements through disease development and multilevel correlations providing valuable and interesting findings on AD progression and the effect of early-life choline supplementation. Moreover, the behavioral data that suggest an adverse effect of low choline in WT mice are interesting and important also beyond the context of AD, highlighting the dramatic effect of diet on the phenotypes of animal models.

Weaknesses:

The readability could be improved.

Author response:

The following is the authors’ response to the previous reviews.

We are pleased that Reviewer 3 has deemed our revisions satisfactory; below, we provide responses to the remaining Recommendations for the Authors from Reviewer 2.

Reviewer #2 (Recommendations For The Authors):

Minor corrections:

• Line 91: GWT should be GNWT

Fixed, thank you.

• Figure 2: fix the label "Participationcoefficient rank" (no space between Participation and coefficient)

Fixed, thank you for spotting.

• Line 317: Figure 2 should be Figure 3

Fixed, thank you.

• Line 360: Figure 4D, right?

Fixed, thank you. We also confirm that Figure 4 and its caption are correct. Under anaesthesia, many regions have more Integrated Information than during Recovery (red regions), but the only changes that are consistently observed across all three contrasts are the decreases.

• Line 375: Should be Figure 5A

Fixed, thank you.

• The recovery period of the anesthesia data is not described in Methods.

We have now added the missing information:

“Propofol was discontinued following the deep anaesthesia scan, and participants reached level 2 of the Ramsey scale approximately 11 minutes afterwards, as indicated by clear and rapid responses to verbal commands. This corresponds to the “recovery” period 176.”

We have also expanded our discussion on the interaction between information decomposition and measures of directionality:

“Indeed, transfer entropy can itself be decomposed into information-dynamic atoms through Partial Information Decomposition and Integrated Information Decomposition 33,34,49,151; ΦID can further decompose the Normalised Directed Transfer Entropy measure used by Deco et al 5, as recently demonstrated 152. We look forward to a more refined conceptualization of the synergistic workspace architecture that takes into account both information types and the directionality of information flow – especially in datasets with higher temporal resolution.”

Reviewer #2 (Public Review):

The authors analysed functional MRI recordings of brain activity at rest, using state-of-the-art methods that reveal the diverse ways in which information can be integrated in the brain. In this way, they found brain areas that act as (synergistic) gateways for the 'global workspace', where conscious access to information or cognition would occur, and brain areas that serve as (redundant) broadcasters from the global workspace to the rest of the brain. The results are compelling and are consistent with the already assumed role of several networks and areas within the Global Neuronal Workspace framework. Thus, in a way, this work comes to stress the role of synergy and redundancy as complementary information processing modes, which fulfill different roles in the bigger context of information integration.

In addition, to prove that the identified high-order interactions are relevant to the phenomenon of consciousness, the same analysis was performed in subjects under anesthesia or with disorders of consciousness (DOC), showing that indeed the loss of consciousness is associated with a deficient integration of information within the gateway regions.

Reviewer #3 (Public Review):

The work proposes a model of neural information processing based on a 'synergistic global workspace,' which processes information in three principal steps: a gatekeeping step (information gathering), an information integration step, and finally, a broadcasting step. They provided an interpretation of the reduced human consciousness states in terms of the proposed model of brain information processing, which could be helpful to be implemented in other states of consciousness. The manuscript is well-organized, and the results are important and could be interesting for a broad range of literature, suggesting interesting new ideas for the field to explore.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In their manuscript, Yu et al. describe the chemotactic gradient formation for CCL5 bound to - i.e. released from - glycosaminoglycans. The authors provide evidence for phase separation as the driving mechanism behind chemotactic gradient formation. A conclusion towards a general principle behind the finding cannot be drawn since the work focuses on one chemokine only, which is particularly prone to glycan-induced oligomerisation.

Strengths:

The principle of phase separation as a driving force behind and thus as an analytical tool for investigating protein interactions with strongly charged biomolecules was originally introduced for protein-nucleic acid interactions. Yu et al. have applied this in their work for the first time for chemokine-heparan sulfate interactions. This opens a novel way to investigate chemokine-glycosaminoglycan interactions in general.

Response: Thanks for the encouragement of the reviewer.

Weaknesses:

As mentioned above, one of the weaknesses of the current work is the exemplification of the phase separation principle by applying it only to CCL5-heparan sulfate interactions. CCL5 is known to form higher oligomers/aggregates in the presence of glycosaminoglycans, much more than other chemokines. It would therefore have been very interesting to see, if similar results in vitro, in situ, and in vivo could have been obtained by other chemokines of the same class (e.g. CCL2) or another class (like CXCL8).

Response: We share the reviewer’s opinion that to investigate more molecules/cytokines that interact with heparan sulfate in the system should be of interesting. We expect that researchers in the field will adapt the concept to continue the studies on additional molecules. Nevertheless, our earlier study has demonstrated that bFGF was enriched to its receptor and triggered signaling transduction through phase separation with heparan sulfate (PMID: 35236856; doi: 10.1038/s41467-022-28765-z), which supports the concept that phase separation with heparan sulfate on the cell surface may be a common mechanism for heparan sulfate binding proteins. The comment of the reviewer that phase separation is related to oligomerization is demonstrated in (Figure 1—figure supplement 2C and D), showing that the more easily aggregated mutant, A22K-CCL5, does not undergo phase separation.

In addition, the authors have used variously labelled CCL5 (like with the organic dye Cy3 or with EGFP) for various reasons (detection and immobilisation). In the view of this reviewer, it would have been necessary to show that all the labelled chemokines yield identical/similar molecular characteristics as the unlabelled wildtype chemokine (such as heparan sulfate binding and chemotaxis). It is well known that labelling proteins either by chemical tags or by fusion to GFPs can lead to manifestly different molecular and functional characteristics.

Response: We agree with the reviewer that labeling may lead to altered property of a protein, thus, we have compared chemotactic activity of CCL5 and CCL5-EGFP (Figure 2—figure supplement 1). To further verify this, we performed additional experiment to compare chemotactic activity between CCL5 and Cy3-CCL5 (see Author response image 1). For the convenience of readers, we have combined the original Figure 2—figure supplement 1 with the new data (Figure R1), which replaced original Figure 2—figure supplement 1.

Author response image 1.

Chemotactic function of CCL5-EGFP and CCL5-Cy3. Cy3-Labeled CCL5 has similar activity as CCL5, 50 nM CCL5 or CCL5-Cy3 were added to the lower chamber of the Transwell. THP-1 cells were added to upper chambers. Data are mean ± s.d. n=3. P values were determined by unpaired two-tailed t-tests. NS, Not Significant.

Reviewer #2 (Public Review):

Although the study by Xiaolin Yu et al is largely limited to in vitro data, the results of this study convincingly improve our current understanding of leukocyte migration.

(1) The conclusions of the paper are mostly supported by the data although some clarification is warranted concerning the exact CCL5 forms (without or with a fluorescent label or His-tag) and amounts/concentrations that were used in the individual experiments. This is important since it is known that modification of CCL5 at the N-terminus affects the interactions of CCL5 with the GPCRs CCR1, CCR3, and CCR5 and random labeling using monosuccinimidyl esters (as done by the authors with Cy-3) is targeting lysines. Since lysines are important for the GAG-binding properties of CCL5, knowledge of the number and location of the Cy-3 labels on CCL5 is important information for the interpretation of the experimental results with the fluorescently labeled CCL5. Was the His-tag attached to the N- or C-terminus of CCL5? Indicate this for each individual experiment and consider/discuss also potential effects of the modifications on CCL5 in the results and discussion sections.

Response: We agree with the reviewer that labeling may lead to altered property of a protein, thus, we have compared chemotactic activity of CCL5 and CCL5-EGFP (Figure 2—figure supplement 1). To further verify this, we performed additional experiment to compare chemotactic activity between CCL5 and Cy3-CCL5 (see Author response image 1). For the convenience of readers, we have combined the original Figure 2—figure supplement 1 with the new data (Author response image 1), which replaced original Figure 2—figure supplement 1.

The His-tag is attached to the C-terminus of CCL5, in consideration of the potential impact on the N-terminus.

(2) In general, the authors appear to use high concentrations of CCL5 in their experiments. The reason for this is not clear. Is it because of the effects of the labels on the activity of the protein? In most biological tests (e.g. chemotaxis assays), unmodified CCL5 is active already at low nM concentrations.

Response: We agree with the reviewer that the CCL5 concentrations used in our experiments were higher than reported chemotaxis assays and also higher than physiological levels in normal human plasma. In fact, we have performed experiments with lower concentration of CCL5, where the effect of LLPS was not seen though the chemotactic activity of the cytokine was detected. Thus, LLPS-associated chemotactic activity may represent a scenario of acute inflammatory condition when the inflammatory cytokines can increase significantly.

(3) For the statistical analyses of the results, the authors use t-tests. Was it confirmed that data follow a normal distribution prior to using the t-test? If not a non-parametric test should be used and it may affect the conclusions of some experiments.

Response: We thank the reviewer for pointing out this issue. As shown in Author response table 1, The Shapiro-Wilk normality test showed that only two control groups (CCL5 and 44AANA47-CCL5+CHO K1) in Figure 3 did not conform to the normal distribution. The error was caused by using microculture to count and calculate when there were very few cells in the microculture. For these two groups, we re-counted 100 μL culture medium to calculate the number of cells. The results were consistent with the positive distribution and significantly different from the experimental group (Author response image 3). The original data for the number of cells chemoattractant by 500 nM CCL5 was revised from 0, 247, 247 to 247, 123, 370 and 500 nM 44AANA47 +CHO-K1 was revised from 1111, 1111, 98 to 740, 494, 617. The revised data does not affect the conclusion.

Author response table 1.

Table R1 Shapiro-Wilk test results of statistical data in the manuscript

Author response image 3.

Quantification of THP-1collected from the lower chamber. Data are mean ± s.d. n=3. P values were determined by unpaired two-tailed t-tests.

Recommendations for the authors:

Reviewer #1:

See the weaknesses section of the Public Review. In addition, the authors should discuss the X-ray structure of CCL5 in complex with a heparin disaccharide in comparison with their docked structure of CCL5 and a heparin tetrasaccharide.

Response: Our study, in fact, is strongly influenced by the report (Shaw, Johnson et al., 2004) that heparin disaccharide interaction with CCL5, which is highlighted in the text (page5, line100-102).

Reviewer #2:

(1) Clearly indicate in the results section and figure legends (also for the supplementary figures) which form and concentration of CCL5 is used.

Response: The relevant missing information is indicated across the manuscript.

(2) Clearly indicate which GAG was used. Was it heparin or heparan sulfate and what was the length (e.g. average molecular mass if known) or source (company?)?

Response: Relevant information is added in the section “Materials and Methods.

(3) Line 181: What do you mean exactly with "tiny amounts"?

Response: “tiny amounts” means 400 transfected cells. This is described in the section of Materials and Methods. It is now also indicated in the text and legend to the figure.

(4) Lines 216-217: This is a very general statement without a link to the presented data. No combination of chemokines is used, in vivo testing is limited (and I agree very difficult). You may consider deleting this sentence (certainly as an opening sentence for the Discussion).

Response: We appreciate very much for the thoughtful suggestion of the reviewer. This sentence is deleted in the revised manuscript.

(5) Why was 5h used for the in vitro chemotaxis assay? This is extremely long for an assay with THP-1 cells.

Response: We apologize for the unclear description. The 5 hr includes 1 hr pre- incubation of CCL5 with the cells enable to form phase separation. After transferring the cells into the upper chamber, the actual chemotactic assay was 4 hr. This is clarified in the Materials and Methods section and the legend to each figure.

(6) Define "Sec" in Sec-CCL5-EGFP and "Dil" in the legend of Figure 4.

Response: The Sec-CCL5-EGFP should be “CCL5-EGFP’’, which has now been corrected. Dil is a cell membrane red fluorescent probe, which is now defined.

(7) Why are different cell concentrations used in the experiment described in Figure 5?

Response: The samples were from three volunteers who exhibited substantially different concentrations of cells in the blood. The experiment was designed using same amount of blood, so we did not normalize the number of the cell used for the experiment. Regardless of the difference in cell numbers, all three samples showed the same trend.

(8) Check the text for some typos: examples are on line 83 "ratio of CCL5"; line 142 "established cell lines"; line 196 "peripheral blood mononuclear cells"; line 224 "to mediate"; line 226 "bind"; line 247 "to form a gradient"; line 248 "of the glycocalyx"; line 343 and 346 "tetrasaccharide"; line 409-410 "wild-type"; line 543 "on the surface of CHO-K1 and CHO-677"; line 568 "white".

Response: Thanks for the careful reading. The typo errors are corrected and Manuscript was carefully read by colleagues.

Reviewer #2 (Public Review):

Although the study by Xiaolin Yu et al is largely limited to in vitro data, the results of this study convincingly improve our current understanding of leukocyte migration.

(1) The conclusions of the paper are mostly supported by the data and in the revised manuscript clarification is provided concerning the exact CCL5 forms (without or with a fluorescent label or His-tag) and amounts/concentrations that were used in the individual experiments. This is important since it is known that modification of CCL5 at the N-terminus affects the interactions of CCL5 with the GPCRs CCR1, CCR3 and CCR5 and random labeling using monosuccinimidyl esters (as done by the authors with Cy-3) is targeting lysines. The revised manuscript more clearly indicates for each individual experiment which form is used. However, a discussion on the potential effects of the modifications on CCL5 in the results and discussion sections is still missing.<br /> (2) In general, authors used high concentrations of CCL5 in their experiments. In their reply to the comments they indicate that at lower CCL5 concentrations no LLPS is detected. This is important information since it may indicate the need for chemokine oligomerization for LLPS. This info should be added to the manuscript and comparison with for instance the obligate monomer CCL7 and another chemokine such as CXCL4 that easily forms oligomers may clarify whether LLPS is controlled by oligomerization.<br /> (3) Statistical analyses have been improved in the revised manuscript.

eLife assessment

How the triplicate interaction between chemokines with both GAGs and G protein-coupled receptors (GPCR) works and how gradients are created and potentially maintained in vivo are poorly understood. The authors provide solid evidence to show phase separation can drive chemotactic gradient formation. The paper is a useful advance in the field of chemokine biology.

eLife assessment

This important study indicates a role for linker Histone H1 in protecting heterochromatic regions from certain types of repression. The experiments and data analysis that support the model for the role of linker Histone H1are solid, although additional experiments could provide a deeper mechanistic understanding. The study will be of broad interest to those interested in the role of chromatin in eukaryotic gene expression.

Reviewer #1 (Public Review):

In this study, the authors obtained multiple, novel and compelling datasets to better understand the relationship between histone H1 and RNA-directed DNA methylation in plants. Most of the authors' claims concerning H1 and RNA polymerase V (Pol V) are backed by convincing and independent lines of evidence. However, Pol V produces noncoding transcripts that act as scaffold RNAs, which AGO4-bound siRNAs recognize in plant chromatin to mediate RNA-directed DNA methylation. Detection of Pol V transcript products at the sites of Pol V redistribution in h1 mutants would significantly enhance the impact of this manuscript. Below I have listed several strengths and a weakness of the manuscript.

Strengths:

- The authors report high-quality NRPE1 ChIP-seq data, allowing them to directly test how and where Pol V occupancy depends on histone H1 function in Arabidopsis.<br /> - nrpe1 mutants generated via CRISPR/Cas9 in the h1 mutant background (nrpe1 h1.1-1 h1.2-1 triple mutants), allow the authors to study the role of Pol V in ectopic DNA methylation in H1-deficient plants.<br /> - Pol V recruitment via ZincFinger-DMS3 expression (a modified version of Pol V's DMS3 recruitment factor) sends Pol V to new genomic loci and thus provides the authors with an innovative dataset for understanding H1 function at these sites.

Weakness:

- The manuscript does not include detection or quantification of Pol V transcripts generated at ectopic sites in the h1 mutant background. Pol V encroachment into heterochromatin in the h1 mutant is indirectly shown by NRPE1-dependent methylation at such ectopic sites.

Previous studies have charted the relationship between H1 function and RNA-directed DNA methylation (RdDM) via analyses of Pol IV-dependent 24 nt siRNAs and factors that recruit Pol IV (Choi et al., 2021 and Papareddy et al., 2020). Harris and colleagues have extended this work and shown that histone H1 function also antagonizes Pol V occupancy in the context of constitutive heterochromatin. The authors thus provide important evidence to show that H1 limits the encroachment of both polymerases Pol IV and Pol V into plant heterochromatin.

Reviewer #2 (Public Review):

Summary:

The main conclusion of the manuscript is that the presence of linker Histone H1 protects Arabidopsis pericentromeric heterochromatic regions and longer transposable elements from encroachment by other repressive pathways. The manuscript focuses on the RNA-dependent DNA-methylation (RdDM) pathway but indirectly finds that other pathways must also be ectopically enriched.

Strengths:

The authors present diverse sets of genomic data comparing Arabidopsis wild-type and h1 mutant background allowing an analysis of differential recruitment of RdDM component NPRE1, which is related to changes in DNA methylation and H1 coverage. The manuscript also contains recruitment data for SUVH1 in wild-type and h1 mutant backgrounds.<br /> Furthermore, the authors make use of a line that recruits NRPE1 ectopically to show that H1 occupancy is not altered because of this recruitment. These data clearly show that there is a hierarchy in which DNA-methylation is impacted by presence of H1 while H1 distribution is independent of DNA-methylation.

Weaknesses:

The manuscript is driven by a strong and reasonable hypothesis that absence of H1 results increased access of chromatin binding factors and that this explains how the RdDM machinery is restricted from encroaching heterochromatic regions, which are particularly enriched in H1. Indeed, increased binding of NPRE1 at pericentromeric sites is observed; however, the major DNA-methylation changes at these sites are symmetric and not related to the RdDM pathway. Thus, the authors propose that many factors redistribute, which is again reasonable. The authors show redistribution of SUVH1 and relate their data to a previous report showing redistribution of the PcG machinery in H1 depletion mutants (Teano et al. in Cell reports (Volume 42, Issue 8, 29 August 2023), but the manuscript provides limited mechanistic insight as to why there is a strong increase in heterochromatin symmetric DNA-methylation.

Author response:

The following is the authors’ response to the original reviews.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Pg. 3 - lines 51-53: "Once established, the canonical RdDM pathway takes over, whereby small RNAs are generated by the plant-specific polymerase IV (Pol IV). In both cases, a second plant-specific polymerase, Pol V, is an essential downstream component." The authors' intro omits an important aspect of Pol V's function in RdDM, which is quite relevant to their study. Pol V transcribes DNA to synthesize noncoding RNA scaffolds, to which AGO4-bound 24 nt siRNAs are thought to base pair, leading to DRM2 recruitment for cytosine methylation near to these nascent Pol V transcripts (Wierzbicki et al 2008 Cell; Wierzbicki et al. 2009 Nat Genet). I recommend that the authors cite these key studies.

These citations have now been added (see line 57).

The authors provide compelling evidence that Pol V redistributes to ectopic heterochromatin regions in h1 mutants (e.g., Fig1a browser shot). Presumably, this would allow Pol V to transcribe these regions in h1 mutants, whereas it could not transcribe them in WT plants. Have the authors detected and/or quantified Pol V transcripts in the h1 mutant compared to WT plants at the sites of Pol V redistribution (detected via NRPE1 ChIP)?

Robust detection of Pol V transcripts can be experimentally challenging, and instead we quantify and detect NRPE1 dependent methylation at these regions (Fig 5), which occurs downstream of Pol V transcript production. However, we note detecting Pol V transcripts as a potential future direction in the discussion (see line 263).

Pg. 5 - lines 101-102: Figure 1e - "The preferential enrichment of NRPE1 in h1 was more pronounced at TEs that overlapped with heterochromatin associated mark, H3K9me2 (Fig. 1e). Was a statistical test performed to determine that the overall differences are significant only at TE sites with H3K9me2? Can the sites without H3K9me2 also be differentiated statistically?

Yes, there is a statistically significant difference between WT and h1 at both the H3K9me2 marked and unmarked TEs (Wilcoxon rank sum tests, see updated Fig 1e). The size of the effect is larger for the H3K9me2 marked TEs (median difference of 0.41 vs 0.16). Median values have now been added to the boxplots so that this is directly viewable to the reader (Fig 1e). This reflects the general increase in NRPE1 occupancy in h1 mutants through the genome, with the effect consistently stronger in heterochromatin. In our initial version of the manuscript, we summarise the effect as follows “We found that h1 antagonizes NRPE1 occupancy throughout the genome, particularly at heterochromatic regions” (previous version line 83, current version line 95). Although important exceptions exist (see Fig 5, NRPE1 and DNA methylation loss in h1), we now make this point even more explicit, and have updated the manuscript at several locations (abstract line 26, results line 245, discussion line 265).

Pg. 5 - lines 108-110: The authors state, "Importantly, we found no evidence for increased NRPE1 expression at the mRNA or protein level in the h1 mutant (Suppl. Fig. 2)." But the authors did observe reduced NRPE1 transcript levels in h1 mutants, in their re-analysis of RNA-seq data and reduced NRPE1 protein signals via western blot in (Suppl. Fig. 2), which should be reported here in the results.

As described further below, we reanalysed h1 RNA-seq from scratch, and see no evidence for significant differential gene expression of NRPE1. This table and analysis are now provided in Supplementary Table 1.

More importantly, the above logic about NRPE1 expression in h1 mutants assumes that NRPE1 is the stoichiometrically limiting subunit for Pol V assembly and function in vivo, but this is not known to be the case:

(1) While NRPE1's expression is somewhat reduced (and not increased) in h1 mutant plants, we cannot be certain that other genes influencing Pol V stability or recruitment are unaffected by h1 mutants. I thus recommend that the authors perform RT-qPCR directly on the WT and h1 mutant materials used in their current study, quantifying NRPE1, NRPE2, NRPE5, DRD1, DMS3, RDM1, SUVH2 and SUVH9 transcript levels.

(2) Normalizations used to compare samples should be included with RT-qPCR and western assays. An appropriate house-keeping gene like Actin2 or Ubiquitin could be used to normalize the RT-qPCR. Protein sample loading in Suppl. Fig. 2 could be checked by Coomassie staining and/or an antibody detection of a house-keeping protein.

We have now included a full re-analysis of h1 RNA-seq (data from Choi et al 2020) focusing on transcriptional changes of DNA methylation machinery genes in the h1 mutant. Of the 61 genes analysed, only AGO6 and AGO9 were found to be differentially expressed (2-3 fold upregulation). This analysis is now included as a table

(Supplementary Table 1). The western blot has been moved to Supplementary Fig 3 to now illustrate antibody specificity and H1 loss in the h1 mutant lines, so NRPE1 itself serves as a loading control (Supplementary Fig 3a).

Pg. 6 - lines 129-131: The authors state that "over NRPE1 defined peaks (where NRPE1 occupancy is strongest in WT) we observed no change in H1 occupancy in nrpe1 (Fig 2b). The results indicate that H1 does not invade RdDM regions in the nrpe1 mutant background." This conclusion assumes that the author's H1 ChIP is successfully detecting H1 occupancy. However, in Fig 2d there does not appear to be H1 enrichment or peaks as visualized across the 10766 ZF-DMS3 off-target loci, or even at the selected 451 ZFDMS3 off-target hyper DMRs, where the putative signal for H1 enrichment on the metaplot center is extremely weak/non-existent.

As a reference for H1 enrichment in chromatin (e.g., looking where H2A.W antagonizes H1 occupancy) one can compare analyses in Bourguet et al (2021) Nat Commun, involving co-authors of the current study. Bourguet et al (2021) Fig 5b show a metaplot of H1 levels centered on H2A.W peaks with H1 ChIP signal clearly tapering away from the metaplot center point peak. To my eye, the H1 ChIP metaplots for ZF-DMS3 offtarget loci in the current manuscript (Fig 2d) resemble "shuffled peaks" controls like those in Fig 5b of Bourguet et al (2021).

Can one definitively interpret Fig 2d as showing RdDM "not reciprocally affecting H1 localization" without first showing the specificity of the ChIP-seq results in a genotype where H1 occupancy changes? Alternatively, could this dataset be displayed with Deeptools heatmaps to strengthen the evidence that the authors are detecting H1 occupancy/enrichment genome-wide, before diving into WT/nrpe1 mutant analysis at ZF-DMS3 off-target loci?

This is an excellent suggestion from the reviewer. We have now included several analyses that assess and demonstrate the quality of our H1 ChIP-seq profiles. First, as suggested by the reviewer, we show that our H1 profiles peak over H2A.W enriched euchromatic TEs as defined by Bourguet et al, mirroring these published findings. Next, we investigated whether our H1 profiles match Teano’s recently described pattern over genes, confirming a similar pattern with 3’ enrichment of H1 over H3K27me3 unmarked genes. Furthermore, we show that the H1 peaks defined here are similarly enriched with GFP tagged H1.2 from the Teano et al. 2023 study. These analyses that validate the quality of our H1 ChIP-seq datasets and bolster the conclusion that NRPE1 redistribution does not affect H1 occupancy. These new analysis are now presented in Supplementary Figure 3 and see line 153.

Pg. 8 - lines 228-230: The authors state that, "As with NRPE1, SUVH1 increased in the h1 background significantly more in heterochromatin, with preferential enrichment over long TEs, cmt2 dependent hypo CHH DMRs, and heterochromatic TEs (Fig. 6b)."

Contrary to the above statement, the violin plots in Fig. 6c show SUVH1 occupancy increasing at euchromatic TEs in the h1 mutant. What statistical test allowed the authors to determine that the increase in h1 occurs "significantly more in heterochromatin"? The authors should critically interpret Fig. 6c and 6d, which are not currently referenced in the results section. More support is needed for the claim that SUVH1 specifically encroaches into heterochromatin in the h1 mutant, rather than just TEs generally (euchromatic and heterochromatic alike).

Similar to what we see for NRPE1, statistical tests that we have now performed show that SUVH1 is significantly enriched in h1 in all classes. Importantly however, the effect size is larger in all of the heterochromatin associated classes. We display these statistical tests and the median values on the plots so that effects are immediately viewable (see updated Fig 6).

In addition, the authors should verify that SUVH1-3xFLAG transgenes (in the WT and h1 mutant backgrounds, respectively) and endogenous Arabidopsis genes encoding the transcriptional activator complex (SUVH1-SUVH3-DNAJ1-DNAJ2) are not overexpressed in the h1 mutant vs. WT. Higher expression of SUVH1 or limiting factors in the larger complex could explain the observation of increased SUVH1 occupancy in the h1 background.

We do not see a difference in SUVH1/3/DNAJ1/2 complex gene expression in the h1 background (see Supplementary Table 1). However, we cannot rule out that that our SUVH1-FLAG line in h1 is more highly expressed than the corresponding SUVH1-FLAG line in WT. We now note this point in line 248.

Pg. 8 - lines 231-232: Here the authors make a sweeping conclusion about H1 demarcating, "the boundary between euchromatic and heterochromatic methylation pathways, likely through promoting nucleosome compaction and restricting heterochromatin access." I do not see how a H1 boundary between euchromatic and heterochromatic methylation pathways is revealed based on the SUVH1-3xFLAG occupancy data, which shows increased enrichment at every category interrogated in the h1 mutant (Fig 6b,c,d) and all along the baseline too in the h1 mutant browser tracks (Fig 6a). Can the authors provide more examples of this phenomenon (similar to Fig 6a) and better explain why their SUVH1-3xFLAG ChIP supports this demarcation model?

The general conclusion from SUVH1 about H1’s agnostic role in preventing heterochromatin access is now further supported from our findings with H3K27me3 (see Figure 6e and description from line 250). However, we agree that the demarcation model as initially presented was overly simplistic. This point was also raised by reviewer 2. We have removed the line highlighted by the reviewer in the revised version of the manuscript. In the revised version we clarify that H1 impedes RdDM and associated machinery throughout the genome (consistent with H1’s established broad occupancy across the genome) but this effect is most pronounced in heterochromatin, corresponding to maximal H1 occupancy (abstract line 26, results line 245, discussion line 265). 

Corrections:

Pg. 8 - lines 226-227: "We therefore wondered whether complex's occupancy might also be affected by H1." The sentence contains a typo, where I assume the authors mean to refer to occupancy by the SUVH1-SUVH3-DNAJ1-DNAJ2 transcriptional activator complex. This needs to be specified more clearly.

The paragraph has been updated (see from line 237).

Pg. 13 - lines 393-405: There are minor errors in the capitalization of titles and author initials in the References. I recommend that the authors proofread all the references to eliminate these issues:

Thank you, these have been corrected.

Choi J, Lyons DB, Zilberman D. 2021. Histone H1 prevents non-cg methylation-mediated small RNA biogenesis in arabidopsis heterochromatin. Elife 10:1-24. doi:10.7554/eLife.72676 (...)

Du J, Johnson LM, Groth M, Feng S, Hale CJ, Li S, Vashisht A a., Gallego-Bartolome J, Wohlschlegel J a., Patel DJ, Jacobsen SE. 2014. Mechanism of DNA methylation-directed histone methylation by KRYPTONITE. Mol Cell 55:495-504. doi:10.1016/j.molcel.2014.06.009 (...)

Du J, Zhong X, Bernatavichute Y V, Stroud H, Feng S, Caro E, Vashisht A a, Terragni J, Chin HG, Tu A, Hetzel J, Wohlschlegel J a, Pradhan S, Patel DJ, Jacobsen SE. 2012. Dual binding of chromomethylase domains to H3K9me2-containing nucleosomes directs DNA methylation in plants. Cell 151:167-80. doi:10.1016/j.cell.2012.07.034

Reviewer #2 (Recommendations For The Authors):

As for a normal review, here are our major and minor points.

Major:

(1) Lines 38 to 45 of the introduction are important for the subsequent definition of heterochromatic and non-heterochromatic transposons, but the definition is ambiguous. Is heterochromatin defined by surrounding context such as pericentromeric position or is this an autonomous definition? Can a TE with the chromosomal arms be considered heterochromatic provided that it is long enough and recruits the right machinery? These cases should be more explicitly introduced. Ideally, a supplemental dataset should provide a key to the categories, genomic locations and overlapping TEs as they were used in this analysis, even if some of the categories were taken from another study.

We have now added all the regions used for analysis in this study to Supplementary Table 3.

(2) Line 80: This would be the first chance to cite Teno et al. and the "encroachment" of

PcG complexes to TEs in H1 mutants

Done - “H1 also plays a key role in shaping nuclear architecture and preventing ectopic polycomb-mediated H3K27me3 deposition in telomeres (Teano et al., 2023).” See line 83

(3) It is "only" a supplemental figure but S2 but it should still follow the rules: Indicate the number of biological replicates for the RNA-seq data, and perform a statistical test. In case of WB data, provide a loading control.

We are now using the western blot to illustrate antibody specificity and H1 loss in the h1 mutant lines, so NRPE1 itself serves as a loading control (Supplementary Fig 3a). For NRPE1 mRNA expression, we have now replaced this with a more comprehensive transcriptome analysis of methylation machinery in h1 (see Supplementary Table 1). 

(4) Lines 115 to 124 and corresponding data: Here, the goal is to exclude other changes to heterochromatin structure other than "increased access" in H1 mutants; however, only one feature, H3K9me2, is tested. Testing this one mark does not necessarily prove that the nature of the chromatin does not change, e.g. H2A.W could be differently redistributed, DDM1 may change, VIM protein, and others. Either more comprehensive testing for heterochromatin markers should be performed, or the conclusions moderated.

We have moderated the text accordingly (see line 135).

(5) Lines 166ff and Figure 1, a bit out of order also Figure 5: The general hypothesis is that NRPE1 redistributes to heterochromatic regions in h1 mutants (as do other chromatin modifiers), but the data seem to only support a higher occurrence at target sites.

a. The way the NRPE1 data is displayed makes it seem like there is much more NRPE1 in the h1 samples, even at peaks that should not be recruiting more as they do not represent "long" TEs. It would be good to present more gbrowse shots of all peak classes.

We now clarify that h1 does result in a general increase of NRPE1 throughout the genome, but the effect is strongest at heterochromatin. In our initial version of the manuscript, we summarise the effect as follows “We found that h1 antagonizes NRPE1 occupancy throughout the genome, particularly at heterochromatic regions” (previous version line 83, current version line 95). We have modified the language at several locations throughout the manuscript to make this point more clearly (abstract line 26, results line 245, discussion line 265). We include several browser shots in Supp Fig. 8.

b. The data are "normalized" how exactly?

c. One argument of observing "gaining" and "losing" peaks is that there is redistribution of NRPE1 from euchromatic to heterochromatic sites. There should be an analysis and figure to corroborate the point (e.g. by comparing FRIP values). Figure 1b shows lower NRPE1 signals at the TE flanking regions. This could reflect a redistribution or a flawed normalization procedure.

The data are normalised using a standardised pipeline by log2 fold change over input, after scaling each sample by mapped read depth using the bamCompare function in deepTools. This is now described in detail in the Materials and Methods line 365, with full code and pipelines available from GitHub (https://github.com/Zhenhuiz/H1-restrictseuchromatin-associated-methylation-pathways-from-heterochromatic-encroachment).

d. Figure 1d and f show similar profiles comparing "long" and "short" TEs or "CMT2 dependent hypo-CHH" and "DRM2 dependent CHH". How do these categories relate to each other, how many fragments are redundant?

The short vs long TEs were defined in Liu et al 2018 (doi: 10.1038/s41477-017-0100-y) and the DMRs were defined in Zhang et al. 2018 (DOI: 10.1073/pnas.1716300115). There is likely to be some degree of overlap between the categories, but numbers are very different (short TEs (n=820), long TEs (n=155), drm2 DMRs (n=5534), CMT (n=21784)) indicating that the different categories are informative. We have now listed all the regions used for analysis in this study as in Supplementary Table 3.

e. The purpose of the data presented in Figure 1 b is to compare changes of NRPE1 association in H3K9me3 non-overlapping and overlapping TEs between wild-type and background, yet the figure splits the categories in two subpanels and does neither provide a fold-change number nor a statistical test of the comparison. As before, the figure does not really support the idea that NPRE1 somehow redistribute from its "normal" sites towards heterochromatin as both TE classes seem to show higher NRPE1 binding in h1 mutants.

There is a statistically significant difference between WT and h1 at both the H3K9me2 marked and unmarked TEs, however, the size of the effect is larger for the H3K9me2 marked TEs (median difference of 0.41 vs 0.16). Median values have now been added to the boxplots so that this is directly viewable to the reader (Fig 1e). Although important exceptions exist (see Fig 5 – regions that lose NRPE1 and DNA methylation), this reflects the general increase in NRPE1 occupancy in h1 mutants throughput the genome, with a consistently stronger effect in heterochromatin. As noted above, we have updated the manuscript to make this point more clearly (abstract line 26, results line 245, discussion line 265).

f. Panel g is the only attempt to corroborate the redistribution towards heterochromatic regions, but at this scale, the apparent reduction of binding in the chromosome arms may be driven by off-peak differences and normalization problems between different ChIP samples with different signal-to-noise-ratio.

We describe our normalisation and informatic pipeline in more detail in the Materials and Methods line 365. It is also important to note that the reduction is not only observed at the chromosomal level, but also at specific sites. We called differential peaks between WT and h1 mutant. The "Regions that gain NRPE1 in h1" peaks are more enriched in heterochromatic regions, while " Regions that lose NRPE1 in h1" peaks are more enriched outside heterochromatic regions.

g. Figure 5: how many regions gain vs lose NRPE1 in h1 mutants? If the "redistribution causes loss" scenario applies, the numbers should overall be balanced but that does not seem the case. The loss case appears to be rather exceptional judging from the zigzagging meta-plot. Are these sites related to the sites taken over by PcG-mediated repression in h1 mutants?

As described in line 222 (previous version of the manuscript line 206), there are 15,075 sites that gain and 1,859 sites that lose NRPE1 in h1. Comparing these sites to

H3K27me3 in the Teano et al. study was an excellent suggestion. We compared sites that gain NRPE1 to sites that gain H3K27me3 in h1, finding a statistically significant overlap (2.4 fold enrichment over expected, hypergeometric test p-value 2.1e-71). Reciprocally, sites that lose NRPE1 were significantly enriched for overlap with H3K27me3 loss regions (1.6 fold over expected, hypergeometric test p-value 1.4e-4). This indicates that RdDM and H3K27me3 patterning are similarly modulated by H1. To directly test this, we reanalysed the H3K27me3 ChIP-seq data from Teano et al., finding coincident gain and loss of H3K27me3 at sites that gain and lose NRPE1 in h1. These results are described from line 250 and in Fig 6e, which supports a general role for H1 in preventing heterochromatin encroachment.

(6) Lines 166ff and Figure 3: The data walk towards the scenario of pathway redistribution but actually find that RdDM plays a minor role overall as a substantial increase in heterochromatin regions occurs in all contexts and is largely independent of RdDM.

a. How exactly are DNA-methylation data converted across regions to reach a fraction score from 0 to 1? There is no explanation in the legend for the methods that allow to recapitulate.

We now explain our methods in full in the Materials and Methods and all the code for generating these has now been deposited on GitHub (https://github.com/Zhenhuiz/H1restricts-euchromatin-associated-methylation-pathways-from-heterochromaticencroachment). Briefly, BSMAP is used to calculate the number of reads that are methylated vs unmethylated on a per-cytosine basis across the genome. Next, the DNA methylation fraction in each region is calculated by adding all the methylation fractions per cytosine in a given window, and divided by the total number of cytosines in that same window (ie mC/(unmC+mC)) i.e. this is expressed as a fraction ranging from 0 to 1.

“0” indicates this region is not methylated, and “1” indicates this region is fully methylated (every cytosine is 100% methylated).  

b. Kernel plots? These are slang for experts and should be better described. In addition, nothing is really concluded from these plots in the text, although they may be quite informative.

Kernel density plots show the proportion of TEs that gain or lose methylation in a particular mutant, rather than the overall average as depicted in the methylation metaplots above. We now describe the kernel density plots in more detail in the Figure 3 legend. 

(7) Figure 4: This could be a very interesting analysis if the reader could actually understand it.

a. The legend is minimal. What is the meaning of hypo and hyper regions indicated to the right of Figure 4c?

b. The color scale represents observed/expected values. What exactly does this mean? Mutant vs WT?

c. Some comparisons in 4a are cryptic, e.g. h1 nrpe1 nrpe1 vs CHH?

d. Figure 4d focuses on a correlation square of relevance, but why? Interestingly the square does not correspond to any "hypo" or "hyper" label?

Thank you, we have revised Figure 4 and legend based on these suggestions to clarify all of the above.

(8) Lines 226 and Figure 6B. De novo (or increased) targeting of SUVH1 to heterochromatic sites in h1 mutants, similar to NRPE1, is used to support the argument that more access allows other chromatin modifiers to encroach. SUVH1 strongly depends on RdDM for its in vivo binding and may be the least conclusive factor to argue for a "general" encroachment mechanism.

We appreciate the reviewers point here. Something that is entirely independent of RdDM following the same pattern would be stronger evidence in favour of general encroachment. Excitingly, this is exactly what we provide evidence for when investigating the interrelationship with H3K27me3 and we appreciate the reviewer’s suggestion to check this! This data is now described in Figure 6e and line 250.

Minor:

(1) Line 23: "Loss of H1 resulted in heterochromatic TE enrichment by NRPE1." This does not seem right. NRPE enrichment as TEs

Modified, (line 26) thank you.

(2) Lines 73-74: The idea that DDM1 displaces H1 in heterochromatic TEs is somewhat counterintuitive to model that heterochromatic TEs are unavailable for RdDM because of the presence of H1. Is this displacement non-permanent and directly linked to interaction with CMT2/3 Met1?

This is a very good question and we agree with the reviewer that the effect of DDM1 may only be transient or insufficient to allow for full RdDM assembly, or indeed there may be a direct interaction between DDM1 and CMTs/MET1. During preparation of these revisions, a structure of Arabidopsis nucleosome bound DDM1 was published, which provides some insight by showing that DDM1 promotes DNA sliding. This is at least consistent with the idea of DDM1 causing transient / non-permanent displacement of H1 that would be insufficient for RdDM establishment. We incorporate discussion of these ideas at line 80.

(3) Line 85: A bit more background on the Reader activator complex should be given. In fact, the reader may not really care that it was more recently discovered (not really recent btw) but what does it actually do?

We have quite extensively reconfigured this paragraph to take into account our new finding with H3K27me3, such that there is less emphasis on the reader activator complex. The sentence now reads as follows:

“We found that h1 antagonizes NRPE1 occupancy throughout the genome, particularly at heterochromatic regions. This effect was not limited to RdDM,  similarly impacting both the methylation reader complex component, SUVH1 (Harris et al., 2018) and polycomb-mediated H3K27me3 (Teano et al., 2023).” (line 95). 

Also, when describing the experiment the results section (line 241), we now provide more background on SUVH1’s function.

(4) Lines 80-81: Since it is already shown that RdDM associated small RNAs are more enriched in h1 at heterochromatin, help us to know what is precisely the added value of studying the enrichment of NRPE1 at these sites.

Good point. We have the following line: ‘...small RNAs are not a direct readout of functional RdDM activity and Pol IV dependent small RNAs are abundant in regions of the genome that do not require RdDM for methylation maintenance and that do not contain Pol V (Stroud et al., 2014).’ (line 90)

(5) Line 99: This seems to be the only time where the connection between long TEs and heterochromatic regions is mentioned but no source is cited.

We have added the following appropriate citations: (Bourguet et al., 2021; Zemach et al., 2013). (line 110).

(6) Line 100: DMRs is used for the first time here without explanation and full text. The abbreviation is introduced later in the text (Line 187).

Thank you, we now describe DMRs upon first use, line 112.

(7) Figure 2: Panels 2 c and d should show metaplots for WT and transgenes in one panel. There is something seriously wrong with the normalization in d or the scale for left and right panel is not the same. Neither legend nor methods describe how normalization was performed.

Thank you for pointing this out, the figure has been corrected. We have updated the Materials and Methods (line 365) and have added codes and pipelines to GitHub to explain the normalisation procedure in more detail (https://github.com/Zhenhuiz/H1restricts-euchromatin-associated-methylation-pathways-from-heterochromaticencroachment).

eLife assessment

This study presents a valuable new behavioral apparatus aimed at differentiating the strategies animals use to orient themselves in an environment. The evidence supporting the claims is solid, with statistical modeling of animal behavior. Overall, this study will attract the interest of researchers exploring spatial learning and memory.

Reviewer #1 (Public Review):

The authors design an automated 24-well Barnes maze with 2 orienting cues inside the maze, then model what strategies the mice use to reach the goal location across multiple days of learning. They consider a set of models and conclude that the animals begin with a large proportion of random choices (choices irrespective of the goal location), which over days of experience becomes a combination of spatial choices (choices targeted around the goal location) and serial choices (successive stepwise choices in a given direction). Moreover, the authors show that after the animal has many days of experience in the maze, they still often began each trial with a random choice, followed by spatial or serial choices.

This study is written concisely and the results are presented concisely. The best fit model provides valuable insight into how the animals solve this task, and therefore offers a quantitative foundation upon which tests of neural mechanisms of the components of the behavioral strategy can be performed. These tests will also benefit from the automated nature of the task.

Reviewer #2 (Public Review):

This paper uses a novel maze design to explore mouse navigation behaviour in an automated analogue of the Barnes maze. A major strength is the novel and clever experimental design which rotates the floor and intramaze cues before the start of each new trial, allowing the previous goal location to become the next starting position. The modelling sampling a Markov chain of navigation strategies is elegant, appropriate and solid, appearing to capture the behavioural data well. This work provides a valuable contribution and I'm excited to see further developments, such as neural correlates of the different strategies and switches between them.

Reviewer #3 (Public Review):

Strength:

The development of an automated Barnes maze allows for more naturalistic and uninterrupted behavior, facilitating the study of spatial learning and memory, as well as the analysis of the brain's neural networks during behavior when combined with neurophysiological techniques. The system's design has been thoughtfully considered, encompassing numerous intricate details. These details include the incorporation of flexible options for selecting start, goal, and proximal landmark positions, the inclusion of a rotating platform to prevent the accumulation of olfactory cues, and careful attention given to atomization, taking into account specific considerations such as the rotation of the maze without causing wire shortage or breakage. When combined with neurophysiological manipulations or recordings, the system provides a powerful tool for studying spatial navigation system.<br /> The behavioral experiment protocols, along with the analysis of animal behavior, are conducted with care, and the development of behavioral modeling to capture the animal's search strategy is thoughtfully executed. It is intriguing to observe how the integration of these innovative stochastic models can elucidate the evolution of mice's search strategy within a variant of the Barnes maze.

Comments on revised version:

The authors have addressed all the points I outlined in the previous round of review, resulting in significant improvements to the manuscript. However, I have one remaining comment. Given the updated inter-animal analysis (Supplementary Figure 8), it appears that male and female mice develop strategies differently across days. Male mice seem to progressively increase their employment of spatial strategy across days, at the expense of the random strategy. Conversely, female mice exhibit both spatial and serial strategies at their highest levels on day 2, with minimal changes observed on the subsequent days.<br /> These findings could alter the interpretation of Figure 5 and the corresponding text in the section "Evolution of search strategy across days".<br /> For instance, this statement on page 6 doesn't hold for female mice: "The spatial strategy was increased across days, ... largely at the expense of the random strategy."

Author response:

The following is the authors’ response to the original reviews.

We are very grateful to the reviewers for their constructive comments. Here is a summary of the main changes we made from the previous manuscript version, based on the reviewers’ comments:

(1) Introduction of a new model, based on a Markov chain, capturing within-trial evolution in search strategy .

(2) Addition of a new figure investigating inter-animal variations in search strategy.

(3) Measurement of model fit consistency across 10 simulation repetitions, to prevent the risk of model overfitting.

(4) Several clarifications have been made in the main text (Results, Discussion, Methods) and figure legends.

(5) We now provide processed data and codes for analyses and models at GitHub repository

(6) Simplification of the previous modeling. We realized that the two first models in the previous manuscript version were simply special cases of the third model. Therefore, we retained only the third model, which has been renamed as the ‘mixture model’.

(7) Modification of Figure 4-6 and Supplementary Figure 7-8 (or their creation) to reflect the aforementioned changes

Public Reviews:

Reviewer #1 (Public Review):

The authors design an automated 24-well Barnes maze with 2 orienting cues inside the maze, then model what strategies the mice use to reach the goal location across multiple days of learning. They consider a set of models and conclude that one of these models, a combined strategy model, best explains the experimental data.

This study is written concisely and the results presented concisely. The best fit model is reasonably simple and fits the experimental data well (at least the summary measures of the data that were presented).

Major points:

(1) One combined strategy (once the goal location is learned) that might seem to be reasonable would be that the animal knows roughly where the goal is, but not exactly where, so it first uses a spatial strategy just to get to the first vestibule, then switches to a serial strategy until it reaches the correct vestibule. How well would such a strategy explain the data for the later sessions? The best combined model presented in the manuscript is one in which the animal starts with a roughly 50-50 chance of a serial (or spatial strategy) from the start vestibule (i.e. by the last session before the reversal the serial and spatial strategies are at ~50-50m in Fig. 5d). Is it the case that even after 15 days of training the animal starts with a serial strategy from its starting point approximately half of the time? The broader point is whether additional examination of the choices made by the animal, combined with consideration of a larger range of possible models, would be able to provide additional insight into the learning and strategies the animal uses.

Our analysis focused on the evolution of navigation strategies across days and trials. The reviewer raises the interesting possibility that navigation strategy might evolve in a specific manner within each trial, especially on the later days once the environment is learned. To address this possibility, we first examined how some of the statistical distributions, previously analyzed across days, evolved within trials. Consistent with the reviewer’s intuition, the statistical distributions changed within trials, suggesting a specific strategy evolution within trials. Second, we developed a new model, where strategies are represented as nodes of a Markov chain. This model allows potential strategy changes after each vestibule visit, according to a specific set of transition probabilities. Vestibules are chosen based on the same stochastic processes as in the previous model. This new model could be fitted to the experimental distributions and captured both the within-trial evolution and the global distributions. Interestingly, the trials were mostly initiated in the random strategy (~67% chance) and to a lesser extent in the spatial strategy (~25% chance), but rarely in the serial strategy (~8% chance). This new model is presented in Figure 6.

(2) To clarify, in the Fig. 4 simulations, is the "last" vestibule visit of each trial, which is by definition 0, not counted in the plots of Fig. 4b? Otherwise, I would expect that vestibule 0 is overrepresented because a trial always ends with Vi = 0.

The last vestibule visit (vestibule 0 by definition) is counted in the plots of Fig.4b. We initially shared the same concern as the reviewer. However, upon further consideration, we arrived at the following explanation: A factor that might lead to an overrepresentation of vestibule 0 is the fact that, unlike other vestibules, it has to be contained in each trial, as trials terminated upon the selection of vestibule 0. Conversely, a factor that might contribute to an underrepresentation of vestibule 0 is that, unlike other vestibules, it cannot be counted more than once per trial. Somehow these two factors seem to counterbalance each other, resulting in no discernible overrepresentation or underrepresentation of vestibule 0 in the random process. 

Reviewer #2 (Public Review):

This paper uses a novel maze design to explore mouse navigation behaviour in an automated analogue of the Barnes maze. Overall I find the work to be solid, with the cleverly designed maze/protocol to be its major strength - however there are some issues that I believe should be addressed and clarified.

(1) Whilst I'm generally a fan of the experimental protocol, the design means that internal odor cues on the maze change from trial to trial, along with cues external to the maze such as the sounds and visual features of the recording room, ultimately making it hard for the mice to use a completely allocentric spatial 'place' strategy to navigate. I do not think there is a way to control for these conflicts between reference frames in the statistical modelling, but I do think these issues should be addressed in the discussion.

It should be pointed out that all cues on the maze (visual, tactile, odorant) remained unchanged across trials, since the maze was rotated together with goal and guiding cues. Furthermore, the maze was equipped with an opaque cover to prevent mice from seeing the surrounding room (the imaging of mouse trajectories was achieved using infrared light and camera). It is however possible that some other cues such as room sounds and odors could be perceived and somewhat interfered with the sensory cues provided inside the maze. We have now mentioned this possibility in the discussion.

(2) Somewhat related - I could not find how the internal maze cues are moved for each trial to demarcate the new goal (i.e. the luminous cues) ? This should be clarified in the methods.

The luminous cues were fixed to the floor of the arena. Consequently, they rotated along with the arena as a unified unit, depicted in figure 1. We have added some clarifications in Figure 1 legend and methods.

(3) It appears some data is being withheld from Figures 2&3? E.g. Days 3/4 from Fig 2b-f and Days 1-5 on for Fig 3. Similarly, Trials 2-7 are excluded from Fig 3. If this is the case, why? It should be clarified in the main text and Figure captions, preferably with equivalent plots presenting all the data in the supplement.

The statistical distributions for all single days/trials are shown in the color-coded panels of Figure2&3. In the line plots of Figure2&3, we show only the overlay of 2-3 lines for the sake of clarity. The days/trials represented were chosen to capture the dynamic range of variability within the distributions. We have added this information in the figure legends.

(4) I strongly believe the data and code should be made freely available rather than "upon reasonable request".

Matrices of processed data and various codes for simulations and analyses are now available at https://github.com/ sebiroyerlab/Vestibule_sequences.

Reviewer #3 (Public Review):

Royer et al. present a fully automated variant of the Barnes maze to reduce experimenter interference and ensure consistency across trials and subjects. They train mice in this maze over several days and analyze the progression of mouse search strategies during the course of the training. By fitting models involving stochastic processes, they demonstrate that a model combined of the random, spatial, and serial processes can best account for the observed changes in mice's search patterns. Their findings suggest that across training days the spatial strategy (using local landmarks) was progressively employed, mostly at the expense of the random strategy, while the serial strategy (consecutive nearby vestibule check) is reinforced from the early stages of training. Finally, they discuss potential mechanistic underpinnings within brain systems that could explain such behavioral adaptation and flexibility.

Strength:

The development of an automated Barnes maze allows for more naturalistic and uninterrupted behavior, facilitating the study of spatial learning and memory, as well as the analysis of the brain's neural networks during behavior when combined with neurophysiological techniques. The system's design has been thoughtfully considered, encompassing numerous intricate details. These details include the incorporation of flexible options for selecting start, goal, and proximal landmark positions, the inclusion of a rotating platform to prevent the accumulation of olfactory cues, and careful attention given to atomization, taking into account specific considerations such as the rotation of the maze without causing wire shortage or breakage. When combined with neurophysiological manipulations or recordings, the system provides a powerful tool for studying spatial navigation system.

The behavioral experiment protocols, along with the analysis of animal behavior, are conducted with care, and the development of behavioral modeling to capture the animal's search strategy is thoughtfully executed. It is intriguing to observe how the integration of these innovative stochastic models can elucidate the evolution of mice's search strategy within a variant of the Barnes maze.

Weakness:

(1) The development of the well-thought-out automated Barnes maze may attract the interest of researchers exploring spatial learning and memory. However, this aspect of the paper lacks significance due to insufficient coverage of the materials and methods required for readers to replicate the behavioral methodology for their own research inquiries.

Moreover, as discussed by the authors, the methodology favors specialists who utilize wired recordings or manipulations (e.g. optogenetics) in awake, behaving rodents. However, it remains unclear how the current maze design, which involves trapping mice in start and goal positions and incorporating angled vestibules resulting in the addition of numerous corners, can be effectively adapted for animals with wired implants.

The reviewer is correct in pointing out that the current maze design is not suitable for performing experiments with wired implant, particularly due to the maze’s enclosed structure and the access to the start/goal boxes through side holes. Instead, pharmacogenetics and wireless approaches for optogenetic and electrophysiology would need to be used. We have now mentioned this limitation in the discussion.

(2) Novelty: In its current format, the main axis of the paper falls on the analysis of animal behavior and the development of behavioral modeling. In this respect, while it is interesting to see how thoughtfully designed models can explain the evolution of mice search strategy in a maze, the conclusions offer limited novel findings that align with the existing body of research and prior predictions.

We agree with the reviewer that our study is weakly connected to previous researches on hippocampus and spatial navigation, as it consists mainly of animal behavior analysis and modeling and addresses a relatively unexplored topic. We hope that the combination of our behavioral approach with optogenetic and electrophysiology will allow in the future new insights that are in line with the existing body of research.

(3) Scalability and accessibility: While the approach may be intriguing to experts who have an interest in or are familiar with the Barnes maze, its presentation seems to primarily target this specific audience. Therefore, there is a lack of clarity and discussion regarding the scalability of behavioral modeling to experiments involving other search strategies (such as sequence or episodic learning), other animal models, or the potential for translational applications. The scalability of the method would greatly benefit a broader scientific community. In line with this view, the paper's conclusions heavily rely on the development of new models using custom-made codes. Therefore, it would be advantageous to make these codes readily available, and if possible, provide access to the processed data as well. This could enhance comprehension and enable a larger audience to benefit from the methodology.

The current approach might indeed extend to other species in equivalent environments and might also constitute a general proof of principle regarding the characterization of animal behaviors by the mixing of stochastic processes. We have now mentioned these points in the discussion.

As suggest by the reviewer, we have now provided model/simulation codes and processed data to replicate the figures, at https://github.com/sebiroyerlab/Vestibule_sequences

(4) Cross-validation of models: The authors have not implemented any measures to mitigate the risk of overfitting in their modeling. It would have been beneficial to include at least some form of cross-validation with stochastic models to address this concern. Additionally, the paper lacks the presence of analytics or measures that assess and compare the performance of the models.

To avoid the risk of model overfitting, the most appropriate solution appeared to be repeating the simulations several times and examining the consistency of the obtained parameters across repetitions. For the mixture model, we now show in Supplementary figure 7 the probabilities obtained from 10 repetitions of the simulation. Similarly, for the Markov chain model, the probabilities obtained from 10 repetitions of the simulation are shown in Figure 6.

Regarding model comparison, we have simplified our mixture model into only one model, as we realized the 2 other models in the previous manuscript version were simply special cases of the 3rd model. Nevertheless, comparison was still needed for the estimation for the best value of N (the number of consecutive segments that a strategy lasts) in the mixture model. We now show the comparison of mean square errors obtained for different values of N, using t-test across 10 repetitions of the simulations (Figure 5c).

(5) Quantification of inter-animal variations in strategy development: It is important to investigate, and address the argument concerning the possibility that not all animals recruit and develop the three processes (random, spatial, and serial) in a similar manner over days of training. It would be valuable to quantify the transition in strategy across days for each individual mouse and analyze how the population average, reflecting data from individual mice, corresponds to these findings. Currently, there is a lack of such quantification and analysis in the paper.

We have added a figure (Supplementary figure 8) showing the mixture model matching analyses for individual animals. A lot of variability is indeed observed across animals, with some animals displaying strong preferences for certain strategies compare to others. The average across mouse population showed a similar trend as the result obtained with the pooled data.

Recommendations for the authors:

Summary of Reviewer Comments:

(1) In its present form, the manuscript lacks sufficient coverage of the materials and methods necessary for readers to replicate the behavioral methodology in their own research inquiries. For instance, it would be beneficial to clarify how the cues are rotated relative to the goal.

(2) The models may be over-fitted, leading to spurious conclusions, and cross-validation is necessary to rule out this possibility.

(3) The specific choice of the three strategies used to fit behavior in this model should be better justified, as other strategies may account for the observed behavior.

(4) The study would benefit from an analysis of behavior on an animal-by-animal basis, potentially revealing individual differences in strategies.

(5) Spatial behavior is not necessarily fully allocentric in this task, as only the two cues in the arena can be used for spatial orientation, unlike odor cues on the floor and sound cues in the room. This should be discussed.

(6) Making the data and code fully open source would greatly strengthen the impact of this study.

In addition, each reviewer has raised both major and minor concerns which should be addressed if possible.

Reviewer #1 (Recommendations For The Authors):

Minor points:

(1) Change "tainted" to "tinted" in Fig. 1a

(2) Should note explicitly in Fig. 2d that the goal is at vestibule 0, and also in the legend

(3) Fig. 3 legend should say "c-e)", not "c-f)"

(4) Supplementary Fig. 8 legend repeats "d)" twice

Reviewer #2 (Recommendations For The Authors):

Packard & McGaugh 1996 is cited twice as refs 5 and 14

Reviewer #3 (Recommendations For The Authors):

- Figure 3: Please correct the labels referenced as "c-f)" in the figure's legend.

- Rounding numbers issue on page 4: 82.62% + 17.37% equals 99.99%, not 100%.

We fixed all minor points. We are very thankful to the reviewers for their constructive comments.

eLife assessment

This study investigates a dietary intervention that employs a smartphone app to promote meal regularity, which may be useful. Despite no observed changes in caloric intake, the authors report significant weight loss. While the concept is very interesting and deserves to be studied due to its potential clinical relevance, the study's rigor needs to be improved, and is currently considered inadequate, notably for its reliance on self-reported food intake, a highly unreliable way to assess food intake. Additionally, the study theorizes that the intervention resets the circadian clock, but the study needs more reliable methods for assessing circadian rhythms, such as actigraphy.

Reviewer #3 (Public Review):

The authors tested a dietary intervention focused on improving meal regularity in this interesting paper. The study, a two-group, single-center, randomized, controlled, single-blind trial, utilized a smartphone application to track participants' meal frequencies and instructed the experimental group to confine their eating to these times for six weeks. The authors concluded that improving meal regularity reduced excess body weight despite food intake not being altered and contributed to overall improvements in well-being.

The concept is interesting, but the need for more rigor is of concern.

A notable limitation is the reliance on self-reported food intake, with the primary outcome being self-reported body weight/BMI, indicating an average weight loss of 2.62 kg. Despite no observed change in caloric intake, the authors assert weight loss among participants.

The trial's reliance on self-reported caloric intake is problematic, as participants tend to underreport intake; for example, in the NEJM paper (DOI: 10.1056/NEJM199212313272701), some participants underreported caloric intake by approximately 50%, rendering such data unreliable and hence misleading. More rigorous methods for assessing food intake are available and should have been utilized. Merely acknowledging the unreliability of self-reported caloric intake is insufficient as it would still leave the reader with the impression that there is no change in food intake when we actually have no idea if food intake was altered. A more robust approach to assessing food intake is imperative. Even if a decrease in caloric intake is observed through rigorous measurement, as I am convinced a more rigorous study would unveil testing this paradigm, this intervention may merely represent another short-term diet among countless others that show that one may lose weight by going on a diet, principally due to heightened dietary awareness.

Furthermore, the assessment of circadian rhythm using the MCTQ, a self-reported measure of chronotype, may not be as reliable as more objective methods like actigraphy.

Given the potential limitations associated with self-reported data in both dietary intake and circadian rhythm assessment, the overall impact of this manuscript is low. Increasing rigor by incorporating more objective and reliable measurement techniques in future studies could strengthen the validity and impact of the findings.

Reviewer #1 (Public Review):

The authors Wilming and colleagues set out to determine the impact of regularity of feeding per se on the efficiency of weight loss. The idea was to determine if individuals who consume 2-3 meals within individualized time frames, as opposed to those who exhibit stochastic feeding patterns throughout the circadian period, will cause weight loss.

The methods are rigorous, and the research is conducted using a two-group, single-center, randomized-controlled, single-blinded study design. The participants were aged between 18 and 65 years old, and a smartphone application was used to determine preferred feeding times, which were then used as defined feeding times for the experimental group. This adds strength to the study since restricting feeding within preferred/personalized feeding windows will improve compliance and study completion. Following a 14-day exploration phase and a 6-week intervention period in a cohort of 100 participants (inclusive of both the controls and the experimental group that completed the study), the authors conclude that when meals are restricted to 45min or less durations (MTVS of 3 or less), this leads to efficient weight loss. Surprisingly, the study excludes the impact of self-reported meal composition on the efficiency of weight loss in the experimental group. In light of this, it is important to follow up on this observation and develop rigorous study designs that will comprehensively assess the impact of changes (sustained) in dietary composition on weight loss. The study also reports interesting effects of regularity of feeding on eating behavior, which appears to be independent of weight loss. Perhaps the most important observation is that personalized interventions that cater to individual circadian needs will likely result in more significant weight loss than when interventions are mismatched with personal circadian structures. One are of concern for the study is its two-group design; however, single-group cross-over designs are tedious to develop, and an adequate 'wash-out' period may be difficult to predict. A second weakness is not considering the different biological variables and racial and ethnic diversity and how that might impact outcomes. In sum, the authors have achieved the aims of the study, which will likely help move the field forward.

Reviewer #2 (Public Review):

Summary:

The authors investigated the effects of the timing of dietary occasions on weight loss and well-being with the aim of explaining if a consistent, timely alignment of dietary occasions throughout the days of the week could improve weight management and overall well-being. The authors attributed these outcomes to a timely alignment of dietary occasions with the body's own circadian rhythms. However, the only evidence the authors provided for this hypothesis is the assumption that the individual timing of dietary occasions of the study participants identified before the intervention reflects the body's own circadian rhythms. This concept is rooted in understanding of dietary cues as a zeitgeber for the circadian system, potentially leading to more efficient energy use and weight management. Furthermore, the primary outcome, body weight loss, was self-reported by the study participants.

Strengths:

The innovative focus of the study on the timing of dietary occasions rather than daily energy intake or diet composition presents a fresh perspective in dietary intervention research. The feasibility of the diet plan, developed based on individual profiles of the timing of dietary occasions identified before the intervention, marks a significant step towards personalised nutrition.

Weaknesses:

Several methodological issues detract from the study's credibility, including unclear definitions not widely recognized in nutrition or dietetics (e.g., "caloric event"), lack of comprehensive data on body composition, and potential confounders not accounted for (e.g., age range, menstrual cycle, shift work, unmatched cohorts, inclusion of individuals with normal weight, overweight, and obesity). The primary outcome's reliance on self-reported body weight and subsequent measurement biases further undermines the reliability of the findings. Additionally, the absence of registration in clinical trial registries, such as the EU Clinical Trials Register or clinicaltrials.gov, and the multiple testing of hypotheses which were not listed a priori in the research protocol published on the German Register of Clinical Trials impede the study's transparency and reproducibility.

Achievement of Objectives and Support for Conclusions:

The study's objectives were partially met; however, the interpretation of the effects of meal timing on weight loss is compromised by the weaknesses mentioned above. The evidence only partially supports some of the claims due to methodological flaws and unstructured data analysis.

Impact and Utility:

Despite its innovative approach, significant methodological and analytical shortcomings limit the study's utility. If these issues were addressed, the research could have meaningful implications for dietary interventions and metabolic research. The concept of timing of dietary occasions in sync with circadian rhythms holds promise but requires further rigorous investigation.

Author response:

The following is the authors’ response to the original reviews.

We are thankful to the reviewers and the editor for their detailed feedback, insightful suggestions, and thoughtful assessment of our work. Our point-by-point responses to the comments and suggestions are below.

The revised manuscript has taken into account all the comments of the three reviewers. Modifications include corrections to errors in spelling and unit notation, additional quantification, improvements to the clarity of the language in some places, as well as additional detail in the descriptions of the methods, and revisions to the figures and figure legends.

We have also undertaken additional analyses and added materials in response to reviewer suggestions. In brief:

In response to a suggestion from Reviewer #1, we added Figure 6-1 to show examples of the calcium traces of individual fish and individual ROIs from the condensed data in Figure 6. We revised Figure 7 as follows:

• We added an analysis of the duration of the response to shock to address comments from Reviewers #2 and #3.

• In response to Reviewer #3, we added histograms showing the distribution of the amplitudes of the calcium signals in the gsc2 and rln3a neurons to show, without relying on the detection of peaks in the calcium trace, that the rln3a neurons have more oscillations in activity.

We added Figure 8-2 in response to the suggestion from Reviewer #3 to analyze turning behavior in larvae with ablated rln3a neurons.

To address Reviewer #2’s suggestion to show how the ablated transgenic animals compare to the non-ablated transgenic animals of the same genotype, we have added this analysis as Figure 8-3.

A detailed point-by-point is as follows:

The reviewers agree that the study of Spikol et al is important, with novel findings and exciting genetic tools for targeting cell types in the nucleus incertus. The conclusions are overall solid. Results could nonetheless be strengthened by performing few additional optogenetic experiments and by consolidating the analysis of calcium imaging and behavioral recordings as summarized below.

(1) Light pulses used for optogenetic-mediated connectivity mapping were very long (5s), which could lead to non specific activation of numerous population of neurons than the targeted ones. To confirm their results, the authors should repeat their experiments with brief 5-50ms (500ms maximum) -long light pulses for stimulation.

As the activity of the gsc2 neurons is already increased by 1.8 fold (± 0.28) within the first frame that the laser is activated (duration ~200 msec), it is unlikely that that the observed response is due to non-specific activation induced by the long light pulse.

(2) In terms of analysis, the authors should improve :

a) The detection of calcium events in the "calcium trace" showing the change in fluorescence over time by detecting the sharp increase in the signal when intracellular calcium rises;

We have added an additional analysis to Figure 7 that does not rely on detection of calcium peaks. See response to Reviewer #3.

b) The detection of bouts in the behavioral recordings by measuring when the tail beat starts and ends, thereby distinguishing the active swimming during bouts from the immobility observed between bouts.

Our recordings capture the entire arena that the larva can explore in the experiment and therefore lack the spatial resolution to capture and analyze the tail beat. Rather, we measured the frequency and length of phases of movement in which the larva shows no more than 1 second of immobility. To avoid confusion with studies that measure bouts from the onset of tail movement, we removed this term from the manuscript and refer to activity as phases of movement.

(3) The reviewers also ask for more precisions in the characterization of the newly-generated knock-in lines and the corresponding anatomy as explained in their detailed reports.

Please refer to the point-by-point request for additional details that have now been added to the manuscript.

Reviewer #1 (Recommendations For The Authors):

The conclusions of this paper are mostly well supported by data, but some technical aspects, especially about calcium imaging and data analysis, need to be clarified.

(1) Both the endogenous gsc2 mRNA expression and Tg(gsc2:QF2) transgenic expression are observed in a neuronal population in the NI, but also in a more sparsely distributed population of neurons located more anteriorly (for example, Fig. 2B, Fig. 5A). The latter population is not mentioned in the text. It would be necessary to clarify whether or not this anterior population is also considered as the NI, and whether this population was included for the analysis of the projection patterns and ablation experiments.

The sparsely distributed neurons had been mentioned in the Results, line 134, but we have now added more detail. In line 328, we have clarified that: “As the sparsely distributed anterior group of gsc2 neurons (Fig. 2B, C) are anatomically distinct from the main cluster and not within the nucleus incertus proper, they were excluded from subsequent analyses.”

(2) Both Tg(gsc2:QF2) and Tg(rln3a:QF2) transgenic lines have the QF genes inserted in the coding region of the targeted genes. This probably leads to knock out of the gene in the targeted allele. Can the authors mention whether or not the endogenous expression of gsc2 and rln3a was affected in the transgenic larvae? Is it possible that the results they obtained using these transgenic lines are affected by the (heterozygous or homozygous) mutation of the targeted genes?

Figure 8-1 includes in situ hybridization for gsc2 and rln3a in heterozygous Tg(gsc2:QF2)c721; Tg(QUAS:GFP)c578 and Tg(rln3a:QF2; he1.1:YFP)c836; Tg(QUAS:GFP)c578 transgenic larvae.

The expression of gsc2 is unaffected in Tg(gsc2:QF2)c721; Tg(QUAS:GFP)c578 heterozygotes

(Fig. 8-1A), whereas the expression of rln3a is reduced in Tg(rln3a:QF2; he1.1:YFP)c836; Tg(QUAS:GFP)c578 heterozygous larvae (Fig. 8-1D), as mentioned in the legend for Figure 8-1. We confirmed these findings by comparing endogenous gene expression between transgenic and non-transgenic siblings that were processed for RNA in situ hybridization in the same tube.

The behavioral results we obtained are not due to rln3a heterozygosity because comparisons were made with sibling larvae that are also heterozygous for Tg(rln3a:QF2; he1.1:YFP)c836; Tg(QUAS:GFP)c578, as stated in the Figure 8 legend.

(3) Optogenetic activation and simultaneous calcium imaging is elegantly designed using the combination of the orthogonal Gal4/UAS and QF2/QUAS systems (Fig. 6). However, I have some concerns about the analysis of calcium responses from a technical point of view. Their definition of ΔF/F in this manuscript is described as (F-Fmin)/(Fmax-Fmin) (see line 1406). This is confusing because it is different from the conventional definition of ΔF/F, which is F-F0/F0, where F0 is a baseline GCaMP fluorescence. Their way of calculating the ΔF/F is inappropriate for measuring the change in fluorescence relative to the baseline signal because it rather normalizes the amplitude of the responses across different ROIs. The same argument applies to the analyses done for Fig. 7.

We have taken a careful look at our analyses and replotted the data using F-F0/F0. However, this only changes Y-axis values and does not change the shape of the calcium trace or the change in signal upon stimulation. Both metrics (F-F0/F0 and (F-Fmin)/(Fmax-Fmin)) adjust the fluorescence values of each ROI to its own baseline.

(4) The %ΔF/F plots shown in Fig.6 are highly condensed showing the average of different ROIs (cells) within one fish and then the average of multiple fish. It would be helpful to see example calcium traces of individual ROIs and individual fish to know the variability across ROIs and fish. Also, It would be helpful to know how much laser power (561 nm laser) was used to photostimulate ReaChR.

Laser power (5%) was added to the section titled Calcium Signaling in Methods.

In Figure 6, shading in the %ΔF/F plots (D, D’, E, E’, F, F’, G, G’, H, H’) represents the variability across ROIs, and the dot plots (D’’, E’’, F’’, G’’, H’’) show the variability across fish (where each data point represents an individual fish). We have now also added Figure 6-1 with examples of calcium traces from individual fish and individual ROIs.

(5) Some calcium traces presented in Fig. 6 (Fig. 6D, D', F, H, H') show discontinuous fluctuations at the onset and offset of the photostimulation period. Is this caused by some artifacts introduced by switching the settings for the photostimulation? The authors should mention if there are some alternative explanations for this discontinuity.

As noted by the reviewer, this artifact does result from switching the settings for photostimulation, which we mention in the legend for Figure 6.

(6) In the introduction, they mention that the griseum centrale is a presumed analogue of the NI (lines 74-75). It would be helpful for the readers to better understand the brain anatomy if the authors could discuss whether or not their findings on the gsc2 and rln3a NI neurons support this idea.

Our findings on the gsc2 and rln3a neurons support the idea that the griseum centrale of fish is the analogue of the mammalian NI. We have now edited the text in the third paragraph of the discussion, line 1271, to make this point more clearly: “By labeling with QUAS-driven fluorescent reporters, we determined that the anatomical location, neurotransmitter phenotype, and hodological properties of gsc2 and rln3a neurons are consistent with NI identity, supporting the assertion that the griseum centrale of fish is analogous to the mammalian NI. Both groups of neurons are GABAergic, reside on the floor of the fourth ventricle and project to the interpeduncular nucleus.”

Reviewer #2 (Recommendations For The Authors):

Major comments:

(1) Throughout the figures a need for more precision and reference in the anatomical evidence:

• Specify how many planes over which height were projected for each Z-projection in Figure 1,2,3, ....

We added this information to the last paragraph of the section titled Confocal Imaging within the Materials and Methods.

• Provide the rhombomere numbers, deliminate the ventricles & always indicate on the panel the orientation (Rostral Caudal, Left Right or Ventral Dorsal) for Figure 1 panels D-F , Figure 2-1B-G, Figure 2-2A-C in the adult brain, Figure 3.

We annotated Figures 2-1 and 2-2 as suggested. We also indicated the orientation (anterior to the top or anterior to the left) in all figure legends. For additional context on the position of gsc2 and rln3a neurons within the larval brain, refer to Fig. 1A-C’, Fig. 1-2A, Fig. 2, Fig. 4 and Fig. 5.

• Add close up when necessary: Figure 2-2A-C, specify in the text & in the figure where are the axon bundles from the gsc2+ neurons in the adult brain- seems interesting and is not commented on?

We added a note to the legend of Figure 2-2: Arrowheads in B and B’ indicate mApple labeling of gsc2 neuronal projections to the hypothalamus. We also refer to Fig 2-2B, B’ in the Results section titled Distinct Projection Patterns of gsc2 and rln3a neurons.

• keep the same color for one transgene within one figure: example, glutamatergic neurons should always be the same color in A,B,C - it is confusing as it is.

We have followed the reviewer’s suggestion and made the color scheme consistent in Figure 3.

• Movies: add the labels (which transgenic lines in which color, orientation & anatomical boundaries for NI, PAG, any other critical region that receives their projections and the brain ventricle boundaries) on the anatomical movies in supplemental (ex Movie 4-1 for gsc2 neurons and 4-2 for rln3 neurons: add cerebellum, IPN, raphe, diencephalon, and rostral and caudal hypothalamus, medulla for 4-1 as well as lateral hypothalamus and optic tectum for 42); add the ablated region when necessary.

We added more detail to the movie legends. Please refer to Figure 4 for additional anatomical details.

• for highlighting projections from NI neurons and distinguish them from the PAG neurons, the authors elegantly used 2 Photon ablation of one versus the other cluster: this method is valid but we need more resolution that the Z stacks added in supplemental by performing substraction of before and after maps.

We are not sure what the author meant by subtraction as there are no before and after images in this experiment. Larvae underwent ablation of cell bodies and were imaged one day later in comparison to unablated larvae.

In particular, it is not clear to me if both PAG and NI rln3a neurons project to medulla - can the authors specify this point & the comparison between intact & PAG vs NI ablation maps? The authors should resolve better the projections to all targeted regions of NI gsc2 neurons and differentiate them from other PAG gsc2 neurons, same for rln3a neurons.

We have clarified this point on line 549.

Make sure to mention in the result section the duration between ablation & observation that is key for the axons to degrade.

We always assessed degeneration of neuronal processes at 1-day post-ablation.

(“2) calcium imaging experiments:

a) with optogenetic connectivity mapping:

the authors combine an impressive diverse set of optogenetic actuators & sensors by taking advantage of the QUAS/QF2 and UAS/GAL4 systems to test connectivity from Hb-IPN onto gsc2 and rln3 neurons.

The experiments are convincing but the choice of the duration of the stimulation (5s) is not adequate to test for direct connectivity: the authors should make sure that response in gsc2 neurons is observed with short duration (50ms-1s max).

As noted above:

“As the activity of the gsc2 neurons is already increased by 1.8 fold (± 0.28) within the first frame that the laser is activated (duration ~200 msec), it is unlikely that that the observed response is due to non-specific activation induced by the long light pulse.”

note: Specify that the gsc2 neurons tested are in NI.

We have edited the text accordingly in the Results section titled Afferent input to the NI from the dHb-IPN pathway.

b) for the response to shock: in the example shown for rln3 neurons, the activity differs before and after the shock with long phases of inhibition that were not seen before. Is it representative? the authors should carefully stare at their data & make sure there is no difference in activity patterns after shock versus before.

We reexamined the responses for each of the rln3a neurons individually and confirmed that, although oscillations in activity are frequent, the apparent inhibition (excursions below baseline) are an idiosyncratic feature of the particular example shown.

(3) motor activity assay:

a) there seems to be a misconception in the use of the word "bout" to estimate in panels H and I bout distance and duration and the analysis should be performed with the criterion used by all in the motor field:

As we know now well based on the work of many labs on larval zebrafish (Orger, Baier, Engert, Wyart, Burgess, Portugues, Bianco, Scott, ...), a bout is defined as a discrete locomotor event corresponding to a distance swam of typically 1-6mm, bout duration is typically 200ms and larvae exhibit a bout every s or so during exploration (see Mirat et al Frontiers 2013; Marques et al Current Biology 2018; Rajan et al. Cell Reports 2022).

Since the larval zebrafish has a low Reynolds number, it does not show much glide and its movement corresponds widely to the active phase of the tail beats.

Instead of detecting the active (moving) frames as bouts, the authors however estimate these values quite off that indicate an error of calibration in the detection of a movement: a bout cannot last for 5-10s, nor can the fish swim for more than 1 cm per bout (in the definition of the authors, bout last for 5-10 s, and bout correspond to 10 cm as 50 cm is covered in 5 bouts).

The authors should therefore distinguish the active (moving) from inactive (immobile) phase of the behavior to define bouts & analyze the corresponding distance travelled and duration of active swimming. They would also benefit from calculating the % of time spent swimming in order to test whether the fish with ablated rln3 neurons change the fraction of the time spent swimming.

As noted above:

Our recordings capture the entire arena that the larva can explore in the experiment and therefore lack the spatial resolution to capture and analyze the tail beat. Rather, we measured the frequency and length of phases of movement in which the larva shows no more than 1 second of immobility. To avoid confusion with studies that measure bouts from the onset of tail movement, we removed this term from the manuscript and refer to activity as phases of movement.

Note that a duration in seconds is not a length and that the corresponding symbol for seconds in a scientific publication is "s" and not "sec".

We have corrected this.

b) controls in these experiments are key as many clutches differ in their spontaneous exploration and there is a lot of variation for 2 min long recordings (baseline is 115s). The authors specify that the control unablated are a mix of siblings; they should show us how the ablated transgenic animals compare to the non ablated transgenic animals of the same clutch.

The unablated Tg(gsc2:QF2)c721; Tg(QUAS:GFP)c578 and Tg(rln3a:QF2, he1.1:YFP)c836; Tg(QUAS:GFP)c578 larvae in the control group are siblings of ablated larvae. We repeated the analyses using either the Tg(gsc2:QF2)c721; Tg(QUAS:GFP)c578 or Tg(rln3a:QF2, he1.1:YFP)c836; Tg(QUAS:GFP)c578 larvae only as controls and added the results in Figure 8-3. Although the statistical power is slightly reduced due to a smaller number of samples in the control group, the conclusions are the same, as the behavior of Tg(gsc2:QF2)c721; Tg(QUAS:GFP)c578 and Tg(rln3a:QF2, he1.1:YFP)c836; Tg(QUAS:GFP)c578 unablated larvae is indistinguishable.

Minor comments:

(1) Anatomy :

• Add precision in the anatomy in Figure 1:

• Improve contrast for cckb.

The contrast is determined by the signal to background ratio from the fluorescence in situ hybridization. Increasing the brightness would increase both the signal and the background, as any modification must be applied to the whole image.

• since the number of neurons seems low in each category, could you quantify the number of rln3+, nmbb+, gsc2+, cckb+ neurons in NI?

Quantification of neuronal numbers has been added to the first Results section titled Identification of gsc2 neurons in the Nucleus Incertus, lines 219-224.

note: indicate duration for the integral of the DF/F in s and not in frames.

We have added this in the legends for Figures 6 and 7 and in Materials and Methods.

(2) Genetic tools:

To generate a driver line for the rln3+ neurons using the Q system, the authors used the promoter for the hatching gland in order to drive expression in a structure outside of the nervous system that turns on early and transiently during development: this is a very elegant approach that should be used by many more researchers.

If the her1 construct was integrate together with the QF2 in the first exon of the rln3 locus as shown in Figure 2, the construct should not be listed with a ";" instead of a "," behind rln3a:QF2 in the transgene name. Please edit the transgene name accordingly.

We have edited the text accordingly.

(3) Typos:

GABAergic neurons is misspelled twice in Figure 3.

Thank you for catching this. We have corrected the misspellings.

Reviewer #3 (Recommendations For The Authors):

• More analysis should be done to better characterize the calcium activity of gsc2 and rln3a populations. Specifically:

Spontaneous activity is estimated by finding peaks in the time-series data, but the example in Fig7 raises concerns about this process: Two peaks for the gsc2 cell are identified while numerous other peaks of apparently similar SNR are not detected. Moreover, the inset images suggest GCaMP7a expression might be weaker in the gsc2 transgenic and as such, differences in peak count might be related to the SNR of the recordings rather than underlying activity. Overall, the process for estimating spontaneous activity should be more rigorous.

To not solely rely on the identification of peaks in the calcium traces, we also plotted histograms of the amplitudes of the calcium signals for the rln3a and gsc2 neurons. The histograms show that the amplitudes of the rln3a calcium signals frequently occur at small and large values (suggesting large fluctuations in activity), whereas the amplitudes of the gsc2 calcium signals occur most frequently at median values. We added this analysis to a revised Figure 7.

Interestingly, there are a number of large negative excursions in the calcium data for the rln3a cell - what is the authors' interpretation of these? Could it be that presynaptic inhibition via GABA-B receptors in dIPN might influence dIPN-innervating rln3a neurons?

As noted above:

We reexamined the responses for each of the rln3a neurons individually and confirmed that, although oscillations in activity are frequent, the apparent inhibition (excursions below baseline) are an idiosyncratic feature of the particular example shown.

Regarding shock-evoked activity, the authors state "rln3a neurons showed ... little response to shock", yet the immediate response after shock appears very similar in gsc2 vs rln3a cells (approx 30 units on the dF/F scale). The subsequent time-course of the response is what appears to distinguish gsc2 versus rln3a; it might thus be useful to separately quantify the amplitude and decay time constant of the shock evoked response for the two populations.

The reviewer is correct that the difference between the gsc2 and rln3a neurons in the response to shock is dependent on the duration of time post-shock that is analyzed. Thus, the more relevant feature is the length of the response rather than the size. To reflect this, we compared the average length of responses for the gsc2 and rln3a neurons. We have now added this analysis to Figure 7 and updated the text accordingly.

• The difference in spontaneous locomotor behavior is interesting and the example tracking data suggests there might also be differences in turn angle distribution and/or turn chain length following rln3 NI ablations. I would recommend the authors consider exploring this.

Thank you for this suggestion. We wrote additional code to quantify turning behavior and found that larvae with rln3a NI neurons ablated do indeed have a statistically significant increase in turning compared to other groups. We now show this analysis as Figure 8-2 and we added an explanation of the quantification of turning behavior to the Methods section titled Locomotor assay.

• I didn't follow the reasoning in the discussion that activity of rln3a cells may control transitions between phases of behavioral activity and inactivity. The events (at least those that are detected) in Fig7 occur with an average interval exceeding 30 s, yet swim bouts occur at a frequency around 1 Hz. The authors should clarify their hypothesis about how these disparate timescales might be connected.

As noted above:

Our recordings capture the entire arena that the larva can explore in the experiment and therefore lack the spatial resolution to capture and analyze the tail beat. Rather, we measure the frequency and length of phases of movement in which the larva shows no more than 1 second of immobility. To avoid confusion with studies that measure bouts from the onset of tail movement, we removed this term from the manuscript and refer to activity as phases of movement.

• Fig2-2: Images are ordered from (A, B, C) anterior to (A', B', C') posterior. Its not clear what this means and images appear to be in sequence A, A', B, B'.... please clarify and consider including a cartoon of the brain in sagittal view showing location of sections indicated.

We clarified the text in the Figure 2-2 legend and added a drawing of the brain showing the location of the sections.

• In Fig7, why are 300 frames analyzed pre/post shock? Even for gsc2, the response appears complete in ~100 frames.

Reviewer #2 also pointed out that the difference between the gsc2 and rln3a neurons in the response to shock is dependent on the duration of time post-shock that is analyzed. Thus, the more relevant feature is the length of the response rather than the size. To reflect this, we compared the average length of response for the gsc2 and rln3a neurons and modified the text and Figure as described above.

• What are the large negative excursions in the calcium signal in the rln3a data (Fig7E)?

See response to Reviewer # 2, repeated below:

We looked through each of the responses of individual rln3a neuron and confirmed that, although oscillations in activity are frequent among the rln3a neurons, the apparent inhibition (excursions below baseline) are an idiosyncratic feature of the particular example shown.

• There are several large and apparently perfectly straight lines in the fish tracking examples (Fig8) suggestive of tracking errors (ie. where the tracked centroid instantaneously jumps across the camera frame). Please investigate these and include analysis of the distribution of swim velocities to support the validity of the tracking data.

The reason for this is indeed imperfect tracking resulting in frames in which the tracker does not detect the larva. The result is that the larva appears to move 1 cm or more in a single frame. However, analysis of the distribution of distances across all frames shows that these events (movement of 1 cm or more in a single frame) are rare (less than 0.04%), and there are no systematic differences that would explain the differences in locomotor behavior presented in Fig. 8. A summary of the data is as follows:

Controls: 0.0249% of distances 1 cm or greater gsc2 neurons ablated: 0.0302% of distances 1 cm or greater rln3a NI neurons ablated: 0.0287% of distances 1 cm or greater rln3a PAG neurons ablated: 0.0241% of distance 1 cm or greater

• Insufficient detail is provided in the methods about how swim bouts are detected (and their durations extracted) from the centroids tracking data. Please expand detail in this section.

We added an explanation to the Methods section titled Locomotor assay.

eLife assessment

This study presents an important finding on the anatomical connectivity and functional roles of the previously uncharacterized neuronal populations in the nucleus incertus. The evidence supporting the conclusions is convincing, with imaging and manipulations of the genetically targeted populations of neurons. The work presents a significant milestone for future mechanistic studies of the nucleus incertus.

Reviewer #1 (Public Review):

Spikol et al. investigate the roles of two distinct populations of neurons in the nucleus incertus (NI). The authors established two new transgenic lines that label gsc2- and rln3a-expressing neurons. They show that the gsc2+ and rln3a+ NI neurons show divergent projection patterns and project to different parts of the interpeduncular nucleus (IPN), which receive inputs from the habenula (Hb). Furthermore, calcium imaging shows that gsc2 neurons are activated by the optogenetic activation of the dorsal Hb-IPN and respond to aversive electric shock stimuli, while rln3a neurons are highly spontaneously active. The ablation of rln3a neurons, but not gsc2 neurons, alters locomotor activity of zebrafish larvae.

The strength of the paper is their genetic approach that enabled the authors to characterize many different features of the two genetically targeted populations in the NI. These two neuronal populations are anatomically closely apposed and would have been indistinguishable without their genetic tools. Their analyses provide valuable information on the diverse anatomical, physiological and behavioral functions of the different NI subtypes. On the other hand, these pieces of evidence are loosely linked with each other to reach a mechanistic understanding of how the NI works in a circuit. For example, the anatomical study revealed the connections from the NI to the IPN, while the optogenetic mapping experiments investigate the other way around, i.e. the connection from the IPN to the NI.

Reviewer #3 (Public Review):

This study uses a range of methods to characterize heterogeneous neural populations within the nucleus incertus (NI). The authors focus on two major populations, expressing gsc2 and rln3a, and present convincing evidence that these cells have different patterns of connectivity, calcium activity and effects on behavior. Although the study does not go as far as clarifying the role of NI in any specific neural computation or aspect of behavioral control, the findings will be valuable in support of future endeavors to do so. In particular, the authors have made two beautiful knock-in lines that recapitulate endogenous expression pattern of gsc2 and rln3a which will be a powerful tool to study the roles of the relevant NI cells. Experiments are well done, data are high quality and most claims are well supported. In this revised version, the authors have added additional analysis that has clarified their results and strengthened some of the claims.

Two points of note:

• The data very clearly show different patterns of neurites for gsc2 and rln3a neurons in the IPN and the authors interpret these are being axonal arbors. However, they do not rule out the possibility that some of the processes might be dendritic in nature. Of relevance to this point, they cite a recent study (Petrucco et al. 2023) that confirmed that, as in other species, tegmental neurons in zebrafish extend spatially segregated dendritic as well as axonal arbors into IPN, and the authors speculate that these GABAergic tegmental cells might in fact be part of NI.

• Although the gsc2 and rln3a populations show differences in calcium activity, there is not as clear a dichotomy as stated in the abstract. For example, both populations clearly respond to electric shocks, albeit with different response time courses.

Reviewer #4 (Public Review):

Summary:

In the present study, Spikol et al. explore the projection patterns and functional characteristics of two distinct and genetically defined populations in the larval zebrafish Nucleus Incertus (NI), expressing the transcription factor gsc2 or the neuropeptide rln3a. To label in vivo these neurons two transgenic lines were generated by CRISPR/Cas9 mediated Knock-in. These genetic tools allowed the analysis of the projection patterns of these neuronal populations showing that the NI neurons expressing gsc2 and rln3a exhibit markedly different projection patterns, targeting separate subregions within the midbrain interpeduncular nucleus (IPN).<br /> Functional imaging and behavioral analysis revealed that while gsc2 neurons respond to electric shock stimuli, rln3a neurons show high spontaneous activity and play a role in regulating locomotor activity.

Strengths:

The paper relies on a series of rigorous experimental approaches including molecular genetic, neuroanatomical, functional and behavioral analysis. The resources generated including the two knock-in transgenic reporter lines will be of great value for the zebrafish neurobiology community as well as inspire further studies of the NI in other model systems.

Weaknesses:

Technical weaknesses present in the first version of the manuscript have largely been addressed in the present revision.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment:

This study uses carefully designed experiments to generate a useful behavioural and neuroimaging dataset on visual cognition. The results provide solid evidence for the involvement of higher-order visual cortex in processing visual oddballs and asymmetry. However, the evidence provided for the very strong claims of homogeneity as a novel concept in vision science, separable from existing concepts such as target saliency, is inadequate.

We appreciate the positive and balanced assessment from the reviewers. We agree that visual homogeneity is similar to existing concepts such as target saliency. We have tried our best to articulate our rationale for defining it as a novel concept. However, the debate about whether visual homogeneity is novel or related to existing concepts is completely beside the point, since that is not the key contribution of our study.

Our key contribution is our quantitative model for how the brain could be solving generic visual tasks by operating on a feature space. In the literature there are no theories regarding the decision-making process by which the brain could be solving generic visual tasks. In fact, oddball search tasks, same-different tasks and symmetry tasks are never even mentioned in the same study because it is tacitly assumed that the underlying processes are completely different! Our work brings together these disparate tasks by proposing a specific computation that enables the brain to solve both types of tasks and providing evidence for it. This specific computation is a well-defined, falsifiable model that will need to be replicated, elaborated and refined by future studies.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The authors define a new metric for visual displays, derived from psychophysical response times, called visual homogeneity (VH). They attempt to show that VH is explanatory of response times across multiple visual tasks. They use fMRI to find visual cortex regions with VH-correlated activity. On this basis, they declare a new visual region in the human brain, area VH, whose purpose is to represent VH for the purpose of visual search and symmetry tasks.

Thank you for your concise summary. We appreciate your careful reading and thoughtful and constructive comments.

Strengths:

The authors present carefully designed experiments, combining multiple types of visual judgments and multiple types of visual stimuli with concurrent fMRI measurements. This is a rich dataset with many possibilities for analysis and interpretation.

Thank you for your accurate assessment of the strengths of our study.

Weaknesses:

The datasets presented here should provide a rich basis for analysis. However, in this version of the manuscript, I believe that there are major problems with the logic underlying the authors' new theory of visual homogeneity (VH), with the specific methods they used to calculate VH, and with their interpretation of psychophysical results using these methods. These problems with the coherency of VH as a theoretical construct and metric value make it hard to interpret the fMRI results based on searchlight analysis of neural activity correlated with VH.

We appreciate your concerns, and have tried our best to respond to them fully against your specific concerns below.

In addition, the large regions of VH correlations identified in Experiments 1 and 2 vs. Experiments 3 and 4 are barely overlapping. This undermines the claim that VH is a universal quantity, represented in a newly discovered area of the visual cortex, that underlies a wide variety of visual tasks and functions.

We agree with you that the VH regions defined using symmetry task and search task do not overlap completely (as we have shown in Figure S13). However this is to be expected for several reasons. First, the images in the symmetry task were presented at fixation, whereas the images in the visual search task were presented peripherally. Second, the lack of overlap could be due to variations across individuals. Indeed, considerable individual variability has been observed in the location of category-selective regions such as VWFA (Glezer and Riesenhuber 2013) and FFA (Weiner and Grill-Spector, 2012). We propose that testing the same participants on both search and symmetry tasks would reveal overlapping VH regions. We now acknowledge these issues in the Results (p. 26).

Maybe I have missed something, or there is some flaw in my logic. But, absent that, I think the authors should radically reconsider their theory, analyses, and interpretations, in light of the detailed comments below, to make the best use of their extensive and valuable datasets combining behavior and fMRI. I think doing so could lead to a much more coherent and convincing paper, albeit possibly supporting less novel conclusions.

We appreciate your concerns. We have tried our best to respond to them fully against your specific concerns below.

THEORY AND ANALYSIS OF VH

(1) VH is an unnecessary, complex proxy for response time and target-distractor similarity. VH is defined as a novel visual quality, calculable for both arrays of objects (as studied in Experiments 1-3) and individual objects (as studied in Experiment 4). It is derived from a center-to-distance calculation in a perceptual space. That space in turn is derived from the multi-dimensional scaling of response times for target-distractor pairs in an oddball detection task (Experiments 1 and 2) or in a same-different task (Experiments 3 and 4).

The above statements are not entirely correct. Experiments 1 & 3 are oddball visual search experiments. Their purpose was to estimate the underlying perceptual space of objects.

Proximity of objects in the space is inversely proportional to response times for arrays in which they were paired. These response times are higher for more similar objects. Hence, proximity is proportional to similarity. This is visible in Fig. 2B as the close clustering of complex, confusable animal shapes.

VH, i.e. distance-to-center, for target-present arrays, is calculated as shown in Fig. 1C, based on a point on the line connecting the target and distractors. The authors justify this idea with previous findings that responses to multiple stimuli are an average of responses to the constituent individual stimuli. The distance of the connecting line to the center is inversely proportional to the distance between the two stimuli in the pair, as shown in Fig. 2D. As a result, VH is inversely proportional to the distance between the stimuli and thus to stimulus similarity and response times. But this just makes VH a highly derived, unnecessarily complex proxy for target-distractor similarity and response time. The original response times on which the perceptual space is based are far more simple and direct measures of similarity for predicting response times.

We agree that VH brings no explanatory power to target-present searches, since target-present response times are a direct estimate of target-distractor similarity. However, we are additionally explaining target-absent response times. Target-absent response times are well known to vary systematically with image properties, but why they do so have not been clear in the literature.

Our key conceptual advance lies in relating the neural response to a search array to the neural response of the constituent elements, and in proposing a decision variable using which participants can make both target-present and target-absent judgements on any search array.

(2) The use of VH derived from Experiment 1 to predict response times in Experiment 2 is circular and does not validate the VH theory.

The use of VH, a response time proxy, to predict response times in other, similar tasks, using the same stimuli, is circular. In effect, response times are being used to predict response times across two similar experiments using the same stimuli. Experiment 1 and the target present condition of Experiment 2 involve the same essential task of oddball detection. The results of Experiment 1 are converted into VH values as described above, and these are used to predict response times in Experiment 2 (Fig. 2F). Since VH is a derived proxy for response values in Experiment 1, this prediction is circular, and the observed correlation shows only consistency between two oddball detection tasks in two experiments using the same stimuli.

We agree that it would be circular to use oddball search times in Experiment 1 to explain only target-present search times in Experiment 2, since they basically involve the same searches. However, we are explaining both target-present and target-absent search times in a unified framework; systematic variations in target-absent search times have been noted in the literature but never really explained. One could still simply say that target-absent search times are some function of the target-present search times, but this still doesn’t provide an explanation for how participants are making target-present and absent decisions. The existing literature contains models for how visual search might occur for a specific target and distractor but does not elucidate how participants might perform generic visual search where target and distractors are not known in advance.

Our key conceptual advance lies in relating the neural response to a search array to the neural response of the constituent elements, and in proposing a decision variable using which participants can make both target-present and target-absent judgements on any search array.

(3) The negative correlation of target-absent response times with VH as it is defined for target-absent arrays, based on the distance of a single stimulus from the center, is uninterpretable without understanding the effects of center-fitting. Most likely, center-fitting and the different VH metrics for target-absent trials produce an inverse correlation of VH with target-distractor similarity.

We see no cause for concern with the center-fitting procedure, for several reasons. First, the best-fitting center remained stable despite many randomly initialized starting points. Second, the best-fitting center derived from one set of objects was able to predict the target-absent and target-present responses of another set of objects. Finally, the VH obtained for each object (i.e. distance from the best-fitting center) is strongly correlated with the average distance of that object from all other objects (Figure S1A). We have now clarified this in the Results (p. 11).

The construction of the VH perceptual space also involves fitting a "center" point such that distances to center predict response times as closely as possible. The effect of this fitting process on distance-to-center values for individual objects or clusters of objects is unknowable from what is presented here. These effects would depend on the residual errors after fitting response times with the connecting line distances. The center point location and its effects on the distance-to-center of single objects and object clusters are not discussed or reported here.

While it is true that the optimal center needs to be found by fitting to the data, there no particular mystery to the algorithm: we are simply performing a standard gradient-descent to maximize the fit to the data. We have described the algorithm clearly and are making our codes public. We find the algorithm to yield stable optimal centers despite many randomly initialized starting points. We find the optimal center to be able to predict responses to entirely novel images that were excluded during model training. We are making no assumption about the location of centre with respect to individual points. Therefore, we see no cause for concern regarding the center-finding algorithm.

Yet, this uninterpretable distance-to-center of single objects is chosen as the metric for VH of target-absent displays (VHabsent). This is justified by the idea that arrays of a single stimulus will produce an average response equal to one stimulus of the same kind. However, it is not logically clear why response strength to a stimulus should be a metric for homogeneity of arrays constructed from that stimulus, or even what homogeneity could mean for a single stimulus from this set. It is not clear how this VHabsent metric based on single stimuli can be equated to the connecting line VH metric for stimulus pairs, i.e. VHpresent, or how both could be plotted on a single continuum.

Most visual tasks, such as finding an animal, are thought to involve building a decision boundary on some underlying neural representation. Even visual search has been portrayed as a signal-detection problem where a particular target is to be discriminated from a distractor. However none of these formulations work in the case of generic visual tasks, where the target and distractor identities are unknown. We are proposing that, when we view a search array, the neural response to the search array can be deduced from the neural responses to the individual elements using well known rules, and that decisions about an oddball target being present or absent can be made by computing the distance of this neural response from some canonical mean firing rate of a population of neurons. This distance to center computation is what we denote as visual homogeneity. We have revised our manuscript throughout to make this clearer and we hope that this helps you understand the logic better.

It is clear, however, what should be correlated with difficulty and response time in the target-absent trials, and that is the complexity of the stimuli and the numerosity of similar distractors in the overall stimulus set. The complexity of the target, similarity with potential distractors, and the number of such similar distractors all make ruling out distractor presence more difficult. The correlation seen in Fig. 2G must reflect these kinds of effects, with higher response times for complex animal shapes with lots of similar distractors and lower response times for simpler round shapes with fewer similar distractors.

You are absolutely correct that the stimulus complexity should matter, but there are no good measures for stimulus complexity. But considering what factors are correlated with target-absent response times is entirely different from asking what decision variable or template is being used by participants to solve the task.

The example points in Fig. 2G seem to bear this out, with higher response times for the deer stimulus (complex, many close distractors in the Fig. 2B perceptual space) and lower response times for the coffee cup (simple, few close distractors in the perceptual space). While the meaning of the VH scale in Fig. 2G, and its relationship to the scale in Fig. 2F, are unknown, it seems like the Fig. 2G scale has an inverse relationship to stimulus complexity, in contrast to the expected positive relationship for Fig. 2F. This is presumably what creates the observed negative correlation in Fig. 2G.

Taken together, points 1-3 suggest that VHpresent and VHabsent are complex, unnecessary, and disconnected metrics for understanding target detection response times. The standard, simple explanation should stand. Task difficulty and response time in target detection tasks, in both present and absent trials, are positively correlated with target-distractor similarity.

Respectfully, we disagree with your assessment. Your last point is not logically consistent though: response times for target-absent trials cannot be correlated with any target-distractor similarity since there is no target in the first place in a target-absent array. We have shown that target-absent response times are in fact, independent of experimental context, which means that they index an image property that is independent of any reference target (Results, p. 15; Section S4). This property is what we define as visual homogeneity.

I think my interpretations apply to Experiments 3 and 4 as well, although I find the analysis in Fig. 4 especially hard to understand. The VH space in this case is based on Experiment 3 oddball detection in a stimulus set that included both symmetric and asymmetric objects. However, the response times for a very different task in Experiment 4, a symmetric/asymmetric judgment, are plotted against the axes derived from Experiment 3 (Fig. 4F and 4G). It is not clear to me why a measure based on oddball detection that requires no use of symmetry information should be predictive of within-stimulus symmetry detection response times. If it is, that requires a theoretical explanation not provided here.

We are using an oddball detection task to estimate perceptual dissimilarity between objects, and construct the underlying perceptual representation of both symmetric and asymmetric objects. This enabled us to then ask if some distance-to-center computation can explain response times in a symmetry detection task, and obtain an answer in the affirmative. We have reworked the text to make this clear.

(4) Contrary to the VH theory, same/different tasks are unlikely to depend on a decision boundary in the middle of a similarity or homogeneity continuum.

We have provided empirical proof for our claims, by showing that target-present response times in a visual search task are correlated with “different” responses in the same-different task, and that target-absent response times in the visual search task are correlated with “same” responses in the same-different task (Section S3).

The authors interpret the inverse relationship of response times with VHpresent and VHabsent, described above, as evidence for their theory. They hypothesize, in Fig. 1G, that VHpresent and VHabsent occupy a single scale, with maximum VHpresent falling at the same point as minimum VHabsent. This is not borne out by their analysis, since the VHpresent and VHabsent value scales are mainly overlapping, not only in Experiments 1 and 2 but also in Experiments 3 and 4. The authors dismiss this problem by saying that their analyses are a first pass that will require future refinement. Instead, the failure to conform to this basic part of the theory should be a red flag calling for revision of the theory.

We respectfully disagree – by no means did we dismiss this problem! In fact, we have explicitly acknowledged this by saying that VH does not explain all the variance in the response times, but nonetheless explains substantial variance and might form the basis for an initial guess or a fast response. The remaining variance might be explained by processes that involve more direct scrutiny. Please see Results, page 10 & 22.

The reason for this single scale is that the authors think of target detection as a boundary decision task, along a single scale, with a decision boundary somewhere in the middle, separating present and absent. This model makes sense for decision dimensions or spaces where there are two categories (right/left motion; cats vs. dogs), separated by an inherent boundary (equal left/right motion; training-defined cat/dog boundary). In these cases, there is less information near the boundary, leading to reduced speed/accuracy and producing a pattern like that shown in Fig. 1G.

The key conceptual advance of our study is that we show that even target/present, same/different or symmetry judgements can be fit into the standard decision-making framework.

This logic does not hold for target detection tasks. There is no inherent middle point boundary between target present and target absent. Instead, in both types of trials, maximum information is present when the target and distractors are most dissimilar, and minimum information is present when the target and distractors are most similar. The point of greatest similarity occurs at the limit of any metric for similarity. Correspondingly, there is no middle point dip in information that would produce greater difficulty and higher response times. Instead, task difficulty and response times increase monotonically with the similarity between targets and distractors, for both target present and target absent decisions. Thus, in Figs. 2F and 2G, response times appear to be highest for animals, which share the largest numbers of closely similar distractors.

Unfortunately, your logic does not boil down to any quantitative account, since you are using vague terms like “maximum information”. Further, any argument based solely on item similarity to explain visual search or symmetry responses cannot explain systematic variations observed for target-absent arrays and for symmetric objects, for the reasons below.

If target-distractor dissimilarity were the sole driver of response times, target-absent judgements should always take the longest time since the target and distractor have zero similarity, with no variation from one image to another. This account does not explain why target-absent response times vary so systematically.

Similarly, if symmetry judgements are solely based on comparing the dissimilarity between two halves of an object, there should be no variation in the response times of symmetric objects since the dissimilarity between their two halves is zero. However we do see systematic variation in the response times to symmetric objects.

DEFINITION OF AREA VH USING fMRI

(1) The area VH boundaries from different experiments are nearly completely non-overlapping.

In line with their theory that VH is a single continuum with a decision boundary somewhere in the middle, the authors use fMRI searchlight to find an area whose responses positively correlate with homogeneity, as calculated across all of their target present and target absent arrays. They report VH-correlated activity in regions anterior to LO. However, the VH defined by symmetry Experiments 3 and 4 (VHsymmetry) is substantially anterior to LO, while the VH defined by target detection Experiments 1 and 2 (VHdetection) is almost immediately adjacent to LO. Fig. S13 shows that VHsymmetry and VHdetection are nearly non-overlapping. This is a fundamental problem with the claim of discovering a new area that represents a new quantity that explains response times across multiple visual tasks. In addition, it is hard to understand why VHsymmetry does not show up in a straightforward subtraction between symmetric and asymmetric objects, which should show a clear difference in homogeneity. • Actually VHsymmetry is apparent even in a simple subtraction between symmetric and asymmetric objects (Figure S10). The VH regions identified using the visual search task and symmetry task have a partial overlap, not zero overlap as you are incorrectly claiming.

We have noted that it is not straightforward to interpret the overlap, since there are many confounding factors. One reason could simply be that the stimuli in the symmetry task were presented at fixation, whereas the visual search arrays contained items exclusively in the periphery. Another that the participants in the two tasks were completely different, and the lack of overlap is simply due to inter-individual variability. Testing the same participants in two tasks using similar stimuli would be ideal but this is outside the scope of this study. We have acknowledged these issues in the Results (p. 26) and in the Supplementary Material (Section S8).

(2) It is hard to understand how neural responses can be correlated with both VHpresent and VHabsent.

The main paper results for VHdetection are based on both target-present and target-absent trials, considered together. It is hard to interpret the observed correlations, since the VHpresent and VHabsent metrics are calculated in such different ways and have opposite correlations with target similarity, task difficulty, and response times (see above). It may be that one or the other dominates the observed correlations. It would be clarifying to analyze correlations for target-present and target-absent trials separately, to see if they are both positive and correlated with each other.

Thanks. The positive correlation between VH and neural response holds even when we do the analysis separately for target-present and -absent searches (correlation between neural response in VH region and visual homogeneity (n = 32, r = 0.66, p < 0.0005 for target-present searches & n = 32, r = 0.56, p < 0.005 for target-absent searches).

(3) The definition of the boundaries and purpose of a new visual area in the brain requires circumspection, abundant and convergent evidence, and careful controls.

Even if the VH metric, as defined and calculated by the authors here, is a meaningful quantity, it is a bold claim that a large cortical area just anterior to LO is devoted to calculating this metric as its major task. Vision involves much more than target detection and symmetry detection. The cortex anterior to LO is bound to perform a much wider range of visual functionalities. If the reported correlations can be clarified and supported, it would be more circumspect to treat them as one byproduct of unknown visual processing in the cortex anterior to LO, rather than treating them as the defining purpose for a large area of the visual cortex.

We totally agree with you that reporting a new brain region would require careful interpretation and abundant and converging evidence. However, this requires many studies worth of work, and historically category-selective regions like the FFA have achieved consensus only after they were replicated and confirmed across many studies. We believe our proposal for the computation of a quantity like visual homogeneity is conceptually novel, and our study represents a first step that provides some converging evidence (through replicable results across different experiments) for such a region. We have reworked our manuscript to make this point clearer (Discussion, p 32).

Reviewer #2 (Public Review):

Summary:

This study proposes visual homogeneity as a novel visual property that enables observers perform to several seemingly disparate visual tasks, such as finding an odd item, deciding if two items are the same, or judging if an object is symmetric. In Experiment 1, the reaction times on several objects were measured in human subjects. In Experiment 2, the visual homogeneity of each object was calculated based on the reaction time data. The visual homogeneity scores predicted reaction times. This value was also correlated with the BOLD signals in a specific region anterior to LO. Similar methods were used to analyze reaction time and fMRI data in a symmetry detection task. It is concluded that visual homogeneity is an important feature that enables observers to solve these two tasks.

Strengths:

(1) The writing is very clear. The presentation of the study is informative.

(2) This study includes several behavioral and fMRI experiments. I appreciate the scientific rigor of the authors.

We are grateful to you for your balanced assessment and constructive comments.

Weaknesses:

(1) My main concern with this paper is the way visual homogeneity is computed. On page 10, lines 188-192, it says: "we then asked if there is any point in this multidimensional representation such that distances from this point to the target-present and target-absent response vectors can accurately predict the target-present and target-absent response times with a positive and negative correlation respectively (see Methods)". This is also true for the symmetry detection task. If I understand correctly, the reference point in this perceptual space was found by deliberating satisfying the negative and positive correlations in response times. And then on page 10, lines 200-205, it shows that the positive and negative correlations actually exist. This logic is confusing. The positive and negative correlations emerge only because this method is optimized to do so. It seems more reasonable to identify the reference point of this perceptual space independently, without using the reaction time data. Otherwise, the inference process sounds circular. A simple way is to just use the mean point of all objects in Exp 1, without any optimization towards reaction time data.

We disagree with you since the same logic applies to any curve-fitting procedure. When we fit data to a straight line, we are finding the slope and intercept that minimizes the error between the data and the straight line, but we would hardly consider the process circular when a good fit is achieved – in fact we take it as a confirmation that the data can be fit linearly. In the same vein, we would not have observed a good fit to the data, if there did not exist any good reference point relative to which the distances of the target-present and target-absent search arrays predicted these response times.

In Section S1, we have already reported that the visual homogeneity estimates for each object is strongly correlated with the average distance of each object to all other objects (r = 0.84, p<0.0005, Figure S1). Second, to confirm that the results we obtained are not due to overfitting, we have already reported a cross-validation analysis, where we removed all searches involving a particular image and predicted these response times using visual homogeneity. This too revealed a significant model correlation confirming that our results are not due to overfitting.

(2) On page 11, lines 214-221. It says: "these findings are non-trivial for several reasons". However, the first reason is confusing. It is unclear to me why "it suggests that there are highly specific computations that can be performed on perceptual space to solve oddball tasks". In fact, these two sentences provide no specific explanation for the results.

We have now revised the text to make it clearer (Results, p. 11).

(3) The second reason is interesting. Reaction times in target-present trials can be easily explained by target-distractor similarity. But why does reaction time vary substantially across target-absent stimuli? One possible explanation is that the objects that are distant from the feature distribution elicit shorter reaction times. Here, all objects constitute a statistical distribution in the feature (perceptual) space. There is certainly a mean of this distribution. Some objects look like outliers and these outliers elicit shorter reaction times in the target-absent trials because outlier detection is very salient.

One might argue that the above account is merely a rephrasing of the idea of visual homogeneity proposed in this study. If so, feature saliency is not a new account. In other words, the idea of visual homogeneity is another way of reiterating the old feature saliency theory.

Thank you for this interesting point. We don’t necessarily see a contradiction. However, we are proposing a quantitative decision variable that the brain could be using to make target present/absent judgements.

(4) One way to reject the feature saliency theory is to compare the reaction times of the objects that are very different from other objects (i.e., no surrounding objects in the perceptual space, e.g., the wheel in the lower right corner of Fig. 2B) with the objects that are surrounded by several similar objects (e.g., the horse in the upper part of Fig. 2B). Also, please choose the two objects with similar distance from the reference point. I predict that the latter will elicit longer reaction times because they can be easily confounded by surrounding similar objects (i.e., four-legged horses can be easily confounded by four-legged dogs). If the density of object distribution per se influences the visual homogeneity score, I would say that the "visual homogeneity" is essentially another way of describing the distributional density of the perceptual space.

We agree with you, and we have indeed found that visual homogeneity estimates from our model are highly correlated with the average distance of an object relative to all other objects. However, we performed several additional experiments to elucidate the nature of target-absent response times. We find that they are unaffected by whether these searches are performed in the midst of similar or dissimilar objects (Section S4, Experiment S6), and even when the same searches are performed among nearby sets of objects with completely uncorrelated average distances (Section S4, Experiment S7). We have now reworked the text to make this clearer.

(5) The searchlight analysis looks strange to me. One can easily perform a parametric modulation by setting visual homogeneity as the trial-by-trial parametric modulator and reaction times as a covariate. This parametric modulation produces a brain map with the correlation of every voxel in the brain. On page 17 lines 340-343, it is unclear to me what the "mean activation" is.

We have done something similar. For each region we took the mean activation at each voxel as the average activation 3x3x3 voxel neighborhood in the brain, and took its correlation with visual homogeneity. We have now reworked this to make it clearer (Results, p. 16).

Minor points

(1) In the intro, it says: "using simple neural rules..." actually it is very confusing what "neural rules" are here. Better to change it to "computational principles" or "neural network models"??

We have now replaced this with “using well-known principles governing multiple object representations”.

(2) In the intro, it says: "while machine vision algorithms are extremely successful in solving feature-based tasks like object categorization (Serre, 2019), they struggle to solve these generic tasks (Kim et al., 2018; Ricci et al. 2021). These are not generic tasks. They are just a specific type of visual task-judging relationship between multiple objects. Moreover, a large number of studies in machine vision have shown that DNNs are capable of solving these tasks and even more difficult tasks. Two survey papers are listed here.

Wu, Q., Teney, D., Wang, P., Shen, C., Dick, A., & Van Den Hengel, A. (2017). Visual question answering: A survey of methods and datasets. Computer Vision and Image Understanding, 163, 21-40.

Małkiński, M., & Mańdziuk, J. (2022). Deep Learning Methods for Abstract Visual Reasoning: A Survey on Raven's Progressive Matrices. arXiv preprint arXiv:2201.12382.

Thank you for sharing these references. In fact, a recent study has shown that specific deep networks can indeed solve the same-different task (Tartaglini et al, 2023). However our broader point remains that the same-different or other such visual tasks are non-trivial for machine vision algorithms.

Reviewer #1 (Recommendations For The Authors):

Nothing to add to the public review. If my concerns turn out to be invalid, I apologize and will happily accept correction. If they are valid, I hope they will point toward a new version of this paper that optimizes the insights to be gained from this impressive dataset.

Reviewer #2 (Recommendations For The Authors):

My suggestions are as follows:

(1) Analyze the fMRI data using the parametric modulation approach first at the single-subject level and then perform group analysis.

To clarify, we have obtained image-level activations from each subject, and used it for all our analyses.

(2) Think about a way to redefine visual homogeneity from a purely image-computable approach. In other words, visual homogeneity should be first defined as an image feature that is independent of any empirical response data. And then use the visual homogeneity scores to predict reaction times.

While we understand what you mean, any image-computable representation such as from a deep network may carry its own biases and may not be an accurate representation of the underlying object representation. By contrast, neural dissimilarities in the visual cortex are strongly predictive of visual search oddball response times. That is why we used visual search oddball response times as a proxy for the underlying neural representation, and then asked whether some decision variable can be derived from this representation to explain both target present and absent judgements in visual search.

eLife assessment

This study uses carefully designed experiments to generate a useful behavioural and neuroimaging dataset on visual cognition. The results provide solid evidence for the involvement of higher-order visual cortex in processing visual oddballs and asymmetry. However, the evidence provided for the very strong claims of homogeneity as a novel concept in vision science, separable from existing concepts such as target saliency, is inadequate.

Reviewer #1 (Public Review):

Summary:

The authors define a new metric for visual displays, derived from psychophysical response times, called visual homogeneity (VH). They attempt to show that VH is explanatory of response times across multiple visual tasks. They use fMRI to find visual cortex regions with VH-correlated activity. On this basis, they declare a new visual region in human brain, area VH, whose purpose is to represent VH for the purpose of visual search and symmetry tasks.

Strengths:

The authors present carefully designed experiments, combining multiple types of visual judgments and multiple types of visual stimuli with concurrent fMRI measurements. This is a rich dataset with many possibilities for analysis and interpretation.

Weaknesses:

The datasets presented here should provide a rich basis for analysis. However, in this version of the manuscript, I believe that there are major problems with the logic underlying the authors' new theory of visual homogeneity (VH), with the specific methods they used to calculate VH, and with their interpretation of psychophysical results using these methods. These problems with the coherency of VH as a theoretical construct and metric value make it hard to interpret the fMRI results based on searchlight analysis of neural activity correlated with VH. In addition, the large regions of VH correlations identified in Experiments 1 and 2 vs. Experiments 3 and 4 are barely overlapping. This undermines the claim that VH is a universal quantity, represented in a newly discovered area of visual cortex, that underlies a wide variety of visual tasks and functions.

Maybe I have missed something, or there is some flaw in my logic. But, absent that, I think the authors should radically reconsider their theory, analyses, and interpretations, in light of detailed comments below, in order to make the best use of their extensive and valuable datasets combining behavior and fMRI. I think doing so could lead to a much more coherent and convincing paper, albeit possibly supporting less novel conclusions.

THEORY AND ANALYSIS OF VH

(1) VH is an unnecessary, complex proxy for response time and target-distractor similarity.

VH is defined as a novel visual quality, calculable for both arrays of objects (as studied in Experiments 1-3) and individual objects (as studied in Experiment 4). It is derived from a center-to-distance calculation in a perceptual space. That space in turn is derived from multi-dimensional scaling of response times for target-distractor pairs in an oddball detection task (Experiments 1 and 2) or in a same different task (Experiments 3 and 4). Proximity of objects in the space is inversely proportional to response times for arrays in which they were paired. These response times are higher for more similar objects. Hence, proximity is proportional to similarity. This is visible in Fig. 2B as the close clustering of complex, confusable animal shapes.

VH, i.e. distance-to-center, for target-present arrays is calculated as shown in Fig. 1C, based on a point on the line connecting target and distractors. The authors justify this idea with previous findings that responses to multiple stimuli are an average of responses to the constituent individual stimuli. The distance of the connecting line to the center is inversely proportional to the distance between the two stimuli in the pair, as shown in Fig. 2D. As a result, VH is inversely proportional to distance between the stimuli and thus to stimulus similarity and response times. But this just makes VH a highly derived, unnecessarily complex proxy for target-distractor similarity and response time. The original response times on which the perceptual space is based are far more simple and direct measures of similarity for predicting response times.

(2) The use of VH derived from Experiment 1 to predict response times in Experiment 2 is circular and does not validate the VH theory.

The use of VH, a response time proxy, to predict response times in other, similar tasks, using the same stimuli, is circular. In effect, response times are being used to predict response times across two similar experiments using the same stimuli. Experiment 1 and the target present condition of Experiment 2 involve the same essential task of oddball detection. The results of Experiment 1 are converted into VH values as described above, and these are used to predict response times in experiment 2 (Fig. 2F). Since VH is a derived proxy for response values in Experiment 1, this prediction is circular, and the observed correlation shows only consistency between two oddball detection tasks in two experiments using the same stimuli.

(3) The negative correlation of target-absent response times with VH as it is defined for target-absent arrays, based on distance of a single stimulus from center, is uninterpretable without understanding the effects of center-fitting. Most likely, center-fitting and the different VH metric for target-absent trials produce an inverse correlation of VH with target-distractor similarity.

The construction of the VH perceptual space also involves fitting a "center" point such that distances to center predict response times as closely as possible. The effect of this fitting process on distance-to-center values for individual objects or clusters of objects is unknowable from what is presented here. These effects would depend on the residual errors after fitting response times with the connecting line distances. The center point location and its effects on distance-to-center of single objects and object clusters are not discussed or reported here.

Yet, this uninterpretable distance-to-center of single objects is chosen as the metric for VH of target-absent displays (VHabsent). This is justified by the idea that arrays of a single stimulus will produce an average response equal to one stimulus of the same kind. But it is not logically clear why response strength to a stimulus should be a metric for homogeneity of arrays constructed from that stimulus, or even what homogeneity could mean for a single stimulus from this set. And it is not clear how this VHabsent metric based on single stimuli can be equated to the connecting line VH metric for stimulus pairs, i.e. VHpresent, or how both could be plotted on a single continuum.

It is clear, however, what *should* be correlated with difficulty and response time in the target-absent trials, and that is the complexity of the stimuli and the numerosity of similar distractors in the overall stimulus set. Complexity of the target, similarity with potential distractors, and number of such similar distractors all make ruling out distractor presence more difficult. The correlation seen in Fig. 2G must reflect these kinds of effects, with higher response times for complex animal shapes with lots of similar distractors and lower response times for simpler round shapes with fewer similar distractors.

The example points in Fig. 2G seem to bear this out, with higher response times for the deer stimulus (complex, many close distractors in the Fig. 2B perceptual space) and lower response times for the coffee cup (simple, few close distractors in the perceptual space). While the meaning of the VH scale in Fig. 2G, and its relationship to the scale in Fig. 2F, are unknown, it seems like the Fig. 2G scale has an inverse relationship to stimulus complexity, in contrast to the expected positive relationship for Fig. 2F. This is presumably what creates the observed negative correlation in Fig. 2G.

Taken together, points 1-3 suggest that VHpresent and VHabsent are complex, unnecessary, and disconnected metrics for understanding target detection response times. The standard, simple explanation should stand. Task difficulty and response time in target detection tasks, in both present and absent trials, are positively correlated with target-distractor similarity.

I think my interpretations apply to Experiments 3 and 4 as well, although I find the analysis in Fig. 4 especially hard to understand. The VH space in this case is based on Experiment 3 oddball detection in a stimulus set that included both symmetric and asymmetric objects. But the response times for a very different task in Experiment 4, a symmetric/asymmetric judgment, are plotted against the axes derived from Experiment 3 (Fig. 4F and 4G). It is not clear to me why a measure based on oddball detection that requires no use of symmetry information should be predictive of within-stimulus symmetry detection response times. If it is, that requires a theoretical explanation not provided here.

(4) Contrary to the VH theory, same/different tasks are unlikely to depend on a decision boundary in the middle of a similarity or homogeneity continuum.

The authors interpret the inverse relationship of response times with VHpresent and VHabsent, described above, as evidence for their theory. They hypothesize, in Fig. 1G, that VHpresent and VHabsent occupy a single scale, with maximum VHpresent falling at the same point as minimum VHabsent. This is not borne out by their analysis, since the VHpresent and VHabsent value scales are mainly overlapping, not only in Experiments 1 and 2 but also in Experiments 3 and 4. The authors dismiss this problem by saying that their analyses are a first pass that will require future refinement. Instead, the failure to conform to this basic part of the theory should be a red flag calling for revision of the theory.

The reason for this single scale is that the authors think of target detection as a boundary decision task, along a single scale, with a decision boundary somewhere in the middle, separating present and absent. This model makes sense for decision dimensions or spaces where there are two categories (right/left motion; cats vs. dogs), separated by an inherent boundary (equal left/right motion; training-defined cat/dog boundary). In these cases, there is less information near the boundary, leading to reduced speed/accuracy and producing a pattern like that shown in Fig. 1G.

This logic does not hold for target detection tasks. There is no inherent middle point boundary between target present and target absent. Instead, in both types of trial, maximum information is present when target and distractors are most dissimilar, and minimum information is present when target and distractors are most similar. The point of greatest similarity occurs at then limit of any metric for similarity. Correspondingly, there is no middle point dip in information that would produce greater difficulty and higher response times. Instead, task difficulty and response times increase monotonically with similarity between targets and distractors, for both target present and target absent decisions. Thus, in Figs. 2F and 2G, response times appear to be highest for animals, which share the largest numbers of closely similar distractors.

DEFINITION OF AREA VH USING fMRI

(1) The area VH boundaries from different experiments are nearly completely non-overlapping.

In line with their theory that VH is a single continuum with a decision boundary somewhere in the middle, the authors use fMRI searchlight to find an area whose responses positively correlate with homogeneity, as calculated across all of their target present and target absent arrays. They report VH-correlated activity in regions anterior to LO. However, the VH defined by symmetry Experiments 3 and 4 (VHsymmetry) is substantially anterior to LO, while the VH defined by target detection Experiments 1 and 2 (VHdetection) is almost immediately adjacent to LO. Fig. S13 shows that VHsymmetry and VHdetection are nearly non-overlapping. This is a fundamental problem with the claim of discovering a new area that represents a new quantity that explains response times across multiple visual tasks. In addition, it is hard to understand why VHsymmetry does not show up in a straightforward subtraction between symmetric and asymmetric objects, which should show a clear difference in homogeneity.

(2) It is hard to understand how neural responses can be correlated with both VHpresent and VHabsent.

The main paper results for VHdetection are based on both target-present and target-absent trials, considered together. It is hard to interpret the observed correlations, since the VHpresent and VHabsent metrics are calculated in such different ways and have opposite correlations with target similarity, task difficulty, and response times (see above). It may be that one or the other dominates the observed correlations. It would be clarifying to analyze correlations for target-present and target-absent trials separately, to see if they are both positive and correlated with each other.

(3) Definition of the boundaries and purpose of a new visual area in the brain requires circumspection, abundant and convergent evidence, and careful controls.

Even if the VH metric, as defined and calculated by the authors here, is a meaningful quantity, it is a bold claim that a large cortical area just anterior to LO is devoted to calculating this metric as its major task. Vision involves much more than target detection and symmetry detection. Cortex anterior to LO is bound to perform a much wider range of visual functionalities. If the reported correlations can be clarified and supported, it would be more circumspect to treat them as one byproduct of unknown visual processing in cortex anterior to LO, rather than treating them as the defining purpose for a large area of visual cortex.

Reviewer #3 (Public Review):

Summary:

This study proposes visual homogeneity as a novel visual property that enables observers perform to several seemingly disparate visual tasks, such as finding an odd item, deciding if two items are same, or judging if an object is symmetric. In Exp 1, the reaction times on several objects were measured in human subjects. In Exp 2, visual homogeneity of each object was calculated based on the reaction time data. The visual homogeneity scores predicted reaction times. This value was also correlated with the BOLD signals in a specific region anterior to LO. Similar methods were used to analyze reaction time and fMRI data in a symmetry detection task. It is concluded that visual homogeneity is an important feature that enables observers to solve these two tasks.

Strengths:

(1) The writing is very clear. The presentation of the study is informative.<br /> (2) This study includes several behavioral and fMRI experiments. I appreciate the scientific rigor of the authors.

Weaknesses:

(1) My main concern with this paper is the way visual homogeneity is computed. On page 10, lines 188-192, it says: "we then asked if there is any point in this multidimensional representation such that distances from this point to the target-present and target-absent response vectors can accurately predict the target-present and target-absent response times with a positive and negative correlation respectively (see Methods)". This is also true for the symmetry detection task. If I understand correctly, the reference point in this perceptual space was found by deliberating satisfying the negative and positive correlations in response times. And then on page 10, lines 200-205, it shows that the positive and negative correlations actually exist. This logic is confusing. The positive and negative correlations emerge only because this method is optimized to do so. It seems more reasonable to identify the reference point of this perceptual space independently, without using the reaction time data. Otherwise, the inference process sounds circular. A simple way is to just use the mean point of all objects in Exp 1, without any optimization towards reaction time data.

(2) Visual homogeneity (at least given the current from) is an unnecessary term. It is similar to distractor heterogeneity/distractor variability/distractor statics in literature. However, the authors attempt to claim it as a novel concept. The title is "visual homogeneity computations in the brain enable solving generic visual tasks". The last sentence of the abstract is "a NOVEL IMAGE PROPERTY, visual homogeneity, is encoded in a localized brain region, to solve generic visual tasks". In the significance, it is mentioned that "we show that these tasks can be solved using a simple property WE DEFINE as visual homogeneity". If the authors agree that visual homogeneity is not new, I suggest a complete rewrite of the title, abstract, significance, and introduction.

(3) Also, "solving generic tasks" is another overstatement. The oddball search tasks, same-different tasks, and symmetric tasks are only a small subset of many visual tasks. Can this "quantitative model" solve motion direction judgment tasks, visual working memory tasks? Perhaps so, but at least this manuscript provides no such evidence. On line 291, it says "we have proposed that visual homogeneity can be used to solve any task that requires discriminating between homogeneous and heterogeneous displays". I think this is a good statement. A title that says "XXXX enable solving discrimination tasks with multi-component displays" is more acceptable. The phrase "generic tasks" is certainly an exaggeration.

(4) If I understand it correctly, one of the key findings of this paper is "the response times for target-present searches were positively correlated with visual homogeneity. By contrast, the response times for target-absent searches were negatively correlated with visual homogeneity" (lines 204-207). I think the authors have already acknowledged that the positive correlation is not surprising at all because it reflects the classic target-distractor similarity effect. But the authors claim that the negative correlations in target-absent searches is the true novel finding.

(5) I would like to make it clear that this negative correlation is not new either. The seminal paper by Duncan and Humphreys (1989) has clearly stated that "difficulty increases with increased similarity of targets to nontargets and decreased similarity between nontargets" (the sentence in their abstract). Here, "similarity between nontargets" is the same as the visual homogeneity defined here. Similar effects have been shown in Duncan (1989) and Nagy, Neriani, and Young (2005). See also the inconsistent results in Nagy& Thomas, 2003, Vicent, Baddeley, Troscianko&Gilchrist, 2009.<br /> More recently, Wei Ji Ma has systematically investigated the effects of heterogeneous distractors in visual search. I think the introduction part of Wei Ji Ma's paper (2020) provides a nice summary of this line of research.

I am surprised that these references are not mentioned at all in this manuscript (except Duncan and Humphreys, 1989).

(6) If the key contribution is the quantitative model, the study should be organized in a different way. Although the findings of positive and negative correlations are not novel, it is still good to propose new models to explain classic phenomena. I would like to mention the three studies by Wei Ji Ma (see below). In these studies, Bayesian observer models were established to account for trial-by-trial behavioral responses. These computational models can also account for the set-size effect, behavior in both localization and detection tasks. I see much more scientific rigor in their studies. Going back to the quantitative model in this paper, I am wondering whether the model can provide any qualitative prediction beyond the positive and negative correlations? Can the model make qualitative predictions that differ from those of Wei Ji's model? If not, can the authors show that the model can quantitatively better account for the data than existing Bayesian models? We should evaluate a model either qualitatively or quantitatively.

(7) In my opinion, one of the advantages of this study is the fMRI dataset, which is valuable because previous studies did not collect fMRI data. The key contribution may be the novel brain region associated with display heterogeneity. If this is the case, I would suggest using a more parametric way to measure this region. For example, one can use Gabor stimuli and systematically manipulate the variations of multiple Gabor stimuli, the same logic also applies to motion direction. If this study uses static Gabor, random dot motion, object images that span from low-level to high-level visual stimuli, and consistently shows that the stimulus heterogeneity is encoded in one brain region, I would say this finding is valuable. But this sounds like another experiment. In other words, it is insufficient to claim a new brain region given the current form of the manuscript.

REFERENCES<br /> - Duncan, J., & Humphreys, G. W. (1989). Visual search and stimulus similarity. Psychological Review, 96(3), 433-458. doi: 10.1037/0033-295x.96.3.433<br /> - Duncan, J. (1989). Boundary conditions on parallel processing in human vision. Perception, 18(4), 457-469. doi: 10.1068/p180457<br /> - Nagy, A. L., Neriani, K. E., & Young, T. L. (2005). Effects of target and distractor heterogeneity on search for a color target. Vision Research, 45(14), 1885-1899. doi: 10.1016/j.visres.2005.01.007<br /> - Nagy, A. L., & Thomas, G. (2003). Distractor heterogeneity, attention, and color in visual search. Vision Research, 43(14), 1541-1552. doi: 10.1016/s0042-6989(03)00234-7<br /> - Vincent, B., Baddeley, R., Troscianko, T., & Gilchrist, I. (2009). Optimal feature integration in visual search. Journal of Vision, 9(5), 15-15. doi: 10.1167/9.5.15<br /> - Singh, A., Mihali, A., Chou, W. C., & Ma, W. J. (2023). A Computational Approach to Search in Visual Working Memory.<br /> - Mihali, A., & Ma, W. J. (2020). The psychophysics of visual search with heterogeneous distractors. BioRxiv, 2020-08.<br /> - Calder-Travis, J., & Ma, W. J. (2020). Explaining the effects of distractor statistics in visual search. Journal of Vision, 20(13), 11-11.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

The authors provide convincing experimental evidence of extended motivational signals encoded in the mouse anterior cingulate cortex (ACC) that are implemented by the orbitofrontal cortex (OFC)-to-ACC signaling during learning. The results are valuable to the field of motivation and cognition. The experimental methods used were state-of-the-art. The manuscript would further benefit from theory-driven analyses to inform a mechanistic understanding, particularly for the single-cell calcium imaging results. These results will be of interest to those interested in cortical function, learning, and/or motivation.

We thank the reviewers for their thoughtful reading of our paper and providing constructive feedback. We have made the relevant changes to the manuscript to improve the writing and figures. We provide responses below to each of the reviewer’s comments.

Reviewer #1 (Public Review):

(1) An important conclusion (Figure 4) is that when mice are trained to run through no reward (N) cues in order to reach reward (R) cues, the OFC neurons projecting to ACC each respond to different specific events in a manner that ensures that collectively they tile the extended behavioural sequence. What I was less sure of was whether the ACC neurons do the same or not. Figure 3 suggests that on average ACC neurons maintain activity across N cues in order to get to R cues but I was not sure whether this was because all individual neurons did this or whether some had activity patterns like the OFC neurons projecting to ACC.

We agree that it remains uncertain what individual ACC neurons do during the extended behavioral sequence. We now include a few sentences in the discussion about what we hypothesize, as we did not perform the cellular resolution imaging to determine this:

“While we did not perform single-cell imaging of ACC in our task, we hypothesize that individual ACC neurons could encode the distribution of actions/opportunities47 (i.e. stop, run, lick, suppress lick) taken during R or N cues. ACC neurons could compute the relative value of the action taken such that more ACC neurons become recruited once mice learn to run out of N cues. The sustained increase in bulk ACC activity across N cue trials (Figure 2) could come from a stable sequence of individual neurons that encode the timescale of the actions taken. In this way, OFC projections would encode current motivation across N cues before learning, which then triggers ACC to compute the valuebased actions. Motivational signals in OFC would thus represent state since past rewards/goals, while in ACC these signals represent actions taken to pursue rewards/goals in the future.”

(2) Figure 1 versus Figure 2: There does not seem to be a particular motivation for whether chemogenetic inactivation or optogenetic inhibition were used in different experiments. I think that this is not problematic but, if I am wrong and there were specific reasons for performing each experiment in a certain way, then further clarification as to why these decisions were made would be useful. If there is no particular reason, then simply explaining that this is the case might stop readers from seeking explanations.

Thank you for this comment and we agree that clarification on this is important. We performed chemogenetic inhibition of ACC in Figure 1 to take a broad survey of behavioral effects throughout a 40-min long behavioral session, and performed optogenetic inhibition in Figure 2 because we wanted to restrict our inhibition to the few seconds of cue presentation during a behavioral session and across days. Furthermore, we wanted to combat any potential off-target effects that would come from repeated administration of CNO over the several days of training (Manvich et al 2018). We have included a couple sentences on page 4 to clarify this:

“We proceeded to test whether these motivation related signals in ACC are required for learning. To restrict our inhibition to cue presentation portions of our task, and combat any potential off-target effects of CNO31 from repeated administration across several days of training, we used optogenetic inhibition.”

(3) P5, paragraph 2. The authors argue that OFC and anteriomedial (AM) thalamic inputs into ACC are especially important for mediating motivation through N cues in order to reach R cues. Is this based on a statistical comparison between the activity in OFC or AM inputs as opposed to the other inputs?

We determined that OFC and AM thalamic inputs to ACC are particularly important by comparing the pre-cue activity in a reward-no reward-reward trial sequence (RNR; Figure 3B). Specifically, we performed paired t-tests comparing pre-cue activity between N and R cues, and found a statistically significant increase for R cues but only for the OFC and AM inputs, not for the BLA or LC inputs.

(4) P3, paragraph 2. Some papers by Khalighinejad and colleagues (eg Neuron 2020, Current Biology, 2022) might be helpful here in as much as they assess ACC roles in determining action frequency, initiation, and speed and mediating the relationship between reward availability and action frequency and speed.

We thank the reviewer for bringing these relevant papers to our attention. We have included these papers in our citations in this paragraph.

(5) Paragraph 1 "This learning is of a more deliberate, informed nature than habitual learning, as they are sensitive to the current value of outcomes and can lead to a novel sequence of actions for a desired outcome1-3." Should "they" be "it"?

This is correct, we have edited this in the manuscript.

Reviewer #2 (Public Review):

Impact:

The findings will be valuable for further research on the impact of motivational states on behaviour and cognition. The authors provided a promising concept of how persistent motivational states could be maintained, as well as established a novel, reproducible task assay. While experimental methods used are currently state-of-the-art, theoretical analysis seems to be incomplete/not extensive. We thank the reviewer for these comments. In our paper, we performed single-cell calcium imaging of OFC projection neurons to ACC to build a mechanistic understanding for the bulk ramp-like response we identified in these neurons with photometry. We identified ensembles of neurons that tile sequences of trials that match the bulk response, in particular a subset of neurons that are active at the time a reward (R) cue is reached after 2 no-reward (N) cues. We included a paragraph in the discussion to address future theory-driven analyses to address how computation is achieved by OFC projection neurons:

“We linked the ramp-like increase in neural activity in OFC to motivation, but several questions still remain about how motivation is computed and why it would be represented as a ramp. Motivation could be computed as a combination of several variables such as time since last reward, value of reward, and effort to reach future rewards. Future theorydriven analyses could determine how motivation is computed, and whether individual variables of time, value, and effort, are encoded as clusters of similar tuned neurons, or mixed and collectively represented at the population level. In either case, it is likely that a combined map of task space and value-information carried by OFC are being used to inform downstream regions, such as ACC, for adjusting behavior. ”

Reviewer #2 (Recommendations for the Authors):

Overall, the layout of the figures seems a little bit chaotic and makes it hard to understand the boundaries between panels.

We agree that the figure layout could be improved upon to aid the reader in moving from panel to panel. We have edited two of the main figures with layouts that are most irregular (Figures 2 and 4) to help with this.

Figures/text should include the promoters used for protein expression so that readers understand which cell types would be affected.

We have made sure to edit the figures to include the promoter of the viruses we used, and edited the text to include both the AAV serotype and promoter.

Discuss why it is necessary for multiple prefrontal areas to be involved in maintaining motivational signals.

We thank the reviewer for this comment. We believe that prefrontal areas would be recruited as tasks to study motivational states become more complex and require animals to keep track of task structure and perform value-guided actions. We have included a couple sentences in the final paragraph of the discussion about this:

“Our work showed the recruitment of multiple frontal cortical areas in this process, which is to be expected as animals are required to build, maintain, and use representations of task structure and value to drive learned, motivated behaviors47. Future work can build upon the task we developed here to determine how the frontal cortex maintains motivational states across many more cue-outcome associations, and how these associations may dynamically change across time48”.

Additionally, we included a short discussion on how in motivational signals differ between OFC and ACC in our work. We suggest OFC encodes current motivation before and after learning, which then leads ACC to represent learned actions taken and thus have a longer timescale motivational response (see response to Reviewer 1).

Minor: Page 4, Line 1: "increase" instead of "increases".

This is correct, we have edited this in the manuscript.

eLife assessment

This important manuscript provides compelling experimental evidence of extended motivational signals encoded in the mouse anterior cingulate cortex (ACC) that are implemented by orbitofrontal cortex (OFC)-to-ACC signaling during learning. The experimental methods used were state-of-the-art. These results will be of interest to those interested in cortical function, learning, and/or motivation.

Reviewer #1 (Public Review):

This is an interesting report examining activity patterns in mouse ACC and in the OFC neurons projecting to ACC. In addition, the effects of inactivation are examined. In aggregate, the results provide new and interesting information about these two brain areas and they translate motivation into action - a function that it seems intuitively plausible that ACC might perform but, despite this intuition, there have been comparatively few direct tests of the idea and little is known of the specific mechanisms. The study is performed carefully and is written up clearly.

The combination of recording and inactivation/inhibition experiments and the combination of investigation of ACC neurons and of OFC regions projecting to ACC are very impressive.

Reviewer #2 (Public Review):

Summary:

Regalado et al. studied how an extended motivational state, necessary for maintaining behavioural drive despite unrewarding experiences, could be encoded in the ACC and its potential causal implications for learning discriminatory behaviour and avoiding unrewarding stimuli. They designed a self-initiated learning task and identified bulk neural responses tuned specifically to reward delivery as well as trial initiation. Interestingly, in both cases, neural activity precedes behavioural onset, indicating the encoding of a motivational signal. To investigate the neural encoding of motivational signals during unrewarded, distracting stimuli presentation, they created a discrimination task by introducing 'no reward' cues, during which animals need to learn not to reduce running speed and not engage in licking. Interestingly, with mice learning to increase running speed and reduce licking rates after 'no reward' cues, the preceding ACC activity also gradually increased. Importantly, only the increase in running speed after 'no reward' cues was impaired upon optogenetic inhibition of ACC activity during early training, linking the extended motivational signal in ACC and learning to maximise rewards by actively avoiding distracting and unrewarded stimuli. Such motivational signals could also be observed in OFC-ACC projecting neurons. Especially the continuous ramping of activity upon repeated 'non-reward' cues, which could be exclusively observed in the 'fast learner' subgroup, provides an interesting concept of how an extended motivational signal necessary for learning avoidance of unrewarded stimuli could be implemented in ACC. The shift in the temporal activity of initially reward-responsive neurons towards the preceding 'no reward' cue, provides a potential mechanism linking extended motivation to reward maximisation. This mechanism seems to be particularly important in periods of persistent 'non-reward' cues, as demonstrated in the impairment of running speed increase after two consecutive 'non-reward' cues.

Appraisal:

The authors provide convincing experimental evidence to support their claims of an extended motivational signal encoded in the ACC that is implemented by OFC-ACC signalling and critically involved in learning avoidance of unrewarded stimuli. The newly designed task seems appropriate to identify correlates of relevant cognitive and behavioural variables (e.g. sustained motivation). The combination of recording Ca2+ transients (bulk as well as longitudinal single neuron recordings) to identify potential neural responses and subsequent evaluation of their causal role in establishing and maintaining this persistent motivational state using opto- and pharmacogenetic manipulations is generally accepted.

Impact:

The findings will be valuable for further research on the impact of motivational states on behaviour and cognition. The authors provided a promising concept of how persistent motivational states could be maintained, as well as established a novel, reproducible task assay. While experimental methods used are currently state-of-the-art, theoretical analysis seems to be incomplete/not extensive.

eLife assessment

This study provides important insights into the role of neurexins as regulators of synaptic strength and timing at the glycinergic synapse between neurons of the medial nucleus of the trapezoid body and the lateral superior olive, key components of the auditory brainstem circuit involved in computing sound source location from differences in the intensity of sounds arriving at the two ears. Through an elegant combination of genetic manipulation, fluorescence in-situ hybridization, ex vivo slice electrophysiology, pharmacology and optogenetics, the authors provide compelling and rigorous evidence to support their claims. While further work is needed to reveal the mechanistic basis by which neurexins influence glycinergic neurotransmission, this work will be of interest to both auditory and synaptic neuroscientists.

Reviewer #1 (Public Review):

Jiang et al. demonstrated that ablating Neurexins results in alterations to glycinergic transmission and its calcium sensitivity, utilizing a robust experimental system. Specifically, the authors employed rAAV-Cre-EGFP injection around the MNTB in Nrxn1/2/3 triple conditional mice at P0, measuring Glycine receptor-dependent IPSCs from postsynaptic LSO neurons at P13-14. Notably, the authors presented a clear reduction of 60% and 30% in the amplitudes of opto- and electric stimulation-evoked IPSCs, respectively. Additionally, they observed changes in kinetics, alterations in PPR, and sensitivity to lower calcium and the calcium chelator, EGTA, indicating solid evidence for changes in presynaptic properties of glycinergic transmission.

Furthermore, the authors uncovered an unexpected increase in sIPSC frequency without altering amplitude. Although the precise mechanism remains unknown, the authors discussed this complex phenotype by considering various possibilities, including the potential scenario where the augmentation in synapses may result from Nrxn deletion rather than being a causal effect.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Jiang et al., explore the role of neurexins at glycinergic MNTB-LSO synapses. The authors utilize elegant and compelling ex vivo slice electrophysiology to assess how the genetic conditional deletion of Nrxns1-3 impacts inhibitory glycinergic synaptic transmission and found that TKO of neurexins reduced electrically and optically evoked IPSC amplitudes, slowed optically evoked IPSC kinetics and reduced presynaptic release probability. The authors use classic approaches including reduced [Ca2+] in ACSF and EGTA chelation to propose that changes in these evoked properties are likely driven by the loss of calcium channel coupling. Intriguingly, while evoked transmission was impaired, the authors reported that spontaneous IPSC frequency was increased, due to an increase in the number of synapses in LSO. Overall, this manuscript provides important insight into the role of neurexins at the glycinergic MNTB-LSO synapse and further emphasizes the need for continued study of both the non-redundant and redundant roles of neurexins.

The authors have addressed all of my previous concerns.

Reviewer #3 (Public Review):

Summary:

The authors investigate the hypothesis that neurexins serve a crucial role as regulators of the synaptic strength and timing at the glycinergic synapse between neurons of the medial nucleus of the trapezoid body (MNTB) and the lateral superior olivary complex (LSO). It is worth mentioning that LSO neurons are an integration station of the auditory brainstem circuit displaying high reliability and temporal precision. These features are necessary for computing interaural cues to derive sound source location from comparing the intensities of sounds arriving at the two ears. In this context, the authors' findings build up according to the hypothesis first by displaying that neurexins were expressed in the MNTB at varying levels. They followed this up with deletion of all neurexins in the MNTB through the employment of a triple knock-out (TKO). Using electrophysiological recordings in acute brainstem slices of these TKO mice, they gathered solid evidence for the role of neurexins in synaptic transmission at this glycinergic synapse primarily by ensuring tight coupling of Ca2+ channels and vesicular release sites. Additionally, the authors uncovered a connection between the deletion of neurexins and a higher number of glycinergic synapses of TKO mice, for which they provided evidence in the form of immunostainings and related it to electrophysiological data on spontaneous release. Consequently, this investigation expands our knowledge on the molecular regulation of synaptic transmission at glycinergic synapses, as well as on the auditory processing at the level of the brainstem.

Strengths:

The authors demonstrate substantial results in support of the hypothesis of a critical role of neurexins for regulating glycinergic transmission in the LSO using various techniques. They provide evidence for the expression of neurexins in the MNTB and consecutively successfully generate and characterize the neurexin TKO. For their study on LSO IPSCs the authors transduced MNTB neurons by co-injection of virus carrying Cre and ChR2 and subsequently optogenetically evoke release of glycine. As a result, they observed a significant reduction in amplitude and significantly slower rise and decay times of the IPSCs of the TKO in comparison with control mice in which MNTB neurons were only transduced with ChR2. Furthermore, they observed an increased paired pulse ratio (PPR) of LSO IPSCs in the TKO mice, indicating lower release probability. Elaborating on the hypothesis that neurexins are essential for the coupling of synaptic vesicles to Ca2+ channels, the authors show lowered Ca2+ sensitivity in the TKO mice. Additionally, they reveal convincing evidence for the connection between the increased frequency of spontaneous IPSC and the higher number of glycinergic synapses of the LSO in the TKO mice, revealed by immunolabeling against the glycinergic presynaptic markers GlyT2 or VGAT.

Weaknesses:

A concern is on novelty as this work on the effects of pan-neurexin deletion in a glycinergic synapse is quite consistent with the authors prior work on glutamatergic synapses (Luo et al., 2020).

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This study provides important insights into the role of neurexins as regulators of synaptic strength and timing at the glycinergic synapse between neurons of the medial nucleus of the trapezoid body and the lateral superior olive, key components of the auditory brainstem circuit involved in computing sound source location from differences in the intensity of sounds arriving at the two ears. Through an elegant combination of genetic manipulation, fluorescence in-situ hybridization, ex vivo slice electrophysiology, pharmacology, and optogenetics, the authors provide convincing evidence to support their claims. While further work is needed to reveal the mechanistic basis by which neurexins influence glycinergic neurotransmission, this work will be of interest to both auditory and synaptic neuroscientists.

We appreciate the recognition of the significance of our study in shedding light on the role of neurexins in regulating synaptic strength and timing at the glycinergic synapse. Indeed, further investigations are warranted to delve deeper into the specific role of each different variant of neurexins in the future. We hope that our work will spark more interest and collaboration in unraveling the complexities of molecular codes of synaptic function.

Public Reviews:

Reviewer #1 (Public Review):

Jiang et al. demonstrated that ablating Neurexins results in alterations to glycinergic transmission and its calcium sensitivity, utilizing a robust experimental system. Specifically, the authors employed rAAV-Cre-EGFP injection around the MNTB in Nrxn1/2/3 triple conditional mice at P0, measuring Glycine receptor-dependent IPSCs from postsynaptic LSO neurons at P13-14. Notably, the authors presented a clear reduction of 60% and 30% in the amplitudes of opto- and electric stimulation-evoked IPSCs, respectively. Additionally, they observed changes in kinetics, alterations in PPR, and sensitivity to lower calcium and the calcium chelator, EGTA, indicating solid evidence for changes in presynaptic properties of glycinergic transmission.

Furthermore, the authors uncovered an unexpected increase in sIPSC frequency without altering amplitude. Despite the reduction in evoked IPSC, immunostaining revealed an increase in GlyT2 and VGAT in TKO mice, supporting the notion of an increase in synapse number. However, the reviewer expresses caution regarding the authors' conclusion that "glycinergic neurotransmission likely by promoting the synapse formation/maintenance, which is distinct from the phenotypes observed in glutamatergic and GABAergic neurons (Chen et al., 2017; Luo et al., 2021)", as outlined in lines 173-175. The reviewer suggests that this statement may be overstated, pointing out the authors' own discussion in lines 254-265, which acknowledges multiple possibilities, including the potential that the increase in synapses is a consequence rather than a causal effect of Nrxn deletion.

We appreciate the reviewer’s thoughtful evaluation of our study. We agree that our conclusion regarding the promotion of synapse formation/maintenance may have been overstated and recognize the need for a more nuanced interpretation of our findings. Accordingly, we have revised our interpretation by discussing carefully the various possibilities that may cause the observed increase in synapse number in line 256-266.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Jiang et al., explore the role of neurexins at glycinergic MNTB-LSO synapses. The authors utilize elegant and compelling ex vivo slice electrophysiology to assess how the genetic conditional deletion of Nrxns1-3 impacts inhibitory glycinergic synaptic transmission and found that TKO of neurexins reduced electrically and optically evoked IPSC amplitudes, slowed optically evoked IPSC kinetics and reduced presynaptic release probability. The authors use classic approaches including reduced [Ca2+] in ACSF and EGTA chelation to propose that changes in these evoked properties are likely driven by the loss of calcium channel coupling. Intriguingly, while evoked transmission was impaired, the authors reported that spontaneous IPSC frequency was increased, potentially due to an increased number of synapses in LSO. Overall, this manuscript provides important insight into the role of neurexins at the glycinergic MNTP-LSO synapse and further emphasizes the need for continued study of both the non-redundant and redundant roles of neurexins.

We thank the reviewer for the strong comments and support of our work.

Strengths:

This well-written manuscript seamlessly incorporates mouse genetics and elegant ex vivo electrophysiology to identify a role for neurexins in glycinergic transmission at MNTB-LSO synapses. Triple KO of all neurexins reduced the amplitude and timing of evoked glycinergic synaptic transmission. Further, spontaneous IPSC frequency was increased. The evoked synaptic phenotype is likely a result of reduced presynaptic calcium coupling while the spontaneous synaptic phenotype is likely due to increased synapse numbers. While neuroligin-4 has been identified at glycinergic synapses, this study, to the best of my knowledge, is the first to study Nrxn function at these synapses.<br />

We again appreciate the positive feedback on the strengths of our study. We agree that the observed reduction in evoked synaptic transmission and the increase in spontaneous IPSC frequency provide intriguing insights into the function of neurexins in regulating glycinergic synaptic activity.

Weaknesses:

The data are compelling and report an intriguing functional phenotype. The role of Neurexins redundantly controls calcium channel coupling has been previously reported. Mechanistic insight would significantly strengthen this study.

We wholeheartedly agree with the reviewer that understanding how neurexins control calcium channel coupling at the presynaptic active zone is crucial for elucidating their role in synaptic transmission. While our current study has provided compelling evidence for the functional phenotypes of pan-neurexin deletion, we recognize the importance of investigating the underlying molecular mechanisms in future research. Exploring these mechanisms would undoubtedly enhance our understanding of neurexin function at various synapses and contribute to advancing the field.

The claim that triple KO of Nrxns from MNTB increases the number of synapses in LSO is not strongly supported.

We agree. Echoing the suggestion made by reviewer 1 (as mentioned above), we acknowledge that the claim regarding the increase in synapse numbers in the LSO following the triple knockout of neurexins from the MNTB was overstated. Consequently, we have revised our conclusions more carefully to reflect this adjustment.

Despite the stated caveats of measuring electrically evoked currents and the more robust synaptic phenotypes observed using optically evoked transmission, the authors rely heavily on electrical stimulation for most measurements.

We acknowledge that optogenetic stimulation offers crucial advantages, and we have provided a balanced discussion of the caveats associated with both methods in our manuscript. Additionally, we have conducted new optogenetic experiments specifically for measuring the paired-pulse ratio in control and Nrxn123 TKO mice. These results have been included as a new supplementary figure (Figure S2).

For experiments involving EGTA and low Ca2+ manipulations, we opted for electrical stimulation due to concerns regarding potential side effects of optogenetics, including the phototoxicity and photobleaching during prolonged light exposure.

The differential expression of individual neurexins might indicate that specific neurexins may dominantly regulate synaptic transmission, however, this possibility is not discussed in detail.

We thank the reviewer for bringing up this important point. The differential expression of individual neurexins indeed suggests that specific neurexins may play dominant roles in regulating synaptic transmission. While our study primarily focused on the collective impact of ablating all neurexins, we acknowledge the significance of exploring the specific contributions of individual neurexin isoforms in the future. Understanding the distinct roles of each neurexin isoform could provide valuable insights into the precise mechanisms underlying synaptic function and plasticity. We have added discussion in our revised manuscript Line223-230.

Reviewer #3 (Public Review):

Summary:

The authors investigate the hypothesis that neurexins serve a crucial role as regulators of the synaptic strength and timing at the glycinergic synapse between neurons of the medial nucleus of the trapezoid body (MNTB) and the lateral superior olivary complex (LSO). It is worth mentioning that LSO neurons are an integration station of the auditory brainstem circuit displaying high reliability and temporal precision. These features are necessary for computing interaural cues to derive sound source location from comparing the intensities of sounds arriving at the two ears. In this context, the authors' findings build up according to the hypothesis first by displaying that neurexins were expressed in the MNTB at varying levels. They followed this up with the deletion of all neurexins in the MNTB through the employment of a triple knock-out (TKO). Using electrophysiological recordings in acute brainstem slices of these TKO mice, they gathered solid evidence for the role of neurexins in synaptic transmission at this glycinergic synapse primarily by ensuring tight coupling of Ca2+ channels and vesicular release sites. Additionally, the authors uncovered a connection between the deletion of neurexins and a higher number of glycinergic synapses in TKO mice, for which they provided evidence in the form of immunostainings and related it to electrophysiological data on spontaneous release. Consequently, this investigation expands our knowledge on the molecular regulation of synaptic transmission at glycinergic synapses, as well as on the auditory processing at the level of the brainstem.

Strengths:

The authors demonstrate substantial results in support of the hypothesis of a critical role of neurexins for regulating glycinergic transmission in the LSO using various techniques. They provide evidence for the expression of neurexins in the MNTB and consecutively successfully generate and characterize the neurexin TKO. For their study on LSO IPSCs the authors transduced MNTB neurons by co-injection of virus-carrying Cre and ChR2 and subsequently optogenetically evoke release of glycine. As a result, they observed a significant reduction in amplitude and significantly slower rise and decay times of the IPSCs of the TKO in comparison with control mice in which MNTB neurons were only transduced with ChR2. Furthermore, they observed an increased paired pulse ratio (PPR) of LSO IPSCs in the TKO mice, indicating lower release probability. Elaborating on the hypothesis that neurexins are essential for the coupling of synaptic vesicles to Ca2+ channels, the authors show lowered Ca2+ sensitivity in the TKO mice. Additionally, they reveal convincing evidence for the connection between the increased frequency of spontaneous IPSC and the higher number of glycinergic synapses of the LSO in the TKO mice, revealed by immunolabeling against the glycinergic presynaptic markers GlyT2 or VGAT.

We thank the reviewer for the thoughtful and thorough evaluation of the significance of investigating the role of neurexins in glycinergic transmission at the MNTB-LSO synapse, particularly in the context of auditory processing and sound localization. The positive feedback is greatly appreciated.

Weaknesses:

The major concern is novelty as this work on the effects of pan-neurexin deletion in a glycinergic synapse is quite consistent with the authors' prior work on glutamatergic synapses (Luo et al., 2020). The authors might want to further work out novel aspects and strengthen the comparative perspective. Conceptually, the authors might want to be more clear about interpreting the results on the altered dependence of release on voltage-gated Ca2+ influx (Ca2+ sensitivity, coupling).

Regarding the reviewer’s concern about the novelty of our work, we acknowledge that our previous work has explored the effects of pan-neurexin deletion on glutamatergic synapses (Luo et al., 2020). However, we would like to point out that a novelty of our present study indeed stems from the exploration of how different types of synapses converge to employ the same mechanism of synaptic function, particularly in the context of neurexin-mediated regulation. Our previous study focused on glutamatergic synapses, the current study delves into the realm of glycinergic synapses, which represent a distinct population with unique properties and functions. Despite the differences between these synapse types, our findings reveal a commonality in the underlying mechanisms of synaptic regulation mediated by neurexins. This convergence of mechanisms across different synapse types highlights the fundamental role of neurexins in synaptic function and plasticity. By elucidating how neurexins regulate synaptic transmission at both excitatory and inhibitory synapses, we provide valuable insights into the general principles governing synaptic function. In addition, this comparative perspective may shed light on the complex interplay between excitatory and inhibitory neurotransmission, which is crucial for maintaining the balance of neuronal activity and network dynamics.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

During the developmental period spanning P3-P12, the MNTB-LSO synapses undergo a transition from GABAergic to glycinergic transmission. It is well-established that Neurexin plays a role in modulating GABAergic transmission. In the authors' experimental system, AAV was injected at P0, likely impacting GABAergic transmission, including potentially influencing synapse number, before subsequently affecting glycinergic transmission. A thoughtful discussion of how the experimental interventions might have influenced this developmental process and glycinergic transmission would enhance the clarity and interpretation of their findings.

We thank the reviewer for raising the interesting topic of the transmitter switch during neurodevelopment. Strong evidence using gerbils and rats as animal models demonstrates that the MNTB-LSO synapses undergo a shift from GABAergic to glycinergic during the early development. However, in a more recent study by Friauf and colleagues (Fisher et al., 2019), patch-clamp recordings in acute mouse brainstem slices at P4-P11 combined with pharmacological blockade of GABAA receptors and/or glycine receptors clearly demonstrated no GABAergic synaptic component on LSO principal neurons, suggesting the transmitter subtype switch may be species different. We add a discussion in our revision to clarify this topic.

Reviewer #2 (Recommendations For The Authors):

The data are compelling and report an intriguing functional phenotype. Mechanistic insight into how this phenotype manifests would significantly strengthen this study. For example, which neuroligin is found at these MNTB-LSO synapses?

We agree that investigating the underlying molecular mechanisms, particularly the specific function of each variant of neurexins and their respective ligands on the postsynaptic neurons, is crucial. Exploring these mechanisms, which extend beyond the scope of our current study, would undoubtedly enhance our understanding of neurexin function at various synapses and foster advancements in the field.

Does the TKO alter the ability of MNTB inputs to induce AP firing in LSO neurons?

Activation of the MNTB inputs does not directly induce AP firing in LSO neurons, because the MNTB-LSO synapses are glycinergic and serve to inhibit neuronal activity.

We think the reviewer was to ask whether pan-neurexin deletion in the MNTB neurons alter their ability to impact the firing of LSO neurons. Indeed, the weakening of glycinergic transmission due to pan-neurexin ablation in MNTB neurons could potentially alter the excitation-inhibition (E/I) balance, thereby impacting the overall excitability of LSO neurons. We have conducted preliminary experiments to investigate this aspect and found that the E/I balance at LSO neurons was notably increased in TKO mice. We are currently preparing a manuscript to comprehensively address the role of neurexins at the auditory circuit and behavior levels.

Additional calcium measurements using GECIs would provide insight into whether nanodomain calcium or total calcium is altered at these synapses.

We appreciate the valuable suggestion provided by the reviewer. However, distinguishing between Ca2+ nanodomain and Ca2+ microdomain using Ca2+ imaging techniques requires advanced systems such as two-photon STED microscopy, which are beyond the scope of our current research.

It is unclear why fluorescence intensity is quantified instead of the number of synaptic clusters in LSO. In addition to changes in synapse numbers, fluorescent intensity can indicate a number of other possible morphological changes.

We appreciate the valuable suggestion from the reviewer. We have re-analyzed our imaging data to compare synaptic density. The results, as included in Fig.3f and 3h, confirm an increase in the number of glycinergic synapses after pan-neurexin deletion.

The most robust synaptic phenotypes were produced by measuring light-evoked oIPSCs and the authors acknowledge that electrically-evoked eIPSCs might be contaminated by uninfected fibers or by other sources of glycinergic inputs. I suggest that IPSC PPRs, EGTA, and low Ca2+ experiments be performed using optogenetics.

As discussed in our response to Public Reviews, we acknowledge that optogenetic stimulation offers crucial advantages, and we have provided a balanced discussion of the caveats associated with both methods in our manuscript. Additionally, following the reviewer’s suggestion, we have conducted new optogenetic experiments specifically for measuring the paired-pulse ratio in control and Nrxn123 TKO mice. We included this new dataset in supplementary Figure S2, which is consistent with our result obtained with electrically fiber stimulation.

For experiments involving EGTA and low Ca2+ manipulations, we opted for electrical stimulation due to major concerns regarding potential side effects of optogenetics, including the phototoxicity and photobleaching during prolonged light exposure.

It is sometimes confusing which type of evoked stimulation is being used (e.g. PPR, EGTA, and low Ca2+ experiments). To aid in the interpretations of these experiments, it would help to clarify.

We appreciate the reviewer's suggestion regarding the clarity of the evoked stimulation methods used in our experiments. We have revised the manuscript to provide clearer descriptions of the specific types of evoked stimulation employed in each experiment. Thank you for guiding towards this clarification.

The comparisons to Chen et al 2017 and the senior author's 2020 paper seem disjointed and do not contribute to the findings, which alone, are quite interesting. Given the prevailing notion that neurexins control different synaptic properties depending on the brain region and/or synapse studied, is it surprising that the findings observed here differ from previous studies of different synapses (glutamatergic and GABAergic)?

By comparing previous studies at different types of neurons/synapses, our findings reveal a commonality in the underlying mechanisms of synaptic regulation mediated by neurexins. This convergence of mechanisms across different synapse types highlights the fundamental role of neurexins in synaptic function and plasticity. In addition, this comparative perspective may shed light on the complex interplay between excitatory and inhibitory neurotransmission, which is crucial for maintaining the balance of neuronal activity and network dynamics.

Despite Nrxn3 being the most abundant Nrxn mRNA in MNTB neurons, the possible contributions of this highly expressed protein are not discussed.

We thank the reviewer for bringing up this important point. The differential expression of individual neurexins indeed suggests that specific neurexins may play dominant roles in regulating synaptic transmission. While our study primarily focused on the collective impact of ablating all neurexins, we acknowledge the significance of exploring the specific contributions of individual neurexin isoforms in the future. Understanding the distinct roles of each neurexin isoform could provide valuable insights into the precise mechanisms underlying synaptic function and plasticity. We have added discussion in our revised manuscript Line223-230.

Reviewer #3 (Recommendations For The Authors):

• There are several instances of spaces missing and typos, please carefully check the manuscript.

We greatly appreciate the reviewer's helpful feedback on the text that could be clarified or improved. We have meticulously edited the manuscript to address these concerns.

• While studying the properties of IPSC, apart from optogenetic stimulation, the authors performed experiments with electrical fiber stimulation. Their findings showed a slightly significant reduction of the IPSC amplitude and no effect on the IPSCs kinetics when comparing the TKO and control. One weakness is the discrepancy between the results from the optogenetic and fiber stimulation experiments, which the authors contribute to inefficient transfection in the fiber stimulation experiments. The authors state that they tried to optimize their protocols for virus injection protocols. However, they do not elaborate on how the transfection rates could be improved in the discussion section. Moreover, it would be good to further address the reasons for the difference in amplitude between the control IPSCs in the optogenetic and fiber stimulation experiments.

Echoing the suggestion by Reviewer 2 (see above), we acknowledge that optogenetic stimulation offers certain advantages, and we have provided a balanced discussion of the caveats associated with both methods in our manuscript. In addition, we have performed a new set of optogenetic experiment for the paired-pulse ratio measurement in control and Nrxn123 TKO mice and included as a new figure in supplementary figure S2.

For experiments involving EGTA and low Ca2+ manipulations, we opted for electrical stimulation due to major concerns regarding potential side effects of optogenetics, including the phototoxicity and photobleaching during prolonged light exposure.

We added the detail of virus injection strategy that optimized the transfection rates in the method section “To enhance virus infection efficiency, we decreased the dosage per injection while increasing the frequency of injections. Additionally, we ensured the pipette remained immobilized for 20-30 seconds to guarantee virus absorption at injection sites. As a result of this strategy, we estimated that the vast majority of MNTB neurons were inoculated by AAVs.” See line288-290.

• Abstract: "ablation of all neurexins in MNTB neurons reduced not only the amplitude but also altered the kinetics of the glycinergic synaptic transmission at LSO neurons."

Changed as suggested.

• Consider revising to "The synaptic dysfunctions primarily resulted from an altered dependence of release on voltage-gated Ca2+ influx."

We appreciate the reviewer's suggestion, which helps improve the clarity of our manuscript. We have revise the phrasing as follows: "The synaptic dysfunctions primarily resulted from an impaired calcium sensitivity of release and a loosened coupling between voltage-gated calcium channels and synaptic vesicles."

• Line 39 should be vertebrates.

Revised as suggested.

• Line 49 it would sound better to say "which further points to the diverse actions of neurexins in specific neurons."

Revised as suggested.

• Line 60 - this paragraph could include information about GABA signaling from the MNTB to the LSO, because on line 113 you mention LSO neurons receive inhibitory GABAergic/glycinergic inputs, but when you do not mention blocking of GABA currents to isolate the glycinergic ones.

We thank the reviewer for the thoughtful and detailed suggestion. We revised the text in line 60 to “In the mature mammalian auditory brainstem” and in line 113, we removed GABAergic to emphasize the nature of glycinergic synapse, particularly in the mouse brainstem where no GABAergic components are found (Fisher et al., 2019).

• Line 72/73 it should be adeno-associated virus; line 73: "combining this with the RNAScope technique" sounds better.

Changed as suggested.

• Line 91 using the RNAScope technique; lines 97, 119 as a control; line 108 the functional organization.<br />

Changed as suggested.

• Line 113 should be a pharmacological approach; line 122 optogenetically evoked.

Changed as suggested.

• Line 132, 160: the control.

Changed as suggested.

• Line 147 thus were infected; line 148 likely to be present but were obscured .

Changed as suggested.

• Line 154 which has been routinely used.

Changed as suggested.

• Line 155 It is not supposed to be Figure 2h but 2i; following that Figure 2i should be 2j; in my opinion, Figure 2i does not display a strong depression for the TKO mice.

Changed as suggested.

• Line 171 a better flow is achieved by saying: together these data show.

Changed as suggested.

• EC50 rather than IC50 of [Ca2+].

Changed as suggested.

• 180 it is better to say "we approached the matter by..."; line 183 while recording;

Changed as suggested.

• Line 203 were much stronger than the effect at control synapses; line 206 tightly clustering.

Changed as suggested.

• Line 212 sounds like they provide evidence for retina and spinal cord as well, should be made clear.

Changed as suggested.

• Line 289 previously.

Changed as suggested.

• Line 295 should be 30 min.

Changed as suggested.

• Line 336, 337 confocal microscope.

Changed as suggested.

• Please provide the number of data points also in figure captions or in the results section.

Added in the captions as suggested.

• Line 533, a better phrasing would be: the blocking effect of 0.2 mM Ca on IPSC amplitude.

Changed as suggested.

• Explain either in the methods or result section how was the EC50 of Ca2+ calculated.

Added in the methods as suggested.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This study provides important evidence supporting the ability of a new type of neuroimaging, OPM-MEG system, to measure beta-band oscillation in sensorimotor tasks on 2-14 years old children and to demonstrate the corresponding development changes, since neuroimaging methods with high spatiotemporal resolution that could be used on small children are quite limited. The evidence supporting the conclusion is solid but lacks clarifications about the much-discussed advantages of OPM-MEG system (e.g., motion tolerance), control analyses (e.g., trial number), and rationale for using sensorimotor tasks. This work will be of interest to the neuroimaging and developmental science communities.

We thank the editors and reviewers for their time and comments on our manuscript. We have responded in detail to the comments, on a point-by-point basis, below. Included in our responses (and our revised manuscript) are additional analyses to control for trial count, clarification of the advantages of OPM-MEG, and justification of our use of sensory (as distinct from motor) stimulation. In what follows, our responses are in bold typeface; additions to our manuscript are in bold italic typeface. 

Reviewer #1 (Public Review):

Summary:

Compared with conventional SQUID-MEG, OPM-MEG offers theoretical advantages of sensor configurability (that is, sizing to suit the head size) and motion tolerance (the sensors are intrinsically in the head reference frame). This study purports to be the first to experimentally demonstrate these advantages in a developmental study from age 2 to age 34. In short, while the theoretical advantages of OPM-MEG are attractive - both in terms of young child sensitivity and in terms of motion tolerance - neither was in fact demonstrated in this manuscript. We are left with a replication of SQUID-MEG observations, which certainly establishes OPM-MEG as "substantially equivalent" to conventional technology but misses the opportunity to empirically demonstrate the much-discussed theoretical advantages/opportunities.

Thank you for reviewing our manuscript. We agree that our results demonstrate substantial equivalence with conventional MEG. However, as mentioned by Reviewer 3, most past studies have “focused on older children and adolescents (e.g., 9-15 years old)” whereas our youngest group is 25 years. We believe that by obtaining data of sufficient quality in these age groups, without the need for any restriction of head movement, we have demonstrated the advantage of OPM-MEG. We now have made this clear in our discussion:

“…our primary aim was to test the feasibility of OPM-MEG for neurodevelopmental studies. Our results demonstrate we were able to scan children down to age 2 years, measuring high-fidelity electrophysiological signals and characterising the neurodevelopmental trajectory of beta oscillations. The fact that we were able to complete this study demonstrates the advantages of OPM-MEG over conventional-MEG, the latter being challenging to deploy across such a large age range…”

Strengths:

A replication of SQUID-MEG observations, which certainly establishes OPM-MEG as "substantially equivalent" to conventional technology but misses the opportunity to empirically demonstrate the much-discussed theoretical advantages/opportunities.

As noted above the demonstration of equivalence was one of our primary aims. We have elaborated further on the advantages below.

Weaknesses:

The authors describe 64 tri-axial detectors, which they refer to as 192 channels. This is in keeping with some of the SQUID-MEG description, but possibly somewhat disingenuous. For the scientific literature, perhaps "64 tri-axial detectors" is a more parsimonious description.

The number of channels in a MEG system refers to the number of independent measurements of magnetic field. This, in turn, tells us the number of degrees of freedom in the data that can be exploited by algorithms like signal space separation or beamforming. E.g. the MEGIN (cryogenic) MEG system has 306 channels, 102 magnetometers and 204 planar gradiometers. Sensors are constructed as “triple sensor elements” with one magnetometer and 2 gradiometers (in orthogonal orientations) centred on a single location. In our system, each sensor has three orthogonal metrics of magnetic field which are (by definition) independent. We have 64 such sensors, and therefore 192 independent channels – indeed when implementing algorithms like SSS we have shown we can exploit this number of degrees of freedom.1 192 channels is therefore an accurate description of the system.

A small fraction (<20%) of trials were eliminated for analysis because of "excess interference" - this warrants further elaboration.

We agree that this is an important point. We now state in our methods section:

“…Automatic trial rejection was implemented with trials containing abnormally high variance (exceeding 3 standard deviations from the mean) removed. All experimental trials were also inspected visually by an experienced MEG scientist, to exclude trials with large spikes/drifts that were missed by the automatic approach. In the adult group, there was a significant overlap between automatically and manually detected bad trials (0.7+-1.6 trials were only detected manually). In the children 10.0 +-9.4 trials were only detected manually)…”

We also note that the other reviewers and editor questioned whether the higher rejection rate in children had any bearing on results. This is an extremely important question. In revising the manuscript this has also been taken into account with all data reanalysed with equal trial counts in children and adults. Results are presented in Supplementary Information Section 5.

Figure 3 shows a reduced beta ERD in the youngest children. Although the authors claim that OPMMEG would be similarly sensitive for all ages and that SQUID-MEG would be relatively insensitive to young children, one trivial counterargument that needs to be addressed is that OPM has NOT in fact increased the sensitivity to young child ERD. This can possibly be addressed by analogous experiments using a SQUID-based system. An alternative would be to demonstrate similar sensitivity across ages using OPM to a brain measure such as evoked response amplitude. In short, how does Figure 3 demonstrate the (theoretical) sensitivity advantage of OPM MEG in small heads ?

We completely understand the referees’ point – indeed the question of whether a neuromagnetic effect really changes with age, or apparently changes due to a drop in sensitivity (caused by reduced head size or - in conventional MEG and fMRI - increased subject movement) is a question that can be raised in all neurodevelopmental studies.

Our authors have many years’ experience conducting studies using conventional MEG (including in neurodevelopment) and agreed that the idea of scanning subjects down to age two in conventional MEG would not be practical; their heads are too small and they typically fail to tolerate an environment where they are forced to remain still for long periods. Even if we tried a comparative study using conventional MEG, the likely data exclusion rate would be so high that the study would be confounded. This is why most conventional MEG studies only scan older children and adolescents. For this reason, we cannot undertake the comparative study the reviewer suggests. There are however two reasons why we believe sensitivity is not driving the neurodevelopmental effects that we observe:

Proximity of sensors to the head: 

For an ideal wearable MEG system, the distance between the sensors and the scalp surface (sensor proximity) would be the same regardless of age (and size), ensuring maximum sensitivity in all subjects. To test how our system performed in this regard, we undertook analyses to compute scalp-to-sensor distances. This was done in two ways:

(1) Real distances in our adaptable system: We took the co-registered OPM sensor locations and computed the Euclidean distance from the centre of the sensitive volume (i.e. the centre of the vapour cell) to the closest point on the scalp surface. This was measured independently for all sensors, and an average across sensors calculated. We repeated this for all participants (recall participants wore helmets of varying size and this adaptability should help minimise any relationship between sensor proximity and age).

(2) Simulated distances for a non-adaptable system: Here, the aim was to see how proximity might have changed with age, had only a single helmet size been used. We first identified the single example subject with the largest head (scanned wearing the largest helmet) and extracted the scalpto-sensor distances as above. For all other subjects, we used a rigid body transform to co-register their brain to that of the example subject (placing their head (virtually) inside the largest helmet). Proximity was then calculated as above and an average across sensors calculated. This was repeated for all participants.

In both analyses, sensor proximity was plotted against age and significant relationships probed using Pearson correlation. 

In addition, we also wanted to probe the relation between sensor proximity and head circumference. Head circumference was estimated by binarising the whole head MRI (to delineate volume of the head), and the axial slice with the largest circumference around was selected. We then plotted sensor proximity versus head circumference, for both the real (adaptive) and simulated (nonadaptive) case (expecting a negative relationship – i.e. larger heads mean closer sensor proximity). The slope of the relationship was measured and we used a permutation test to determine whether the use of adaptable helmets significantly lowered the identified slope (i.e. do adaptable helmets significantly improve sensor proximity in those with smaller head circumference).

Results are shown in Figure R1. We found no measurable relationship between sensor proximity and age (r = -0.195; p = 0.171) in the case of the real helmets (panel A). When simulating a non-adaptable helmet, we did see a significant effect of age on scalp-to-sensor distance (r = -0.46; p = 0.001; panel B). This demonstrates the advantage of the adaptability of OPM-MEG; without the ability to flexibly locate sensors, we would have a significant confound of sensor proximity. 

Plotting sensor proximity against head circumference we found a significant negative relationship in both cases (r = -0.37; p = 0.007 and  r = -0.78; p = 0.000001); however, the difference between slopes was significant according to a permutation test (p < 0.025) suggesting that adaptable has indeed improved sensor proximity in those with smaller head circumference. This again shows the benefits of adaptability to head size.

Author response image 1.

Scalp-to-sensor distance as a function of age (A/B) and head circumference (C/D). A and C show the case for the real helmets; B and D show the simulated non-adaptable case.

In sum, the ideal wearable system would see sensors located on the scalp surface, to get as close as possible to the brain in all subjects. Our system of multiple helmet sizes is not perfect in this regard (there is still a significant relationship between proximity and head circumference). However, our solution has offered a significant improvement over a (simulated) non-adaptable system. Future systems should aim to improve even further on this, either by using additively manufactured bespoke helmets for every subject (this is a gold standard, but also costly for large studies), or potentially adaptable flexible helmets.

Burst amplitudes:

The reviewer suggested to “demonstrate similar sensitivity across ages using OPM to a brain measure”. We decided not to use the evoked response amplitude (as suggested), since this would be expected to change with age. Instead, we used the amplitude of the bursts.

Our manuscript shows a significant correlation between beta modulation and burst probability – implying that the stimulus-related drop in beta amplitude occurs because bursts are less likely to occur. Further, we showed significant age-related changes in both beta amplitude and burst probability leading to a conclusion that the age dependence of beta modulation was caused by changes in the likelihood of bursts (i.e. bursts are less likely to ’switch off’ during sensory stimulation in children). We have now extended these analyses to test whether burst amplitude also changes significantly with age – we reasoned that if burst amplitude remained the same in children and adults, this would not only suggest that beta modulation is driven by burst probability (distinct from burst amplitude), but also show directly that the beta effects we see are not attributable to a lack of sensitivity in younger people. 

We took the (unnormalized) beamformer projected electrophysiological time series from sensorimotor cortex and filtered it 5-48 Hz (the motivation for the large band was because bursts are known to be pan-spectral and have lower frequency content in children; this band captures most of the range of burst frequencies highlighted in our spectra). We then extracted the timings of the bursts, and for each burst took the maximum projected signal amplitude. These values were averaged across all bursts in an individual subject, and plotted for all subjects against age.

Author response image 2.

Beta burst amplitude as a function of age; A) shows index finger simulation trials; B shows little finger stimulation trials. In both case there was no significant modulation of burst amplitude with age.

Results (see Figure R2) showed that the amplitude of the beta burst showed no significant age-related modulation (R2 = 0.01, p = 0.48 for index finger and R2 = 0.01, p = 0.57 for the little finger). This is distinct from both burst probability and task induced beta modulation. This adds weight to the argument that the diminished beta modulation in children is not caused by a lack of sensitivity to the MEG signal and supports our conclusion that burst probability is the primary driver of the agerelated changes in beta oscillations.

Both of the above analyses have been added to our supplementary information and mentioned in the main manuscript. The first shows no confound of sensor proximity to the scalp with age in our study. The second shows that the bursts underlying the beta signal are not significantly lower amplitude in children – which we reasoned they would be if sensitivity was diminished at younger ages. We believe that the two together suggest that we have mitigated a sensitivity confound in our study.

The data do not make a compelling case for the motion tolerance of OPM-MEG. Although an apparent advantage of a wearable system, an empirical demonstration is still lacking. How was motion tracked in these participants?

We agree that this was a limitation of our experiment. 

We have the equipment to track motion of the head during an experiment, using IR retroreflective markers placed on the helmet and a set of IR cameras located inside the MSR. However, the process takes a long time to set up, it lacks robustness, and would have required an additional computer (the one we typically use was already running the somatosensory stimulus and video). When the study was designed, we were concerned that the increased set up time for motion tracking would cause children to get bored, and result in increased participant drop out. For this reason we decided not to capture motion of the head during this study.

With hindsight this was a limitation which – as the reviewer states – makes us unable to prove that motion robustness was a significant advantage for this study. That said, during scanning there was both a parent and an experimenter in the room for all of the children scanned, and anecdotally we can say that children tended to move their head during scans – usually to talk to the parent. Whilst this cannot be quantified (and is therefore unsatisfactory) we thought it worth mentioning in our discussion, which reads:

“…One limitation of the current study is that practical limitations prevented us from quantitatively tracking the extent to which children (and adults) moved their head during a scan. Anecdotally however, experimenters present in the room during scans reported several instances where children moved, for example to speak to their parents who were also in the room. Such levels of movement could not be tolerated in conventional MEG or MRI and so this again demonstrates the advantages afforded by OPM-MEG…”

As a note, empirical demonstrations of the motion tolerance of OPM-MEG have been published previously: Early demonstrations included Boto et al. 2 who captured beta oscillations in adults playing a ball game and Holmes et al. who measured visual responses as participants moved their head to change viewing angle3. In more recent demonstrations, Seymour et al. measured the auditory evoked field in standing mobile participants4; Rea et al. measured beta modulation as subjects carried out a naturalistic handwriting task5 and Holmes et al measured beta modulation as a subject walked around a room.6

Furthermore, while the introduction discusses at some length the phenomenon of PMBR, there is no demonstration of the recording of PMBR (or post-sensory beta rebound). This is a shame because there is literature suggesting an age-sensitivity to this, that the optimal sensitivity of OPM-MEG might confirm/refute. There is little evidence in Figure 3 for adult beta rebound. Is there an explanation for the lack of sensitivity to this phenomenon in children/adolescents? Could a more robust paradigm (button-press) have shed light on this?

We understand the question. There are two limitations to the current study in respect to measuring the PMBR:

Firstly, sensory tasks generally do not induce as strong a PMBR as motor tasks and with this in mind a stronger rebound response could have been elicited using a button press. However, it was our intention to scan children down to age 2 and we were sceptical that the youngest children would carry out a button press as instructed. For this reason we opted for entirely passive stimulation, requiring no active engagement from our participants. The advantages of this was a stimulus that all subjects could engage with. However, this was at the cost of a diminished rebound.

The second limitation relates to trial length. Multiple studies have shown that the PMBR can last over ~10 s 7,8. Indeed, Pfurtscheller et al. argued in 1999 that it was necessary to leave 10 s between movements to allow the PMBR to return to a true baseline9, though this has rarely been adhered to in the literature. Here, we wanted to keep recordings short for the comfort of the younger participants, so we adopted a short trial duration. However, a consequence of this short trial length is that it becomes impossible to access the PMBR directly; one can only measure beta modulation with the task. This limitation has now been addressed explicitly in our discussion:

“…this was the first study of its kind using OPM-MEG, and consequently aspects of the study design could have been improved. Firstly, the task was designed for children; it was kept short while maximising the number of trials (to maximise signal to noise ratio). However, the classical view of beta modulation includes a PMBR which takes ~10 s to reach baseline following task cessation7–9. Our short trial duration therefore doesn’t allow the rebound to return to baseline between trials, and so conflates PMBR with rest. Consequently, we cannot differentiate the neural generators of the task induced beta power decrease and the PMBR; whilst this helped ensure a short, child friendly task, future studies should aim to use longer rest windows to independently assess which of the two processes is driving age related changes…”

Data on functional connectivity are valuable but do not rely on OPM recording. They further do not add strength to the argument that OPM MEG is more sensitive to brain activity in smaller heads - in fact, the OPM recordings seem plagued by the same insensitivity observed using conventional systems.

Given the demonstration above that bursts are not significantly diminished in amplitude in children relative to adults; and further given the demonstrations in the literature (e.g. Seedat et al.10) that functional connectivity is driven by bursts, we would argue that the effects of connectivity changing with age are not related to sensitivity but rather genuinely reflect a lack of coordination of brain activity.

The discussion of burst vs oscillations, while highly relevant in the field, is somewhat independent of the OPM recording approach and does not add weight to the OPM claims.

We agree that the burst vs. oscillations discussion does not add weight to the OPM claims per se. However, we had two aims of our paper, the second being to “investigate how task-induced beta modulation in the sensorimotor cortices is related to the occurrence of pan-spectral bursts, and how the characteristics of those bursts change with age.” As the reviewer states, this is highly relevant to the field, and therefore we believe adds impact, not only to the paper, but also by extension to the technology.

In short, while the theoretical advantages of OPM-MEG are attractive - both in terms of young child sensitivity and in terms of motion tolerance, neither was in fact demonstrated in this manuscript. We are left with a replication of SQUID-MEG observations, which certainly establishes OPM-MEG as "substantially equivalent" to conventional technology but misses the opportunity to empirically demonstrate the much-discussed theoretical advantages/opportunities.

We thank the referee for the time and important contributions to this paper. We believe the fact that we were able to record good data in children as young as two years old was, in itself, an experimental realisation of the ‘theoretical advantages’ of OPM-MEG. Our additional analyses, inspired by the reviewers comments, help to clarify the advantages of OPM-MEG over conventional technology. The reviewers’ insights have without doubt improved the paper.

Reviewer #2 (Public Review):

Summary:

The authors introduce a new 192-channel OPM system that can be configured using different helmets to fit individuals from 2 to 34 years old. To demonstrate the veracity of the system, they conduct a sensorimotor task aimed at mapping developmental changes in beta oscillations across this age range. Many past studies have mapped the trajectory of beta (and gamma) oscillations in the sensorimotor cortices, but these studies have focused on older children and adolescents (e.g., 9-15 years old) and used motor tasks. Thus, given the study goals, the choice of a somatosensory task was surprising and not justified. The authors recorded a final sample of 27 children (2-13 years old) and 24 adults (21-34 years) and performed a time-frequency analysis to identify oscillatory activity. This revealed strong beta oscillations (decreases from baseline) following the somatosensory stimulation, which the authors imaged to discern generators in the sensorimotor cortices. They then computed the power difference between 0.3-0.8 period and 1.0-1.5 s post-stimulation period and showed that the beta response became stronger with age (more negative relative to the stimulation period). Using these same time windows, they computed the beta burst probability and showed that this probability increased as a function of age. They also showed that the spectral composition of the bursts varied with age. Finally, they conducted a whole-brain connectivity analysis. The goals of the connectivity analysis were not as clear as prior studies of sensorimotor development have not conducted such analyses and typically such whole-brain connectivity analyses are performed on resting-state data, whereas here the authors performed the analysis on task-based data. In sum, the authors demonstrate that they can image beta oscillations in young children using OPM and discern developmental effects.

Thank you for this summary and for taking the time to review our manuscript.

Strengths:

Major strengths of the study include the novel OPM system and the unique participant population going down to 2-year-olds. The analyses are also innovative in many respects.

Thank you – we also agree that the major strength is in the unique cohort.

Weaknesses:

Several weaknesses currently limit the impact of the study. 

First, the choice of a somatosensory stimulation task over a motor task was not justified. The authors discuss the developmental motor literature throughout the introduction, but then present data from a somatosensory task, which is confusing. Of note, there is considerable literature on the development of somatosensory responses so the study could be framed with that.

We completely understand the referee’s point, and we agree that the motivation for the somatosensory task was not made clear in our original manuscript.

Our choice of task was motivated completely by our targeted cohort; whilst a motor task would have been our preference, it was generally felt that making two-year-olds comply with instructions to press a button would have been a significant challenge. In addition, there would likely have been differences in reaction times. By opting for a passive sensory stimulation we ensured compliance, and the same stimulus for all subjects. We have added text on this to our introduction as follows:

“…Here, we combine OPM-MEG with a burst analysis based on a Hidden Markov Model (HMM) 10–12 to investigate beta dynamics. We scanned a cohort of children and adults across a wide age range (upwards from 2 years old). Because of this, we implemented a passive somatosensory task which can be completed by anyone, regardless of age…”

We also state in our discussion:

“…here we chose to use passive (sensory) stimulation. This helped ensure compliance with the task in subjects of all ages and prevented confounds of e.g. reaction time, force, speed and duration of movement which would be more likely in a motor task.7,8 However, there are many other systems to choose and whether the findings here regarding beta bursts and the changes with age also extend to other brain networks remains an open question.…”

Regarding the neurodevelopmental literature – we are aware of the literature on somatosensory evoked responses – particularly median nerve stimulation – but we can find little on the neurodevelopmental trajectory of somatosensory induced beta oscillations (the topic of our paper). We have edited our introduction as follows:

“…All these studies probed beta responses to movement execution; in the case of tactile stimulation (i.e. sensory stimulation without movement) both task induced beta power loss, and the post stimulus rebound have been consistently observed in adults9,13–18. Further, beta amplitude in sensory cortex has been related to attentional processes19 and is broadly thought to carry top down top down influence on primary areas20. However, there is less literature on how beta modulation changes with age during purely sensory tasks.…”

We would be keen for the reviewer to point to any specific papers in the literature that we may have missed.

Second, the primary somatosensory response actually occurs well before the time window of interest in all of the key analyses. There is an established literature showing mechanical stimulation activates the somatosensory cortex within the first 100 ms following stimulation, with the M50 being the most robust response. The authors focus on a beta decrease (desynchronization) from 0.3-0.8 s which is obviously much later, despite the primary somatosensory response being clear in some of their spectrograms (e.g., Figure 3 in older children and adults). This response appears to exhibit a robust developmental effect in these spectrograms so it is unclear why the authors did not examine it. This raises a second point; to my knowledge, the beta decrease following stimulation has not been widely studied and its function is unknown. The maps in Figure 3 suggest that the response is anterior to the somatosensory cortex and perhaps even anterior to the motor cortex. Since the goal of the study is to demonstrate the developmental trajectory of well-known neural responses using an OPM system, should the authors not focus on the best-understood responses (i.e., the primary somatosensory response that occurs from 0.0-0.3 s)?

We understand the reviewer’s point. The original aim of our manuscript was to investigate the neurodevelopmental trajectory of beta oscillations, not the evoked response. In fact, the evoked response in this paradigm is complicated by the fact that there are three stimuli in a very short (<500 ms) time window. For this reason, we prefer the focus of our paper to remain on oscillations.

Nevertheless, we agree that not including the evoked responses was a missed opportunity.  We have now added evoked responses to our analysis pipeline and manuscript. As surmised by the reviewer, the M50 shows neurodevelopmental changes (an increase with age). Our methods section has been updated accordingly and Figure 3 has been modified. The figure and caption are copied below for the convenience of the reviewer.

Author response image 3.

Beta band modulation with age: (A) Brain plots show slices through the left motor cortex, with a pseudo-T-statistical map of beta modulation (blue/green) overlaid on the standard brain. Peak MNI coordinates are indicated for each subgroup. Time frequency spectrograms show modulation of the amplitude of neural oscillations (fractional change in spectral amplitude relative to the baseline measured in the 2.5-3 s window). Vertical lines indicate the time of the first braille stimulus. In all cases results were extracted from the location of peak beta desynchronisation (in the left sensorimotor cortex). Note the clear beta amplitude reduction during stimulation. The inset line plots show the 4-40 Hz trial averaged phase-locked evoked response, with the expected prominent deflections around 20 and 50 ms. (B) Maximum difference in beta-band amplitude (0.3-0.8 s window vs 1-1.5 s window) plotted as a function of age (i.e., each data point shows a different participant; triangles represent children, circles represent adults). Note significant correlation (𝑅2 \= 0.29, 𝑝 = 0.00004 *). (C) Amplitude of the P50 component of the evoked response plotted against age. There was no significant correlation (𝑅2 \= 0.04, 𝑝 = 0.14 ). All data here relate to the index finger stimulation; similar results are available for the little finger stimulation in Supplementary Information Section 1.

Regarding the developmental effects, the authors appear to compute a modulation index that contrasts the peak beta window (.3 to .8) to a later 1.0-1.5 s window where a rebound is present in older adults. This is problematic for several reasons. First, it prevents the origin of the developmental effect from being discerned, as a difference in the beta decrease following stimulation is confounded with the beta rebound that occurs later. A developmental effect in either of these responses could be driving the effect. From Figure 3, it visually appears that the much later rebound response is driving the developmental effect and not the beta decrease that is the primary focus of the study. Second, these time windows are a concern because a different time window was used to derive the peak voxel used in these analyses. From the methods, it appears the image was derived using the .3-.8 window versus a baseline of 2.5-3.0 s. How do the authors know that the peak would be the same in this other time window (0.3-0.8 vs. 1.0-1.5)? Given the confound mentioned above, I would recommend that the authors contrast each of their windows (0.3-0.8 and 1.0-1.5) with the 2.5-3.0 window to compute independent modulation indices. This would enable them to identify which of the two windows (beta decrease from 0.3-0.8 s or the increase from 1.0-1.5 s) exhibited a developmental effect. Also, for clarity, the authors should write out the equation that they used to compute the modulation index. The direction of the difference (positive vs. negative) is not always clear.

We completely understand the referee’s point; referee 1 made a similar point. In fact, there are two limitations of our paradigm regarding the measurement of PMBR versus the task-induced beta decrease:

Firstly, sensory tasks generally do not induce as strong a PMBR as motor tasks and with this in mind a stronger rebound response could have been elicited using a button press. However, as described above it was our intention to scan children down to age 2 and we were sceptical that the youngest children would carry out a button press as instructed.

The second limitation relates to trial length. Multiple studies have shown that the PMBR can last over ~10 s7,8. Indeed, Pfurtscheller et al. argued in 1999 that it was necessary to leave 10 s between movements to allow the PMBR to return to a true baseline9 Here, we wanted to keep recordings relatively short for the younger participants, and so we adopted a short trial duration. However, a consequence of this short trial length is that it becomes impossible to access the PMBR directly because the PMBR of the nth trial is still ongoing when the (n+1)th trial begins. Because of this, there is no genuine rest period, and so the stimulus induced beta decrease and subsequent rebound cannot be disentangled. This limitation has now been made clear in our discussion as follows:

“…this was the first study of its kind using OPM-MEG, and consequently aspects of the study design could have been improved. Firstly, the task was designed for children; it was kept short while maximising the number of trials (to maximise signal to noise ratio). However, the classical view of beta modulation includes a PMBR which takes ~10 s to reach baseline following task cessation7–9. Our short trial duration therefore doesn’t allow the rebound to return to baseline between trials, and so conflates PMBR with rest. Consequently, we cannot differentiate the neural generators of the task induced beta power decrease and the PMBR; whilst this helped ensure a short, child friendly task, future studies should aim to use longer rest windows to independently assess which of the two processes is driving age related changes…”

To clarify our method of calculating the modulation index, we have added the following statement to the methods:

“The beta modulation index was calculated using the equation , where , and are the average Hilbert-envelope-derived amplitudes in the stimulus (0.3-0.8s), post-stimulus (1-1.5s) and baseline (2.5-3s) windows, respectively.”

Another complication of using a somatosensory task is that the literature on bursting is much more limited and it is unclear what the expectations would be. Overall, the burst probability appears to be relatively flat across the trial, except that there is a sharp decrease during the beta decrease (.3-.8 s). This matches the conventional trial-averaging analysis, which is good to see. However, how the bursting observed here relates to the motor literature and the PMBR versus beta ERD is unclear.

Again, we agree completely; a motor task would have better framed the study in the context of existing burst literature – but as mentioned above, making 2-year-olds comply with the instructions for a motor task would have been difficult. Interestingly in a recent paper, Rayson et al. used EEG to investigate burst activity in infants (9 and 12 months) and adults during observed movement execution, with results showing stimulus induced decrease in beta burst rate at all ages, with the largest effects in adults21. This paper was not yet published when we submitted our article but does help us to frame our burst results since there is strong agreement between their study and ours. We now mention this study in both our introduction and discussion. 

Another weakness is that all participants completed 42 trials, but 19% of the trials were excluded in children and 9% were excluded in adults. The number of trials is proportional to the signal-to-noise ratio. Thus, the developmental differences observed in response amplitude could reflect differences in the number of trials that went into the final analyses.

This is an important observation and we thank the reviewer for raising the issue. We have now re-analysed all of our data, removing trials in the adults such that the overall number of trials was the same as for the children. All effects with age remained significant. We chose to keep the Figures in the main manuscript with all good trials (as previously) and present the additional analyses (with matched trial numbers) in supplementary information. However, if the reviewer feels strongly, we could do it the other way around (there is very little difference between the results).

Reviewer #3 (Public Review):

This study demonstrated the application of OPM-MEG in neurodevelopment studies of somatosensory beta oscillations and connections with children as young as 2 years old. It provides a new functional neuroimaging method that has a high spatial-temporal resolution as well wearable which makes it a new useful tool for studies in young children. They have constructed a 192-channel wearable OPM-MEG system that includes field compensation coils which allow free head movement scanning with a relatively high ratio of usable trials. Beta band oscillations during somatosensory tasks are well localized and the modulation with age is found in the amplitude, connectivity, and panspectral burst probability. It is demonstrated that the wearable OPM-MEG could be used in children as a quite practical and easy-to-deploy neuroimaging method with performance as good as conventional MEG. With both good spatial (several millimeters) and temporal (milliseconds) resolution, it provides a novel and powerful technology for neurodevelopment research and clinical applications not limited to somatosensory areas.

We thank the reviewer for their summary, and their time in reviewing our manuscript.

The conclusions of this paper are mostly well supported by data acquired under the proper method. However, some aspects of data analysis need to be improved and extended.

(1) The colour bars selected for the pseudo-T-static pictures of beta modulation in Figures 2 and 3, which are blue/black and red/black, are not easily distinguished from the anatomical images which are grey-scale. A colour bar without black/white would make these figures better. The peak point locations are also suggested to be marked in Figure 2 and averaged locations in Figure 3 with an error bar.

Thank you for this comment which we certainly agree with. The colour scheme used has now been changed to avoid black. We have also added peak locations. 

(2) The data points in plots are not constant across figures. In Figures 3 and 5, they are classified into triangles and circles for children and adults, but all are circles in Figures 4 and 6.

Thank you! We apologise for the confusion. Data points are now consistent across plots.

(3) Although MEG is much less susceptible to conductivity inhomogeneity of the head than EEG, the forward modulating may still be impacted by the small head profile. Add more information about source localization accuracy and stability across ages or head size.

This is an excellent point. We have added to our discussion relating to the accuracy of the forward model. 

“…We failed to see a significant difference in the spatial location of the cortical representations of the index and little finger; there are three potential reasons for this. First, the system was not designed to look for such a difference – sensors were sparsely distributed to achieve whole head coverage (rather than packed over sensory cortex to achieve the best spatial resolution in one area22). Second, our “pseudo-MRI” approach to head modelling (see Methods) is less accurate than acquisition of participantspecific MRIs, and so may mask subtle spatial differences. Third, we used a relatively straightforward technique for modelling magnetic fields generated by the brain (a single shell forward model). Although MEG is much less susceptible to conductivity inhomogeneity of the head than EEG, the forward model may still be impacted by the small head profile. This may diminish spatial resolution and future studies might look to implement more complex models based on e.g. finite element modelling23. Finally, previous work 24 suggested that, for a motor paradigm in adults, only the beta rebound, and not the power reduction during stimulation, mapped motortopically. This may also be the case for purely sensory stimulation. Nevertheless, it remains the case that by placing sensors closer to the scalp, OPM-MEG should offer improved spatial resolution in children and adults; this should be the topic of future work…”

Recommendations for the authors:

Reviewer #2 (Recommendations For The Authors):

Major items to further test include the differing number of trials, the windowing issue, and the focus on motor findings in the intro and discussion. First, I would recommend the authors adjust the number of trials in adults to equate them between groups; this will make their developmental effects easier to interpret.  

Thank you for raising this important point. This has now been done and appears in our supplementary information as discussed above.

Second, to discern which responses are exhibiting developmental effects, the authors need to contrast the 0.3-0.8 window with the later window (2.5-3.0), not the window that appears to have the PMBR-like response. This artificially accentuates the response. I also think they should image the 1.0-1.5 vs 2.5-3.0s window to determine whether the response in this time window is in the same location as the decrease and then contrast this for beta differences. 

We completely understand this point, which relates to separating the reduction in beta amplitude during stimulation and the rebound post stimulation. However, as explained above, doing so unambiguously would require the use of much longer trials. Here we were only able to measure stimulus induced beta modulation (distinct from the separate contributions of the task induced beta power reduction and rebound). It may be that future studies, with >10 s trial length, could probe the role of the PMBR, but such studies require long paradigms which are challenging to implement with children.

Third, changing the framing of the study to highlight the somatosensory developmental literature would also be an improvement.

We have added to our introduction a stated in the responses above.

Finally, the connectivity analysis on data from a somatosensory task did not make sense given the focus of the study and should be removed in my opinion. It is very difficult to interpret given past studies used resting state data and one would expect the networks to dynamically change during different parts of the current task (i.e., stimulation versus baseline).

We appreciate the point regarding connectivity. However, it was our intention to examine the developmental trajectory of beta oscillations, and a major role of beta oscillations is in mediating connectivity. It is true that most studies are conducted in the resting state (or more recently – particularly in children – during movie watching). The fact that we had a sensory task running is a confound; nevertheless, the connectivity we derived in adults bears a marked similarity to that from previous papers (e.g. 25) and we do see significant changes with age. We therefore believe this to be an important addition to the paper and we would prefer to keep it.

References

(1) Holmes, N., Bowtell, R., Brookes, M. J. & Taulu, S. An Iterative Implementation of the Signal Space Separation Method for Magnetoencephalography Systems with Low Channel Counts.

Sensors 23, 6537 (2023).

(2) Boto, E. et al. Moving magnetoencephalography towards real-world applications with a wearable system. Nature (2018) doi:10.1038/nature26147.

(3) Holmes, M. et al. A bi-planar coil system for nulling background magnetic fields in scalp mounted magnetoencephalography. NeuroImage 181, 760–774 (2018).

(4) Seymour, R. A. et al. Using OPMs to measure neural activity in standing, mobile participants. NeuroImage 244, 118604 (2021).

(5) Rea, M. et al. A 90-channel triaxial magnetoencephalography system using optically pumped magnetometers. annals of the new york academy of sciences 1517, https://doi.org/10.1111/nyas.14890 (2022).

(6) Holmes, N. et al. Enabling ambulatory movement in wearable magnetoencephalography with matrix coil active magnetic shielding. NeuroImage 274, 120157 (2023).

(7) Pakenham, D. O. et al. Post-stimulus beta responses are modulated by task duration. NeuroImage 206, 116288 (2020).

(8) Fry, A. et al. Modulation of post-movement beta rebound by contraction force and rate of force development. Human Brain Mapping 37, 2493–2511 (2016).

(9) Pfurtscheller, G. & Lopes da Silva, F. H. Event-related EEG/MEG synchronization and desynchronization: Basic principles. Clin Neurophysio 110, 1842–1857 (1999).

(10) Seedat, Z. A. et al. The role of transient spectral ‘bursts’ in functional connectivity: A magnetoencephalography study. NeuroImage 209, 116537 (2020).

(11) Baker, A. P. et al. Fast transient networks in spontaneous human brain activity. eLife 2014, 1867 (2014).

(12) Vidaurre, D. et al. Spectrally resolved fast transient brain states in electrophysiological data. NeuroImage 126, 81–95 (2016).

(13) Gaetz, W. & Cheyne, D. Localization of sensorimotor cortical rhythms induced by tactile stimulation using spatially filtered MEG. NeuroImage 30, 899–908 (2006).

(14) Cheyne, D. et al. Neuromagnetic imaging of cortical oscillations accompanying tactile stimulation. Cognitive Brain Research 17, 599–611 (2003).

(15) van Ede, F., Jensen, O. & Maris, E. Tactile expectation modulates pre-stimulus β-band oscillations in human sensorimotor cortex. NeuroImage 51, 867–876 (2010).

(16) Salenius, S., Schnitzler, A., Salmelin, R., Jousmäki, V. & Hari, R. Modulation of Human Cortical Rolandic Rhythms during Natural Sensorimotor Tasks. NeuroImage 5, 221–228 (1997).

(17) Cheyne, D. O. MEG studies of sensorimotor rhythms: A review. Experimental Neurology 245, 27–39 (2013).

(18) Kilavik, B. E., Zaepffel, M., Brovelli, A., MacKay, W. A. & Riehle, A. The ups and downs of beta oscillations in sensorimotor cortex. Experimental Neurology 245, 15–26 (2013).

(19) Bauer, M., Oostenveld, R., Peeters, M. & Fries, P. Tactile Spatial Attention Enhances Gamma-Band Activity in Somatosensory Cortex and Reduces Low-Frequency Activity in Parieto-Occipital Areas. J. Neurosci. 26, 490–501 (2006).

(20) Barone, J. & Rossiter, H. E. Understanding the Role of Sensorimotor Beta Oscillations. Frontiers in Systems Neuroscience 15, (2021).

(21) Rayson, H. et al. Bursting with Potential: How Sensorimotor Beta Bursts Develop from Infancy to Adulthood. J Neurosci 43, 8487–8503 (2023).

(22) Hill, R. M. et al. Optimising the Sensitivity of Optically-Pumped Magnetometer Magnetoencephalography to Gamma Band Electrophysiological Activity. Imaging Neuroscience (2024) doi:10.1162/imag_a_00112.

(23) Stenroos, M., Hunold, A. & Haueisen, J. Comparison of three-shell and simplified volume conductor models in magnetoencephalography. NeuroImage 94, 337–348 (2014).

(24) Barratt, E. L., Francis, S. T., Morris, P. G. & Brookes, M. J. Mapping the topological organisation of beta oscillations in motor cortex using MEG. NeuroImage 181, 831–844 (2018).

(25) Rier, L. et al. Test-Retest Reliability of the Human Connectome: An OPM-MEG study. Imaging Neuroscience (2023) doi:10.1162/imag_a_00020.

Reviewer #3 (Public Review):

This study demonstrated the application of OPM-MEG in neurodevelopment studies of somatosensory beta oscillations and connections with children as young as 2 years old. It provides a new functional neuroimaging method which has high spatial-temporal resolution as well wearable which makes it a new useful tool for studies in young children. They have constructed a 192-channel wearable OPM-MEG system includes field compensation coils which allows free head movement scanning with relatively high ratio of usable trials. Beta band oscillations during somatosensory tasks are well localized and the modulation with age are found in the amplitude, connectivity, and pan-spectral burst probability. It is demonstrated that the wearable OPM-MEG could be used in children as a quite practical and easy to deploy neuroimaging method with performance as good as conventional MEG. With both good spatial (several millimeter) and temporal (milliseconds) resolution, it provides a novel and powerful technology to neurodevelopment research and clinical application not limited to somatosensory areas.

The conclusions of this paper are mostly well supported by data acquired under proper method.

eLife assessment

This study provides important evidence supporting the ability of a new type of neuroimaging, OPM-MEG system, to measure beta-band oscillation in sensorimotor tasks in 2-14 years old children and to demonstrate the corresponding development changes, since neuroimaging methods with high spatiotemporal resolution that could be used on small children are quite limited. The evidence supporting the conclusion is compelling. This work will be of interest to the neuroimaging and developmental science communities.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

This important study investigates the transcriptional changes in neurons that underlie loss of learning and memory with age in C. elegans, and how cognition is maintained in insulin/IGF-1-like signaling mutants. The presented evidence is convincing, utilizing a cutting-edge method to isolate neurons from worms for genomics that is clearly conveyed with a rigorous experimental approach. Overall, this study supports that older daf-2 worms maintain cognitive function via mechanisms that are unique from younger wild type worms, which will be of interest to neuroscientists and researchers studying ageing.

Thank you, we appreciate the positive comments.

Public Reviews: 

Reviewer #1 (Public Review): 

The authors perform RNA-seq on FACS-isolated neurons from adult worms at days 1 and 8 of adulthood to profile the gene expression changes that occur with cognitive decline. Supporting data are included indicating that by day 7 of adulthood, learning and memory are reduced, indicating that this time point or after represents cognitively aged worms. Neuronal identity genes are reduced in expression within cognitively aged worms, whereas genes involved in proteostasis, transcription/chromatin, and stress response are elevated. A number of specific examples are provided, representing markers of specific neuronal subtypes, and correlating expression changes to the erosion of particular functions (e.g. motor neurons, chemosensory neurons, aversive learning neurons, etc). 

To investigate whether the upregulation of genes in neurons with age is compensatory or deleterious, the authors reduced the expression of a set of three significantly upregulated genes and performed behavioral assays in young adults. In each case, reduction of expression improved memory, consistent with a model in which age-associated increases impair neuronal function. This claim would be bolstered by an experiment elevating the expression of these genes in young neurons, which should reduce the learning index if the hypothesis is correct. 

This is an interesting suggestion. Our long-term goal is to find ways to improve memory, and to better understand the “rules” that might govern changes with age. In this case, were interested in addressing the hypothesis that genes that rise with age must be compensatory, which is a frequently stated theory that is not often tested. Here we showed that knocking down three genes that are upregulated in aged animals improved memory; our results suggest that the wild-type functions of these genes are likely deleterious for learning and memory functions, and further, that their increased expression with age is not a compensatory function. Certainly for future work, it might be interesting to better understand how and why these specific genes have a deleterious function that increases with age, and whether that function is different in younger animals where they are not highly expressed.

The authors then characterize learning and memory in wild-type, daf-2, and daf-2/daf-16 worms with age and find that daf-2 worms have an extended ability to learn for approximately 10 days longer than wild types. This was daf-16 dependent. Memory was extended in daf-2 as well, and strikingly, daf-2;daf-16 had no short-term memory even at day 1. Transcriptomic analysis of FACS-sorted neurons was performed on the three groups at day 8. The authors focus their analysis on daf-2 vs. daf-2;daf-16 and present evidence that daf-2 neurons express a stress-resistance gene program. One question that remains unanswered is how well the N2 and daf-2;daf-16 correlate overall, and are there differences? This may be informative as wild type and daf-2;daf-16 mutants are not phenotypically identical when it comes to memory, and there may be differences that can be detected despite the overlap in the PCA. This analysis could reveal the daf-16 targets involved in memory. 

Re. daf-2;daf-16 vs N2: This is a good suggestion. Our analysis in Fig. S5 showed that the daf-2 vs N2 comparison shows similar results with the daf-2 vs daf-16;daf-2 comparison, but some additional genes are differentially expressed. Interestingly, the daf-2 vs N2 comparison shows that the bZip transcription factors are upregulated in daf-2 compared with N2 worms (Fig. S6f). This may indicate that additional transcription factors are controlled by the daf-2 mutation in the nervous system in addition to the DAF-16/FOXO transcription factor.

Author response image 1.

We also identified the differentially expressed genes in the Day 8 neuronal daf-16;daf-2 to N2 comparison, as the reviewer is asking about. The samples from different genotypes do separate from one another in the PCA plot, indicating there are differences between daf-16,daf-2 and N2 neurons. However, the difference is smaller and there are fewer genes differentially expressed between daf-16;daf-2 and N2: only 38 genes are significantly higher in daf-16;daf-2, and only 53 genes are significantly higher in N2 (log2FC > 0.5, p-adj<0.05). The genes higher in N2 are enriched in endopeptidase inhibitors, and the genes higher in daf-16;daf-2 are not enriched in any gene ontology terms. These results indicate that there are some differences between daf-16;daf-2 and N2 neurons, which correlates with the behavioral differences we see, but the difference is small compared to daf-2 neurons. We have added these data to the paper (Fig. S4e,f); thank you for the suggestion.

The authors tested eight candidate genes that were more highly expressed in daf-2 neurons vs. daf-2;daf-16 and showed that reduction of 2 and 5 of these genes impaired learning and memory, respectively, in daf-2 worms. This finding implicates specific neuronal transcriptional targets of IIS in maintaining cognitive ability in daf-2 with age, which, importantly, are distinct from those in young wild type worms. 

Reviewer #2 (Public Review): 

Weng et al. perform a comprehensive study of gene expression changes in young and old animals, in wild-type and daf-2 insulin receptor mutants, in the whole animal, and specifically in the nervous system. Using this data, they identify gene families that are correlated with neuronal ageing, as well as a distinct set of genes that are upregulated in neurons of aged daf-2 mutants. This is particularly interesting as daf-2 mutants show both extended lifespans and healthier neurons in aged animals, reflected by better learning/memory in older animals compared with wild-type controls. Indeed, the knockdown of several of these upregulated genes resulted in poorer learning and memory. In addition, the authors showed that several genes upregulated during ageing in wild-type neurons also contribute to learning and memory; specifically knockdown of these genes in young animals resulted in improved memory. This indicates that (at least in this small number of cases), genes that show increased transcript levels with age in the nervous system somehow suppress memory, potentially by having damaging effects on neuronal health. 

Finally, from a resource perspective, the neuronal transcriptome provided here will be very useful for C. elegans researchers as it adds to other existing datasets by providing the transcriptome of older animals (animals at day 8 of adulthood) and demonstrating the benefits of performing tissue-specific RNAseq instead of whole-animal sequencing. 

Thank you!

The work presented here is of high quality and the authors present convincing evidence supporting their conclusions.

Thanks!

I only have a few comments/suggestions: 

(1) Do the genes identified to decrease learning/memory capacity in daf-2 animals (Figure 4d/e) also impact neuronal health? daf-2 mutant worms show delayed onset of age-related changes to neuron structure (Tank et al., 2011, J Neurosci). Does knockdown of the genes shown to affect learning also affect neuron structure during ageing, potentially one mechanism through which they modulate learning/memory? 

Thank you for this suggestion, which would be good for a future direction, particularly for genes that might have some relationship to previously-identified cellular structural process. The genes we tested here include dod-24, alh-2, mtl-1, F08H9,4, C44B7.5, hsp-12.3, hsp-12.6, and cpi-1, which are related to stress response, proteolysis inhibitor, metabolic, and innate immunity GO categories, thus associated with stress resistance, proteolysis, lipid metabolism processes; none are obvious choices for morphological effects.

However, it is worth noting that learning and memory decline much faster (Days 4-8) than morphological differences are observed (generally after Day 12-15). Moreover, those morphological differences have been studied primarily in mechanosensory neurons (touch neurons) rather than the chemosensory neurons that are involved in learning and memory, so additional genes may be required for those differences that we were not focusing on in thisi study.

(2) The learning and memory assay data presented in this study uses the butanone olfactory learning paradigm, which is well established by the same group. Have the authors tried other learning assays when testing for learning/memory changes after the knockdown of candidate genes? Depending on the expression pattern of these genes, they may have more or less of an effect on olfactory learning versus for example gustatory or mechanosensory-based learning. 

The reason that we use the butanone olfactory learning paradigm is because it is more similar to learning of information (neutral odorant association with positive cue (food)) – the kind of memory we would like to preserve in humans - rather than a stress-induced memory, such as starvation or pathogenesis-associated aversive learning paradigms, which are more like PTSD. (There is likely to be quite a bit of overlap in mechanism, however, including the role of genes such as magi-1 and casy-1, so it would not be surprising if many of these genes also were required for other learning paradigms.)

(3) I have a comment on the 'compensatory vs dysregulatory' model as stated by the authors on page 7. I understand that this model presents the two main options, but perhaps this is slightly too simplistic: the gene expression that rises during ageing may be detrimental for memory (= dysregulatory), but at the same time may also be beneficial for other physiological roles in other tissues (=compensatory). 

This is a good point, and we made the clarification that in the text: “There may be other scenarios in which a gene with multiple functions may be detrimental for some behaviors but beneficial for other physiological roles.”

Reviewer #3 (Public Review): 

Summary: 

In this manuscript, Weng et al. detect a neuron-specific transcriptome that regulates aging. The authors first profile neuron-specific responses during aging at a time point where a loss in memory function is present. They discover signatures unique to neurons which validate their pipeline and reveal the loss of neuron identity with age. For example, old neurons reduce the expression of genes related to synaptic function and neuropeptide signaling and increase the expression of chromatin regulators, insulin peptides, and glycoproteins. The authors discover the detrimental effect of selected upregulated genes (utx-1, ins-19, and nmgp-1) by knocking them down in the whole body and detecting improvement of short memory functions. They then use their pipeline to test neuronal profiles of long-lived insulin/IGF mutants. They discover that genes related to stress response pathways are upregulated upon longevity (e.g. dod-24, F08H9.4) and that they are required for improved neuron function in long-lived individuals. 

Strengths: 

Overall, the manuscript is well-written, and the experiments are well-described. The authors take great care to explain their reasoning for performing experiments in a specific way and guide the reader through the interpretation of the results, which makes this manuscript an enjoyable and interesting read. Using neuron-specific transcriptomic analysis in aged animals the authors discover novel regulators of learning and memory, which underlines the importance of cell-specific deep sequencing. The time points of the transcriptomic profiling are elegantly chosen, as they coincide with the loss of memory and can be used to specifically reveal gene expression profiles related to neuron function. The authors showcase on the dod-24 example how powerful this approach is. In long-lived insulin/IGF-1 receptor mutants body-wide dod-24 expression differs from neuron-specific profiles. Importantly, the depletion of dod-24 has an opposing effect on lifespan and learning memory. The dataset will provide a useful resource for the C. elegans and aging community. 

Thank you, we do hope people will find the data useful.

Weaknesses: 

While this study nicely describes the neuron-specific profiles, the authors do not test the relevance in a tissue-specific way. It remains unclear if modifying the responses only in neurons has implications for either memory or potentially for lifespan. The authors point to this in the text and refer to tissue-specific datasets. However, it is possible that the tissue-specific profile changes with age. The authors should consider mining publicly available cell-specific aging datasets and performing neuron-specific RNAi to test the functional relevance of the neuron-specific response. This would strengthen the importance of cell-specific profiling.

Thank you for your suggestions. As we have mentioned in the text, our candidate genes are either (1) only expressed in the neurons (alh-2 and F08H9.4), or they are only more highly expressed in daf-2 compared to wild type only in the nervous system (C44B7.5 or dod-24). Thus, the effect we see from knocking down these genes in daf-2 are likely neuron-specific. Additionaly, we performed our assays with neuron-sensitive RNAi strain CQ745: daf-2(e1370) III; vIs69 [pCFJ90(Pmyo-2::mCherry + Punc-119::sid-1)] V. It has been previously shown that neuronal expression of sid-1 decreases non-neuronal RNAi, suggesting that neurons expressing transgenic sid-1(+) served as a sink for dsRNA (Calixto et al., 2010). Thus, this neuron-sensitive RNAi is likely neuron-specific and our results is unlikely from knocking down these genes in non-neuronal tissues. However, we do acknowledge this issue.

To identify the expression pattern of these genes in a more cell-specific way in the adults, we examined the expression of our candidate genes that affected learning and memory, namely dod-24, F08H9.4, C44B7.9, alh-2, and mtl-1, in the Calico database (Roux et al., 2023). From that database, we can see that dod-24 is mainly expressed in the PHC and PVM neurons, and F08H9.4 is largely expressed in various neurons. Both have only slight expression outside the nervous system. C44B7.5 and mtl-1 are more broadly expressed, but C44B7.5 was not found to be differentially expressed in other tissues in daf-2, and mtl-1 only had a slight effect on learning and memory. Perhaps due to their sequencing depth and detection limit, Roux et al. didn’t detect alh-2 expression anywhere in their data.

Thus, the neuron-specific expression and daf-2 differential expression pattern of these genes indicate that the learning and memory improvement in aged daf-2 is unlikely due to neuronal non-autonomous effects.

To better address this concern (that for the genes that we found only expressed in the neurons, the neuron-confined expression may change with age) we examined the expression pattern change of these genes with age. As is shown below, from the Calico database, we can see that the expression in the nervous system persists, and even slightly increases, with age, thus age-related expression pattern change is not a concern to our analysis.

Author response image 2.

Author response image 3.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors): 

Most of my comments are in the public section. A few additional recommendations for the authors regarding the formatting/presentation: 

The presentation of Figure S6e-h in the introduction is somewhat confusing and feels out of order. If presented first, it should be S1. Otherwise, discussion of this figure should go at the end of the results section or in the discussion if appropriate. 

Thank you for pointing this out. We have moved the discussion of this figure to the Discussion section.

I do not see Figure S5 described in the text.

Good catch, thank you. We have added the descriptions for Figure S5 in the text.

In general, check the figures, figure legends, and how they are referenced in the text, particularly the supplemental figures and legends.

Minor comments:

There is a typo in the Figure 4 legend: Neuronal IIX should be IIS. 

Thanks for pointing this out. We have corrected it in the text.

Reviewer #2 (Recommendations For The Authors): 

• There are multiple instances throughout the manuscript where there are statements in brackets that provide justification or explanation for some of the approaches used. There is no reason for 'side note' brackets to be used. I suggest removing them and incorporating these statements into the narrative.

Thank you, we have now incorporated these points into the main text.

• Introduction: page 4 "here we RNA-sequenced FACS-isolated neurons" should be "here we performed RNA sequencing on FACS-isolated neurons...".

Thank you, we have changed the text accordingly.

• Figure 2A: I do not understand the legend for this panel "Tissue Query for wild-type genes expressed at higher levels in aged worms show lower nervous system and neuron prediction score." Please clarify.

We have clarified the Figure 2A legend:

(A)  Tissue prediction score for wild-type genes expressed at higher levels in aged worms.

• Page 8: "We previously observed that loss of single genes that play a role in complex behaviors like learning and memory can have a large impact on function 60, unlike the additive roles of longevity-promoting genes 11." - a large impact on what function?

Thank you for noting, we have clarified it in the text accordingly:

“We previously observed that for genes that play a role in complex behaviors like learning and memory, the loss of single genes can have a large impact on these complex behaviors 60, unlike the additive roles of longevity-promoting genes 11.”

• Next line "Therefore, one mechanism by which wild-type worms lose their function with age..." - again, what function?

Thank you for noting this, we have clarified the text to say we refer to the learning and memory functions.

• Page 9: "Thus, daf-2 mutants maintain their higher cognitive quality of life longer than wild-type worms, while daf-16;daf-2 mutants spend their whole lives without memory ability (Figure 3d), in contrast to claims that daf-2 mutants are less healthy than wild-type or daf-16 worms23." - since ref 23 did not perform any learning/memory tests, the definition of 'health' in ref 23 is different to 'cognitive health' as studied here. So the findings in this study are not 'in contrast' to ref 23 but rather add to these findings.

Learning and memory ability is an important function for a healthy individual, thus we would assert that indeed, cognitive health is an important part of the “health” of daf-2 worms. In ref 23, Bansal et al. claim that daf-2 worms are less healthy without assessing their learning and memory ability; their lack of data is an insufficient reason for us to remove our statement, as cognitive health is part of healthspan. Here we find that the “learning span” of daf-2 lasts at least proportionally if not longer than that of wild type. We have also previously shown that daf-2 worms also have longer maximum velocity span with age (Hahm et al., 2015), in direct contrast with Bansal et al.’s claim that daf-2 worms move less well and thus are less healthy – daf-2 worms simply stop sooner when presented with food and switch to feeding, due to their higher odr-10 levels. The Bansal paper continues to be frequently cited as finding that daf-2 mutants are less healthy than wild type, a claim for which we can still find no experimental evidence to support. Therefore, it is important that we make the point that daf-2 worms have extended cognitive health, which is part of health span.

• Page 13: I feel like the sentence "Furthermore, memory maintenance with age might require additional functions that were not previously uncovered in analyses of young animals" is both vague (what functions are referred to?) and a little bit obvious (obvious that age-related changes would not be revealed in analyses of young animals). Perhaps rephrase to make the desired point clearer? 

We have clarified the sentence in the text:

“Furthermore, memory maintenance with age might require additional genes that function in promoting stress resistance and neuronal resilience, which were not previously uncovered in analyses of young animals.”

Reviewer #1 (Public Review):

The authors perform RNA-seq on FACS isolated neurons from adult worms at days 1 and 8 of adulthood to profile the gene expression changes that occur with cognitive decline. Supporting data are included indicating that by day 7 of adulthood, learning and memory are reduced, indicating that this timepoint or after represents cognitively aged worms. Neuronal identity genes are reduced in expression within the cognitively aged worms, whereas genes involved in proteostasis, transcription/chromatin, and the stress response are elevated. A number of specific examples are provided, representing markers of specific neuronal subtypes, and correlating expression changes to the erosion of particular functions (e.g. motor neurons, chemosensory neurons, aversive learning neurons, etc).

To investigate whether upregulation of genes in neurons with age is compensatory or deleterious, the authors reduced expression of a set of three significantly upregulated genes and performed behavioral assays in young adults. In each case, reduction of expression improved memory, consistent with a model in which age-associated increases impair neuronal function.

The authors then characterize learning and memory in wild type, daf-2, and daf-2/daf-16 worms with age and find that daf-2 worms have an extended ability to learn for approximately 10 days longer that wild types. This was daf-16 dependent. Memory was extended in daf-2 as well, and strikingly, daf-2;daf-16 had no short term memory even at day 1. Transcriptomic analysis of FACS-sorted neurons was performed on the three groups at day 8. The authors focus their analysis on daf-2 vs. daf-2;daf-16 and present evidence that daf-2 neurons express a stress-resistance gene program. They also find small differences between the N2 and daf-2;daf-16 neurons, which correlate with the observed behavioral differences, though these differences are modest.

The authors tested eight candidate genes that were more highly expressed in daf-2 neurons vs. daf-2;daf-16 and showed that reduction of 2 and 5 of these genes impaired learning and memory, respectively, in daf-2 worms. This finding implicates specific neuronal transcriptional targets of IIS in maintaining cognitive ability in daf-2 with age, which, importantly, are distinct from those in young wild type worms.

Overall, this is a strong study with rigorously performed experiments. The authors achieved their aim of identifying transcriptional changes in neurons that underlie loss of learning and memory in C. elegans, and how cognition is maintained in insulin/IGF-1-like signaling mutants.

Reviewer #2 (Public Review):

Weng et al. perform a comprehensive study of gene expression changes in young and old animals, in wild-type and daf-2 insulin receptor mutants, in the whole animal and specifically in the nervous system. Using this data, they identify gene families that are correlated with neuronal ageing, as well as a distinct set of genes that are upregulated in neurons of aged daf-2 mutants. This is particularly interesting as daf-2 mutants show both extended lifespan and healthier neurons in aged animals, reflected by better learning/memory in older animals compared with wild-type controls. Indeed, knockdown of several of these upregulated genes resulted in poorer learning and memory. In addition, the authors showed that several genes upregulated during ageing in wild-type neurons also contribute to learning and memory; specifically, knockdown of these genes in young animals resulted in improved memory. This indicates that (at least in this small number of cases), genes that show increased transcript levels with age in the nervous system somehow suppress memory, potentially by having damaging effects on neuronal health.

Finally, from a resource perspective, the neuronal transcriptome provided here will be very useful for C. elegans researchers as it adds to other existing datasets by providing the transcriptome of older animals (animals at day 8 of adulthood) and demonstrating the benefits of performing tissue-specific RNAseq instead of whole-animal sequencing.

The work presented here is of high quality and the authors present convincing evidence supporting their conclusions. I only have a few comments/suggestions:

(1) Do the genes identified to decrease learning/memory capacity in daf-2 animals (Figure 4d/e) also impact neuronal health? daf-2 mutant worms show delayed onset of age-related changes to neuron structure (Tank et al., 2011, J Neurosci). Does knockdown of the genes shown to affect learning also affect neuron structure during ageing, potentially one mechanism through which they modulate learning/memory?

(2) The learning and memory assay data presented in this study uses the butanone olfactory learning paradigm, which is well established by the same group. Have the authors tried other learning assays when testing for learning/memory changes after knockdown of candidate genes? Depending on the expression pattern of these genes, they may have more or less of an effect on olfactory learning versus for e.g. gustatory or mechanosensory-based learning.

(3) A comment on the 'compensatory vs dysregulatory' model as stated by the authors on page 7 - I understand that this model presents the two main options, but perhaps this is slightly too simplistic: gene expression that rises during ageing may be detrimental for memory (= dysregulatory), but at the same time may also be beneficial other physiological roles in other tissues (=compensatory).

Comments on revised version:

I am satisfied with how the authors have addressed all my comments/suggestions.

Reviewer #3 (Public Review):

Summary

In this manuscript, Weng et al. identify the neuron specific transcriptome that impacts age dependent cognitive decline. The authors design a pipeline to profile neurons from wild type and long-lived insulin receptor/IGF-1 mutants using timepoints when memory functions are declining. They discover signatures unique to neurons which validates their approach. The authors identify that genes related to neuronal identity are lost with age in wild type worms. For example, old neurons reduce the expression of genes linked to synaptic function and neuropeptide signaling and increase the expression of chromatin regulators, insulin peptides and glycoproteins. Depletion of selected genes which are upregulated in old neurons (utx-1, ins-19 and nmgp-1) leads to improved short memory function. This indicates that some genes that increase with age have detrimental effects on learning and memory. The pipeline is then used to test neuronal profiles of long-lived insulin/IGF-1 daf-2 mutants. Genes related to stress response pathways are upregulated in long lived daf-2 mutants (e.g. dod-24, F08H9.4) and those genes are required for improved neuron function.

Strengths

The manuscript is well written, and the experiments are well described. The authors take great care to explain their reasoning for performing experiments in a specific way and guide the reader through the interpretation of the results, which makes this manuscript an enjoyable and interesting read. The authors discover novel regulators of learning and memory using neuron-specific transcriptomic analysis in aged animals, which underlines the importance of cell specific deep sequencing. The timepoints of the transcriptomic profiling are elegantly chosen, as they coincide with the loss of memory and can be used to specifically reveal gene expression profiles related to neuron function. The authors discuss on the dod-24 example how powerful this approach is. In daf-2 mutants whole-body dod-24 expression differs from neuron specific profiles, which underlines the importance of precise cell specific approaches. This dataset will provide a very useful resource for the C. elegans and aging community as it complements existing datasets with additional time points and neuron specific deep profiling.

Weakness

This study nicely describes the neuron specific profiles of aged long-lived daf-2 mutants. Selected neuronal genes that were upregulated in daf-2 mutants (e.g. F08H9.4, mtl-1, dod-24, alh-2, C44B7.5) decreased learning/memory when knocked down. However, the knock down of these genes was not specific to neurons. The authors use a neuron-sensitive RNAi strain to address this concern and acknowledge this caveat in the text. While it is likely that selected candidates act only in neurons it is possible that other tissues participate as well.

Reviewer #1 (Public Review):

Summary:

The authors utiilze the model organism C. elegans to interrogate cell non-autonomous signaling between GABAergic neurons and somatic tissues. They demonstrate that RNAi of isp-1 or spg-7 in GABAergic neurons leads to lifespan extension and improved healthspan (by resistance to paraquat or heat stress), which are dependent on the transcription factor daf-16/FOXO3a.

Strengths:

The authors are clear and straightforward in their study. They examine the healthspan of C. elegans at days 3, 6, and 9 to give a wide perspective on how the phenotypes changes with aging. They use two methods to specifically knockdown isp-1 or spg-7 in GABAergic neurons: (1) a previously published rde-1 mutant that has rde-1 and sid-1 restored only in GABAergic neurons and (2) a novel model uses a sid-1 mutant that makes dsRNA of isp-1 or spg-7 in GABAergic neurons. They use multiple methods to examine healthspan. They identified daf-16/FOXO3a as the mechanism of their phenotype and ruled out other transcription factors. The authors do not use FUdR in their studies, which is known to confound experiments.

Weaknesses:

(1) Incomplete validation of GABAergic knockdown. The study relies on the specific knockdown of isp-1 or spg-7 in GABAergic neurons, but in the opinion of this reviewer, the authors do not adequately validate their models to demonstrate GABAergic specificity. For the previously published rde-1 mutant model, a simple validation of specific knockdown of GFP in GFP-labeled GABAergic neurons should be included. They should also show that GFP RNAi would not be effective in knocking down intestinal GFP, for example.

Their second model is poorly explained and not validated and this reviewer could not find similar previously published models of its kind. This model claims that dsRNA of isp-1 was made in the GABAergic neurons of a sid-1 mutant, but no evidence is shown to support this claim. The authors point to changes in phenotypes such as lifespan extension and reduced lipofuscin in the intestines as proof that knockdown is occurring in the GABAergic neurons, but this is indirect evidence. Rigorous validation of this model is needed, especially if it is the first model of its kind.

(2) Lifespan. The control lifespans using the rde-1 mutants are very short-lived and no explanation for this is provided (eg. Figure 1D, E). The authors use two RNAis in their lifespan with daf-16 and isp-1. For their controls, they should use empty vector mixed with isp-1, not only isp-1 RNAi.

(3) Cell non-autonomous effects. The claims that GABAergic mitochondrial dysfunction have effects on somatic tissues is weak. More specific tests on somatic stress resistance are warranted for their claims. Better quality images of intestinal mitochondria are needed. Examining additional tissues, such as muscle, would also strengthen their claims. For example, they could examine muscle mitochondria and determine if muscle strength is improved in their models.

(4) Dependence on daf-16/FOXO3a. The authors show that loss of daf-16 reverses the lifespan and healthspan effects in their model. Next, they show that loss of daf-16 reverses the effects of isp-1 in the intestines and in the germline. However, they only show the daf-16 mutant data and not the positive control (EV and isp-1 alone), which should be included. Furthermore, the phenotypes they examine are only a subset of somatic phenotypes, and this reviewer would be more convinced with the additional controls and with more parameters examined.

eLife assessment

This study interrogates cell non-autonomous signaling between GABAergic neurons and somatic tissues in the nematode C. elegans. The authors report that mitochondrial stress in only GABAergic neurons extends lifespan and improves healthspan, phenotypes that are dependent on the transcription factor daf-16/FOXO3a. However, while the findings may be valuable to furthering our understanding of neuronal control of aging and health, the current evidence is incomplete and additional experiments are needed to support their claims.

Reviewer #2 (Public Review):

Summary:

In this work, the authors show that GABAergic neurons play a role in sensing mitochondrial stress and regulating organismal aging. Thus, disrupting the mitochondrial mitochondria function in GABAergic neurons induces resistance to thermal and paraquat stresses, promotes longevity, and affects reproduction. This mechanism is regulated by the iron-sulfur subunit of complex III of the mitochondrial electron transport chain, ISP-1, and a mitochondrial quality control m-AAA protease, SPG-7, which in turn requires DAF-16/FoxO activity in GABAergic neurons.

Strengths:

A strength of this work is that the authors identify the specific site where mitochondrial stress promotes health and longevity, i.e., GABAergic neurons. In addition, the paper corroborates the findings with the appropriate experiments. How neuronal regulation of mitochondrial function impacts systemic health and aging is of interest to cell biology and neuroscience fields.

Weaknesses:

The entire paper is based on tissue-specific RNAi in GABAergic neurons, which was achieved using two different conditions of RNAi (although not for all experiments). However, multiple studies have shown deficiencies in the tissue-specific RNAi in C. elegans, especially for the rde-1(ne219) mutant used in this study. Therefore, it is necessary to repeat critical experiments by rescuing the isp-1 or spg-7 mutants in GABAergic neurons. Additionally, it is clear in the paper that perturbing mitochondrial function requires DAF-16/FoxO activity in GABAergic neurons to promote longevity, yet the downstream cellular pathways are not described.

Reviewer #3 (Public Review):

Summary:

This manuscript describes RNAi depletion of isp-1 or spg-7 in the GABAergic neurons of C. elegans leads to: lifespan extension; increased resistance to paraquat oxidative stress and heat stress; decreased brood size and mitotic germ cell numbers in the gonad and increased DNA aggregates in the oocytes; increased mitochondrial membrane potential, ATP levels, mitochondrial mass, mitochondrial DNA copies, mitochondrial DNA polymerase gamma polg-1 levels, and decreased ROS levels. The authors further show that daf-16 is necessary for GABAergic depletion of isp-1 mediated lifespan extension, stress resistance, increased mitochondrial membrane potential, mitochondrial mass and DNA copies, and decreased brood size. Unc-25 for GABA synthesis, unc-31 for neuropeptide secretion, and flp-13 neuropeptide are all in the same pathway of isp-1 RNAi in GABAergic neurons for lifespan extension and stress resistance.

Strengths:

The topic is interesting and relatively novel in terms of GABAergic mitochondrial dysfunction. The data provided support the conclusions well.

Weaknesses:

The mechanistic evidence needs to be improved substantially.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Comments to the Author):

Summary:

In this study, Xie and colleagues aimed to explore the function and potential mechanisms of the gut microbiota in a hamster model of severe leptospirosis. The results demonstrated that Leptospira infection was able to cause intestine damage and inflammation. Leptospira infection promoted an expansion of Proteobacteria, increased gut barrier permeability, and elevated LPS levels in the serum. Thus, they proposed an LPS-neutralization therapy which improved the survival rate of moribund hamsters combined with antibody therapy or antibiotic therapy.

Strengths:

The work is well-designed and the story is interesting to me. The gut microbiota is essential for immunity and systemic health. Many life-threatening pathogens, such as SARS-CoV-2 and other gut-damaged infection, have the potential to disrupt the gut microbiota in the later stages of infection, causing some harmful gut microbiota-derived substances to enter the bloodstream. It is emphasized that in addition to exogenous pathogenic pathogens, harmful substances of intestinal origin should also be considered in critically ill patients.

Weaknesses:

Q1: There are many serotypes of Leptospira, it is suggested to test another pathogenic serotype of Leptospira to validate the proposed therapy.

That’s a constructive suggestion. We have tested another pathogenic serotype of Leptospira (L. interrogans serovar Autumnalis strain 56606) to verify the LPS-neutralization therapy combined with antibiotic therapy (Supplementary Fig. S9B). The results showed that the combination of the LPS-neutralization therapy with antibody therapy or antibiotic therapy also significantly improved the survival rate of hamsters infected by 56606.

Q2: Authors should explain why the infective doses of leptospires was not consistent in different study.

Thank you for your comment. To examine the role of the gut microbiota on acute leptospirosis, the infective doses of leptospires was chosen for 106, while in other sections of the study, the infective doses of leptospires was chosen for 107. In fact, we also used 107 leptospires to infect hamsters, however, the infective doses of 107 leptospires might be overdose, there was no significant difference on the survival rate between the control group and the Abx-treated group. A previous study also highlighted that the infective doses of leptospires was important in the investigating the sex on leptospirosis, as male hamsters infected with L. interrogans are more susceptible to severe leptospirosis after exposure to lower infectious doses than females (103 leptospires but not 104 leptospires) (1).

Reference

(1) GOMES C K, GUEDES M, POTULA H H, et al. Sex Matters: Male Hamsters Are More Susceptible to Lethal Infection with Lower Doses of Pathogenic Leptospira than Female Hamsters (J). Infect Immun, 2018, 86(10).

Q3: In the discussion section, it is better to supplement the discussion of the potential link between the natural route of infection and leptospirosis.

Thank for your suggestion. We have supplemented it in the discussion (line 523-527 in the track change PDF version).

Q4: Line 231, what is the solvent of thioglycolate?

We have supplemented it in the manuscript (line 242-243 in the track change PDF version).

Q5: Lines 962-964, there are some mistakes which are not matched to Figure 7.

Thank you for pointing that out, we have corrected it in the manuscript.

Reviewer #2 (Comments to the Author):

Summary:

Severe leptospirosis in humans and some mammals often meet death in the endpoint. In this article, authors explored the role of the gut microbiota in severe leptospirosis. They found that Leptospira infection promoted a dysbiotic gut microbiota with an expansion of Proteobacteria and LPS neutralization therapy synergized with antileptospiral therapy significantly improved the survival rates in severe leptospirosis. This study is well-organized and has potentially important clinical implications not only for severe leptospirosis but also for other gut-damaged infections.

Weaknesses:

Q1: In the Introduction section and Discussion section, the authors should describe and discuss more about the differences in the effect of Leptospira infection between mice and hamsters, so that the readers can follow this study better.

Thank you for your suggestion, we have supplemented it in the manuscript (line 62-66 in the track change PDF version).

Q2: Lines 92-95, the authors should explain why they chose two different routines of infection.

Thank you for your comment, we have explained it in the manuscript (line 100 in the track change PDF version).

Q3: Line 179-180, the concentration of PMB and Dox is missed, and 0.016 μg/L is just ok.

We have corrected it in the manuscript.

Q4: "μL" or "μl" and "mL" or "ml' should be uniform in the manuscript.

Thank you for your suggestion, we have revised it in the manuscript.

Q5: In the culture of primary macrophages, how many cells are inoculated in the plates should be described clearly.

We have supplemented it in the manuscript (line 250 in the track change PDF version).

Q6: Line 271, it is better to list primers used for leptospiral detection in the text. Because it allows readers to find the information they need more directly.

Thank you for your suggestions, we have supplemented it in the manuscript (line 281-284 in the track change PDF version).

Q7: Line 366-369, Lactobacillus seems to be a kind of key bacteria during Leptospira infection. A previous study (doi: 10.1371/journal.pntd.0005870) also demonstrated that pre-treatment with Lactobacillus plantarum prevented severe pathogenesis in mice. The authors should discuss the potential probiotic for leptospirosis prevention.

We have discussed it in the manuscript (line 564-566 in the track change PDF version).

Q8: Lines 450-451, not all concentrations of fecal filtration from two groups upregulated all gene expression mentioned in the text, the authors should correct it.

Thank you for pointing that out, we have corrected it in the manuscript (line 461-462 in the track change PDF version).

Reviewer #3 (Comments to the Author):

Summary:

This is a well-prepared manuscript that presented interesting research results. The only defect is that the authors should further revise the English language.

Strengths:

The omics method produced unbiased results.

Weaknesses:

Q1: LPS neutralization is not a new method for treating leptospiral infection.

Thank you for your comment. Yes, LPS neutralization is not a new method for treating leptospiral infection, most of which might focus on leptospiral LPS. In addition, Leptospira seemed to be naturally resistant to polymyxin B (1). Recently, neutralizing gut-derived LPS was applied in other diseases which significantly relieved diseases (2-3). In this study, we found that Leptospira infection promoted an expansion of Proteobacteria, increased gut barrier permeability, and elevated LPS levels in the serum. Thus, we proposed an LPS-neutralization therapy which improved the survival rate of moribund hamsters combined with antibody therapy or antibiotic therapy.

Reference

(1) LIEGEON G, DELORY T, PICARDEAU M. Antibiotic susceptibilities of livestock isolates of leptospira (J). Int J Antimicrob Agents, 2018, 51(5):693-699.

(2) MUNOZ L, BORRERO M J, UBEDA M, et al. Intestinal Immune Dysregulation Driven by Dysbiosis Promotes Barrier Disruption and Bacterial Translocation in Rats With Cirrhosis (J). Hepatology, 2019, 70(3):925-938.

(3) ZHANG X, LIU H, HASHIMOTO K, et al. The gut-liver axis in sepsis: interaction mechanisms and therapeutic potential (J). Crit Care, 2022, 26(1):213.

Q2: The authors should further revise the English language used in the text.

Thank you for your suggestion, our manuscript has been polished by American Journal Experts (certificate number: 81C8-C5C1-9D5D-109D-3F23).

eLife assessment

The gut microbiota influences many infectious diseases; however, its role Leptospirosis remains unclear. In this fundamental work, Xie et al. use a hamster model to show that Leptospira infection leads to gut pathology, an altered gut microbiota, and increased translocation. A combined use of antibiotics and LPS neutralization prolonged survival, providing a potential new therapeutic approach. This study utilizes compelling methods to provide new insights into this emerging disease, which could be dissected further in future studies aimed at gaining mechanistic insight and assessing the translational relevance of these discoveries.

Reviewer #1 (Public Review)

Summary:

In this study, Xie and colleagues aimed to explore the function and potential mechanisms of the gut microbiota in a hamster model of severe leptospirosis. The results demonstrated that Leptospira infection was able to cause intestine damage and inflammation. Leptospira infection promoted an expansion of Proteobacteria, increased gut barrier permeability, and elevated LPS levels in the serum. Thus, they proposed an LPS-neutralization therapy which improved the survival rate of moribund hamsters combined with antibody therapy or antibiotic therapy.

Strengths:

The work is well-designed and the story are interesting to me. The gut microbiota is essential for immunity and systemic health. Many life-threatening pathogens, such as SARS-CoV-2 and other gut-damaged infection, have the potential to disrupt the gut microbiota in the later stages of infection, causing some harmful gut microbiota-derived substances to enter the bloodstream. It is emphasized that in addition to exogenous pathogenic pathogens, harmful substances of intestinal origin should also be considered in critically ill patients.

Reviewer #2 (Public Review):

Severe leptospirosis in humans and some mammals often meet death in the endpoint. In this article, authors explored the role of the gut microbiota in severe leptospirosis. They found that Leptospira infection promoted a dysbiotic gut microbiota with an expansion of Proteobacteria and LPS neutralization therapy synergized with antileptospiral therapy significantly improved the survival rates in severe leptospirosis. This study is well-organized and has potentially important clinical implications not only for severe leptospirosis but also for other gut-damaged infections.

Reviewer #3 (Public Review):

Summary:

This is a well prepared manuscript which presented interesting research result.

Strengths:

The omics method produced unbiased results.

Weaknesses:

LPS neutralization is not new method for treating leptospiral infection.

eLife assessment

This study introduces a useful deep learning-based algorithm that tracks animal postures with reduced drift by incorporating transformers for more robust keypoint detection. The efficacy of this new algorithm for single-animal pose estimation was demonstrated through comparisons with two popular algorithms. However, the analysis is incomplete and would benefit from comparisons with other state-of-the-art methods and consideration of multi-animal tracking.

Reviewer #1 (Public Review):

Summary:

In this paper, the authors introduce a new deep learning-based algorithm for tracking animal poses, especially in minimizing drift effects. The algorithm's performance was validated by comparing it with two other popular algorithms, DeepLabCut and LEAP.

Strengths:

The authors showcased the effectiveness of their new algorithm in a systematic manner, covering individual levels of mice, drosophilas, macaques, and multi-animal poses.

Weaknesses:

(1) The accessibility of this tool for biological research is not clearly addressed, despite its potential usefulness. Researchers in biology often have limited expertise in deep learning training, deployment, and prediction. A detailed, step-by-step user guide is crucial, especially for applications in biological studies.

(2) The proposed algorithm focuses on tracking and is compared with DLC and LEAP, which are more adept at detection rather than tracking.

Reviewer #2 (Public Review):

Summary:

The authors present a new model for animal pose estimation. The core feature they highlight is the model's stability compared to existing models in terms of keypoint drift. The authors test this model across a range of new and existing datasets. The authors also test the model with two mice in the same arena. For the single animal datasets the authors show a decrease in sudden jumps in keypoint detection and the number of undetected keypoints compared with DeepLabCut and SLEAP. Overall average accuracy, as measured by root mean squared error, generally shows similar but sometimes superior performance to DeepLabCut and better performance compared to SLEAP. The authors confusingly don't quantify the performance of pose estimation in the multi (two) animal case instead focusing on detecting individual identity. This multi-animal model is not compared with the model performance of the multi-animal mode of DeepLabCut or SLEAP.

Strengths:

The major strength of the paper is successfully demonstrating a model that is less likely to have incorrect large keypoint jumps compared to existing methods. As noted in the paper, this should lead to easier-to-interpret descriptions of pose and behavior to use in the context of a range of biological experimental workflows.

Weaknesses:

There are two main types of weaknesses in this paper. The first is a tendency to make unsubstantiated claims that suggest either model performance that is untested or misrepresents the presented data, or suggest excessively large gaps in current SOTA capabilities. One obvious example is in the abstract when the authors state ADPT "significantly outperforms the existing deep-learning methods, such as DeepLabCut, SLEAP, and DeepPoseKit." All tests in the rest of the paper, however, only discuss performance with DeepLabCut and SLEAP, not DeepPoseKit. At this point, there are many animal pose estimation models so it's fine they didn't compare against DeepPoseKit, but they shouldn't act like they did. Similar odd presentation of results are statements like "Our method exhibited an impressive prediction speed of 90{plus minus}4 frames per second (fps), faster than DeepLabCut (44{plus minus}2 fps) and equivalent to SLEAP (106{plus minus}4 fps)." Why is 90{plus minus}4 fps considered "equivalent to SLEAP (106{plus minus}4 fps)" and not slower? I agree they are similar but they are not the same. The paper's point of view of what is "equivalent" changes when describing how "On the single-fly dataset, ADPT excelled with an average mAP of 92.83%, surpassing both DeepLabCut and SLEAP (Figure 5B)" When one looks at Figure 5B, however, ADPT and DeepLabCut look identical. Beyond this, oddly only ADPT has uncertainty bars (no mention of what uncertainty is being quantified) and in fact, the bars overlap with the values corresponding to SLEAP and DeepPoseKit. In terms of making claims that seem to stretch the gaps in the current state of the field, the paper makes some seemingly odd and uncited statements like "Concerns about the safety of deep learning have largely limited the application of deep learning-based tools in behavioral analysis and slowed down the development of ethology" and "So far, deep learning pose estimation has not achieved the reliability of classical kinematic gait analysis" without specifying which classical gait analysis is being referred to. Certainly, existing tools like DeepLabCut and SLEAP are already widely cited and used for research.

The other main weakness in the paper is the validation of the multi-animal pose estimation. The core point of the paper is pose estimation and anti-drift performance and yet there is no validation of either of these things relating to multi-animal video. All that is quantified is the ability to track individual identity with a relatively limited dataset of 10 mice IDs with only two in the same arena (and see note about train and validation splits below). While individual tracking is an important task, that literature is not engaged with (i.e. papers like Walter and Couzin, eLife, 2021: https://doi.org/10.7554/eLife.64000) and the results in this paper aren't novel compared to that field's state of the art. On the other hand, while multi-animal pose estimation is also an important problem the paper doesn't engage with those results either. The two methods already used for comparison in the paper, SLEAP and DeepPoseKit, already have multi-animal modes and multi-animal annotated datasets but none of that is tested or engaged with in the paper. The paper notes many existing approaches are two-step methods, but, for practitioners, the difference is not enough to warrant a lack of comparison. The authors state that "The evaluation of our social tracking capability was performed by visualizing the predicted video data (see supplement Videos 3 and 4)." While the authors report success maintaining mouse ID, when one actually watches the key points in the video of the two mice (only a single minute was used for validation) the pose estimation is relatively poor with tails rarely being detected and many pose issues when the mice get close to each other.

Finally, particularly in the methods section, there were a number of places where what was actually done wasn't clear. For example in describing the network architecture, the authors say "Subsequently, network separately process these features in three branches, compute features at scale of one-fourth, one-eight and one-sixteenth, and generate one-eight scale features using convolution layer or deconvolution layer." Does only the one-eight branch have deconvolution or do the other branches also? Similarly, for the speed test, the authors say "Here we evaluate the inference speed of ADPT. We compared it with DeepLabCut and SLEAP on mouse videos at 1288 x 964 resolution", but in the methods section they say "The image inputs of ADPT were resized to a size that can be trained on the computer. For mouse images, it was reduced to half of the original size." Were different image sizes used for training and validation? Or Did ADPT not use 1288 x 964 resolution images as input which would obviously have major implications for the speed comparison? Similarly, for the individual ID experiments, the authors say "In this experiment, we used videos featuring different identified mice, allocating 80% of the data for model training and the remaining 20% for accuracy validation." Were frames from each video randomly assigned to the training or validation sets? Frames from the same video are very correlated (two frames could be just 1/30th of a second different from each other), and so if training and validation frames are interspersed with each other validation performance doesn't indicate much about performance on more realistic use cases (i.e. using models trained during the first part of an experiment to maintain ids throughout the rest of it.)

eLife assessment

This study provides valuable insights into the mechanism of axonal directional changes, utilizing sLNv neurons as a model. The data were collected and analysed using solid methodology, although the conceptual framing of the work and contextualization of the results require revision and reassessment. The study holds potential interest for neurobiologists focusing on axonal growth and development.

Reviewer #1 (Public Review):

Summary:

The mechanisms of how axonal projections find their correct target requires the interplay of signalling pathways, and cell adhesion that act over short and long distances. The current study aims to use the small ventral lateral clock neurons (s-LNvs) of the Drosophila clock circuit as a model to study axon projections. These neurons are born during embryonic stages and are part of the core of the clock circuit in the larval brain. Moreover, these neurons are maintained through metamorphosis and become part of the adult clock circuit. The authors use the axon length by means of anti-Pdf antibody or Pdf>GFP as a read-out for the axonal length. Using ablation of the MB- the overall target region of the s-LNvs, the authors find defects in the projections. Next, by using Dscam mutants or knock-down they observe defects in the projections. Manipulations by the DNs - another group of clock neurons- can induce defects in the s-LNvs axonal form, suggesting an active role of these neurons in the morphology of the s-LNvs.

Strengths:

The use of Drosophila genetics and a specific neural type allows targeted manipulations with high precision.

Proposing a new model for a small group of neurons for axonal projections allows us to explore the mechanism with high precision.

Weaknesses:

It is unclear how far the proposed model can be seen as developmental.

The study of changes in fully differentiated and functioning neurons may affect the interpretation of the findings.

Reviewer #2 (Public Review):

Summary:

The paper from Li et al shows a mechanism by which axons can change direction during development. They use the sLNv neurons as a model. They find that the appearance of a new group of neurons (DNs) during post-embryonic proliferation secretes netrins and repels horizontally towards the midline, the axonal tip of the LNvs.

Strengths:

The experiments are well done and the results are conclusive.

Weaknesses:

The novelty of the study is overstated, and the background is understated. Both things need to be revised.

eLife assessment

This important manuscript uses circuit mapping, chemogenetics, and optogenetics to demonstrate a novel hippocampal lateral septal circuit that regulates social novelty behaviours and shows that downstream of the hippocampal septal circuit, septal projections to the ventral tegmental area are necessary for general novelty discrimination. The strength of the evidence supporting the claims is convincing but would be strengthened by the inclusion of additional functional assays. The work will be of interest to systems and behavioural neuroscientists who are interested in the brain mechanisms of social behaviours.

Reviewer #1 (Public Review):

Summary:

The study investigated the neural circuits underlying social novelty preference in mice. Using viral circuit tracing, chemogenetics, and optogenetics in the vHPC, LS, and VTA, the authors found that vHPC to LS projections may contribute to the salience of social novelty investigations. In addition, the authors identify LS projections to the VTA involved in social novelty and familiar food responses. Finally, via viral tracing, they demonstrate that vHPC-LS neurons may establish direct monosynaptic connections with VTA dopaminergic neurons. The experiments are well-designed, and the conclusions are mostly very clear. The manuscript is well-written and logically organized, and the content will be of interest to specialists in the field and to the broad readership of the journal.

Strengths:

(1) The vHPC has been involved in social memory for novel and familiar conspecifics. Yet, how the vHPC conveys this information to drive motivation for novel social investigations remains unclear. The authors identified a pathway from the vHPC to the LS and eventually the VTA, that may be involved in this process.

(2) Mice became familiar with a novel conspecific by co-housing for 72h. This represents a familiarization session with a longer duration as compared to previous literature. Using this new protocol, the authors found robust social novelty preference when animals were given a choice between a novel and familiar conspecific.

(3) The effects of vHPC-LS inhibition are specific to novel social stimuli. The authors included novel food and novel object control experiments and those were not affected by neuronal manipulations.

(4) For optogenetic studies, the authors applied closed-loop photoinhibition only when the animals investigated either the novel conspecific or the familiar. This optogenetic approach allowed for the investigation of functional manipulations to selective novel or familiar stimuli approaches.

Weaknesses:

(1) The abstract and the overall manuscript pose that the authors identified a novel vHPC-LS-VTA pathway that is necessary for mice to preferentially investigate novel conspecifics. However, the authors assessed the functional manipulations of vHPC-LS and LS-VTA circuits independently and the sentence could be misleading. Therefore, a viral strategy specifically designed to target the vHPC-LS-VTA circuit combined with optogenetic/chemogenetic tools and behavior may be necessary for the statement of this conclusion.

(2) The authors combined males and females in their analysis, as neural circuit manipulation affected novelty discrimination ratios in both sexes. However, supplementary Figure 1 demonstrates the chemogentic inhibition of vHPC-LS circuit may cause stronger effects in male mice as compared to females.

(3) In most experiments, the same animals were used for social novelty preference, for food or object novelty responses but washout periods between experiments are not mentioned in the methods section. In this line, the authors did not mention the time frame between the closed-loop optogenetic experiments that silenced the vHPC-LS only during familiar and then only novel social investigations. When using the same animals tested for social experiments in the same context there may be an effect of context-dependent social behaviors that could affect future outcomes.

(4) All the experiments were performed in a non-cell-type-specific manner. The viral strategies used targeted multiple neuronal subpopulations that could have divergent effects on social novelty preference. This constraint could be added in the discussion section.

(5) The authors' assumptions were all based on experiments of necessity. The authors could use an experiment of sufficiency by targeting for instance the LS-VTA circuit and assess if animals reduce novel social investigations with LS-VTA photostimulation.

Reviewer #2 (Public Review):

Summary:

Rashid and colleagues demonstrate a novel hippocampal lateral septal circuit that is important for social recognition and drives the exploration of novel conspecifics. Their study spans from neural tracing to close-loop optogenetic experiments with clever controls and conditions to provide compelling evidence for their conclusion. They demonstrate that downstream of the hippocampal septal circuit, septal projections to the ventral tegmental area are necessary for general novelty discrimination. The study opens an avenue to study these circuits further to uncover the plasticity and synaptic mechanisms regulating social novelty preference.

Strengths:

Chemogenetic and optogenetic experiments have excellent behavioral controls. The synaptic tracing provides important information that informs the narrative of experiments presented and invites future studies to investigate the effects of septal input on dopaminergic activity.

Weaknesses:

There are unclear methodological important details for circuit manipulation experiments and analyses where multiple measures are needed but missing. Based on the legends, the chemogenetic experiment is done in a within-animal design. That is the same mouse receives SAL and CNO. However, the data is not presented in a within-animal manner such that we can distinguish if the behavior of the same animal changes with drug treatment. Similarly, the methods specify that the optogenetic manipulations were done in three different conditions, but the analyses do not report within-animal changes across conditions nor account for multiple measures within subjects. Finally, it is unclear if the order of drug treatment and conditions were counterbalanced across subjects.

eLife assessment

This study provides solid evidence of coordinated spiking activity of neurons in the anterior cingulate cortex (ACC), and correlated activity in the CA1 subregion of the hippocampus, during observational learning. The authors also show coordinated ACC-CA1 neural activity during rest periods prior to the performance of the observationally learned task. The important findings advance the field's understanding of neural mechanisms underlying social learning.

Reviewer #1 (Public Review):

Summary:

In this manuscript by Mou and Ji, the authors describe the correlation between firing rates in the ACC with that of CA1 ensembles during observational learning. Their main findings include trajectory selective (observational) responses in ACC, correlations between ACC and CA1 place cells for specific trajectories during observational learning, and correlations between ACC and CA1 place cells that are reactivated during SWRs, specifically during CA1 replays.

Strengths:

The study is well designed, the data presented is very clear and the conclusions are appropriate regarding their results. The study is novel and of high relevance for the understanding of social learning.

Weaknesses:

Lack of physiological characterization of the neurons that could have been included, such as regular firing rates of neurons in different regions (not only constrained to behavioral landmarks) or PSTH during sharp-wave ripples. The first experiment, NMDA blockage, is a bit disconnected from the rest of the results. Perhaps clarifying in the text a bit further that this proves that ACC is necessary for social learning would help.

Reviewer #2 (Public Review):

Summary:

In the manuscript, Xiang Mou and Daoyun JI investigate how ACC neurons activated by observational learning communicate with the hippocampus. They assess this line of communication through a complex behavioral technique, in vivo electrophysiology, pharmacological approaches, and data analytical techniques. Firstly, the authors find that observational performance is dependent on the ACC, and that the ACC possesses neurons that show side selectivity (trajectory-related) in both the observation box when shuttling to reward, and during subsequent maze running, shuttling to the corresponding same side for reward. The side-selective activation appears stronger for correct trials compared to error trials specifically during observation of Demo rats. They compare how the CA1 of the hippocampus encodes these two environments and find that ACC side-selective neurons show a correlation with side-selective CA1 ensembles during maze behavior, water consumption, and sharp-wave ripples.

Strengths:

Overall, the paper provides strong evidence that ACC neurons are activated by observational learning and that this activation seems to be correlated with CA1 activity.

Weaknesses:

Concerns, however, surround the strength of evidence that links ACC and CA1 activity during observational learning. Only weak correlations between the two regions are shown, and it is unclear if the ACC may lead to CA1 activity or vice versa. It is possible that these processes reflect two parallel pathways. Without manipulation of ACC, it is difficult to assess whether ACC activity influences hippocampal replay.

Reviewer #3 (Public Review):

Summary:

Mou and Ji investigated neuro-computational mechanisms behind observational spatial learning in rats and reported several signs of functional coupling between the ACC and CA1 at the single neuron level. Using multi-site tetrode recording, they found that ACC cells encoding a path on a maze were activated while a rat observed another rat took that path. This activation was also correlated with the activation of CA1 cells encoding the same path and facilitated their replay during sharp-wave ripples (SWRs) before the recording rat ran on the maze by itself. These activity patterns were associated with correct path choice during self-running and were absent in control conditions where the recording rat did not learn the correct choice through observations. Based on these findings, the authors argue that ACC cells capture the critical information during observation to organize hippocampal cell activity for subsequent spatial decisions.

Strengths:

The authors used multiple outcome measures to build a strong case for path-specific spike coordination between ACC and CA1 cells. The analyses were conducted carefully, and proper control measures were used to establish the statistical significance of the detected effects. The authors also demonstrated tight correlations between the spike coordination patterns and the successful use of observed information for future decisions.

Weaknesses:

(1) As evidence for the activation of path information in the ACC during observation, the authors showed positive correlations between firing rates during observation and those during self-running. This argument will be solidified if the authors use a decoding approach to demonstrate the activation of path-selective ACC ensemble activity patterns during observation. This approach will also open the door to uncovering how the activation of ACC path representation is related to the moment-to-moment position of the demonstrator rat and whether it is coupled with the timing of SWRs.

(2) The authors argued that the ACC biases the content of awake replay in CA1 during SWRs in the observation period. The reviewer wonders if a similar bias also occurs during SWRs in the self-run period (i.e., water consumption after the correct choice). This analysis will be helpful in testing if the biased replay occurs due to the need to convert observed information into future choices.

(3) Although the authors demonstrated the necessity of the ACC for the task, it still remains to be determined firing coordination between the ACC and CA1 during observation is necessary for the correct path choice during self-runs. Some discussion on this point, along with future direction, would be beneficial for readers.

eLife assessment

This study presents valuable insight into a nuclear-encoded transcription factor network and the role of one transcription factor Clifford in mitochondrial biogenesis. The experimental design, data collection, and analyses are solid. Addressing a few points related to mitochondrial and ETC biogenesis will further strengthen the study.

Reviewer #1 (Public Review):

Summary:

In this manuscript, Zhang et al. report a genetic screen to identify novel transcriptional regulators that could coordinate mitochondrial biogenesis. They performed an RNAi-based modifier screen wherein they systematically knocked down all known transcription factors in the developing Drosophila eye, which was already sensitised and had decreased mitochondrial DNA content. Through this screen, they identify CG1603 as a potential regulator of mitochondrial content. They show that protein levels of mitochondrial proteins like TFAM, SDHA, and other mitochondrial proteins and mtDNA content are downregulated in CG1603 mutants. RNA-Seq and ChIP-Seq further show that CG1603 binds to the promoter regions of several known nuclear-encoded mitochondrial genes and regulates their expression. Finally, they also identified YL-1 as an upstream regulator of CG1603. Overall, it is a very important study as our understanding of the regulation of mitochondrial biogenesis remains limited across metazoans. Most studies have focused on PGC-1α as a master regulator of mitochondrial biogeneis, which seems a context-dependent regulator. Also, PGC-1α mediated regulation could not explain the regulation of 1100 genes that are required for mitochondrial biogenesis. Therefore, identifying a new regulator is crucial for understanding the overall regulation of mitochondrial biogenesis.

Reviewer #2 (Public Review):

Summary:

In this study, the authors aim to identify the nuclear genome-encoded transcription factors that regulate mtDNA maintenance and mitochondrial biogenesis. They started with an RNAi screening in developing Drosophila eyes with reduced mtDNA content and identified a number of putative candidate genes. Subsequently, using ChIP-seq data, they built a potential regulatory network that could govern mitochondrial biogenesis. Next, they focused on a candidate gene, CG1603, for further characterization. Based on the expression of different markers, such as TFAM and SDHA, in the RNAi and OE clones in the midgut cells, they argue that CG1603 promotes mitochondrial biogenesis and the expression of ETC complex genes. Then, they used a mutant of CG1603 and showed that both mtDNA levels and mitochondrial protein levels were reduced. Using clonal analyses, they further show a reduction in mitochondrial biogenesis and membrane potential upon loss of CG1603. They made a reporter line of CG1603, showed that the protein is localized to the mitochondria, and binds to polytene chromosomes in the salivary gland. Based on the RNA-seq results from the mutants and the ChIP data, the authors argue that the nucleus-encoded mitochondrial genes that are downregulated >2 folds in the CG1603 mutants and that are bound by CG1603 are related to ETC biogenesis. Finally, they show that YL-1, another candidate in the network, is an upstream regulator of CG1603.

Strengths:

This is a valuable study, which identifies a potential regulator and a network of nucleus-encoded transcription factors that regulate mitochondrial biogenesis. Through in-vivo and in-vitro experimental evidence, the authors identify the role of CG1603 in this process. The screening strategy was smart, and the follow-up experiments were nicely executed.

Weaknesses:

Some additional experiments showing the effects of CG1603 loss on ETC integrity and functionality would strengthen the work.

eLife assessment

This study provides a methodological report on a modified adaptive sampling approach, multiple walker supervised molecular dynamics (mwSuMD), and its application to G protein-coupled receptors (GPCR), which are the most abundant membrane proteins and key targets for drugs. The mwSuMD approach assists in sampling complex binding processes, leading to some useful findings for GPCR activity, although results may be considered incomplete because the approach may have limited convergence to high-resolution structural data and is lacking other validation. The manuscript explores perhaps too many case studies at the expense of depth of description of methods, reference to existing computational literature, and deeper insight into GPCR activity.

Reviewer #1 (Public Review):

Summary:

The authors investigate ligand and protein-binding processes in GPCRs (including dimerization) by the multiple walker supervised molecular dynamics method. The paper is interesting and it is very well written.

Strengths:

The authors' method is a powerful tool to gain insight into the structural basis for the pharmacology of G protein-coupled receptors.

Weaknesses:

Cholesterol may play a fundamental role in GPCR dimerization (as cited by the authors, Prasanna et al, "Cholesterol-Dependent Conformational Plasticity in GPCR Dimers"). Yet they do not use cholesterol in their simulations of the dimerization.

Reviewer #2 (Public Review):

The study by Deganutti and co-workers is a methodological report on an adaptive sampling approach, multiple walker supervised molecular dynamics (mwSuMD), which represents an improved version of the previous SuMD.

Case-studies concern complex conformational transitions in a number of G protein Coupled Receptors (GPCRs) involving long time-scale motions such as binding-unbinding and collective motions of domains or portions. GPCRs are specialized GEFs (guanine nucleotide exchange factors) of heterotrimeric Gα proteins of the Ras GTPase superfamily. They constitute the largest superfamily of membrane proteins and are of central biomedical relevance as privileged targets of currently marketed drugs.

MwSuMD was exploited to address:<br /> (1) Binding and unbinding of the arginine-vasopressin (AVP) cyclic peptide agonist to the V2 vasopressin receptor (V2R);<br /> (2) Molecular recognition of the β2-adrenergic receptor (β2-AR) and heterotrimeric GDP-bound Gs protein;<br /> (3) Molecular recognition of the A1-adenosine receptor (A1R) and palmitoylated and geranylgeranylated membrane-anchored heterotrimeric GDP-bound Gi protein;<br /> (4) The whole process of GDP release from membrane-anchored heterotrimeric Gs following interaction with the glucagon-like peptide 1 receptor (GLP1R), converted to the active state following interaction with the orthosteric non-peptide agonist danuglipron;<br /> (5) The heterodimerization of D2 dopamine and A2A adenosine receptors (D2R and A2AR, respectively) and binding to a bi-valent ligand.

The mwSuMD method is solid and valuable, has wide applicability, and is compatible with the most world-widely used MD engines. It may be of interest to the computational structural biology community.

The huge amount of high-resolution data on GPCRs makes those systems suitable, although challenging, for method validation and development.

While the approach is less energy-biased than other enhanced sampling methods, knowledge, at the atomic detail, of binding sites/interfaces and conformational states is needed to define the supervised metrics, the higher the resolution of such metrics is the more accurate the outcome is expected to be. The definition of the metrics is a user- and system-dependent process.

The too many and ambitious case-studies undermine the accuracy of the output and reduce the important details needed for a methodological report. In some cases, the available CryoEM structures could have been exploited better.

The most consistent example concerns AVP binding/unbinding to V2R. The consistency with CryoEM data decreases with an increase in the complexity of the simulated process and involved molecular systems (e.g. receptor recognition by membrane-anchored G protein and the process of nucleotide exchange starting from agonist recognition by an inactive-state receptor). The last example, GPCR hetero-dimerization, and binding to a bi-valent ligand, is the most speculative one as it does not rely on high-resolution structural data for metrics supervision.

Reviewer #3 (Public Review):

Summary:

In the present work, Deganutti et al. report a structural study on GPCR functional dynamics using a computational approach called supervised molecular dynamics.

Strengths:

The study has the potential to provide novel insight into GPCR functionality. An example is the interaction between loops of GPCR and G proteins, which are not resolved experimentally, or the interaction between D344 and R385 identified during the Gs coupling by GLP-1R. However, validation of the findings, even computationally through for instance in silico mutagenesis study, is advisable.

Weaknesses:

In its current form, the manuscript seems immature and in particular, the described results grasp only the surface of the complex molecular mechanisms underlying GPCR activation. No significant advance of the existing structural data on GPCR and GPCR/G protein coupling is provided. Most of the results are a reproduction of the previously reported structures.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

In their valuable study, Chen et al. aim to define the neuronal role of HMMR, a microtubule-associated protein typically associated with cell division. Their findings suggest that HMMR is necessary for proper neuronal morphology and the generation of polymerizing microtubules within neurites, potentially by promoting the function of TPX2. While the study is recognized as a first step in deciphering the influence of HMMR on microtubule organization in neurons, reviewers note the current work has important gaps and would benefit from further exploration of the mechanism of microtubule stability by HMMR, the link between HMMR-mediated microtubule generation and morphogenesis, and the physiological implications of disrupting HMMR during neuronal morphogenesis.

Public Reviews:

Reviewer #1 (Public Review):

The microtubule cytoskeleton is essential for basic cell functions, enabling intracellular transport, and establishment of cell polarity and motility. Microtubule-associated proteins (MAPs) contribute to the regulation of microtubule dynamics and stability - mechanisms that are specifically important for the development and physiological function of neurons. Here, the authors aimed to elucidate the neuronal function of the MAP Hmmr, which they had previously identified in a quantitative study of the proteome associated with neuronal microtubules.

The authors conduct well-controlled experiments to demonstrate the localization of endogenous as well as exogenous Hmmr on microtubules within the soma as well as all neurites of hippocampal neurons. Functional analysis using gain- and loss-of-function approaches demonstrates that Hmmr levels are crucial for neuronal morphogenesis, as the length of both dendrites and axons decreases upon loss of Hmmr and increases upon Hmmr overexpression. In addition to length alterations, the branching pattern of neurites changes with Hmmr levels. To uncover the mechanism of how Hmmr influences neuronal morphology, the authors follow the lead that Hmmr overexpression induces looped microtubules in the soma, indicative of an increase in microtubule stability. Microtubule acetylation indeed decreases and increases with Hmmr LOF and GOF, respectively. Together with a rescue of nocodazole-induced microtubule destabilization by Hmmr GOF, these results argue that Hmmr regulates microtubule stability. Highlighted by the altered movement of a plus-end-associated protein, Hmmr also has an effect on the dynamic nature of microtubules. The authors present evidence suggesting that the nucleation frequency of neuronal microtubules depends on Hmmr's ability to recruit the microtubule nucleator Tpx2. Together, these data add novel insight into MAP-mediated regulation of microtubules as a prerequisite for neuronal morphogenesis. While the data shown support the author's conclusions, the study also has several weaknesses:

• The study appears incomplete as the initial proteomics analysis which is referenced as an entry into the study is not presented. This surely is the authors' choice, however, without presenting this data set, it would make more sense if the authors first showed the localization of Hmmr on neuronal microtubules and then started with the functional analysis.

The reviewer suggests moving the Hmmr localization data in front of the loss- and gain-of-function data because we did not present the proteomics data. However, we still believe placing the loss- and gain-of-function data in the beginning is the better arrangement. This is because it allows the audience to see the drastic changes on neuronal morphology when HMMR is depleted or overly abundant. It also provides a better linkage between HMMR’s localization on microtubules and its effect on the stability and dynamics of microtubules.

• Neurite branching is quantified, but the methods used are not consistent (normalized branch density vs. Sholl analysis) and there is no distinction between alterations of branching in dendrites vs. axons. This information should be added as it could prove informative with respect to the physiological function of Hmmr in neurite branching.

Sholl analysis is considered the gold standard in neurite branching analyses. However, in the knockdown experiment (Figure 1A~1E), HMMR-depleted neurons exhibited extremely short axons (<100 μm) and dendrites (<40 μm). Using Sholl analysis to assess the branching of these Hmmrdepleted neurons became unsuitable. That is why we used normalized branch density (Figure 1E) in the knockdown experiment and Sholl analysis (Figure 1J) in the overexpression experiment.

Regarding the branching difference between axons and dendrites, only axons exhibit branches at 4 DIV. Therefore, the branching analysis focuses on axons rather than on dendrites. We have revised the manuscript to clarify this.

• The authors show that altered Hmmr levels affect neurite branching and identify an effect on microtubule stability and dynamics as a molecular mechanism. However, how branching correlates with or is regulated by Hmmr-mediated microtubule dynamics is neither addressed experimentally nor discussed by the authors. The physiological significance of altered neuronal morphogenesis also lacks discussion.
• To discuss how branching correlates with or is regulated by HMMR-mediated microtubule dynamics, we have added the following paragraph into the Discussion section:

“It has been shown that compromising microtubule nucleation in neurons by SSNA1 mutant overexpression prevents proper axon branching (Basnet et al., 2018). Additionally, dendritic branching in Drosophila sensory neurons depends on the orientation of microtubule nucleation. Nucleation that results in an anterograde microtubule growth leads to increased branching, while nucleation that results in a retrograde microtubule growth leads to decreased branching (Yalgin et al., 2015). These results demonstrate the importance of microtubule nucleation on neurite branching. It is conceivable that overexpressing a microtubule nucleation promoting protein such as HMMR results in an increase of branching complexity.”

• In terms of discussing the physiological significance of altered neuronal morphogenesis. We have added the following paragraph to the Discussion section:

“Neurons are the communication units of the nervous system. The formation of their intricate shape is therefore crucial for the physiological function. Alterations in neuronal morphogenesis have a profound impact on how nerve cells communicate, leading to a variety of physiological consequences. These consequences include impaired neural circuit formation and function, compromised signal transmission between neurons, as well as altered anatomical structure of the CNS. Depending on the specific type and location of the morphogenetically altered neurons, the physiological consequences can include neurological disorders such as autism spectrum disorder (Berkel et al., 2012) and schizophrenia (Goo et al., 2023), as well as learning and memory deficits (Winkle et al., 2016). However, due to the involvement of HMMR on mitosis, most HMMR mutations are associated with familial cancers (based on ClinVar data).”

• Multiple times, the manuscript lacks a rationale for an experimental approach, choice of cell type, time points, regions of interest, etc. Also, a meaningful description of the methods and for how data were analyzed is missing, making the paper hard to read for someone not directly from the field.

We understand the reviewer’s comments regarding the lack of rationale for choosing the experimental approach, choice of cell type, time points, regions of interest, etc. As a result, we have added the rationales where appropriate to help readers from other fields to better understand the choice of cell type, time points, regions of interest, etc. A brief explanation is shown below:

• Approach and timing: We employed both electroporation (immediate but milder expression) and lipofectamine transfection (delayed but stronger expression). We prioritized knocking down HMMR early in development, so electroporation was used. For overexpression experiments, we chose lipofectamine which allows high protein expression level to be achieved.

• Cell selection: Hippocampal neurons were chosen in experiments that involve morphological quantification due to their homogeneous morphology. On the other hand, cortical neurons were selected in experiments that require large amounts of neurons and/or experiments where we want to demonstrate the universality of a proposed hypothesis.

• Regions of interest (ROIs): In our previous publication (Chen et al., 2017), it was discovered that a significant reduction of EB3 emanation frequency can be detected at the tip and the base of the neurite but not in the middle of the neurite in TPX2-depleted neurons. The reason for this difference is due to the presence of GTP-bound Ran GTPase (RanGTP) at the tip and the base of the neurite. Since RanGTP has also been shown to regulate the interaction between HMMR and TPX2 in the cell-free system (Scrofani et al., 2015), it is possible that the same phenomenon can be observed in HMMR-depleted neurons. This is why we examined those 3 ROIs in Figure 4.

Reviewer #2 (Public Review):

The mechanism of microtubule formation, stabilization, and organization in neurites is important for neuronal function. In this manuscript, the authors examine the phenotype of neurons following alteration in the level of the protein HMMR, a microtubule-associated protein with established roles in mitosis. Neurite morphology is measured as well as microtubule stability and dynamic parameters using standard assays. A binding partner of HMMR, TPX2, is localized. The results support a role for HMMR in neurons.

The work presented in this manuscript seeks to determine if a MAP called HMMR contributes to microtubule dynamics in neurons. Several steps, including validation of the RNAi, additional statistical analysis, use of cells at the same age in culture, and better documentation in figures, would increase the impact of the work.

In many places, the data can be improved which might make the story more convincing. As presented, the results show that HMMR is distributed as puncta on neurons with data coming from a single HMMR antibody, and some background staining that was not discussed. In the discussion the authors state that HMMR impacts microtubule stability, which was evaluated by the presence of post-translational modification and resistance to nocodazole; the data are suggestive but not entirely convincing. The discussion also states that HMMR increases the “amount” of growing microtubules which was measured as the frequency of comet appearance. The authors did not comment on how the number of growing microtubules results in the observed morphological changes.

We actually tested several HMMR antibodies, including E-19 (Santa Cruz, sc-16170), EPR4054 (Abcam, ab124729), and a variety of antibodies provided by Prof. Eva Turley. E-19 performed the best in immunofluorescence (IF) staining and knockdown validation. The other antibodies either failed to detect HMMR in IF staining or generate excessive background signal. We understand that the final images are produced using a single antibody. But since we meticulous validated this antibody and that the localization of overexpressed HMMR is consistent with the endogenous HMMR, we are very confident about our data generated using this single antibody.

We have added the following paragraph in the Discussion section to elucidate how the number of growing microtubules result in the observed morphological changes such as an increase of axon branches:

“It has been shown that compromising microtubule nucleation in neurons by SSNA1 mutant overexpression prevents proper axon branching (Basnet et al., 2018). Additionally, dendritic branching in Drosophila sensory neurons depends on the orientation of microtubule nucleation. Nucleation that results in an anterograde microtubule growth leads to increased branching, while nucleation that results in a retrograde microtubule growth leads to decreased branching (Yalgin et al., 2015). These results demonstrate the importance of microtubule nucleation on neurite branching. It is conceivable that overexpressing a microtubule nucleation promoting protein such as HMMR results in an increase of branching complexity.

Reviewer #1 (Recommendations for The Authors):

(1) The manuscript jumps extensively between main figures and supplementary figures. Please check whether parts of the supplement could be moved to the main figures.

We understand the frustration of moving back and forth between the main figures and supplementary figures. After examining the manuscript, we decided to combine Figure 2A with Figure S3.

(2) In Figure 1, total neurite length between days 3 and 4 DIV does not appear to change - can this be true?

Please check or else explain.

We carefully re-examined our raw data and found out the total neurite length of 4 DIV hippocampal neurons expressing non-targeting shRNA (Figure 1B) and that of 3 DIV hippocampal neurons expressing AcGFP (Figure 1G) are indeed very similar. The explanation is that the 3 DIV hippocampal neurons used for Figure 1G was cultured in low-density and in the presence of cortical neuron-conditioned neurobasal medium (as written in Methods, Neuron culture and transfection section). The low-density culture with minimal overlapping neurites allowed us to better quantify total neurite length, because neurons expressing AcGFP-mHMMR sprouted long and highly branched axons. However, the addition of cortical neuron-conditioned neurobasal medium promoted neurite elongation. This is the reason why the total neurite length of 4 DIV hippocampal neurons expressing non-targeting shRNA (Figure 1B) and that of 3 DIV hippocampal neurons expressing AcGFP (Figure 1G) is similar.

(3) Groen et al. have shown that Hmmr also bundles microtubules, a mechanism that surely is important for neuronal microtubules. Please discuss.

We thank the reviewer for pointing out that HMMR also bundles microtubules and have added this to our revised Discussion section:

“It has been shown that the Xenopus HMMR homolog XRHAMM bundles microtubules in vitro (Groen et al., 2004). In addition, deleting proteins which promote microtubule bundling (e.g., doublecortin knockout, MAP1B/MAP2 double knockout) leads to impaired neurite outgrowth (Bielas et al., 2007; Teng et al., 2001). These observations are consistent with our data that overexpressing HMMR leads to the increased axon and dendrite outgrowth, while depleting it results in the opposite phenotype (Figure 1).”

(4) Please explain why in Figure 4, cortical neurons were chosen for analysis and why and how the three different ROIs were picked.

To answer the question why we chose cortical neurons for the analyses in Figure 4, it will be important to explain why we used hippocampal neurons for other figures. Primary hippocampal neurons have a high homogeneity in terms of their morphology. This uniform morphology allows more consistent morphological quantification. Figure 4, however, does not involve morphological quantification. We are more confident to conclude that HMMR regulates microtubule dynamics if this effect can be detected in the relatively heterogeneous cortical neurons. These are the reasons why we chose to analyze cortical neurons in Figure 4.

In our previous publication (Chen et al., 2017), it was discovered that a significant reduction of EB3 emanation frequency can be detected at the tip and the base of the neurite but not in the middle of the neurite in TPX2-depleted neurons. The reason for this difference is due to the presence of GTP-bound Ran GTPase (RanGTP) at the tip of the neurite and in the soma. Since RanGTP has also been shown to regulate the interaction between HMMR and TPX2 in the cell-free system (Scrofani et al., 2015), it is possible that the same phenomenon can be observed in HMMR-depleted neurons. This was why we examined those 3 ROIs in Figure 4.

(5) Microtubule looping has been shown to occur in regions prior to branch formation (e.g. Dent et al. 2004). As the authors identify increased looping upon Hmmr GOF, this should be discussed.

We thank the reviewer for pointing out that microtubule looping occurs in regions of branch formation and have added this to our revised discussion:

“It is worth noting that the elevated level of HMMR increases the branching density of axons (Figure 1J) and promotes the formation of looped microtubules (Figure 3A). This is consistent with the observations that looped microtubules are often detected in regions of axon branch formation (Dent et al., 1999; Dent and Kalil, 2001; Purro et al., 2008).”

Reviewer #2 (Recommendations for The Authors):

(1) The work seeks to gain insight into microtubule behavior in neurons, an important issue.

(2) Several steps, including validation of the RNAi, additional statistical analysis, use of cells at the same age in culture, and better documentation in figures, would increase the impact of the work.

(3) Figure 1 documents the results of experiments in which the HMMR protein was depleted using shRNA. A western blot of cell extracts from control and depleted cells is needed to verify that the protein level is reduced; alternatively, documentation of the reduction in RNA levels in treated cells could be provided. Neurite, axon, and dendrite length and branch density are measured. The neurite length is in microns, and the axon length is normalized to 100% of the non-treated cells. Please use the same for measures for easier comparison. Looking at the images in Figure 1, the length of the dendrites does not look different in the examples shown, whereas the axon appears shorter. This impression is not supported by the quantification. Are representative images shown? Additionally, the authors should report the values for each replicate of the experiment and compare the three averages rather than comparison of lengths from all measurements. A related issue is that the dendrites do not look longer in panel F, following overexpression of HMMR. For examples of using averages of replicates see: https://pubmed.ncbi.nlm.nih.gov/32346721/

The reviewer mentioned that Western blot of cell extracts or RNA quantification from control and depleted cells are needed to verify that the protein level is reduced.

Unfortunately, these assays are extremely difficult to perform in primary neurons due to the low transfection efficiency. We believe that the consistent knockdown phenotype from 3 different shRNA sequences (Figure 1A-D) and the immunofluorescence staining in depleted primary neurons (Figure S2) are sufficient to confirm that HMMR level is reduced.

We revised Figure 1C, 1D, 1H, 1I so that axon and dendrite lengths are all in micron.

We selected another image for the non-targeting control in Figure 1A to better demonstrate the reduction of dendrite length when HMMR is knocked down.

We thank the reviewer for the suggestion of comparing the three average values rather than comparing all measurements. We have performed statistical analyses for all our data using the average values and revised the graphs accordingly. While the P-values changed, our conclusions remain the same.

We thank the reviewer for pointing out this discrepancy and have selected another image of the AcGFP control for Figure 1F to better demonstrate the increase of dendrite length when HMMR is overexpressed.

(4) Given the changes in neurite morphology, the authors examine the localization of endogenous and overexpressed. The supplemental figures (see S2 and S3) show evidence that HMMR is present in a punctate pattern by conventional immunofluorescence. This is reasonable evidence that the protein is in a linear pattern along cytoskeletal microtubules and that the signal is present in puncta. Please move this to the main text, perhaps replacing Figure 2A, which is low magnification and very hard to see the HMMR staining. Additionally, the level of overexpression of HMMR is not mentioned. Please address this; were cells with similar levels of overexpression selected? Did the result depend on the overexpression? A related issue is the DIV for the cells - some are examined earlier and some at later times; does this impact the results? Please provide information or perform experiments with consistent timing. For the immunofluorescence, were multiple antibodies tried to see if the result was the same with each? Were different fixations, in addition to methanol, utilized?

We have replaced Figure 2A with Figure S3 based on the reviewer’s suggestion.

In the HMMR overexpression experiments, we used HMMR antibody and immunofluorescence staining to confirm that the overexpression is achieved. However, we did not quantify to what extend HMMR was overexpressed.

We performed all the depletion experiments on 4 DIV to maximize knockdown efficiency and performed all the overexpression experiments on 3 DIV to prevent excessive axon fasciculation. Nonetheless, we examined the effect of HMMR depletion on neuronal morphology on 3 DIV. The trend of reduced total neurite length, axon length, and dendrite length can be observed, but no statistical significance can be detected. We also examined the effect of HMMR overexpression on neuronal morphology on 4 DIV and did observe an increase of total neurite length, axon length, and dendrite length. But the overlapping and bundled axons made reliable quantification extremely difficult.

We actually tested multiple HMMR antibodies, such as E-19 (Santa Cruz, sc-16170), EPR4054 (Abcam, ab124729), and a variety of antibodies provided by Prof. Eva Turley. E19 performed the best in immunofluorescence (IF) staining and knockdown validation. The other antibodies either failed to detect HMMR in IF staining or generate excessive background signal. We also tested various fixation methods, including 37°C formaldehyde fixation, -20°C methanol fixation, 37°C formaldehyde followed by -20°C methanol fixation. All fixation methods generated similar IF staining pattern using the E-19 antibody, but 3.7% formaldehyde fixation produced the highest signal.

(5) In Figure 2 C it is hard to see DAPI fluorescence. Are the white areas in the merge with bright cell nuclei? Is Figure 2C control or overexpressing cells? If this is endogenous, is there less signal in PLA compared with S4, which was in culture longer and is overexpressed prior to using PLA for detection?

The white areas in Figure 2C the reviewer mentioned are not cell nuclei, they are actually bubbles formed within the mounting medium.

HMMR detected in Figure 2C is endogenous. We did not quantitatively compare the PLA signals in Figure 2C and those in Figure S4. This is because the PLA signals in Figure 2C are generated using anti-HMMR (to detect endogenous HMMR) and anti-β-III-tubulin antibodies while those in Figure S4 are generated using anti-AcGFP (to detect overexpressed AcGFP-mHMMR) and anti-β-III-tubulin antibodies. Since the affinity of the two antibodies (i.e., anti-HMMR and anti-AcGFP) toward their antigens is different, comparing the PLA signals is not informative.

(6) The images of the endogenous HMMR (Fig S3) and the PLA with tubulin and HMMR antibodies are not the same (2C). The "dots" in PLA are widely separated; gauging from the marker bar length of 50 μm, the small clusters of dots are about 10 μm apart. In Figure S3, the puncta are much more closely spaced, appearing almost in a linear fashion along the microtubules. Enlarging the PLA image shows that each dot is very small - just a few pixels - please provide additional explanation including the minimal detection limit for the method, and why the images differ. If the standard immunofluorescence signal was enhanced, for example with the use of two secondaries, what is observed? Is the distribution of HMMR similar for both dendrites and axons? Microtubule polarity differs in these locations, so greater attention to this point seems of interest. There is a significant amount of punctate HMMR in the cytoplasm (or outside the cytoplasm?) in Figure S5; this is concerning. Please outline the cell edge for ease of visualization. What is the distribution of HMMR in a cell that has been treated with cold and/or nocodazole to disassemble the microtubules? is the signal lost?

The reasons images of the endogenous HMMR (Figure S3) and the PLA with tubulin and HMMR antibodies (Figure 2C) differ are due to the following reasons. o PLA utilizes two primary antibodies to target two different epitopes on HMMR and βIII-tubulin. It is conceivable that not every anti-HMMR antibody has the correct orientation and/or proximity (<40 nm) toward the anti-β-III-tubulin antibody to enable DNA amplification. This results in the shortage of PLA puncta compared to immunofluorescence signals.

• The creator of PLA has pointed out that in situ PLA is a method based upon equilibrium reactions and several enzymatic steps. Therefore, only a fraction of the inter-acting molecules is detected (Weibrecht et al., 2010).

We have not used signal enhancing immunofluorescence staining methods [e.g., using tertiary antibodies or tyramide signal amplification (TSA)] to detect HMMR. This is mainly because HMMR signal is strong enough to be detected using standard immunofluorescence staining.

Regarding the question “Is the distribution of HMMR similar for both dendrites and axons?” The reviewer raised a very important issue about the polarity difference of microtubules in axons (uniform) and dendrites (mixed). We were aware of such issue and very carefully examined the distribution and signal intensity of HMMR in axons vs dendrites. However, no differences were detected.

The reviewer mentioned that “there is a significant amount of punctate HMMR in the cytoplasm (or outside the cytoplasm?) in Figure S5; this is concerning. Please outline the cell edge for ease of visualization.” Instead of outlining the cell edge, we have selected another image to facilitate the visualization of HMMR signals. There are indeed HMMR signals outside the cell. However, these outside signals are usually weaker and smaller in size compared to those inside the cell.

After the examination of neurons expressing AcGFP-mHMMR with or without 100 nM nocodazole treatment, we did not notice any difference of AcGFP-mHMMR in distribution. We did not examine the distribution and signal intensity of the endogenous HMMR.

(7) To determine if HMMR alters microtubule stability, the authors examine the distribution of acetylated tubulin and resistance to nocodazole-induced microtubule disassembly. In Figure 3 please show immunofluorescence images of the acetylated tubulin staining, not just the ratio images; the color is not obviously different in the various panels shown. For statistical analysis, see the comment above for Figure 1. For the nocodazole experiment, a similar change in neurite length following drug treatment was observed (Figure 3H), for the experimental and control, even though the starting length was greater in the overexpressing cells. Please consider the possibility that in both cases the microtubules are only partially resistant to nocodazole and that HMMR is not changing the fraction of microtubules that are sensitive to the drug. The cells were treated at 3 DIV; the authors note that more stable microtubules accumulate with time; how does time in culture impact stability? Often, acute treatment with a high concentration of nocodazole is used to assay microtubule stability; here the authors used a low (nM) concentration for 2 days (chronic). Why not use a higher concentration (1-10 μM) for a shorter incubation? The data show that overexpression of HMMR results in curved, buckled microtubules are these microtubules more acetylated and/or retained after nocodazole treatment?

The reviewer suggested that we show immunofluorescence images of the acetylated tubulin staining, not just the ratio images. But we still believe showing the ratio images is the better approach. This is because the microtubules density can be different from neuron to neuron. Showing acetylated tubulin may provide a false impression when the overall microtubule density is higher or lower in a particular neuron. We realized that “16 colors” pseudo-color scheme has the cyan color at the lower intensity which can sometimes be confused with the white color at the higher intensity. Therefore, we changed the pseudocolor from “16 colors” to “fire” for Figure 3B and 3E to better visualize these images so that they appear more consistent with the quantitative data.

The reviewer raised a very good question regarding the possibility that HMMR is not changing the fraction of microtubules that are sensitive to nocodazole. We re-conducted the same experiment and used a series of different nocodazole concentrations. While the addition of nocodazole causes a concentration-dependent reduction of total neurite length in both AcGFP and AcGFP-mHMMR expressing neurons, there are subtle differences in the susceptibility of neurite length to the concentration of nocodazole. 1) 10 nM nocodazole treatment causes a significant reduction of neurite length in AcGFP expressing neurons, but not in AcGFP-mHMMR expressing neurons. This result indicates that AcGFP-mHMMR expression increases the tolerance of neurite elongation toward 10 nM nocodazole treatment. 2) 50 nM and 100 nM nocodazole treatment exhibits no statistical significance in AcGFP expressing neurons, suggesting that 50 nM nocodazole has reached maximal effectiveness. In AcGFP-mHMMR expressing neurons, 100 nM nocodazole further reduces the neurite length compared to the 50 nM group. These results argue against the possibility that HMMR does not change the fraction of microtubules that are sensitive to nocodazole. We have revised Figure 3H accordingly.

The reviewer asked why we did not use the acute nocodazole treatment (μM concentration) to assess the effect of Hmmr on microtubule stability. This is because we used the neurite length as an indicator for microtubule stability. That is why the chronic treatment was chosen to produce a more detectable effect on neurite length.

The reviewer asked whether the looped microtubules caused by HMMR overexpression are more acetylated and/or nocodazole resistant. While we do not have direct evidence to answer the reviewer’s question, we can deduce the answer from our observations. We noticed that looped microtubules are only present when HMMR is highly expressed (i.e., using lipofection to introduce HMMR-expressing plasmid) but not when HMMR is mildly expressed (i.e., using electroporation to introduce HMMR-expressing plasmid). From these observations, we can conclude that HMMR is more abundantly present on looped microtubules. Since HMMR overexpression leads to higher microtubule acetylation (Figure 3E), looped microtubules which contains more HMMR are most likely to be more acetylated.

(8) An additional measure of microtubule dynamics is to measure the growth of microtubules using a live cell marker for microtubule plus ends. Such experiments were performed, using tagged EB3. The images are rather fuzzy. Parameters of microtubule dynamics were measured at three locations - is there data that the authors can cite about any differences in dynamics in control cells at these locations? They look very similar, so it is not clear why the different locations were used. It is not possible to learn much from the kymographs which look similar for all panels; I would remove these unless they can be changed or labeled to help the reader. Data is presented for three shRNA reagents. No data are presented to document the extent to which the protein is depleted with these reagents. This should be fixed. Alternatively, an RNAi pool could be utilized. Is there a control for off-target effects? For the analysis were all the comets used to generate the average values? What about a comparison of the average of each trial - not each comet?

In our previous publication (Chen et al., 2017), it was discovered that a significant reduction of EB3 emanation frequency can be detected at the tip and the base of the neurite but not in the middle of the neurite in TPX2-depleted neurons. The reason for this difference is due to the presence of RanGTP at the tip and the base of the neurite. Since RanGTP has also been shown to regulate the interaction between HMMR and TPX2 in the cell-free system (Scrofani et al., 2015), it is possible that the same phenomenon can be observed in HMMR-depleted neurons. This is why we examined those 3 ROIs in Figure 4.

We notice that photobleaching causes the EB3-mCherry signal to diminish at later time points, which made it difficult to observe the differences amongst kymographs. In the revised Figure 4B and 4D, we removed the second half of all the kymographs to make the differences more obvious.

The reviewer mentioned that there are no data documenting the extent to which the protein is depleted with the shRNAs. These data are shown in Figure S2, in which we quantified the HMMR protein level in the soma and along the neurite in neurons expressing different shRNA molecules.

The reviewer asked whether there is a control for off-target effects. The answer is yes. We performed the rescue experiment to control for off-target effects, which is shown in Figure S1.

We revised Figure 4 so that the dynamic properties of EB3 are quantified using the average of each experimental repetition.

(9) In a final experiment, the authors examine the distribution of TPX2, a binding partner of HMMR. Include a standard immunofluorescence in addition to PLA to illustrate the distribution of TPX2. The quantification used was the inter puncta distance; please quantify the signal in control and treated cells.

The reviewer asked us to include a standard immunofluorescence staining to illustrate the distribution of TPX2. We have done that in our previous publication (Chen et al., 2017) and TPX2 localizes primarily to the centrosome (https://www.nature.com/articles/srep42297/figures/2). In order to enhance the weak signal of TPX2 along the neurite, we actually needed to use PLA in that publication (https://www.nature.com/articles/srep42297/figures/3).

Proximity ligation assay (PLA) generates fluorescent signals based on a local enzymatic reaction which catalyzes the amplification of a specific DNA sequence that can then be detected using a red fluorescent probe. Because this enzymatic reaction is not linear, the amount of amplified DNA nor the intensity of the fluorescence does not correlate with the strength of the interaction (Soderberg et al., 2006). As a result, quantification of PLA is typically done by counting the number of fluorescent puncta per unit area or by calculating the area containing fluorescent signal (not signal intensity) per unit area in the case that PLA signals are too strong and coalesced. That is why our quantification is based on the distance between PLA fluorescent puncta, not the fluorescent signal intensity.

References

Basnet, N., H. Nedozralova, A.H. Crevenna, S. Bodakuntla, T. Schlichthaerle, M. Taschner, G. Cardone, C. Janke, R. Jungmann, M.M. Magiera, C. Biertumpfel, and N. Mizuno. 2018. Direct induction of microtubule branching by microtubule nucleation factor SSNA1. Nat. Cell Biol. 20:1172-1180.

Berkel, S., W. Tang, M. Trevino, M. Vogt, H.A. Obenhaus, P. Gass, S.W. Scherer, R. Sprengel, G. Schratt, and G.A. Rappold. 2012. Inherited and de novo SHANK2 variants associated with autism spectrum disorder impair neuronal morphogenesis and physiology. Hum. Mol. Genet. 21:344-357.

Bielas, S.L., F.F. Serneo, M. Chechlacz, T.J. Deerinck, G.A. Perkins, P.B. Allen, M.H. Ellisman, and J.G. Gleeson. 2007. Spinophilin facilitates dephosphorylation of doublecortin by PP1 to mediate microtubule bundling at the axonal wrist. Cell. 129:579-591.

Chen, W.S., Y.J. Chen, Y.A. Huang, B.Y. Hsieh, H.C. Chiu, P.Y. Kao, C.Y. Chao, and E. Hwang. 2017. Ran-dependent TPX2 activation promotes acentrosomal microtubule nucleation in neurons. Sci. Rep. 7:42297.

Dent, E.W., J.L. Callaway, G. Szebenyi, P.W. Baas, and K. Kalil. 1999. Reorganization and movement of microtubules in axonal growth cones and developing interstitial branches. J. Neurosci. 19:8894-8908.

Dent, E.W., and K. Kalil. 2001. Axon branching requires interactions between dynamic microtubules and actin filaments. J. Neurosci. 21:9757-9769.

Goo, B.S., D.J. Mun, S. Kim, T.T.M. Nhung, S.B. Lee, Y. Woo, S.J. Kim, B.K. Suh, S.J. Park, H.E. Lee, K. Park, H. Jang, J.C. Rah, K.J. Yoon, S.T. Baek, S.Y. Park, and S.K. Park. 2023. Schizophrenia-associated Mitotic Arrest Deficient-1 (MAD1) regulates the polarity of migrating neurons in the developing neocortex. Mol. Psychiatry. 28:856-870.

Groen, A.C., L.A. Cameron, M. Coughlin, D.T. Miyamoto, T.J. Mitchison, and R. Ohi. 2004. XRHAMM functions in ran-dependent microtubule nucleation and pole formation during anastral spindle assembly. Curr. Biol. 14:1801-1811.

Purro, S.A., L. Ciani, M. Hoyos-Flight, E. Stamatakou, E. Siomou, and P.C. Salinas. 2008. Wnt regulates axon behavior through changes in microtubule growth directionality: a new role for adenomatous polyposis coli. J. Neurosci. 28:8644-8654.

Scrofani, J., T. Sardon, S. Meunier, and I. Vernos. 2015. Microtubule nucleation in mitosis by a RanGTP-dependent protein complex. Curr. Biol. 25:131-140.

Soderberg, O., M. Gullberg, M. Jarvius, K. Ridderstrale, K.J. Leuchowius, J. Jarvius, K. Wester, P. Hydbring, F. Bahram, L.G. Larsson, and U. Landegren. 2006. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat. Methods. 3:995-1000.

Teng, J., Y. Takei, A. Harada, T. Nakata, J. Chen, and N. Hirokawa. 2001. Synergistic effects of MAP2 and MAP1B knockout in neuronal migration, dendritic outgrowth, and microtubule organization. J. Cell Biol. 155:65-76.

Weibrecht, I., K.J. Leuchowius, C.M. Clausson, T. Conze, M. Jarvius, W.M. Howell, M. Kamali-Moghaddam, and O. Soderberg. 2010. Proximity ligation assays: a recent addition to the proteomics toolbox. Expert Rev Proteomics. 7:401-409.

Winkle, C.C., R.H. Olsen, H. Kim, S.S. Moy, J. Song, and S.L. Gupton. 2016. Trim9 Deletion Alters the Morphogenesis of Developing and Adult-Born Hippocampal Neurons and Impairs Spatial Learning and Memory. J. Neurosci. 36:49404958.

Yalgin, C., S. Ebrahimi, C. Delandre, L.F. Yoong, S. Akimoto, H. Tran, R. Amikura, R. Spokony, B. Torben-Nielsen, K.P. White, and A.W. Moore. 2015. Centrosomin represses dendrite branching by orienting microtubule nucleation. Nat. Neurosci. 18:1437-1445.

Reviewer #1 (Public Review):

The microtubule cytoskeleton is essential for basic cell functions, enabling intracellular transport, and establishment of cell polarity and motility. Microtubule-associated proteins (MAPs) contribute to the regulation of microtubule dynamics and stability - mechanisms that are specifically important for the development and physiological function of neurons. Here, the authors aimed to elucidate the neuronal function of the MAP Hmmr, which they had previously identified in a (yet unpublished) quantitative study of the proteome associated with neuronal microtubules. The authors conduct well-controlled experiments to demonstrate the localization of endogenous as well as exogenous Hmmr on microtubules within the soma as well as all neurites of hippocampal neurons. Functional analysis using gain- and loss-of-function approaches demonstrates that Hmmr levels are crucial for neuronal morphogenesis, as the length of both dendrites and axons decreases upon loss of Hmmr and increases upon Hmmr overexpression. In addition to length alterations, the branching pattern of neurites changes with Hmmr levels. To uncover the mechanism of how Hmmr influences neuronal morphology, the authors follow the lead that Hmmr overexpression induces looped microtubules in the soma, indicative of an increase in microtubule stability. Microtubule acetylation indeed decreases and increases with Hmmr LOF and GOF, respectively. Together with a rescue of nocodazole-induced microtubule destabilization by Hmmr GOF, these results argue that Hmmr regulates microtubule stability. Highlighted by the altered movement of a plus-end-associated protein, Hmmr also has an effect on the dynamic nature of microtubules. The authors present evidence suggesting that the nucleation frequency of neuronal microtubules depends on Hmmr's ability to recruit the microtubule nucleator Tpx2. The authors discuss how branching may be regulated by Hmmr-mediated microtubule dynamics and speculate about the physiological significance of altered neuronal morphogenesis. Together, their work adds novel insight into MAP-mediated regulation of microtubules as a prerequisite for neuronal morphogenesis.

eLife assessment

In their valuable study, Chen et al. investigate the neuronal role of HMMR, a microtubule-associated protein typically associated with cell division. Their findings indicate that HMMR is necessary for proper neuronal morphology and the generation of polymerizing microtubules within neurites, potentially by promoting the function of TPX2. This solid body of work is the first step in deciphering the influence of a mitotic microtubule-associated protein in organizing microtubules in neurons and will be of interest to the neurobiology and cytoskeleton fields.

Reviewer #2 (Public Review):

The mechanism of microtubule formation, stabilization, and organization in neurites is important for neuronal function. In this manuscript, the authors examine the phenotype of neurons following alteration in the level of the protein HMMR, a microtubule-associated protein with established roles in mitosis. Neurite morphology is measured as well as microtubule stability and dynamic parameters using standard assays. A binding partner of HMMR, TPX2, is localized. The results support a role for HMMR in microtubule stabilization in neurons.

The results show that HMMR is distributed as puncta on neurons using standard immunofluorescence and PLA. Depletion of HMMR reduced neurite length and extent of branching; reduced post-translational acetylation of neurite microtubules. Conversely, overexpression of HMMR increased resistance to nocodazole. The parameters of microtubule dynamics were also impacted by reduction or overexpression of HMMR. The authors discuss the possibility HMMR regulates neurite morphological changes via regulation of microtubule nucleation and dynamics.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

Summary:

The authors set up a pipeline for automated high-throughput single-molecule fluorescence imaging (htSMT) in living cells and analysis of molecular dynamics

Strengths:

htSMT reveals information on the diffusion and bound fraction of molecules, dose-response curves, relative estimates of binding rates, and temporal changes of parameters. It enables the screening of thousands of compounds in a reasonable time and proves to be more sensitive and faster than classical cell-growth assays. If the function of a compound is coupled to the mobility of the protein of interest, or affects an interaction partner, which modulates the mobility of the protein of interest, htSMT allows identifying the modulator and getting the first indication of the mechanism of action or interaction networks, which can be a starting point for more in-depth analysis.

Weaknesses:

While elegantly showcasing the power of high-throughput measurements, the authors disclose little information on their microscope setup and analysis procedures. Thus, reproduction by other scientists is limited. Moreover, a critical discussion about the limits of the approach in determining dynamic parameters, the mechanism of action of compounds, and network reconstruction for the protein of interest is missing. In addition, automated imaging and analysis procedures require implementing sensitive measures to assure data and analysis quality, but a description of such measures is missing.

The reviewer rightly highlights both the power and complexity in high throughput assay systems, and as such the authors have spent significant effort in first developing quality control checks to support screening. We discuss some of these as part of the description and characterization of the platform. We added additional details into the manuscript to help clarify. The implementation of our workflow for image acquisition, processing and analysis relies heavily on the specifics of our lab hardware and software infrastructure. We have added additional details to the text, particularly in the Methods section, and believe we have added enough information that our results can be reproduced using the suite of tools that already exist for single molecule tracking.

The reviewer also points out that all assays have limitations, and these have not been clearly identified as part of our discussion of the htSMT platform. We have also added some comments on the limitations of the current system and our approach.

Reviewer #2 (Public Review):

Summary:

McSwiggen et al present a high throughput platform for SPT that allows them to identify pharmaceutics interactions with the diffusional behavior of receptors and in turn to identify potent new ligands and cellular mechanisms. The manuscript is well written, it provides a solid new mentor and a proper experimental foundation

Strengths:

The method capitalizes and extends to existing high throughput toolboxes and is directly applied to multiple receptors and ligands. The outcomes are important and relevant for society. 10^6 cells and >400 ligands per is a significant achievement.

The method can detect functionally relevant changes in transcription factor dynamics and accurately differentiate the ligand/target specificity directly within the cellular environment. This will be instrumental in screening libraries of compounds to identify starting points for the development of new therapeutics. Identifying hitherto unknown networks of biochemical signaling pathways will propel the field of single-particle live cell and quantitative microscopy in the area of diagnostics. The manuscript is well-written and clearly conveys its message.

Weaknesses:

There are a few elements, that if rectified would improve the claims of the manuscript.

The authors claim that they measure receptor dynamics. In essence, their readout is a variation in diffusional behavior that correlates to ligand binding. While ligand binding can result in altered dynamics or /and shift in conformational equilibrium, SPT is not recording directly protein structural dynamics, but their effect on diffusion. They should correct and elaborate on this.

This is an excellent clarifying question, and we have tried to make it more explicit in the text. The reviewer is absolutely correct; we’re not using SPT to directly measure protein structural dynamics, but rather the interactions a given protein makes with other macromolecules within the cell. So when an SHR binds to ligand it adopts conformations that promote association with DNA and other protein-protein interactions relevant to transcription. This is distinct from assays that directly measure conformational changes of the protein.

L 148 What do the authors mean 'No correlation between diffusion and monomeric protein size was observed, highlighting the differences between cellular protein dynamics versus purified systems'. This is not justified by data here or literature reference. How do the authors know these are individual molecules? Intensity distributions or single bleaching steps should be presented.

The point we were trying to make is that the relative molecular weights for the monomer protein (138 kDa for Halo-AR, 102 kDa for ER-Halo, 122 kDa for Halo-GR, and 135 kDa for Halo-PR) is uncorrelated with its apparent free diffusion coefficient. Were we to make this measurement on purified protein in buffer, where diffusion is well described by the Stokes Einstein equation, one would expect to see monomer size and diffusion related. We’ve clarified this point in the manuscript.

Along the same lines, the data in Figs 2 and 4 show that not only the immobile fraction is increased but also that the diffusion coefficient of the fast-moving (attributed to free) is reduced. The authors mention this and show an extended Fig 5 but do not provide an explanation.

This is an area where there is still more work to do in understanding the estrogen receptor and other SHRs. As the reviewer says, we see not only an increase in chromatin binding but also a decrease in the diffusion coefficient of the “free” population. A potential explanation is that this is a greater prevalence of freely-diffusing homodimers of the receptor, or other protein-protein interactions (14-3-3, P300, CBP, etc) that can occur after ligand binding. Nothing in our bioactive compound screen shed light on this in particular, and so we can only speculate and have refrained from drawing further conclusions in the text.

How do potential transient ligand binding and the time-dependent heterogeneity in motion (see comment above) contribute to this? Also, in line 216 the authors write "with no evidence" of transient diffusive states. How do they define transient diffusive states? While there are toolboxes to directly extract the existence and abundance of these either by HMM analysis or temporal segmentation, the authors do not discuss or use them.

Throughout the analysis in this work, we consider all of tracks with a 2-second FOV as representative of a single underlying population and have not looked at changes in dynamics within a single movie. As we show in the supplemental figures we added (see Figure 3, figure supplement 1), this appears to be a reasonable assumption, at least in the cases we’ve encountered in this manuscript. For experiments involving changes in dynamics over time, these are experiments where we’ve added compound simultaneous with imaging and collect many 2-second FOVs in sequence to monitor changes in ER dynamics. In this case when we refer to “transient states,” we are pointing out that we don’t observe any new states in the State Array diagram that exist in early time points but disappear at later time point.

The reviewer suggests track-level analysis methods like hidden Markov models or variational Bayesian approaches which have been used previously in the single molecule community. These are very powerful techniques, provided the trajectories are long (typically 100s of frames). In the case of molecules that diffuse quickly and can diffuse out of the focal plane, we don’t have the luxury of such long trajectories. This was demonstrated previously (Hansen et al 2017, Heckert el al 2022) and so we’ve adopted the State Array approach to inferring state occupations from short trajectories. As the reviewer rightly points out, this approach potentially loses information about state transitions or changes over time, but as of now we are not aware of any robust methods that work on short trajectories.

The authors discuss the methods for extracting kinetic information of ligand binding by diffusion. They should consider the temporal segmentation of heterogenous diffusion. There are numerous methods published in journals or BioRxiv based on analytical or deep learning tools to perform temporal segmentation. This could elevate their analysis of Kon and Koff.

We’re aware of a number of approaches for analyzing both high framerate SMT as well as long exposure residence time imaging. As we say above, we’re not aware of any methods that have been demonstrated to work robustly on short trajectories aside from the approaches we’ve taken. Similarly, for residence time imaging there are published approaches, but we’re not aware of any that would offer new insight into the experiments in this study. If the reviewer has specific suggestions for analytical approaches that we’re not aware of we would happily consider them.

Reviewer #3 (Public Review):

Summary:

The authors aim to demonstrate the effectiveness of their developed methodology, which utilizes super-resolution microscopy and single-molecule tracking in live cells on a high-throughput scale. Their study focuses on measuring the diffusion state of a molecule target, the estrogen receptor, in both ligand-bound and unbound forms in live cells. By showcasing the ability to screen 5067 compounds and measure the diffusive state of the estrogen receptor for each compound in live cells, they illustrate the capability and power of their methodology.

Strengths:

Readers are well introduced to the principles in the initial stages of the manuscript with highly convincing video examples. The methods and metrics used (fbound) are robust. The authors demonstrate high reproducibility of their screening method (R2=0.92). They also showcase the great sensitivity of their method in predicting the proliferation/viability state of cells (R2=0.84). The outcome of the screen is sound, with multiple compounds clustering identified in line with known estrogen receptor biology.

Weaknesses:

• Potential overstatement on the relationship of low diffusion state of ER bound to compound and chromatin state without any work on chromatin level.

We appreciate the reviewers caution in over-interpreting the relationship between an increase in the slowest diffusing states that we observe by SMT and bona fide engagement with chromatin. In the case of the estrogen receptor there is strong precedent in the literature showing increases in chromatin binding and chromatin accessibility (as measured by ChIP-seq and ATAC-seq) upon treatment with either estradiol or SERM/Ds. Taken together with the RNA-seq, we felt it reasonable to assume all the trajectories with a diffusion coefficient less that 0.1 µm2/sec were chromatin bound.

• Could the authors clarify if the identified lead compound effects are novel at any level?

Most of the compounds we characterize in the manuscript have not previously been tested in an SMT assay, but many are known to functionally impact the ER or other SHRs based on other biochemical and functional assays. We have not described here any completely novel ER-interacting compounds, but to our knowledge this is the first systematic investigation of a protein showing that both direct and indirect perturbation can be inferred by observing the protein’s motion. Especially for the HSP90 inhibitors, the observation that inhibiting this complex would so dramatically increase ER chromatin-binding as opposed to increasing the speed of the free population is counterintuitive and novel.

• More video example cases on the final lead compounds identified would be a good addition to the current data package.

Reviewer #1 (Recommendations For The Authors):

General:

• More information on the microscope setup and analysis procedures should be given. Since custom code is used for automated image registration, spot detection, tracking, and analysis of dynamics, this code should be made publicly available.

Results:

• line 97: more details about the robotic system and automatic imaging, imaging modalities, and data analysis procedures should be given directly in the text.

Additional information added to text and methods

• line 100: we generated three U2OS cell lines --> how?

Additional information added to text and methods

• line 101: ectopically expressing HaloTag fused proteins --> how much overexpression did cells show?

The L30 promoter tends to produce fairly low expression levels. The same approach was used for all ectopic expression plasmids, and for the SHRs the expression levels were all comparable to endogenous levels. We have not checked this for H2B, Caax and free Halo but given that the necessary dye concentration to achieve similar spot densities is within a 10-fold range for all constructs, its reasonable to say that those clonal cell lines will also have modest Halotag expression.

• line 107: Single-molecule trajectories measured in these cell lines yielded the expected diffusion coefficients --> how was data analysis performed?

Additional information added to text and methods

• line 109: how was the localization error determined?

Additional information added to text and methods

• line 155: define occupation-weighted average diffusion coefficient.

Additional information added to text and methods

• line 157: with 34% bound in basal conditions and 87% bound after estradiol treatment  contradicts figure 2b, where the bound fraction is up to 50% after estradiol treatment.

Line 157 is the absolute fraction bound, figure 2b is change in fbound

• line 205: Figure 2c is missing.

Fixed

• line 215: within minutes --> how was this data set obtained? which time bins were taken?

Additional information added to text and methods

• line 216: with no evidence of transient diffusive states  What is meant by transient diffusive state? It seems all time points have a diffusive component, which decreases over time.

Additional information added to text and methods

The diffusive peak decreases, the bound peak increases but no other peaks emerge during that time (e.g. neither super fast nor super slow)

• line 225: it seems that fbound of GDC-0810 and GDC-0927 are rather similar in FRAP experiments, please comment, how was FRAP done?

FRAP is in the methods section. The curves and recovery times are quite distinct, is the reviewer looking at

• line 285: reproducibly: how often was this repeated?

Information added to the manuscript

• line 285: it would be necessary to name all of the compounds that were tested, e.g. with an ID number in the graph and a table. This also refers to extended data 7 and 8.

Additional supplemental file with the list of bioactive compounds tested will be included.

• line 290/1: what is meant by vendor-provided annotation was poorly defined?

Additional information added to text and methods. Specifically, the “other” category is the most common category, and it includes both compounds with unknown targets/functions as well as compound where the target and pathway are reasonably well documented. Hence, we applied our own analysis to better understand the list of active compounds.

Figures:

• fig. 2-6: detailed statistics are missing (number of measured cells, repetitions, etc.).

We have added clarifying information, including an “experiment design and sample size” section in the Methods.

• fig. 3: the authors need to give a list with details about the 5067 compounds tested,

Additional supplemental file with the list of bioactive compounds tested will be included.

• extended data 1c: time axis does not correspond to the 1.5s of imaging in the text, results line 127.

Axes fixed

• extended data 3: panel c and d are mislabeled.

Panel labels fixed

Methods:

• line 746: HILO microscope: the authors need to explain how they can get such large fields of view using HILO

Additional details added to the materials and methods. The combination of the power of the lasers, the size of the incident beam out of the fiber optic coupling device and the sCMOS camera are the biggest components that enable detection over a larger field of view.

• line 761: it is common practice to publish the analysis code. Since the authors wrote their own code, they should publish it

Our software contains proprietary information that we cannot yet release publicly. Comparable results can be achieved with HILO data using publicly-available tools like utrack. State Arrays code is distributed and the parameters used are listed in the M&M.

Reviewer #2 (Recommendations For The Authors):

The writing and presentation are coherent, concise, and easy to follow.

The authors should consider justifying the following:

Why is 1.5s imaging time selected? Topological and ligand variations may last significantly longer than this. The authors should present at least for one condition the same effect images for longer.

Related to the similar comment above, we added a figure examining the jump length distribution as a function of frame. Over the 6 seconds of data collection the jump length distribution is unchanged, suggesting it is reasonable to consider all the trajectories within an FOV as representative of the same underlying dynamical states.

The authors miss the k test or T test in their graphs.

We chose to apply the Kurskal-Wallis test in the context of the bioactive screen to assess whether a grouping of compounds based on their presumed cellular target was significantly different from the control even when individual compounds might not by themselves raise to significance. In this case many of the pathway inhibitors are subtle and not necessarily obvious in their difference. In the other cases throughout the manuscript, whether two conditions are statistically distinguishable is rarely in question and of far less importance to the conclusions in the manuscript than the magnitude of the difference. We’ve added statistical tests where appropriate.

The overall integrated area of Fig 4a appears to reduce upon ligand addition. Data appear normalized but the authors should also add N (number of molecules) on top of the graphs.

While the integrated area may appear to decrease, all State Array analysis is performed by first randomly sampling 10,000 trajectories from the assay well and inferring state distribution on those 10,000. This has been clarified in the figure legend and in the Methods.

Minor

Extended Figure 3 legend c, d appear swapped and incorrectly named in the text.

Panel labels fixed

L 197 but this appears not to BE a general feature of SHRs (maybe missing Be).

Error fixed

L205 authors refer to Figure 2c, which does not exist.

Panel reference fixed

Reviewer #3 (Recommendations For The Authors):

Among minor issues:

In Figure 1B, if the authors could specify how they discriminate the specific cell lines from the mixed context, it would enhance clarity. Could they perform additional immunofluorescence to understand how the assignment is determined? Alternatively, could they also show the case with isolated cell lines in an unmixed context?

Immunofluorescence would be a challenge given that there is not a good epitope to distinguish the three ectopically-expressed genes from each other or from endogenous proteins in the case of H2B and CaaX. We are really reliant on the single cell dynamics to determine the likely cell identity. That said, we’ve added graphs of a number of individual cell State Arrays from the same data graphed in 1A which support the notion that it’s reasonable to assume a cells identity given the observed dynamics.

In Extended Figure 2F: possibly a CHip-Seq experiment would be more directly qualified to state the effect of ER ligand on ER ability to bind chromatin.

This is true. Presumably ER that is competent at activating transcription of ER-responsive genes is also capable of binding DNA. ChIP would be the more direct measure, but would not address whether the protein was functional. We chose to balance these measuring these two aspects of ER biology by pairing dynamics with the end-point transcription readout.

In Figure 3: A representation with plate-by-plate orientation along the x-axis, with controls included in each plate, would be more appropriate to reflect the consistency of the controls used in the assay across different plates. Currently, all controls are pooled in one location, and we cannot appreciate how the controls vary from plate to plate.

Figure added to the supplement

Also in this figure, a general workflow of the screen down to segmentation/analysis would be a great add-on.

New figure added to the supplement and reflected in the textual description of the platform

In Extended Figures 3B and C an add-on of the positive and negative control would make the figure more convincing.

Addressed as part of figure added to the supplement

Is there any description of compound leads identified that is novel in nature in relation to impact on ER, and if so could it be stated more clearly in the text as novel finding?

To our knowledge, the impact of HSP inhibition in increasing ER-chromatin association has never been described, neither has the link between inhibition post-translation modifying enzymes like the CDKs or mTOR and ER dynamics ever been described. We added clarifying text to the manuscript

eLife assessment

This work presents an important technological advance, in the form of a high throughput platform for Single Particle Tracking allowing us to measure millions of cells and thousands of compounds per day. Analysis of the diffusional behaviour of fluorescently-tagged targets permits the identification of, and differentiation between, small molecules that bind directly or affect the target indirectly. The methodology and metrics employed are compelling, leading to the identification of multiple compounds that effectively change the diffusive state of the estrogen receptor, the POC target of the study.

Reviewer #1 (Public Review):

Summary:

The authors set up a pipeline for automated high-through single-molecule fluorescence imaging (htSMT) in living cells and analysis of molecular dynamics.

Strengths:

htSMT reveals information on the diffusion and bound fraction of molecules, dose-response curves, relative estimates on binding rates, and temporal changes of parameters. It enables the screening of thousands of compounds in a reasonable time and proves to be more sensitive and faster than classical cell-growth assays. If the function of a compound is coupled to the mobility of the protein of interest or affects an interaction partner, which modulates the mobility of the protein of interest, htSMT allows identifying the modulator and getting the first indication on the mechanism of action or interaction networks, which can be a starting point for more in-depth analysis. The authors describe their automated imaging and analysis procedures as well as the measures taken to assure data and analysis quality.

Weaknesses:

While elegantly showcasing the power of high-throughput measurements, htSMT relies on a sophisticated robot-based workflow and several microscopes for parallel imaging, thus limiting wide-spread application of htSMT by other scientists.

Reviewer #3 (Public Review):

Summary:

The authors aim to demonstrate the effectiveness of their developed methodology, which utilizes super-resolution microscopy and single-molecule tracking in live cells on a high-throughput scale. Their study focuses on measuring the diffusion state of a molecule target, the estrogen receptor, in both ligand-bound and unbound forms in live cells. By showcasing the ability to screen 5067 compounds and measure the diffusive state of the estrogen receptor for each compound in live cells, they illustrate the capability and power of their methodology.

Readers are well introduced to the principles in the initial stages of the manuscript with highly convincing video examples. The methods and metrics used (fbound) are robust. The authors demonstrate high reproducibility of their screening method (R2=0.92). They also showcase the great sensitivity of their method in predicting the proliferation/viability state of cells (R2=0.84). The outcome of the screen is sound, with multiple compounds clustering identified in line with known estrogen receptor biology.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1:

Detection of early-stage colorectal cancer is of great importance. Recently, both laboratory scientists and clinicians have reported different exosomal biomarkers to identify colorectal cancer patients.

Here, the authors exhibited a full RNA landscape for plasma exosomes of 60 individuals, including 31 colorectal cancer (CRC) patients, 19 advanced adenoma (AA) patients, and 10 noncancerous controls. RNAs with high fold change, high absolute abundance, and various module attribution were used to construct RT-qPCR-based RNA models for CRC and AA detection.

Overall, this is a well-performed proof-of-concept study to highlight exosomal RNAs as potential biomarkers of early-stage colorectal cancer and its precancerous lesions.

Thank you for your careful evaluation and valuable suggestions, which have provided valuable guidance for the improvement of our paper. In response to your feedback, we have implemented the following improvements.

(1) Depicting the full RNA landscape of circulating exosomes is still quite challenging. The authors annotated 58,333 RNA species in exosomes, most of which were lncRNAs, but the authors do not explain how they characterized those RNAs.

Author response and action taken: Thanks for your comments. In the Supplementary Methods section titled "Identification of mRNAs and lncRNAs", we have provided a comprehensive explanation on the characterization of mRNAs and lncRNAs to address the concerns you raised. Characterization of long-chain RNAs is a great challenge. For lncRNA analysis, the transcriptome was assembled using the Cufflinks and Scripture based on the reads mapped to the reference genome. The assembled transcripts were annotated using the Cuffcompare program from the Cufflinks package. The unknown transcripts were used to screen for putative lncRNAs.

(2) The authors tested their models in a medium size population of 124 individuals, which is not enough to obtain an accurate evaluation of the specificity and sensitivity of the biomarkers proposed here. External validation would be required.

Author response and action taken: Thanks for your comments. We fully acknowledge the significance of external validations in the evaluation of diagnostic model performance. Unfortunately, as a pilot study, we currently do not have the conditions for a multicenter investigation. To mitigate result bias and overfitting effects, we implemented a rigorous variable selection strategy and enhanced model stability through 10-fold cross-validation. In the meantime, we will persist in our efforts to elevate the quality of our research and seek additional resources for external validation in future studies.

Reviewer #2:

The authors present an important study on the potential of small extracellular vesicle (sEV)-derived RNAs as biomarkers for the early detection of colorectal cancer (CRC) and precancerous adenoma (AA). The authors provide a detailed analysis of the RNA landscape of sEVs isolated from participants, identifying differentially expressed sEV-RNAs associated with T1a stage CRC and AA compared to normal controls. The paper further categorises these sEV-RNAs into modules and constructs a 60-gene model that successfully distinguishes CRC/AA from NC samples. The authors also validate their findings using RT-qPCR and propose an optimised classifier with high specificity and sensitivity. Additionally, the authors discuss the potential of sEV-RNAs in understanding CRC carcinogenesis and suggest that a comprehensive biomarker panel combining sEV-RNAs and proteins could be promising for identifying both early and advanced CRC patients. Overall, the study provides valuable insights into the potential clinical application of sEV-RNAs in liquid biopsy for the early detection of CRC and AA.

Major strengths:

(1) Comprehensive sEV RNA profiling: The study provides a valuable dataset of the whole-transcriptomic profile of circulating sEVs, including miRNA, mRNA, and lncRNA. This approach adds to the understanding of sEV-RNAs' role in CRC carcinogenesis and facilitates the discovery of potential biomarkers.

(2) Detection of early-stage CRC and AA: The developed 60-gene t-SNE model successfully differentiated T1a stage CRC/AA from normal controls with high specificity and sensitivity, indicating the potential of sEV-RNAs as diagnostic markers for early-stage colorectal lesions.

(3) Independent validation cohort: The study combines RNA-seq, RT-qPCR, and modelling algorithms to select and validate candidate sEV-RNAs, maximising the performance of the developed RNA signature. The comparison of different algorithms and consideration of other factors enhance the robustness of the findings.

Thank you for your careful evaluation and valuable suggestions. These comments have been highly valuable for the performance evaluation and clinical applications of our work. In response to your feedback, we have implemented the following improvements.

(1). Lack of analysis on T1-only patients in the validation cohort: While the study identifies key sEV-RNAs associated with T1a stage CRC and AA, the validation cohort is only half of the patients in T1(25 out of 49). It would be better to do an analysis using only the T1 patients in the validation cohort, so the conclusion is not affected by the T2-T3 patients.

Author response and action taken: Thanks for your comments. This feedback is essential for ensuring consistency in the results with our previous findings. In this context, we revalidated various diagnostic panels using exclusively Stage I patients (Figure 7—figure supplement 2). To minimize the potential overfitting effect due to the reduction in sample size after partitioning, we implemented a 10-fold cross-validation for each panel and these panels exhibit promising performance in Stage I colorectal cancer (CRC) patients.

Author response image 1.

The ROC analysis of different sEV-RNA signatures in the prediction of Stage I CRC patients by different algorithms (a: 6-gene panel; b: 7-gene panel; c: 8-gene panel; d: 9-gene panel).

(2). Lack of performance analysis across different demographic and tumor pathology factors listed in Supplementary Table 12. It's important to know if the sEV-RNAs identified in the study work better/worse in different age/sex/tumor size/Yamada subtypes etc.

Author response and action taken: Thanks for your comments. This feedback will be immensely beneficial for clinical diagnosis. Similarly, cross-validation was performed in this section. We assessed the discriminative effects of CRC on NC, taking into account different age groups, genders, tumor sizes, and anatomical locations (Figure 7—figure supplement 3). Overall, these sEV RNA panels perform better in individuals under the age of 55 and in female patients. There is no significant difference in discriminative effects across different tumor sizes. Compared to rectal cancer, the discriminative effects are better in colon cancer.

Author response image 2.

The ROC analysis of different sEV-RNA signatures for predicting CRC patients using the Lasso regression algorithm in different clinical parameters (ab: age; cd: gender; ef: tumor size; gh: anatomical position).

eLife assessment

This study presents a useful description of RNA in extracellular vesicles (EV-RNAs) and highlights the potential to develop biomarkers for the early detection of colorectal cancer (CRC) and precancerous adenoma (AA). The data were analysed using overall solid methodology and would benefit from further validation of predicted lncRNAs and biomarker validation at each stage of CRC/AA to evaluate the potential application to early detection of CRC and AA.

Joint Public Review:

Detection of early-stage colorectal cancer is of great importance. Laboratory scientists and clinicians have reported different exosomal biomarkers to identify colorectal cancer patients. This is a proof-of-principle study of whether exosomal RNAs, and particularly predicted lncRNAs, potential biomarkers of early-stage colorectal cancer and its precancerous lesions.

Strengths:

The study provides a valuable dataset of the whole-transcriptomic profile of circulating sEVs, including miRNA, mRNA, and lncRNA. This approach adds to the understanding of sEV-RNAs' role in CRC carcinogenesis and facilitates the discovery of potential biomarkers.

The developed 60-gene t-SNE model successfully differentiated T1a stage CRC/AA from normal controls with high specificity and sensitivity, indicating the potential of sEV-RNAs as diagnostic markers for early-stage colorectal lesions.

The study combines RNA-seq, RT-qPCR, and modelling algorithms to select and validate candidate sEV-RNAs, maximising the performance of the developed RNA signature. The comparison of different algorithms and consideration of other factors enhance the robustness of the findings.

Weaknesses:

Validation in larger cohorts would be required to establish as biomarkers, and to demonstrate whether the predicted lncRNAs implicated in these biomarkers are indeed present, and whether they are robustly predictive/prognostic.

eLife assessment

This important study, which presents novel data on variation in sperm whale communication, contributes to a richer understanding of the social transmission of vocal styles across neighbouring clans. The evidence is solid but could be further improved with some clarification of the specialized measurements and terms used, particularly for comparisons to other taxa. This research will be of interest for bioacoustics and animal communication specialists, particularly those working on social learning and culture.

Reviewer #1 (Public Review):

Summary:

This manuscript presents evidence of 'vocal style' in sperm whale vocal clans. Vocal style was defined as specific patterns in the way that rhythmic codas were produced, providing a fine-scale means of comparing coda variations. Vocal style effectively distinguished clans similar to the way in which vocal repertoires are typically employed. For non-identity codas, vocal style was found to be more similar among clans with more geographic overlap. This suggests the presence of social transmission across sympatric clans while maintaining clan vocal identity.

Strengths:

This is a well-executed study that contributes exciting new insights into cultural vocal learning in sperm whales. The methodology is sound and appropriate for the research question, building on previous work and ground-truthing much of their theories. The use of the Dominica dataset to validate their method lends strength to the concept of vocal style and its application more broadly to the Pacific dataset. The results are framed well in the context of previous works and clearly explain what novel insights the results provide to the current understanding of sperm whale vocal clans. The discussion does an overall great job of outlining why horizontal social learning is the best explanation for the results found.

Weaknesses:

The primary issues with the manuscript are in the technical nature of the writing and a lack of clarity at times with certain terminology. For example, several tree figures are presented and 'distance' between trees is key to the results, yet 'distance' is not clearly defined in a way for someone unfamiliar with Markov chains to understand. However, these are issues that can easily be dealt with through minor revisions with a view towards making the manuscript more accessible to a general audience.

I also feel that the discussion could focus a bit more on the broader implications - specifically what the developed methods and results might imply about cultural transmission in other species. This is specifically mentioned in the abstract but not really delved into in detail during the discussion.

Reviewer #2 (Public Review):

Summary:

The current article presents a new type of analytical approach to the sequential organisation of whale coda units.

Strengths:

The detailed description of the internal temporal structure of whale codas is something that has been thus far lacking.

Weaknesses:

It is unclear how the insight gained from these analyses differs or adds to the voluminous available literature on how codas varies between whale groups and populations. It provides new details, but what new aspects have been learned, or what features of variation seem to be only revealed by this new approach?<br /> The theoretical basis and concepts of the paper are problematical and indeed, hamper potentially the insights into whale communication that the methods could offer. Some aspects of the results are also overstated.

Reviewer #3 (Public Review):

Summary:

The study presented by Leitao et al., represents an important advancement in comprehending the social learning processes of sperm whales across various communicative and socio-cultural contexts. The authors introduce the concept of "vocal style" as an addition to the previously established notion of "vocal repertoire," thereby enhancing our understanding of sperm whale vocal identity.

Strengths:

A key finding of this research is the correlation between the similarity of clan vocal styles for non-ID codas and spatial overlap (while no change occurs for ID codas), suggesting that social learning plays a crucial role in shaping symbolic cultural boundaries among sperm whale populations. This work holds great appeal for researchers interested in animal cultures and communication. It is poised to attract a broad audience, including scholars studying animal communication and social learning processes across diverse species, particularly cetaceans.

Weaknesses:

In terms of terminology, while the authors use the term "saying" to describe whale vocalizations, it may be more conservative to employ terms like "vocalize" or "whale speech" throughout the manuscript. This approach aligns with the distinction between human speech and other forms of animal communication, as outlined in prior research (Hockett, 1960; Cheney & Seyfarth, 1998; Hauser et al., 2002; Pinker & Jackendoff, 2005; Tomasello, 2010).

Author response:

We thank the reviewers for their positive assessments and constructive feedback.

In light of their comments, we will aim to improve the explanation of the methods and interpretation of results, as well as their relation to well-established literature in this research area.

The major contributions of our work are threefold:

• First, we introduce a novel way of analyzing codas that specifically targets subcoda structures by considering inter-click intervals within codas in terms of transition probabilities. By describing codas’ click patterns via Variable Length Markov Chains, we do not need to consider codas in their entirety, but we can detect coda subunits.This enables a new dimension for quantitatively comparing differences among various individuals, social units, and clans; which we term ‘vocal style’.

• Using this approach, we reinforce findings from past research, including the idea that identity codas function as symbolic markers of vocal clan identity (Hersh et al., 2022; Sharma et al., 2024). More importantly, we offer new insights into the function of non-identity codas, which comprise the majority of coda types produced by sperm whales but have been largely uncharacterized. 

• Our work reveals that non-identity coda vocal styles are more similar for spatially overlapped clans, and suggests that this similarity in style may be maintained by social learning across clan boundaries. This opens up a paradigm shift in our understanding of between-clan acoustic interactions.

From a broader perspective, our work builds on two well-established research areas: the form and function of sperm whale codas, and statistical generative models, specifically Variable Length Markov Chains on finite data spaces. Our methods, results, and interpretations are grounded in theories and concepts from these fields.

For clarity, we will ensure that our terminology aligns with field standards and existing research. We will clearly introduce each key theory or concept at first mention and justify its relevance. In particular, we will clarify the definition and meaning of the distance between subcoda trees for a general audience. We agree with the reviewers’ comments on the broader implications and will refine our work accordingly.

eLife assessment

This important study highlights the role of SLAM-SAP signaling in shaping innate-like γδ T cell subsets, providing compelling evidence for the importance of SLAM-SAP in immune system regulation, and the potential implications of the findings for tumor surveillance and infectious disease management. The work will be of broad interest to immunologists.

Reviewer #1 (Public Review):

Summary:

In this study, the authors advance their previous findings on the role of the SLAM-SAP signaling pathway in the development and function of multiple innate-like gamma-delta T-cell subsets. Using a high throughput single-cell proteogenomics approach, the authors uncover SAP-dependent developmental checkpoints, and the role of SAP signaling in regulating the diversion of γδ T cells into the αβ T cell developmental pathway. Finally, the authors define TRGV4/TRAV13-4(DV7)-expressing T cells as a novel, SAP-dependent Vγ4 γδT1 subset.

Strengths:

This study furthers our understanding of the importance and complexity of the SLAM-SAP signaling pathway not only in the development of innate-like γδ T cells but also in how it potentially balances the γδ/αβ T cell lineage commitment. Additionally, this study reveals the role of SAP-dependent events in the generation of γδ TCR repertoire.

The conclusions of the study are supported by well-thought-out experiments and compelling data.

Weaknesses:

No major weaknesses in the study were identified.

Reviewer #2 (Public Review):

Summary:

Mistri et al explore the role of SLAM-SAP signaling in the developmental programming of innate-like gd T cell subsets. Using proteo-genomics, they determined that abrogation of SLAM-SAP signaling altered that programming, reducing some IL-17-producing subsets, including a novel Vγ4 γδT1 subset, and diverting gdTCR-expressing precursors to the ab fate. Altogether, this is a very thorough, thoughtfully interpreted study that adds significantly to our understanding of the contribution of the SLAM-SAP pathway to lineage specification. A particularly interesting element is the role of SLAM-SAP in preventing gd17 progenitors from switching fates and adopting the ab fate.

One thing to keep in mind in assessing the ultimate fate of the "ab wannabe cells" is that mechanisms exist to silence the gd TCR as cells differentiate to the DP stage and so their presence as diverted DP cells may not be evident by staining for gdTCR expression - and will only be evident transcriptomically.

Strengths:

This is an exceedingly well-designed and thorough study that significantly enriches our understanding of gd T cell development.

Weaknesses:

There are no major weaknesses identified by this reviewer.

Author response:

The following is the authors’ response to the original reviews.

eLife assessment

Both reviewers positively received the manuscript, in general. The agreement was that the manuscript presented valuable findings, using solid techniques and approaches, that shed additional light into how the canine distemper virus hemagglutinin might engage cellular receptors and how that engagement impacts host tropism. While both reviewers appreciated the X-ray crystallographic data, they also felt that the AFM experiments could have been performed at a higher standard and that the interpretation of the results ensuing from those AFM experiments could have been explained more thoroughly and in simpler terms. An additional missed opportunity of the current manuscript is the lack of comparison of the crystal structure to that of the already published cryo-EM structure, for context.

Thank you very much for constructive comments of the editor and reviewers. Following your comments, we have changed the text related to the AFM experiments with simpler terms as follows.

“When CDV-H was loaded onto a mica substrate and scanned with a cantilever to acquire images of attached molecules, the CDV-H dimer was observed as two globules clustered together in most cases, but sometimes, each domain moved independently (Fig. 7B and Supplementary Movie). Time-course analysis of the dynamics of the representative CDV-H dimer showed that CDV-H could adopt both associated and dissociated forms (Fig. 7C). The distances between the domains were calculated by measuring those between the centers of mass of each domain. Finally, the distribution of distances between each head domain in the CDV-H dimers showed approximately 15 nm as a major peak (Fig. 7D). This is a reasonable length for the linker between the head domain dimers.” in Page 11, Lines 8-17.

With regards to the structural comparison between cryo-EM structure published in Proc. Natl. Acad. Sci. U. S. A. (2023) 120, e2208866120 and our crystal structure, we have compared these structures for Cα on page 6 and added the following text. “A recent cryo-EM structure of the wild-type CDV-H ectodomain revealed that the head dimer is located on one side of the stalk region in solution (Proc. Natl. Acad. Sci. U. S. A. (2023) 120, e2208866120)” in Page 14, Lines 22-24.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

Fukuhara, Maenaka, and colleagues report a crystal structure of the canine distemper virus (CDV) attachment hemagglutinin protein globular domain. The structure shows a dimeric organization of the viral protein and describes the detailed amino-acid side chain interactions between the two protomers. The authors also use their best judgement to comment on predicted sites for the two cellular receptors - Nectin-4 and SLAM - and thus speculate on the CDV host tropism. A complementary AFM study suggests a breathing movement at the hemagglutinin dimer interface.

Strengths:

The study of CDV and related Paramyxoviruses is significant for human/animal health and is very timely. The crystallographic data seem to be of good quality.

Thank you very much for the constructive comment of the reviewer.

Weaknesses:

While the recent CDV hemagglutinin cryo-EM structure is mentioned, it is not compared to the present crystal structure, and thus the context of the present study is poorly justified. Additionally, the results of the AFM experiment are not unexpected. Indeed, other paramyxoviral RBP/G proteins also show movement at the protomer interface.

Thank you very much for constructive comments of the reviewer. When we submitted our manuscript to e-life, cryo-EM structure just published in Proc. Natl. Acad. Sci. U. S. A. (2023) 120, e2208866120 a week ago was not able to be available. Following the comment of the reviewer, we have added the text about the structural comparison between the cryo-EM structure and our crystal structure. We also have changed the text related to the AFM experiments to tone down the movement of the protomer interfaceas follows.

“This observation raises the possibility that each head domain of CDV-H also dissociates and moves flexibly, as shown in the structure of Nipah virus (NiV)-G protein, previously (Science (2022) 375, 1373–1378).” in Page 11, Lines 4-6.

Reviewer #2 (Public Review):

Summary:

The authors solved the crystal structure of CDV H-protein head domain at 3,2 A resolution to better understand the detailed mechanism of membrane fusion triggering. The structure clearly showed that the orientation of the H monomers in the homodimer was similar to that of measles virus H and different from other paramyxoviruses. The authors used the available co-crystal strictures of the closely related measles virus H structures with the SLAM and Nectin4 receptors to map the receptor binding site on CDV H. The authors also confirmed which N-linked sites were glycosylated in the CDV H protein and showed that both wildtype and vaccine strains of CDV H have the same glycosylation pattern. The authors documented that the glycans cover a vast majority of the H surface while leaving the receptor binding site exposed, which may in part explain the long-term success of measles virus and CDV vaccines. Finally, the authors used HS-AFM to visualize the real-time dynamic characteristics of CDV-H under physiological conditions. This analysis indicated that homodimers may dissociate into monomers, which has implications for the model of fusion triggering.

The structural data and analysis were thorough and well-presented. However, the HS-AFM data, while very exciting, was not presented in a manner that could be easily grasped by readers of this manuscript. I have some suggestions for improvement.

(1) The authors claim their structure is very similar to the recently published croy-EM structure of CDV H. Can the authors provide us with a quantitative assessment of this statement?

Thank you very much for constructive comments of the reviewer. When we submitted our manuscript to e-life, cryo-EM structure just published in Proc. Natl. Acad. Sci. U. S. A. (2023) 120, e2208866120 a week ago was not able to be available. Following the comment of the reviewer, we have added the text about the structural comparison between the cryo-EM structure and our crystal structure. We also have changed the text related to the AFM experiments to tone down the movement of the protomer interface as follows.

“This observation raises the possibility that each head domain of CDV-H also dissociates and moves flexibly, as shown in the structure of Nipah virus (NiV)-G protein, previously (Science (2022) 375, 1373–1378).” in Page 11, Lines 4-6.

(2) The results for the HS-AFM are difficult to follow and it is not clear how the authors came to their conclusions. Can the authors better explain this data and justify their conclusions based on it?

Thank you very much for constructive comments of the reviewer. Following your comments, we have changed the text related to the AFM experiments with simpler terms as follows.

“When CDV-H was loaded onto a mica substrate and scanned with a cantilever to acquire images of attached molecules, the CDV-H dimer was observed as two globules clustered together in most cases, but sometimes, each domain moved independently (Fig. 7B and Supplementary Movie). Time-course analysis of the dynamics of the representative CDV-H dimer showed that CDV-H could adopt both associated and dissociated forms (Fig. 7C). The distances between the domains were calculated by measuring those between the centers of mass of each domain. Finally, the distribution of distances between each head domain in the CDV-H dimers showed approximately 15 nm as a major peak (Fig. 7D). This is a reasonable length for the linker between the head domain dimers.” in Page 11, Lines 8-17.

(3) The fusion triggering model in Figure 8 is ambiguous as to when H-F interactions are occurring and when they may be disrupted. The authors should clarify this point in their model.

Thank you very much for constructive comments of the reviewer. Following your comments, we have changed the Figure 8 and its legend.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) AFM experiments with SLAM or Nectin-4 immobilized on the cantilever would be much more informative.

Thank you very much for the constructive comment of the reviewer. We will try this experiment in the next paper.

(2) The authors should compare their crystal structure to that of the reported cryo-EM structure.

With regards to the structural comparison between cryo-EM structure published in Proc. Natl. Acad. Sci. U. S. A. (2023) 120, e2208866120 and our crystal structure, we have added the text.

(3) Figure 1D - why does the beta2 MG negative control have such a high SPR signal?

Thank you very much for the constructive comment of the reviewer. The immobilization levels for b 2-microglobulin (beta2 MG), CDV-OP-H and CDV-5VD-H were similar, 1204.7 RU, 1235.7 RU, and 1504.5 RU, respectively. We applied relatively high concentrations (5 mM) of dNectin4 and hNectin4 onto the chip to determine low-affinity dissociation constants. Then, the signals for beta2 MG (negative control) were high. In other SPR experiments for cell surface receptors, such high signals for beta2 MG were often observed in our previous paper, Kuroki et al., J. Immunol. 2019 Dec 15;203(12):3386-3394. doi: 10.4049/jimmunol.1900562. Therefore, we think that these SPR signals are not unusual.

(4) Figure 1C - please indicate the Ve volume for the peak and add in Ve for standard.

Thank you very much for the constructive comment of the reviewer. We have indicated the Ve volume for the peak and added in Ve for standard in Figure 1C.

(5) The authors mention that one of the chains in the asymmetric unit was better resolved than the other. Please show regions of the atomic model fit regions of the electron density to convince the reader of the quality of your data.

Thank you very much for the constructive comment of the reviewer. We have added new Supplementary figure 2 for comparison of electron density maps of chains A and B.

(6) Table 2 indicates that the difference between Rw and Rf values is larger than 5% which indicates slight overfitting during refinement. Please provide details of your refinement strategy and attempt simulated annealing as a strategy to reduce this delta.

Thank you very much for the constructive comment of the reviewer. We further introduced TLS and NCS parameters for the refinement. Consequently, the R/Rfree factors became 0.2645/0.3092. Simulated annealing had been already carried out. All the refinement statistics in the table 2 are updated.

Reviewer #2 (Recommendations For The Authors):

(1) The authors' fusion triggering model was difficult to follow. For example, this sentence was difficult to understand: "The other possible models may include the monomer-dimer-tetramer transition facilitated by receptor binding for the fusion."

Thank you very much for the constructive comment of the reviewer. Following your comments, we have removed the above sentences and have added the detail mechanism of the proposed model in Discussion. Furthermore, we have changed the Figure 8 and its legend for readers to understand more clearly.

(2) Figure 5A is not called out in the main text.

Thank you very much for the constructive comment of the reviewer. Following your comments, we have added the text as follows.

“the crystal structure of MeV-H in complex with hNectin-4 showed that the H-SLAM interaction consists of three main sites (Fig. 5A) (Nat. Struct. Mol. Biol. (2013) 20, 67–72).” in Page 11, Lines 4-6.

(3) Page 9, Line 4: interspaces? Perhaps interphases.

Thank you very much for the constructive comment of the reviewer. We have changed the term “interspaces” to “internal spaces”.

(4) Page 12, penultimate line: The authors mention "epitopes for anti-MeV-H Abs." Do they mean anti-CDV-H Abs?

Thank you very much for the constructive comment of the reviewer. Following your comments, we have changed the “anti-MeV-H Abs” to “anti-morbillivirus H neutralizing antibodies”.

(5) The paper will benefit from an English language editor to help clarify what the authors are trying to convey.

Thank you very much for the constructive comment of the reviewer.

We have asked a English proof reading company to check.

eLife assessment

The manuscript presents valuable findings, using solid techniques and approaches, that shed additional light into how the canine distemper virus (CDV) hemagglutinin might engage cellular receptors and how that engagement impacts host tropism. The structural data and their analysis were thorough and well-presented. The HS-AFM data, which indicate that homodimers may dissociate into monomers - and thus have significant implications for the model of fusion triggering - are very exciting, but require further validation, perhaps by alternate approaches, to bolster the current molecular model of the CDV fusion triggering.

Reviewer #2 (Public Review):

The authors solved the crystal structure of CDV H-protein head domain at 3,2 A resolution to better understand the detailed mechanism of membrane fusion triggering. The structure clearly showed that the orientation of the H monomers in the homodimer was similar to that of measles virus H and different from other paramyxoviruses. The authors used the available co-crystal strictures of the closely related measles virus H structures with the SLAM and Nectin4 receptors to map the receptor binding site on CDV H. The authors also confirmed which N-linked sites were glycosylated in the CDV H protein and showed that both wildtype and vaccine strains of CDV H have the same glycosylation pattern. The authors documented that the glycans cover a vast majority of the H surface while leaving the receptor binding site exposed, which may in part explain the long-term success of measles virus and CDV vaccines. Finally, the authors used HS-AFM to visualize the real-time dynamic characteristics of CDV-H under physiological conditions. This analysis indicated that homodimers may dissociate into monomers, which has implications for the model of fusion triggering.

The structural data and analysis were thorough and well-presented. The HS-AFM data, while very exciting, needs to be further validated, perhaps by alternate approaches to further support the authors' model describing the molecular dynamics of fusion triggering.

Reviewer #2 (Public Review):

Summary:

The manuscript by Djebar et al investigated the role and the underlying mechanism of the ciliary transition zone protein Rpgrip1l in zebrafish spinal alignment. They showed that rpgrip1l mutant zebrafish develop a nearly full penetrance of body curvature at juvenile stages. The mutant fish have cilia defects associated with ventricular dilations and loss of the Reissner fibers. Scoliosis onset and progression are also strongly associated with astrogliosis and neuroinflammation, and anti-inflammatory drug treatment prevents scoliosis in mutant zebrafish, suggesting a novel pathogenic mechanism for human idiopathic scoliosis. This study is quite comprehensive with high-quality data, and the manuscript is well written, providing important information on how the ciliary transition zone protein functions in maintaining the zebrafish body axis straightness.

Strengths:

Very clear and comprehensive analysis of the mutant zebrafish.

Weaknesses:

(1) In Figures 1D-G, magnified high-resolution pictures are required to show there are indeed no vertebral malformations.

(2) Are the transcriptome data and proteomic data consistent? Consistent targets in both analyses should be highlighted.

(3) What is the role of Anxa2 in neuroinflammation? Is increased Anxa2 expression in rpgrip1l mutant zebrafish reduced after anti-inflammatory drug treatment? What is the expression level of anxa2 in cep290 mutant zebrafish?

(4) More background about Rpgrip1l should be provided in the introduction, particularly the past studies of the mammalian homolog of Rpgrip11, if there are any.

(5) Is there any human disease associated with Rpgrip1l? Do these patients have scoliosis phenotype?

(6) A summary diagram at the end would be helpful for understanding the main findings.

eLife assessment

This valuable study analyzes the role of the ciliary transition zone protein rpgrip1l in the development of the scoliotic phenotype in zebrafish. Through convincing proteomic and experimental validation in vivo, the authors demonstrated increased Annexin A2 expression in the brain and increased LCP1+ immune cell infiltration in scoliosis fish. These findings provide additional evidence for the previously proposed role of neuroinflammation in the development of idiopathic scoliosis in zebrafish.

Reviewer #1 (Public Review):

Summary:

In this study, Djebar et al. perform a comprehensive analysis of mutant phenotypes associated with the onset and progression of scoliosis in zebrafish ciliary transition zone mutants rpgrip1l and cep290. They determine that rpgrip1l is required in foxj1a-expressing cells for normal spine development, and that scoliosis is associated with brain ventricle dilations, loss of Reissner fiber polymerization, and the loss of 'tufts' of multi-cilia surrounding the subcommissural organ (the source of Reissner substance). Informed by transcriptomic and proteomic analyses, they identify a neuroinflammatory response in rpgrip1l and cep290 mutants that is associated with astrogliosis and CNS macrophage/microglia recruitment. Furthermore, anti-inflammatory drug treatment reduced scoliosis penetrance and severity in rpgrip1l mutants. Based on their data, the authors propose a feed-forward loop between astrogliosis, induced by perturbed ventricular homeostasis, and immune cell recruitment as a novel pathogenic mechanism of scoliosis in zebrafish ciliary transition zone mutants.

Strengths:

(1) Comprehensive characterization of the causes of scoliosis in ciliary transition zone mutants rpgrip1l and cep290.

(2) Comparison of rpgrip1l mutants pre- and post-scoliosis onset allowed authors to identify specific phenotypes as being correlated with spine curvature, including brain ventricle dilations, loss of Reissner fiber, and loss of cilia in proximity to the sub-commissural organ.

(3) Elegant genetic demonstration that increased urotensin peptide levels do not account for spinal curvature in rpgrip1l mutants.

(4) The identification of astrogliosis and Annexin over-expression in glial cells surrounding diencephalic and rhombencephalic ventricles as being correlated with scoliosis onset and severe curve progression is a very interesting finding, which may ultimately inform pathogenic mechanisms driving spine curvature

Weaknesses:

(1) The fact that cilia loss/dysfunction and Reissner fiber defects cause scoliosis in zebrafish is already well established in the literature, as is the requirement for cilia in foxj1a-expressing cells.

(2) Neuroinflammation has already been identified as the underlying pathogenic mechanism in at least 2 previously published scoliosis models (zebrafish ptk7a and sspo mutants).

(3) Anti-inflammatory drugs like aspirin, NAC, and NACET have also previously been demonstrated to suppress scoliosis onset and severe curve progression in these models.

Therefore, although similar observations in rpgrip1l and cep290 mutants (as reported here) add to a growing body of literature that supports a common biological mechanism underlying spine curvature in zebrafish, the novelty of reported findings is diminished.

(4) Although authors demonstrate that astrogliosis and/or macrophage or microglia cell recruitment are correlated with scoliosis, they do not formally demonstrate that these events are sufficient to drive spine curvature. Thus, the functional consequences of astrogliosis and microglia infiltration remain uncertain.

(5) The authors do not investigate the effect of anti-inflammatory treatments on other phenotypes they have correlated with spinal curve onset (like ventricle dilation, Reissner fiber loss, and multi-cilia loss around the subcommissural organ). This would help to identify causal events in scoliosis.

Reviewer #3 (Public Review):

Summary:

This paper describes a new mechanism of clearance of protein aggregates occurring during mitosis.

The authors have observed that animal cells can clear misfolded aggregated proteins at the end of mitosis. The images and data gathered are solid, convincing, and statistically significant. However, there is a lack of insight into the underlying mechanism. They show the involvement of the ER, ATPase-dependent, BiP chaperone, and the requirement of Cdk1 inactivation (a hallmark of mitotic exit) in the process. They also show that the mechanism seems to be independent of the APC/C complex (anaphase-promoting complex). Several points need to be clarified regarding the mechanism that clears the aggregates during mitosis:

• What happens in the cell substructure during mitosis to explain the recruitment of BiP towards the aggregates, which seem to be relocated to the cytoplasm surrounded by the ER membrane.

• How the changes in the cell substructure during mitosis explain the relocation of protein aggregates during mitosis.

• Why BiP seems to be the main player of this mechanism and not the cyto Hsp70 first described to be involved in protein disaggregation.

Strengths:

Experimental data showing clearance of protein aggregates during mitosis is solid, statistically significant, and very interesting.

Weaknesses:

Weak mechanistic insight to explain the process of protein disaggregation, particularly the interconnection between what happens in the cell substructure during mitosis to trigger and drive clearance of protein aggregates.

eLife assessment

How misfolded proteins are segregated and cleared is a significant question in mechanistic cell biology, since clearance of these aggregates can protect against pathologies that may otherwise arise. The authors discover a cell cycle stage-dependent clearing mechanism that involves the ER chaperone BiP, the proteosome, and CDK inactivation, but is curiously independent of the APC. These are valuable and interesting new observations, but the evidence supporting these claims is incomplete, and needs to be strengthened and further validated.

Reviewer #1 (Public Review):

Strengths:

The manuscript utilizes a previously reported misfolding-prone reporter to assess its behaviour in ER in different cell line models. They make two interesting observations:

(1) Upon prolonged incubation, the reporter accumulates in nuclear aggregates.

(2) The aggregates are cleared during mitosis. They further provide some insight into the role of chaperones and ER stressors in aggregate clearance. These observations provide a starting point for addressing the role of mitosis in aggregate clearance. Needless to say, going ahead understanding the impact of aggregate clearance on cell division will be equally important.

Weaknesses:

The study almost entirely relies on an imaging approach to address the issue of aggregate clearance. A complementary biochemical approach would be more insightful. The intriguing observations pertaining to aggregates in the nucleus and their clearance during mitosis lack mechanistic understanding. The issue pertaining to the functional relevance of aggregation clearance or its lack thereof has not been addressed. Experiments addressing these issues would be a terrific addition to this manuscript.

Reviewer #2 (Public Review):

Summary:

The authors provide an interesting observation that ER-targeted excess misfolded proteins localize to the nucleus within membrane-entrapped vesicles for further quality control during cell division. This is useful information indicating transient nuclear compartmentalization as a quality control strategy for misfolded ER proteins in mitotic cells, although endogenous substrates of this pathway are yet to be identified.

Strengths:

This microscopy-based study reports unique membrane-based compartments of ER-targeted misfolded proteins within the nucleus. Quarantining aggregating proteins in membrane-less compartments is a widely accepted protein quality control mechanism. This work highlights the importance of membrane-bound quarantining strategies for aggregating proteins. These observations open up multiple questions on proteostasis biology. How do these membrane-bound bodies enter the nucleus? How are the single-layer membranes formed? How exactly are these membrane-bound aggregates degraded? Are similar membrane-bound nuclear deposits present in post-mitotic cells that are relevant in age-related proteostasis diseases? Etc. Thus, the observations reported here are potentially interesting.

Weaknesses:

This study, like many other studies, used a set of model misfolding-prone proteins to uncover the interesting nuclear-compartment-based quality control of ER proteins. The endogenous ER-proteins that reach a similar stage of overdose of misfolding during ER stress remain unknown.

The mechanism of disaggregation of membrane-trapped misfolded proteins is unclear. Do these come out of the membrane traps? The authors report a few vesicles in living cells. This may suggest that membrane-untrapped proteins are disaggregated while trapped proteins remain aggregates within membranes.

The authors figure out the involvement of proteasome and Hsp70 during the disaggregation process. However, the detailed mechanisms including the ubiquitin ligases are not identified. Also, is the protein ubiquitinated at this stage?

This paper suffers from a lack of cellular biochemistry. Western blots confirming the solubility and insolubility of the misfolded proteins are required. This will also help to calculate the specific activity of luciferase more accurately than estimating the fluorescence intensities of soluble and aggregated/compartmentalized proteins. Microscopy suggested the dissolution of the membrane-based compartments and probably disaggregation of the protein. This data should be substantiated using Western blots. Degradation can only be confirmed by Western blots. The authors should try time course experiments to correlate with microscopy data. Cycloheximide chase experiments will be useful.

The cell models express the ER-targeted misfolded proteins constitutively that may already reprogram the proteostasis. The authors may try one experiment with inducible overexpression.

It is clear that a saturating dose of ER-targeted misfolded proteins activates the pathway. The authors performed a few RT-PCR experiments to indicate the proteostasis-sensitivity. Proteome-based experiments will be better to substantiate proteostasis saturation.

The authors should immunostain the nuclear compartments for other ER-membrane resident proteins that span either the bilayer or a single layer. The data may be discussed.

All microscopy figures should include control cells with similarly aggregating proteins or without aggregates as appropriate. For example, is the nuclear-targeted FlucDM-EGFP similarly entrapped? A control experiment will be interesting. Expression of control proteins should be estimated by western blots.

There are few more points that may be out of the scope of the manuscript. For example, how do these compartments enter the nucleus? Whether similar entry mechanisms/events are ever reported? What do the authors speculate? Also, the bilayer membrane becomes a single layer. This is potentially interesting and should be discussed with probable mechanisms. Also, do these nuclear compartments interfere with transcription and thereby deregulate cell division? What about post-mitotic cells? Similar deposits may be potentially toxic in the absence of cell division. All these may be discussed.

Reviewer #3 (Public Review):

Summary:

In this manuscript, Boudjerna and Balagé et al. aim to elucidate the spatial origin of centriole amplification and the mechanisms behind the formation of an apical-basal body patch in multiciliated cells (MCCs). To this end, they focused on the role of microtubules and developed new tools for spatiotemporal and high-resolution analysis of different stages of centriole amplification, including the centrosome stages, A-stage, G-stage, and MCC-stage. Among these tools, the MEF-MCC cells grown on micropatterns stands out for its versatility as it is not tissue-specific and does not require epithelial cell-to-cell contact for differentiation. Additionally, the Cen2-GFP; mRuby-Deup1 knock-in mouse model was used to study different stages of centriole amplification in physiological brain MCCs. This model offers an advantage over the previously described Cen2-GFP model by enabling the resolution of early events in centriole amplification through the visualization of Deup1-positive structures and their dynamics. Finally, the authors leveraged powerful imaging techniques, including super-resolution microscopy, the U-ExM, and high-resolution live cell imaging in order to detect and track centriole amplification, elongation, disengagement, and migration.

By combining the MEF-MCC and knock-in mouse model with spatiotemporal imaging in control and nocodazole-treated cells(treated acutely or chronically), the authors define the sequence of events during centriole amplification, revealing the critical roles of microtubules for the first time. Initially, the centrosome-mediated microtubule network forms, organizing a pericentrosomal nest from which procentrioles and deuterosomes emerge. Their findings indicate the importance of microtubules in recruiting and maintaining pericentriolar material clouds that contain DEUP1, PCNT, SAS6, PLK1, PLK4, and tubulins. Following the amplification stage, the procentrioles mature, leading to cells displaying numerous MTOCs, as demonstrated by regrowth experiments. Mature centrioles then disengage from deuterosomes, attach to the nuclear envelope, and migrate to the apical surface facilitated by microtubules.

Strengths:

The manuscript provides new insights into the regulatory function of microtubules in centriole amplification. Addressing the role of microtubules during different stages of centriole amplification required the development of new tools to study brain MCCs, which will be useful in future studies of MCCs. A notable strength of this manuscript is the authors' thorough and quantitative analysis of highly dynamic processes in MCCs. The precision and detail in describing these dynamic events are impressive. This comprehensive analysis advances our understanding of MCC biology.

Weaknesses:

The role of microtubules and other molecular players during different stages of centriole amplification in brain MCCs can be further studied and strengthened using the tools developed in the manuscript. A more quantitative description of some of the analysis performed in the manuscript is required to strengthen the conclusions.

eLife assessment

In this important study, Boudjema et al. use cell culture models and advanced microscopic imaging to provide detailed analyses of the cellular events underlying centriole amplification, apical migration, and assembly of hundreds of motile cilia in multi-ciliated cells. This largely descriptive work provides a better understanding of this process that is of interest to cell biologists studying centrioles and cilia. Most of the claims are supported by the data, but the study would benefit from additional analyses regarding the roles of microtubules, which are currently incomplete, and from text editing to improve accessibility and readability, especially for a wider audience.

Reviewer #1 (Public Review):

The manuscript by Boudjema et al. describes the cellular events underlying centriole amplification and apical migration to allow the assembly of hundreds of motile cilia in multi-ciliated cells. For this, they use cell culture models in combination with fixed and live cell imaging using antibody staining and fluorescence from endogenously tagged centriole and deuterostome markers, respectively. The work is largely descriptive and functional analyses are restricted to treatment with the microtubule depolymerizing drug nocodazole. The imaging is state-of-the-art including confocal microscopy, live imaging with optical sectioning and high optical and temporal resolution, as well as super-resolution imaging by ultra-expansion microscopy.

The study does a good job of providing a very detailed description of the dynamics of centrioles and deuterostomes that lead to centriole amplification and apical migration in multiciliated cells. This detailed view was missing in previous work. It also reveals the involvement of microtubules at multiple steps: the formation of a cloud of deuterostome precursors, the nuclear envelope tethering of newly formed centrioles, their separation, and their migration to the apical surface.

It would have been useful to expand the analysis of the role of microtubules by including analyses of the requirement for specific microtubule motors, for a better understanding and additional evidence that microtubule-based transport is involved. A weak point is that there is no visualization of microtubules together with deuterosomes and centrioles at the different steps of centriole amplification and migration, to directly address how these structures may interact with and move along microtubules.

Overall, apart from experimental aspects and since this is largely a descriptive study, the manuscript would benefit from more precise language and a better description of the complex events underlying centriole amplification and movements.

Reviewer #2 (Public Review):

This important work will be of interest to centriole and cilia cell biologists. It describes in detail how microtubules control multiple aspects of centriole amplification in brain multiciliated cells. This study provides a greater time-resolved and molecular proteomic mapping of the different steps involved, with or without microtubule disruption. Boudjema et al. show that microtubules are important throughout the centriole amplification process, from the early stages, where the procentrioles emerge from a pericentriolar "nest", through the growth stage where microtubules maintain the perinuclear localisation, to the detachment stage, where microtubules assist in perinuclear disengagement and apical migration. The results are generally well supported by the evidence, but the manuscript would benefit significantly from some heavy editing to introduce more niche terms, standardize abbreviations in text, and labels on figures to help bring the readers, especially non-specialists, along with them - increasing the accessibility of their work.

Reviewer #3 (Public Review):

Summary:

The authors want to prove that there is a redox potential between germline stem cells (GSCs) and somatic cyst stem cells (CySCs) in the Drosophila testis, with ROS being higher in the former compared to the latter. They also want to prove that ROS travels from CySCs to GSCs. Finally, they begin to characterize the phenotypes caused by loss of SOD (which normally lowers ROS levels) in the tj- lineage and how this impacts the germline.

Strengths:

The role of SOD in somatic support cells is an under-explored area.

Weaknesses:

The authors fall short of accomplishing their goals. There are issues with the concept of the paper (ROS gradient between cells that causes a transfer of ROS across membranes for homeostasis), the data, the figures, and the scholarship of the testis. I have discussed each of the points in detail below. These weaknesses negatively impact the conclusions put forward by the authors. In short, their data is not compelling: there is no evidence provided by the authors that ROS diffuses from CySCs to GSCs as most of the claims about stem cells are founded on data about differentiating germ and somatic cells. The somatic SOD depletion phenotype is incompletely characterized and several pathways appear to change in these cells, including reduced Egfr signaling, increased Tor signaling, and increased Hh signaling. None of these results are sufficiently followed up on. And none of them are considered relative to their known roles in the testis. For example, high Hh signaling in CySCs increases their competitiveness with GSCs. Increased Tor signaling in all CySCs does not affect the CySC lineage. Reduced Egfr signaling in CySCs reduces the number of CySCs and reduces/inhibits abscission between GSCs-gonialblasts.

Major issues:

(1) Data<br /> a. Problems proving which mitochondria are associated with which lineage.<br /> b. There is no evidence that ROS diffuses from CySCs into GSCs.<br /> c. The changes in gst-GFP (redox readout) are possibly seen in differentiating germ cells (i.e., spermatogonia) but not in GSCs. This weakens their model that ROS in CySC is transferred to GSCs.<br /> d. Most of the paper examines the effect of SOD depletion (which should increase ROS) on the CySC lineage and GSC lineage. One big caveat is that tj-Gal4 is expressed in hub cells (Fairchild, 2016) so the loss of SOD from hub cells may also contribute to the phenotype. In fact, the niche in Figure 2D looks larger than the niche in the control in Figure 2C, arguing that the expression of Tj in niche cells may be contributing to the phenotype. The authors need to better characterize the niche in tj>SOD-RNAi testes.<br /> e. The tj>SOD-RNAi phenotype is an expansion of the Zfh1+ CySC pool, expansion of the Tj+ Zfh1- cyst cells (both due to increased somatic proliferation) and a non-autonomous disruption of the germline.<br /> f. I am not convinced that MAPK signaling is decreased in tj>SOD-i testes. Not only is this antibody finicky, but the authors don't have any follow-up experiments to see if they can restore SOD-depleted CySCs by expressing an Egfr gain of function. Additionally, reduced Egfr activity causes fewer somatic cells (not more) (Amoyel, 2016) and also inhibits abscission between GSCs and gonialblasts (Lenhart 2015), which causes interconnected cysts of 8- to 16 germ cells with one GSC emanating from the hub.<br /> g. The increase in Hh signaling in SOD-depleted CySCs would increase their competitiveness against GSCs and GSCs would be lost (Amoyel 2014). The authors need to validate that Hh protein expression is indeed increased in SOD-depleted CySCs/cyst cells and which cells are producing this Hh. Normally, only hub cells produce Hh (Michel, 2012; Amoyel 2013) to promote self-renewal in CySCs.<br /> h. The increase in p4E-BP is an indication that Tor signaling is increased, but an increase in Tor in the CySC lineage does not significantly affect the number of CySCs or cyst cells (Chen, 2021). So again I am not sure how increased Tor factors into their phenotype.<br /> i. The over-expression of SOD in CySCs part is incomplete. The authors would need to monitor ROS in these testes. They would also need to examine with tj>SOD affects the size of the hub.

(2) Concept<br /> Why would it be important to have a redox gradient across adjacent cells? The authors mention that ROS can be passed between cells, but it would be helpful for them to provide more details about where this has been documented to occur and what biological functions ROS transfer regulates.

(3) Issues with scholarship of the testis<br /> a. Line 82 - There is no mention of BMPs, which are the only GSC-self-renewal signal. Upd/Jak/STAT is required for adhesion of GSCs to the niche but not self-renewal (Leatherman and Dinardo, 2008, 2010). The author should read a review about the testis. I suggest Greenspan et al 2015. The scholarship of the testis should be improved.<br /> b. Line 82-84 - BMPs are produced by both hub cells and CySCs. BMP signaling in GSCs represses bam. So it is not technically correct to say the CySCs repress bam expression in GSCs.<br /> c. Throughout the figures the authors score Vasa+ cells for GSCs. This is technically not correct. What they are counting is single, Vasa+ cells in contact with the niche. All graphs should be updated with the label "GSCs" on the Y-axis.

(4) Issues with the text<br /> a. Line 1: multi-lineage is not correct. Multi-lineage refers to stem cells that produce multiple types of daughter cells. GSCs produce only one type of offspring and CySCs produce only one type of offspring. So both are uni-lineage. Please change accordingly.<br /> b. Lines 62-75 - Intestinal stem cells have constitutively high ROS (Jaspar lab paper) so low ROS in stem cell cells is not an absolute.<br /> c. Line 79: The term cystic is not used in the Drosophila testis. There are cyst stem cells (CySCs) that produce cyst cells. Please revise.<br /> d. Line 90 - perfectly balanced is an overstatement and should be toned down.<br /> e. Line 98 - division of labour is not supported by the data and should be rephrased.<br /> f. Line 200 - the authors provide no data on BMPs - the GSC self-renewal cue - so they should avoid discussing an absence of self-renewal cues.

(5) Issues with the figures<br /> a. The images are too small to appreciate the location of mitochrondria in GSCs and CySCs.<br /> b. Figure 1<br /> i. cell membranes are not marked, reducing the precision of assigning mitochondria to GSC or CySCs. It would be very helpful if the authors depleted ATP5A from GSCs and showed that the puncta are reduced in these cells and did a similar set of experiments for the tj-Gal4 lineage. It would also be very helpful if the authors expressed membrane markers (like myr-GFP) in the GSC and then in the CySC lineage and then stained with ATP5A. This would pinpoint in which cells ATP5A immunoreactivity is occurring.<br /> ii. The presumed changes in gst-GFP (redox readout) are possibly seen in differentiating germ cells (i.e., spermatogonia) but not in GSC.<br /> iii. Panels F, Q, and S are not explained and currently are irrelevant.<br /> c. Figure 3K - The evidence to support less Ecad in GSCs in tj>SOD-i testes is not compelling as the figure is too small and the insets show changes in Ecad in somatic cells, not GSC.<br /> d. Figure 4:<br /> i. Panel A, B The apparent decline (not quantified) may not contribute to the phenotype.<br /> ii. dpERK is a finicky antibody and the authors are showing a single example of each genotype. This is an important experiment because the authors are going to use it to conclude that MAPK is decreased in the tj>SOD-i samples. However, the authors don't have any positive (dominant-active Egfr) or negative (tj>mapk-i). As is standing the data are not compelling. The graph in F does not convey any useful information.<br /> e. Figure S1D - cannot discern green on black. It is critical for the authors to show monochromes (gray scale) for the readouts that they want to emphasize. I cannot see the green on black in Figure S1D.<br /> f. Figure S4 - there is no quantification of the number of Tj cells in K-N.

(6) Issues with Methods<br /> a. Materials and Methods are not described in sufficient depth - please revise.<br /> b. Note that tj-Gal4 has real-time expression in hub cells and this is not considered by the authors. The ideal genotype for targeting CySCs is tjGal4, Gal80TS, hh-Gal80. Additionally, the authors do not mention whether they are depleting throughout development into adulthood or only in adults. If the latter, then they must have used a temperature shift like growing the flies at 18C and then upshifting to 25C or 29C during adult stages.<br /> c. The authors need to show data points in all of the graphs. Some graphs do this but others do not.<br /> d. The authors state that all data points are from three biological replicates. This is not sufficient for GSC and CySC counts. Most labs count GSCs and CySCs from at least 10 testes of the correct genotype.

eLife assessment

This work focuses on the role of Reactive Oxygen Species (ROS) signaling in cyst stem cells of the Drosophila testis. In particular, the authors suggest that ROS can act as signaling molecules between somatic and germ stem cells of the testis. The work is potentially useful, although the evidence that supports the authors' claims is incomplete.

Reviewer #1 (Public Review):

The manuscript by Majhi and colleagues describes the effects of manipulating ROS levels in somatic stem cells of the testis on overall testis architecture, signaling, and function. The conclusions made by the authors are somewhat difficult to judge as the changes to the testis cell types are mostly not apparent in the representative images shown. This is true in examining gstD1-GFP expression and in the analysis of cell types and behaviours (e.g. cell cycle) and cell signaling pathway activity. Thus, the reader is left to try and interpret the quantification of the data to justify the authors' conclusions, but it is often not clear how the quantification was accomplished. For example, it is not clear how CySC vs. GSC quantification is done when the molecular markers used do not define the surface of these cells (plasma membrane) and mark different cellular compartments (Tj is nuclear while Vasa is perinuclear or cytoplasmic). Why the changes reported in quantification are not apparent in the specific example images chosen for the figures is worrisome. I'm much more used to being able to clearly see what the authors are reporting in the images, and then using the quantification to illustrate the range of data observed and demonstrate statistical significance. For this reason, I'm very concerned about the strength and validity of the conclusions. In addition, while many different characteristics of the testis somatic and germline cells are analyzed, a general and consistent view of how ROS affects these cells is not presented. In particular, one of the principle conclusions, that ROS signaling in the CySCs affects ROS signaling in the GSCs, is not well-supported by the data presented.

Specific Comments:

In Figure 1, it is very difficult to identify where CySCs end and GSCs begin without using a cell surface marker for these different cell types. In addition, the methods for quantifying the mitochondrial distribution in GSCs vs. CySCs are very much unclear, and appear to rely on colocalization with molecular markers that are not in the same cellular compartment (Tj-nuclear vs Vasa-perinuclear and cytoplasmic), the reader has no way to determine the validity of the mitochondrial distribution. Similarly, the labeling with gstD1-GFP is also very much unclear - I see little to no GFP signal in either GSCs or CySCs in panels 1G-K. Lastly, while the expression of SOD in CySCs does increase the gstD1-GFP signal in CySCs, the effects on GSCs claimed by the authors are not apparent.

In Figure 2, while the cell composition of the niche region does appear to be different from controls when SOD1 is knocked down in the CySCs, at least in the example images shown in Figures 2A and B, how cell type is quantified in Figures 2E-G is very much unclear in the figure and methods. Are these counts of cells contacting the niche? If so, how was that defined? Or were additional regions away from the niche also counted and, if so, how were these regions defined?

In Figure 3, it is quite interesting that there is an increase in Eya+, differentiating cyst cells in SOD1 knockdown animals, and that these Eya+ cells appear closer to the niche than in controls. However, this seems at odds with the proliferation data presented in Figure 2, since Eya+ somatic cells do not normally divide at all. Are they suggesting that now differentiating cyst cells are proliferative? In addition, it is important for them to show example images of the changes in Socs36E and ptp61F expression.

Overall, the various changes in signaling are quite puzzling-while Jak/Stat signaling from the niche is reduced, hh signaling appears to be increased. Similarly, while the authors conclude that premature differentiation occurs close to the niche, EGF signaling, which occurs from germ cells to cyst cells during differentiation, is decreased. Many times these changes are contradictory, and the authors do not provide a suitable explanation to resolve these contradictions.

Reviewer #2 (Public Review):

Summary:

In this work, the authors investigate the role of the Superoxide disumutase 1 (Sod1) enzyme, which acts to reduce the reactive oxygen species load, in the Drosophila testis. They show that the knockdown of Sod1 in somatic cells impacts stem cell numbers both autonomously in the soma and non-autonomously in the germline. Somatic stem cell numbers are increased, while germline stem cells are decreased and differentiate prematurely. The authors then show that in somatic Sod1 knockdowns, several signalling pathways are disrupted and that these may be responsible at least in part for the phenotypes observed. Finally, over-expression of Sod1 in the soma results in opposite phenotypes, suggesting that ROS levels do play a role in maintaining the balance between both stem cell populations in the testis.

Strengths:

The main strength of this work is to show a previously unappreciated role for Sod1 (and presumably by extension of ROS) in the Drosophila testis and in the regulation of stem cell self-renewal and differentiation. The authors use multiple readouts to show that the knockdown of Sod1 in the soma increases the number of somatic cells and also drives a non-autonomous, premature differentiation of germ cells. They also quantify the early differentiation of the germline using two different methods. Importantly, overexpression of Sod1 produces opposite phenotypes to knockdown, strengthening the conclusions.

Weaknesses:

Although the data presented are interesting, an important weakness of the manuscript as it currently stands is that many statements are not fully supported by the data. In particular, the authors do not provide any evidence of "cell redox-pairs" as indicated in the manuscript title, nor of intercellular redox gradients, as stated in several places throughout. While the data are consistent with non-autonomous regulation of ROS levels, this would not constitute a gradient. However, and crucially, there is no evidence provided to show that Sod1 manipulation in the soma is affecting ROS levels in the germline and that any of the phenotypes observed are a consequence of elevated ROS in the germline, rather than indirect effects caused by dysregulation of somatic self-renewal and differentiation, which is known to impact the germline. Indeed, there are many published reports of autonomous manipulations in the soma that influence either germline stem cell number (eg PMID: 19797664 among others) or differentiation (eg PMID: 17629483). The latter example is particularly relevant as the authors show altered somatic ERK levels, and the role of somatic ERK in promoting germ cell development is well established (PMID: 11048722, 11048723,...). Thus, whether Sod1 plays any non-autonomous role in controlling germ cell fate through ROS in the germline directly, or whether the phenotypes observed can all be explained by autonomous effects on somatic cell behaviour is debatable, but the experiments presented here do not distinguish between these two hypotheses. The only evidence presented by the authors for a non-autonomous role of Sod1 is the expression of a GFP reporter for gstD1. The quantifications and images are not clear and do not show unambiguously that this reporter is expressed in germ cells. Indeed, the quantifications show overlap between somatic and germline markers, suggesting that either the images themselves or the way they are quantified does not allow the authors to distinguish between the two cell types. Similarly, the claim that somatic mitochondria are enriched at the CySC-GSC interface and that this distribution maintains the redox balance in the niche is not supported by any experimental data. CySCs are extremely thin cells and much of the space is occupied by the nucleus (PMID: 114676), therefore it is likely that mitochondria would be enriched at the periphery. A careful analysis would be necessary to show that this enrichment is specific to the interface with GSCs. Moreover, no experiments are conducted to test whether mitochondrial distribution in CySCs has any impact on GSCs. Finally, no experiments are conducted to show definitively that the phenotypes observed upon Sod1 knockdown are indeed due to increased ROS, while this claim is made several times in the text. At present, the data presented here can support a role for Sod1 in somatic CySCs, but much more caution is required in attributing this to either ROS or intercellular ROS signaling. Therefore, several claims made in the title and throughout the text are not supported by evidence.

Besides this central point, there are other areas that should be improved. In particular, the data using the Fucci reporter to show accelerated proliferation do not appear convincing. It would seem that the proportions of cells in each phase are roughly similar, just that there are more cycling cells. A careful analysis of these results would distinguish between these two and determine whether Sod1 knockdown simply impairs differentiation (and therefore results in more somatic cells proliferating) or whether it speeds up the cell cycle (resulting in an increased mitotic index as suggested, but this requires a ratio to be shown). Similarly, several quantifications are not clearly explained, making it hard to understand what is being measured. As an example, while the decrease in pERK in CySCs is clear from the image and matched in the quantification, the increase in cyst cells is not apparent from the fire LUT used. The change in fluorescence intensity therefore may be that more cells have active ERK, rather than an increase per cell (similar arguments apply to the quantifications for p4E-BP or Ptc). Therefore, it is hard to know whether Sod1 knockdown results in increased or decreased signaling in individual cells.

Impact of study:

Demonstrating intercellular communication through ROS and its importance in maintaining the balance between two stem cell populations would be a finding of interest to a broad field. However, it remains to be demonstrated that this is the case, and given this, this study will have a limited impact.

Reviewer #1 (Public Review):

In this work, the authors provide a valuable transcriptomic resource for the intermediate free-living transmission stage (miracidium larva) of the blood fluke. The single-cell transcriptome inventory is beautifully supplemented with in situ hybridization, providing spatial information and absolute cell numbers for many of the recovered transcriptomic states. The identification of sex-specific transcriptomic states within the populations of stem cells was particularly unexpected. The work comprises a rich resource to complement the biology of this complex system, however falls short in some technical aspects of the bioinformatic analyses of the generated sequence data.

(1) Four sequencing libraries were generated and then merged for analysis, however, the authors fail to document any parameters that would indicate that the clustering does not suffer from any batch effects.

(2) Additionally, the authors switch between analysis platforms without a clear motivation or explanation of what the fundamental differences between these platforms are. While in theory, any biologically robust observation should be recoverable from any permutation of analysis parameters, it has been recently documented that the two popular analysis platforms (Seurat - R and scanPy - python) indeed do things slightly differently and can give different results (https://www.biorxiv.org/content/10.1101/2024.04.04.588111v1). For this reason, I don't think that one can claim that Seurat fails to find clusters resolved by SAM without running a similar pipeline on the cluster alone as was done with SAM/scanPy here. The manuscript itself needs to be checked carefully for misleading statements in this regard.

(3) Similarly, the manuscript contains many statements regarding clusters being 'connected to', or forming a 'bridge' on the UMAP projection. One must be very careful about these types of statements, as the relative position of cells on a reduced-dimension cell map can be misleading (see Chari and Pachter 2023). To support these types of interpretations, the authors should provide evidence of gene expression transitions that support connectivity as well as stability estimates of such connections under different parameter conditions. Otherwise, these descriptors hold little value and should be dropped and the transcriptomic states simply defined as clusters with no reference to their positions on the UMAP.

(4) The underlying support for the clusters as transcriptomically unique identities is not well supported by the dot plots provided. The authors used very permissive parameters to generate marker lists, which hampers the identification of highly specific marker genes. This permissive approach can allow for extensive lists of upregulated genes for input into STRING/GO analyses, this is less useful for evaluating the robustness of the cluster states. Running the Seurat::FindAllMarkers with more stringent parameters would give a more selective set of genes to display and thereby increase the confidence in the reader as to the validity of profiles selected as being transcriptomically unique.

(5) Figure 5B shows a UMAP representation of cell positions with a statement that the clustering disappears. As a visual representation of this phenomenon, the UMAP is a very good tool, however, to make this statement you need to re-cluster your data after the removal of this gene set and demonstrate that the data no longer clusters into A/B and C/D. Also, as a reader, these data beg the question: which genes are removed here? Is there an over-representation of any specific 'types' of genes that could lead to any hypotheses of the function? Perhaps the STRING/GO analyses of this gene set could be informative.

(6) How do the proportions of cell types characterized via in situ here compare to the relative proportions of clusters obtained? It does not correspond to the percentages of the clusters captured (although this should be quantified in a similar manner in order to make this comparison direct: 10,686/20,478 = ~50% vs. 7%), how do you interpret this discrepancy? While this is mentioned in the discussion, there is no sufficient postulation as to why you have an overabundance of the stem cells compared to their presence in the tissue. While it is true that you could have a negative selection of some cell types, for example as stated the size of the penetration glands exceeds both that of the 10x capabilities (40uM), and the 30uM filters used in the protocol, this does not really address why over half of the captured cells represent 'stem cells'. A more realistic interpretation would be biological rather than merely technical. For example, while the composition of the muscle cells and the number of muscle transcriptomes captured are quite congruent at ~20%, the organism is composed of more than 50% of neurons, but only 15% of the transcriptomic states are assigned to neuronal. Could it be that a large fraction of the stem cells are actually neural progenitors? Are there other large inconsistencies between the cluster sizes and the fraction of expected cells? Could you look specifically at early transcription factors that are found in the neurons (or other cell types) within the various stem cell populations to help further refine the precursor/cell type relationships?

eLife assessment

This is a valuable study in which the authors provide an expression profile of the human blood fluke, Schistosoma mansoni. A strength of this solid study is in its inclusion of in situ hybridisation to validate the predictions of the transcript analysis.

Reviewer #2 (Public Review):

Summary:

In this manuscript the authors have generated a single-cell atlas of the miracidium, the first free-living stage of an important human parasite, Schistosoma mansoni. Miracidia develop from eggs produced in the mammalian (human) host and are released into freshwater, where they can infect the parasite's intermediate snail host to continue the life cycle. This study adds to the growing single-cell resources that have already been generated for other life-cycle stages and, thus, provides a useful resource for the field.

Strengths:

Beyond generating lists of genes that are differentially expressed in different cell types, the authors validated many of the cluster-defining genes using in situ hybridization chain reaction. In addition to providing the field with markers for many of the cell types in the parasite at this stage, the authors use these markers to count the total number of various cell types in the organism. Because the authors realized that their cell isolation protocols were biasing the cell types they were sequencing, they applied a second method to help them recover additional cell types.

Schistosomes have ZW sex chromosomes and the authors make the interesting observation that the stem cells at this stage are already expressing sex (i.e. W)-specific genes.

Weaknesses:

The sample sizes upon which the in situ hybridization results and cell counts are based are either not stated (in most cases) or are very small (n=3). This lack of clarity about biological replicates and sample sizes makes it difficult for the reader to assess the robustness of the results and the extremely small sample sizes (when provided) are a missed opportunity to explore the variability of the system, or lack thereof.

Although assigning transcripts to a given cell type is usually straightforward via in situ experiments, the authors fail to consider the potential difficulty of assigning the appropriate nuclei to cells with long cytoplasmic extensions, like neurons. In the absence of multiple markers and a better understanding of the nervous system, it seems likely that the authors have overestimated the number of neurons and misassigned other cell types based on their proximity to neural projections.

The conclusion that germline genes are expressed in the miracidia stem cells seems greatly overstated in the absence of any follow-up validation. The expression scales for genes like eled and boule are more than 3 orders of magnitude smaller than those used for any of the robustly expressed genes presented throughout the paper. These scales are undefined, so it isn't entirely clear what they represent, but neither of these genes is detected at levels remotely high (or statistically significant) enough to survive filters for cluster-defining genes. Given that germ cells often develop early in embryogenesis and arrest the cell cycle until later in development, and that these transcripts reveal no unspliced forms, it seems plausible that the authors are detecting some maternally supplied transcripts that have yet to be completely degraded.

Reviewer #1 (Public Review):

Summary:

The article explores the connection between immunogenic cell death (ICD)-related genes and bladder cancer prognosis, immune infiltration, and response to therapy. The study identifies a risk-scoring model involving four ICD-related genes (CALR, IL1R1, IFNB1, IFNG), showing a correlation between higher risk scores and weaker anti-tumor immune function.

Strengths:

The significance lies in the potential for personalized treatment guidance in bladder cancer. The establishment of a risk-scoring model to predict patient survival is noteworthy.

Weaknesses:

However, the identification of ICD-related genes is somewhat conventional, focusing on known genes regulating cancer immune response. To enhance the significance of the risk-scoring model, it would be better if the authors could validate the model across various cancer types. The strength of evidence appears moderate, but broader applicability would strengthen the findings.

eLife assessment

This study investigates the associations of four ICD-related genes in bladder cancer with increased immune cell infiltration and more prolonged survival. The study is valuable because it identifies a risk-scoring model, showing a correlation between high-risk scores based on four ICD-related genes and weak anti-tumour immune function. However, the evidence supporting the association of these genes and immunotherapy response is incomplete.

Reviewer #2 (Public Review):

Immunogenic cell death (ICD) can lead to the release of factors such as DAMPs which promote an adaptive immune response. In the context of cancer, there is clear evidence of anti-tumour benefits as a result of ICD, perhaps induced by chemotherapy.

Lilong et al used TCGA data to explore whether a previously published 34 gene 'ICD-related' signature could stratify bladder cancer patients by prognosis and ultimately predict patient survival. The gene signature contains many genes involved in inflammation and immunity (IFNg, IL6, TNF, IL17A, TLR4, CD8B, etc) and those related to ICD (such as CALR, HMGB1, HSP, NLRP3, etc). The authors divide patients into 'ICD-high' and '-low' based on the expression of this gene set and find that 'ICD-high' is associated with longer survival in TCGA bladder cancer data. The authors further argue that ICD-high group responds better to PD1 therapies. From this 34-gene signature, it appears that LASSO regularisation and Cox analysis identifies a four-gene 'risk' signature (CALR, IL1R1, IFNB1, IFNG) which is associated with shorter patient survival and lower immunotherapy response rates. This is the primary finding. Their methodology is very similar to a publication in 2021 in Frontiers in Immunology instead in the context of head and neck squamous cell carcinoma. This paper is not referenced.

In terms of the strengths of the work, it is certainly plausible that the author's four gene signature has an association with survival in bladder cancer, at least based on the two datasets studied. However, the relatedness of their findings to ICD is unconvincing, and glaring omissions from the manuscript in terms of methods limit confidence in the work. The authors show a potential association with bladder cancer patient survival and their four gene signatures, but substantial revisions are required for this to be appropriately evidenced.

Reviewer #2 (Public Review):

Summary:

The authors provide compelling evidence that stimulation of epidermal cells in Drosophila larvae results in the stimulation of sensory neurons that evoke a variety of behavioral responses. Further, the authors demonstrate that epidermal cells are inherently mechanoresponsive and implicate a role for store-operated calcium entry (mediated by Stim and Orai) in the communication to sensory neurons.

Strengths:

The study represents a significant advance in our understanding of mechanosensation. Multiple strengths are noted. First, the genetic analyses presented in the paper are thorough with appropriate consideration to potential confounds. Second, behavioral studies are complemented by sophisticated optogenetics and imaging studies. Third, identification of roles for store-operated calcium entry is intriguing. Lastly, conservation of these pathways in vertebrates raise the possibility that the described axis is also functional in vertebrates.

Weaknesses:

The study has a few conceptual weaknesses that are arguably minor. The involvement of store-operated calcium entry implicates ER calcium store release. Whether mechanical stimulation evokes ER calcium release in epidermal cells and how this might come about (e.g., which ER calcium channels, roles for calcium-induced calcium release etc.) remains unaddressed. On a related note, the kinetics of store-operated calcium entry is very distinct from that required for SV release. The link between SOC and epidermal cells-neuron transmission is not reconciled. Finally, it is not clear how optogenetic stimulation of epidermal cells results in the activation of SOC.

eLife assessment

This is an important work and provides a significant advance in our understanding of mechanosensation in the epidermis. The evidence presented is solid, however, additional work such as testing whether the activation time can be shorter, addressing the mechanism underlying endoplasmic reticulum calcium release, and improving the clarity of writing and rigor of analysis would strengthen the study. This work will be of broad interest to neurobiologists, epithelial cell biologists, and mechanobiologists.

Reviewer #1 (Public Review):

Summary:

In this meticulously conducted study, the authors show that Drosophila epidermal cells can modulate escape responses to noxious mechanical stimuli. First, they show that activation of epidermal cells evokes many types of behaviors including escape responses. Subsequently, they demonstrate that most somatosensory neurons are activated by activation of epidermal cells, and that this activation has a prolonged effect on escape behavior. In vivo analyses indicate that epidermal cells are mechanosensitive and require stored-operated calcium channel Orai. Altogether, the authors conclude that epidermal cells are essential for nociceptive sensitivity and sensitization, serving as primary sensory noxious stimuli.

Strengths:

The manuscript is clearly written. The experiments are logical and complementary. They support the authors' main claim that epidermal cells are mechanosensitive and that epidermal mechanically evoked calcium responses require the stored-operated calcium channel Orai. Epidermal cells activate nociceptive sensory neurons as well as other somatosensory neurons in Drosophila larvae, and thereby prolong escape rolling evoked by mechanical noxious stimulation.

Weaknesses:

Core details are missing in the protocols, including the level of LED intensity used, which are necessary for other researchers to reproduce the experiments. For most experiments, the epidermal cells are activated for 60 s, which is long when considering that nocifensive rolling occurs on a timescale of milliseconds. It would be informative to know the shortest duration of epidermal cell activation that is sufficient for observing the behavioral phenotype (prolongation of escape behavior) and activation of sensory neurons.

eLife assessment

The study presents important findings on the role of MSI2-HOXA9 translocation in chronic myeloid leukemia. The authors provide convincing evidence supporting the role of this translocation in leukemogenesis by using elegant mouse modeling and in vitro mechanistic studies. Consistent with the reviews, the studies can be strengthened with further murine and cell line experiments.

Reviewer #1 (Public Review):

This is a very interesting study by Kyle Spinler et al., demonstrating the novel role of MSI2-HOXA9 translocation in the development and pathogenesis of blast crisis CML. The authors employed appropriate in vitro and in vivo assays, including a sophisticated transplantation-based model of CML, which is well-established in the field of studying the pathogenesis of CML. Additionally, the authors successfully concluded that the MSI2 RNA binding domain RRM1 has a preferential impact on the growth of blast crisis CML.

The quality of this research article could be significantly enhanced by addressing the following points:

Major:

(1) Do mice with BCR-ABL/MSI2-HOXA9 leukemia have an increased pool of leukemic stem cells (LSC), or do they have an increased propensity to develop blast cells? Is it the number of LSCs that has increased, or is it the function of LSC to give rise to the disease that has increased? It is not clear if the detected differences in Lineage-negative cells (Figure S1D) were detected in vitro in retrovirally transduced cells or were detected in vivo in transplanted mice. If the differences were detected in vitro, could the author confirm the same findings in vivo? This will greatly enhance the understanding of in vivo disease pathogenesis and could directly link the aggressivity of the disease (shortened survival) with an increased stem cell-like population.

(2) The authors suggest that BCR-ABL/MSI2-HOXA9 leads to the development of blast crisis-CML. One of the main characteristics of blast crisis-CML is drug resistance. Is BCR-ABL/MSI2-HOXA9 leukemia resistant to classical CML treatment drugs?

(3) The authors have emphasized the heightened expression of Polrmt in delineating the mitochondrial phenotype of BCR-ABL/MSI2-HOXA9 leukemia cells. However, the regulatory mechanism governing the expression of Polrmt by MSI2-HOXA9 has not been clearly demonstrated by the authors. Unveiling this mechanism would constitute a novel finding and significantly elevate the quality of the research.

(4) Did the authors observe any survival differences between BCR-ABL/NUP98-HOXA9 and BCR-ABL/MSI2-HOXA9?

Reviewer #2 (Public Review):

The manuscript titled, "Identification of a Musashi2 translocation as a novel oncogene in myeloid leukemia" by Spinler et al. studies the functional role of the translocation t(7;17)(p15;q23), resulting in MSI2/HOXA9 fusion gene, as a secondary driver in bcCML. MSI2-HOXA9 forced expression along with BCR-ABL enhances colony formation and leads to a more aggressive disease in vivo. Depletion of the RNA binding domain RRM1 or RRM2 of MSI2 led to a significant reduction in colony formation, with RRM1 depletion specifically impacting differentiation and blast cell counts. Mechanistically, the authors find that MSI2-HOXA9 aberrantly localizes to the nucleus, elevating the expression of mitochondrial polymerase Polrmt, thereby leading to upregulation of mitochondrial components and enhancing mitochondrial function and basal respiration. Overall, this study examines how the rare MSI2-HOXA9 fusion gene can act as a novel cooperating oncogene and could serve as a secondary hit in the progression of CML to blast crisis.

Strengths:

(1) Demonstration that MSI2-HOXA9 contributes to oncogenesis in the BCR-ABL context.

(2) Development of a novel cooperativity model for BCR-ABL and provides additional supporting data for the role of MSI2 in leukemogenesis.

(3) Evidence that MSI2-HOXA9 acts uniquely compared to MSI2 alone through nuclear vs. cytoplasmic localization and activation of mitochondrial polymerase Polrmt.

Weaknesses:

(1) MSI2-HOXA9 fusion is extremely rare as it has been only found in a handful of patients and it is not clear whether other MSI2 fusions function in a similar manner.

(2) The mechanism needs to be strengthened since MSI2 alone or the HOXA9 mutant may not be linked to the mitochondrial mechanism.

(3) It is not clear that the mitochondrial pathway is sufficient for the MSI2-HOXA9 oncogenic mechanism.

Author response:

We are grateful to the reviewers for their interest and enthusiasm about the work, and deeply appreciate their constructive comments and suggestions. Our responses are below

(1) Do mice with BCR-ABL/MSI2-HOXA9 leukemia have an increased pool of leukemic stem cells (LSC), or do they have an increased propensity to develop blast cells? Is it the number of LSCs that has increased, or is it the function of LSC to give rise to the disease that has increased? It is not clear if the detected differences in Lineage-negative cells (Figure S1D) were detected in vitro in retrovirally transduced cells or were detected in vivo in transplanted mice. If the differences were detected in vitro, could the author confirm the same findings in vivo? This will greatly enhance the understanding of in vivo disease pathogenesis and could directly link the aggressivity of the disease (shortened survival) with an increased stem cell-like population.

We find that BCR-ABL/MSI2-HOXA9 leads to a marked increase in Lineage negative (Lin-) cells which contains the LSC fraction. Specifically, the LSC containing fraction represented 14.1% of the BCR-ABL driven disease and 56.7% of the BCR-ABL and MSI2-HOXA9 driven disease (p<.0001). This suggests that MSI2-HOXA9 triggers the expansion of the undifferentiated LSC containing pool. In addition, the blast frequency was also increased albeit to a lesser extent, with 63.8% blasts (SEM 1.1) for BCR-ABL and 83.3% (SEM 3.1) for BCR-ABL/MSI2-HOXA9 (p=.0001). This suggests that the resulting aggressive disease seen with MSI2-HOXA9 is a consequence of a large increase in undifferentiated  LSC containing cells, as well as the resulting increase in the blast count. The Lin- cells were analyzed from fully established leukemias in vivo (Fig. S1D)

(2) The authors suggest that BCR-ABL/MSI2-HOXA9 leads to the development of blast crisis-CML. One of the main characteristics of blast crisis-CML is drug resistance. Is BCR-ABL/MSI2-HOXA9 leukemia resistant to classical CML treatment drugs?

The sensitivity to Imatinib is a very interesting question. In general, while differentiated cells in CML are sensitive to Imatinib, the more undifferentiated cells (LSCs) are resistant1,2. Based on the fact that therapy resistance in blast crisis is largely driven by the undifferentiated fraction of leukemia cells, and given that BCR-ABL/MSI2-HOXA9 driven disease harbors a larger fraction of these undifferentiated cells, we would predict that BCR-ABL/MSI2-HOXA9 leukemia would also be more resistant to imatinib. However, this would need to be experimentally demonstrated and is an important question to address.

(3) The authors have emphasized the heightened expression of Polrmt in delineating the mitochondrial phenotype of BCR-ABL/MSI2-HOXA9 leukemia cells. However, the regulatory mechanism governing the expression of Polrmt by MSI2-HOXA9 has not been clearly demonstrated by the authors. Unveiling this mechanism would constitute a novel finding and significantly elevate the quality of the research.

Since Polrmt and mitochondrial genes are transcribed in the nucleus we explored whether MSI2-HOXA9 may control mitochondrial gene expression by triggering expression of Polrmt and other key transcription factors. Consistent with this possibility, MSI2-HOXA9 was preferentially found in the nucleus relative to MSI2. In addition, there were 10 occurrences of the minimal MSI2 RRM1 consensus binding sequence UAGU within the Polrmt transcript. While this is consistent with the possibility that Polrmt expression can be post-transcriptionally modulated by MSI2-HOXA9, this needs to be experimentally validated using Clip Seq analysis with wild type MSI2 as well as the MSI2-HOXA9 fusion protein in context of blast crisis CML.

(4) Did the authors observe any survival differences between BCR-ABL/NUP98-HOXA9 and BCR-ABL/MSI2-HOXA9?

In previous work from our lab we have found that the median survival for BCR-ABL/NUP98-HOXA9 was 17 days, and with BCR-ABL/ MSI2-HOXA9 was 18.5 days (p value of 0.22). This suggests that there is not a significant difference in survival times between the leukemias driven by the distinct alleles, and they may be equally aggressive.

(1) MSI2-HOXA9 fusion is extremely rare as it has been only found in a handful of patients and it is not clear whether other MSI2 fusions function in a similar manner.

We were very surprised and excited to see the large number of translocations in solid cancers that involve MSI2.  Interestingly, MSI2 translocations occurred both at the N and the C terminus.  Distinct translocations are likely to have unique roles in each disease context. For example, if MSI2’s 5 prime end is part of a translocation, it may functionally contribute via its promoter to drive expression in immature cells and could thus activate oncogenic signals (e.g. controlled by the partner gene) in immature cells which are inherently more susceptible to transformation (Eµ-myc is an example of such a translocation). If Msi2’s RRM domains are part of the fusion, they could bind and target RNAs aberrantly (such as in the wrong cell and the wrong time) and lead to activation of downstream oncogenic mediators. To fully understand the role of each of these translocations in each specific cancer, we would need to experimentally test their impact by ectopic expression in the appropriate cell of origin and domain mapping the basis of any impact in the relevant cancer models as we have done for MSI2-HOXA9 in blast crisis CML in the work we report here.   While this is an intensive undertaking, it is nonetheless important future work as it will undoubtedly lead to new insight about MSI2 linked translocations in diverse solid cancers such as breast cancer and lung cancer.

(2) The mechanism needs to be strengthened since MSI2 alone or the HOXA9 mutant may not be linked to the mitochondrial mechanism. (3) It is not clear that the mitochondrial pathway is sufficient for the MSI2-HOXA9 oncogenic mechanism.

Our observation that MSI2-HOXA9 triggered changes in mitochondrial function was of particular interest as it was (to our knowledge) uncharted in context of Msi2 signaling in cancer, thus leading us to explore this further.  However, multiple other signals are likely downstream regulators and these may well act cooperatively with, or independently of, the heightened­­ mitochondrial function we report here. Among these pathways, the most likely mediators included oncogenic programs related to the Wnt pathway including Wnt, Fzd 3 and Frat1, and those related to the Notch pathway including-Tribbles and Hey1 as well as other stem cell genes such as Aldh1. These programs have been previously implicated in the regulation of myeloid leukemia3-11 and could well mediate the impact of the MSI2-HOXA9 translocation. The relative contribution of mitochondrial metabolism and that of developmental and stem cell signals to the onset of MSI2-HOXA9 driven blast crisis CML is an important avenue of future work.

References

(1) Corbin, A. S., Agarwal, A., Loriaux, M., Cortes, J., Deininger, M. W. & Druker, B. J. 2011. Human chronic myeloid leukemia stem cells are insensitive to imatinib despite inhibition of BCR-ABL activity. J Clin Invest 121: 396-409. PMC3007128.

(2) Graham, S. M., Jørgensen, H. G., Allan, E., Pearson, C., Alcorn, M. J., Richmond, L. & Holyoake, T. L. 2002. Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571 in vitro. Blood 99: 319-325.

(3) Gurska, L. M., Ames, K. & Gritsman, K. 2019. Signaling Pathways in Leukemic Stem Cells. Adv Exp Med Biol 1143: 1-39. PMC7249489.

(4) Narendra, G., Raju, B., Verma, H. & Silakari, O. 2021. Identification of potential genes associated with ALDH1A1 overexpression and cyclophosphamide resistance in chronic myelogenous leukemia using network analysis. Med Oncol 38: 123.

(5) Ran, D., Schubert, M., Pietsch, L., Taubert, I., Wuchter, P., Eckstein, V., Bruckner, T., Zoeller, M. & Ho, A. D. 2009. Aldehyde dehydrogenase activity among primary leukemia cells is associated with stem cell features and correlates with adverse clinical outcomes. Exp Hematol 37: 1423-1434.

(6) Reya, T., Duncan, A. W., Ailles, L., Domen, J., Scherer, D. C., Willert, K., Hintz, L., Nusse, R. & Weissman, I. L. 2003. A role for Wnt signalling in self-renewal of haematopoietic stem cells. Nature 423: 409-414.

(7) Riether, C., Schürch, C. M., Bührer, E. D., Hinterbrandner, M., Huguenin, A. L., Hoepner, S., Zlobec, I., Pabst, T., Radpour, R. & Ochsenbein, A. F. 2017. CD70/CD27 signaling promotes blast stemness and is a viable therapeutic target in acute myeloid leukemia. J Exp Med 214: 359-380. PMC5294846.

(8) Riether, C., Schürch, C. M., Flury, C., Hinterbrandner, M., Drück, L., Huguenin, A. L., Baerlocher, G. M., Radpour, R. & Ochsenbein, A. F. 2015. Tyrosine kinase inhibitor-induced CD70 expression mediates drug resistance in leukemia stem cells by activating Wnt signaling. Sci Transl Med 7: 298ra119.

(9) Venton, G., Pérez-Alea, M., Baier, C., Fournet, G., Quash, G., Labiad, Y., Martin, G., Sanderson, F., Poullin, P., Suchon, P., Farnault, L., Nguyen, C., Brunet, C., Ceylan, I. & Costello, R. T. 2016. Aldehyde dehydrogenases inhibition eradicates leukemia stem cells while sparing normal progenitors. Blood Cancer J 6: e469. PMC5056970.

(10) Yin, D. D., Fan, F. Y., Hu, X. B., Hou, L. H., Zhang, X. P., Liu, L., Liang, Y. M. & Han, H. 2009. Notch signaling inhibits the growth of the human chronic myeloid leukemia cell line K562. Leuk Res 33: 109-114.

(11) Kang, Y. A., Pietras, E. M. & Passegué, E. 2020. Deregulated Notch and Wnt signaling activates early-stage myeloid regeneration pathways in leukemia. J Exp Med 217. PMC7062512.

eLife assessment

This study provides useful data substantiating a role of long noncoding RNAs in liver metabolism and organismal physiology. With murine knockout and knockin models, the authors invoke a previously unidentified role for the lncRNA Snhg3 in fatty liver. While certain findings are backed by solid evidence, other conclusions require more support and should be consolidated with existing paradigms in the field.

Reviewer #1 (Public Review):

Summary:

In this manuscript, the authors investigate the contributions of the long noncoding RNA snhg3 in liver metabolism and MAFLD. The authors conclude that liver-specific loss or overexpression of Snhg3 impacts hepatic lipid content and obesity through epigenetic mechanisms. More specifically, the authors invoke that the nuclear activity of Snhg3 aggravates hepatic steatosis by altering the balance of activating and repressive chromatin marks at the Pparg gene locus. This regulatory circuit is dependent on a transcriptional regulator SNG1.

Strengths:

The authors developed a tissue-specific lncRNA knockout and KI models. This effort is certainly appreciated as few lncRNA knockouts have been generated in the context of metabolism. Furthermore, lncRNA effects can be compensated in a whole organism or show subtle effects in acute versus chronic perturbation, rendering the focus on in vivo function important and highly relevant. In addition, Snhg3 was identified through a screening strategy and as a general rule the authors the authors attempt to follow unbiased approaches to decipher the mechanisms of Snhg3.

Weaknesses:

Despite efforts at generating a liver-specific knockout, the phenotypic characterization is not focused on the key readouts. Notably missing are rigorous lipid flux studies and targeted gene expression/protein measurement that would underpin why the loss of Snhg3 protects from lipid accumulation. Along those lines, claims linking the Snhg3 to MAFLD would be better supported with careful interrogation of markers of fibrosis and advanced liver disease. In other areas, significance is limited since the presented data is either not clear or rigorous enough. Finally, there is an important conceptual limitation to the work since PPARG is not established to play a major role in the liver.

Reviewer #2 (Public Review):

Through RNA analysis, Xie et al found LncRNA Snhg3 was one of the most down-regulated Snhgs by a high-fat diet (HFD) in mouse liver. Consequently, the authors sought to examine the mechanism through which Snhg3 is involved in the progression of metabolic dysfunction-associated fatty liver diseases (MASLD) in HFD-induced obese (DIO) mice. Interestingly, liver-specific Sngh3 knockout was reduced, while Sngh3 over-expression potentiated fatty liver in mice on an HFD. Using the RNA pull-down approach, the authors identified SND1 as a potential Sngh3 interacting protein. SND1 is a component of the RNA-induced silencing complex (RISC). The authors found that Sngh3 increased SND1 ubiquitination to enhance SND1 protein stability, which then reduced the level of repressive chromatin H3K27me3 on PPARg promoter. The upregulation of PPARg, a lipogenic transcription factor, thus contributed to hepatic fat accumulation.

The authors propose a signaling cascade that explains how LncRNA sngh3 may promote hepatic steatosis. Multiple molecular approaches have been employed to identify molecular targets of the proposed mechanism, which is a strength of the study. There are, however, several potential issues to consider before jumping to a conclusion.

(1) First of all, it's important to ensure the robustness and rigor of each study. The manuscript was not carefully put together. The image qualities for several figures were poor, making it difficult for the readers to evaluate the results with confidence. The biological replicates and numbers of experimental repeats for cell-based assays were not described. When possible, the entire immunoblot imaging used for quantification should be presented (rather than showing n=1 representative). There were multiple mislabels in figure panels or figure legends (e.g., Figure 2I, Figure 2K, and Figure 3K). The b-actin immunoblot image was reused in Figure 4J, Figure 5G, and Figure 7B with different exposure times. These might be from the same cohort of mice. If the immunoblots were run at different times, the loading control should be included on the same blot as well.

(2) The authors can do a better job in explaining the logic for how they came up with the potential function of each component of the signaling cascade. Sngh3 is down-regulated by HFD. However, the evidence presented indicates its involvement in promoting steatosis. In Figure 1C, one would expect PPARg expression to be up-regulated (when Sngh3 was down-regulated). If so, the physiological observation conflicts with the proposed mechanism. In addition, SND1 is known to regulate RNA/miRNA processing. How do the authors rule out this potential mechanism? How about the hosting snoRNA, Snord17? Does it involve the progression of NASLD?

(3) The role of PPARg in fatty liver diseases might be a rodent-specific phenomenon. PPARg agonist treatment in humans may actually reduce ectopic fat deposition by increasing fat storage in adipose tissues. The relevance of the findings to human diseases should be discussed.

eLife assessment

The study presents a potentially useful approach to genetically modify cells to produce extracellular matrices with altered compositions. The evidence supporting the authors' conclusions regarding the chondrogenicity of lyophilized constructs is considered incomplete, as the study does not adequately demonstrate the formation of a histologically identifiable cartilaginous matrix. The study also lacks several significant details and does not have sufficient power to support the conclusions.

Reviewer #1 (Public Review):

Summary:

The authors aimed to modify the characteristics of the extracellular matrix (ECM) produced by immortalized mesenchymal stem cells (MSCs) by employing the CRISPR/Cas9 system to knock out specific genes. Initially, they established VEGF-KO cell lines, demonstrating that these cells retained chondrogenic and angiogenic properties. Additionally, lyophilized carriage tissues produced by these cells exhibited retained osteogenic properties.

Subsequently, the authors established RUNX2-KO cell lines, which exhibited reduced COLX expression during chondrogenic differentiation and notably diminished osteogenic properties in vitro. Transplantation of lyophilized carriage tissues produced by RUNX2-KO cell lines into osteochondral defects in rat knee joints resulted in the regeneration of articular cartilage tissues as well as bone tissues, a phenomenon not observed with tissues derived from parental cells. This suggests that gene-edited MSCs represent a valuable cell source for producing ECM with enhanced quality.

Strengths:

The enhanced cartilage regeneration observed with ECM derived from RUNX2-KO cells supports the authors' strategy of creating gene-edited MSCs capable of producing ECM with superior quality. Immortalized cell lines offer a limitless source of off-the-shelf material for tissue regeneration.

Weaknesses:

Most data align with anticipated outcomes, offering limited novelty to advance scientific understanding. Methodologically, the chondrogenic differentiation properties of immortalized MSCs appeared deficient, evidenced by Safranin-O staining of 3D tissues and histological findings lacking robust evidence for endochondral differentiation. This presents a critical limitation, particularly as authors propose the implantation of cartilage tissues for in vivo experiments. Instead, the bulk of data stemmed from type I collagen scaffold with factors produced by MSCs stimulated by TGFβ.

The rationale behind establishing VEGF-KO cell lines remains unclear. What specific outcomes did the authors anticipate from this modification?

Insufficient depth was given to elucidate the disparity in osteogenic properties between those observed in ectopic bone formation and those observed in transplantation into osteochondral defects. While the regeneration of articular cartilage in RUNX2-KO ECM presents intriguing results, the study lacked an exploration into underlying mechanisms, such as histological analyses at earlier time points.

Reviewer #2 (Public Review):

The manuscript submitted by Sujeethkumar et al. describes an alternative approach to skeletal tissue repair using extracellular matrix (ECM) deposited by genetically modified mesenchymal stromal/stem cells. Here, they generate a loss of function mutations in VEGF or RUNX2 in a BMP2-overexpressing MSC line and define the differences in the resulting tissue-engineered constructs following seeding onto a type I collagen matrix in vitro, and following lyophilization and subcutaneous and orthotopic implantation into mice and rats. Some strengths of this manuscript are the establishment of a platform by which modifications in cell-derived ECM can be evaluated both in vitro and in vivo, the demonstration that genetic modification of cells results in complexity of in vitro cell-derived ECM that elicits quantifiable results, and the admirable goal to improve endogenous cartilage repair. However, I recommend the authors clarify their conclusions and add more information regarding reproducibility, which was one limitation of primary-cell-derived ECMs.

Overcoming the limitations of native/autologous/allogeneic ECMs such as complete decellularization and reduction of batch-to-batch variability was not specifically addressed in the data provided herein. For the maintenance of ECM organization and complexity following lyophilization, evidence of complete decellularization was not addressed, but could be easily evaluated using polarized light microscopy and quantification of human DNA for example in constructs pre and post-lyophilization. It would be ideal to see minimization of batch-to-batch variability using this approach, as mitigation of using a sole cell line is likely not sufficient (considering that the sole cell line-derived Matrigel does exhibit batch-to-batch and manufacturer-to-manufacturer variability).

I recommend adding details regarding experimental design and outcomes not initially considered. Inter- and intra-experimental reproducibility was not adequately addressed. The size of in vitro-derived cartilage pellets was not quantified, and it is not clear that more than one independent 'differentiation' was performed from each gene-edited MSC line to generate in vitro replicates and constructs that were implanted in vivo.

The use of descriptive language in describing conclusions may mislead the reader and should be modified accordingly throughout the manuscript. For example, although this reviewer agrees with the comparative statements made by the authors regarding parental and gene-edited MSC lines, non-quantifiable terms such as 'frank' 'superior' (example, line 242) are inappropriate and should rather be discussed in terms of significance. Another example is 'rich-collagenous matrix,' which was not substantiated by uniform immunostaining for type II collagen (line 189).

I have similar recommendations regarding conclusive statements from the rat implantation model, which was appropriately used for the purpose of evaluating the response of native skeletal cells to the different cell-derived ECMs. Interpretations of these results should be described with more accuracy. For example, increased TRAP staining does not indicate reduced active bone formation (line 237). Many would not conclude that GAGs were retained in the RUNX2-KO line graft subchondral region based on the histology. Quantification of % chondral regeneration using histology is not accurate as it is greatly influenced by the location in the defect from which the section was taken. Chondral regeneration is usually semi-quantified from gross observations of the cartilage surface immediately following excision. The statements regarding integration (example line 290) are not founded by histological evidence, which should show high magnification of the periphery of the graft adjacent to the native tissue.

Reviewer #3 (Public Review):

Summary:

In this study, the authors have started off using an immortalized human cell line and then gene-edited it to decrease the levels of VEGF1 (in order to influence vascularization), and the levels of Runx2 (to decrease chondro/osteogenesis). They first transplanted these cells with a collagen scaffold. The modified cells showed a decrease in vascularization when VEGF1 was decreased, and suggested an increase in cartilage formation.

In another study, the matrix generated by these cells was subsequently remodeled into a bone marrow organ. When RUNX2 was decreased, the cells did not mineralize in vitro, and their matrices expressed types I and II collagen but not type X collagen in vitro, in comparison with unedited cells. In vivo, the author claims that remodeling of the matrices into bone was somewhat inhibited. Lastly, they utilized matrices generated by RUNX2 edited cells to regenerate chondro-osteal defects. They suggest that the edited cells regenerated cartilage in comparison with unedited cells.

Strengths:

-The notion that inducing changes in the ECM by genetically editing the cells is a novel one, as it has long been thought that ECM composition influences cell activity.

-If successful, it may be possible to make off-the-shelf ECMS to carry out different types of tissue repair.

Weaknesses:

-The authors have not generated histologically identifiable cartilage or bone in their transplants of the cells with a type I scaffold.

-In many cases, they did not generate histologically identifiable cartilage with their cell-free-edited scaffold. They did generate small amounts of bone but this is most likely due to BMPs that were synthesized by the cells and trapped in the matrix.

-There is a great deal of missing detail in the manuscript.

-The in vivo study is underpowered, the results are not well documented pictorially, and are not convincing.

-Given the fact that they have genetically modified cells, they could have done analyses of ECM components to determine what was different between the lines, both at the transcriptome and the protein level. Consequently, the study is purely descriptive and does not provide any mechanistic understanding of what mixture of matrix components and growth factors works best for cartilage or bone. But this presupposes that they actually induced the formation of bona fide cartilage, at least.

eLife assessment

This is a mechanistic study showing the effect of combining inhibition of autophagy (through ULK1/2) and KRAS (using sotorasib) on KRAS mutant NSCLC making the study valuable to cancer biologists and more broadly in a clinical setting. The evidence generated by GEM mouse models and cell lines is solid but could be further strengthened by increasing the mouse cohort size. This study holds translational relevance beyond NSCLC to other indications that carry KRAS mutations.

Reviewer #1 (Public Review):

Summary:

Given that KRAS inhibition approaches are a relatively new innovation and that resistance is now being observed to such therapies in patients with NSCLC, investigation of combination therapies is valuable. The manuscript furthers our understanding of combination therapy for KRAS mutant non-small cell lung cancer by providing evidence that combined inhibition of ULK1/2 (and therefore autophagy) and KRAS can inhibit KRAS-mutant lung cancer growth. The manuscript will be of interest to the lung cancer community but also to researchers in other cancer types where KRAS inhibition is relevant.

Strengths:

The manuscript combines cell line, cell line-derived xenograft, and genetically-engineered mouse model data to provide solid evidence for the proposed combination therapy.

The manuscript is well written, and experiments are broadly well performed and presented.

Weaknesses:

With 3-4 mice per group in many experiments, experimental power is a concern and some comparisons (e.g. mono vs combination therapy) seem to be underpowered to detect a difference. Both male and female mice are used in experiments which may increase variability.

Reviewer #2 (Public Review):

Summary:

In this manuscript, Ghazi et reported that inhibition of KRASG12C signaling increases autophagy in KRASG12C-expressing lung cancer cells. Moreover, the combination of DCC 3116, a selective ULK1/2 inhibitor, plus sotorasib displays cooperative/synergistic suppression of human KRASG12C-driven lung cancer cell proliferation in vitro and tumor growth in vivo. Additionally, in genetically engineered mouse models of KRASG12C-driven NSCLC, inhibition of either KRASG12C or ULK1/2 decreases tumor burden and increases mouse survival. Additionally, this study found that LKB1 deficiency diminishes the sensitivity of KRASG12C/LKB1Null-driven lung cancer to the combination treatment, perhaps through the emergence of mixed adeno/squamous cell carcinomas and mucinous adenocarcinomas.

Strengths:

Both human cancer cells and mouse models were employed in this study to illustrate that inhibiting ULK1/2 could enhance the responsiveness of KRASG12C lung cancer to sotorasib. This research holds translational importance.

Weaknesses:

Additional validation of certain data is necessary.

(1) mCherry-EGFP-LC3 reporter was used to assess autophagy flux in Figure 1A. Please explain how autophagy status (high, medium, and low) was defined. It's also suggested to show WB of LC3 processing in different treatments as in Figure 1A at 48 hours.

(2) For Figures 1J, K, and L, please provide immunohistochemistry (IHC) images demonstrating RAS downstream signaling blockade by sotorasib and autophagy blockade by DCC 3116 in tumors.

(3) Given that both DCC 3116 and ULK1K46N exhibit the ability to inhibit autophagy and synergize with sotorasib in inhibiting cell proliferation, in addition to demonstrating decreased levels of pATG13 via ELISA assay, please include Western blot analyses of LC3 or p62 to confirm the blockade of autophagy by DCC 3116 and ULK1K46N in Figure 1 & Figure 2.

(4) Since adenocarcinomas, adenosquamous carcinomas (ASC), and mucinous adenocarcinomas were detected in KL lung tumors, please conduct immunohistochemistry (IHC) to detect these tumors, including markers such as p63, SOX2, Katrine 5.

(5) Please provide the sample size (n) for each treatment group in the survival study (Figure 4E). It appears that all mice were sacrificed for tumor burden analysis in Figure 4F. However, there doesn't seem to be a significant difference among the treatment groups in Figure 4F, which contrasts with the survival analysis in Figure 4E. It is suggested to increase the sample size in each treatment group to reduce variation.

(6) In KP mice (Figure 5), it seems that a single treatment alone is sufficient to inhibit established KP lung tumor growth. Combination treatment does not further enhance anti-tumor efficacy. Therefore, this result doesn't support the conclusion generated from human cancer cell lines. Please discuss.

eLife assessment

This fundamental paper reports a new biosensor to study G protein-coupled receptor activation by the pituitary adenylyl cyclase-activating polypeptide (PACAP) in cell culture, ex vivo (mouse brain slices), and in vivo (zebrafish). Convincing data are presented that show the new sensor works, albeit at very high (non-physiological) concentrations of exogenous PACAP. The sensor has not yet been used to detect endogenously released PACAP, raising questions about whether the sensor can be used for its intended purpose. While further work must be pursued to achieve broad in vivo applications under physiological conditions, the new tool will be of interest to cell biologists, especially those studying the large GPCR family.

Reviewer #1 (Public Review):

Summary:

This study assumes but also demonstrates that auditory rhythm processing is produced by internal oscillating systems and evaluates the properties of internal oscillators across individuals. The authors designed an experiment and performed analyses that address individuals' preferred rate and flexibility, with a special focus on how much past rhythms influence subsequent trials. They find evidence for such historical dependence and show that we adapt less well to new rhythms as we age. Furthermore, the revised version of this manuscript includes evidence for detuning; i.e., a gradual reduction in accuracy as the difference between a participant's preferred rate and stimulus rate increases. Such detuning also correlates with modelled oscillator flexibility measures. Such outcomes increase our credence that an entrainment-based interpretation is indeed warranted. Regardless of mechanism though, this work contributes to our understanding of individual differences in rhythm processing.

Strengths:

The inclusion of two tasks -- a tapping and a listening task -- complement each other methodologically. By analysing both the production and tracking of rhythms, the authors emphasize the importance of the characteristics of the receiver, the external world, and their interplay. The relationship between the two tasks and components within tasks are explored using a range of analyses. The visual presentation of the results is very clear. The age-related changes in flexibility are useful and compelling. The paper includes a discussion of the study assumptions, and it contextualizes itself more explicitly as taking entrainment frameworks as a starting point. Finally, the revised versions show creative additional analyses that increase our credence in an entrainment-based interpretation versus an interpretation of timekeeper other models, increasing the theoretical relevance of this study as compared to previous work.

Weaknesses:

The authors have addressed many of the weaknesses of previous peer review rounds. One final point is that our credence in an entrainment-based interpretation of these results could further increase by not only carefully outlining what is expected under entrainment (as is now done), but to also specify more extensively what predictions emerge from a timekeeper or other model, and how these data do not bear out such predictions.

Author response:

The following is the authors’ response to the previous reviews.

We thank the reviewers for their thorough re-evaluation of our revised manuscript. Addressing final issues they raised has improved the manuscript further. We sincerely appreciate the detailed explanations that the reviewers provided in the "recommendations for authors" section. This comprehensive feedback helped us identify the sources of ambiguity within the analysis descriptions and in the discussion where we interpreted the results. Below, you will find our responses to the specific comments and recommendations.

Reviewer #1 (Recommendations):

(1) I find that the manuscript has improved significantly from the last version, especially in terms of making explicit the assumptions of this work and competing models. I think the response letter makes a good case that the existence of other research makes it more likely that oscillators are at play in the study at hand (though the authors might consider incorporating this argumentation a bit more into the paper too). Furthermore, the authors' response that the harmonic analysis is valid even when including x=y because standard correlation analysis were not significant is a helpful response. The key issue that remains for me is that I have confusions about the additional analyses prompted by my review to a point where I find it hard to evaluate how and whether they demonstrate entrainment or not. 

First, I don't fully understand Figure 2B and how it confirms the Arnold tongue slice prediction. In the response letter the authors write: "...indicating that accuracy increased towards the preferred rate at fast rates and decreased as the stimulus rate diverged from the preferred rate at slow rates". The figure shows that, but also more. The green line (IOI < preferred rate) indeed increases toward the preferred rate (which is IOI = 0 on the x-axis; as I get it), but then it continues to go up in accuracy even after the preferred rate. And for the blue line, performance also continues to go up beyond preferred rate. Wouldn't the Arnold tongue and thus entrainment prediction be that accuracy goes down again after the preferred rate has passed? That is to say, shouldn't the pattern look like this (https://i.imgur.com/GPlt38F.png) which with linear regression should turn to a line with a slope of 0?

This was my confusion at first, but then I thought longer about how e.g. the blue line is predicted only using trials with IOI larger than the preferred rate. If that is so, then shouldn't the plot look like this? (https://i.imgur.com/SmU6X73.png). But if those are the only data and the rest of the regression line is extrapolation, why does the regression error vary in the extrapolated region? It would be helpful if the authors could clarify this plot a bit better. Ideally, they might want to include the average datapoints so it becomes easier to understand what is being fitted. As a side note, colours blue/green have a different meaning in 2B than 2D and E, which might be confusing. 

We thank the reviewer for their recommendation to clarify the additional analyses we ran in the previous revision to assess whether accuracy systematically increased toward the preferred rate estimate. We realized that the description of the regression analysis led to misunderstandings. In particular, we think that the reviewer interpreted (1) our analysis as linear regression (based on the request to plot raw data rather than fits), whereas, in fact, we used logistic regression, and (2) the regression lines in Figure 2B as raw IOI values, while, in fact, they were the z-scored IOI values (from trials where stimulus IOI were faster than an individual’s preferred rate, IOI < preferred rate, in green; and from trials stimulus IOI were slower than an individual’s preferred rate, IOI > preferred rate, in blue), as the x axis label depicted. We are happy to have the opportunity to clarify these points in the manuscript. We have also revised Figure 2B, which was admittedly maybe a bit opaque, to more clearly show the “Arnold tongue slice”.  

The logic for using (1) logistic regression with (2) Z-scored IOI values as the predictor is as follows. Since the response variable in this analysis, accuracy, was binary (correct response = 1, incorrect response = 0), we used a logistic regression. The goal was to quantify an acrosssubjects effect (increase in accuracy toward preferred rate), so we aggregated datasets across all participants into the model. The crucial point here is that each participant had a different preferred rate estimate. Let’s say participant A had the estimate at IOI = 400 ms, and participant B had an estimate at IOI = 600 ms. The trials where IOI was faster than participant A’s estimate would then be those ranging from 200 ms to 398 ms, and those that were slower would range from 402 ms to 998 ms. For Participant B, the situation would be different:  trials where IOI was faster than their estimate would range from 200 ms to 598 ms, and slower trials would range between 602 ms to 998 ms. For a fair analysis that assesses the accuracy increase, regardless of a participant’s actual preferred rate, we normalized these IOI values (faster or slower than the preferred rate). Zscore normalization is a common method of normalizing predictors in regression models, and was especially important here since we were aggregating predictors across participants, and the predictors ranges varied across participants. Z-scoring ensured that the scale of the sample (that differs between participant A and B, in this example) was comparable across the datasets. This is also important for the interpretation of Figure 2B. Since Z-scoring involves mean subtraction, the zero point on the Z-scaled IOI axis corresponds to the mean of the sample prior to normalization (for Participant A: 299 ms, for Participant B: 399 ms) and not the preferred rate estimate. We have now revised Figure 2B in a way that we think makes this much clearer.  

The manuscript text includes clarification that the analyses included logistic regression and stimulus IOI was z-scored: 

“In addition to estimating the preferred rate as stimulus rates with peak performance, we investigated whether accuracy increased as a function of detuning, namely, the difference between stimulus rate and preferred rate, as predicted by the entrainment models (Large, 1994; McAuley, 1995; Jones, 2018). We tested this prediction by assessing the slopes of mixed-effects logistic regression models, where accuracy was regressed on the IOI condition, separately for stimulus rates that were faster or slower than an individual’s preferred rate estimate. To do so, we first z-scored IOIs that were faster and slower than the participant’s preferred rate estimates, separately to render IOI scales comparable across participants.” (p. 7)

While thinking through the reviewer’s comment, we realized we could improve this analysis by fitting mixed effects models separately to sessions’ data. In these models, fixed effects were z-scored IOI and ‘detuning direction’ (i.e., whether IOI was faster or slower than the participant’s preferred rate estimate). To control for variability across participants in the predicted interaction between z-scored IOI and direction, this interaction was added as a random effect. 

“Ideally, they might want to include the average datapoints so it becomes easier to understand what is being fitted.”

Although we agree with the reviewer that including average datapoints in a figure in addition to model predictions usually better illustrates what is being fitted than the fits alone, this doesn’t work super well for logistic regression, since the dependent variable is binary. To try to do a better job illustrating single-participant data though, we instead  fitted logistic models to each participant’s single session datasets, separately to conditions where z-scored IOI from fasterthan-preferred rate trials, and those from slower-than-preferred rate trials, predicted accuracy. From these single-participant models, we obtained slope values, we referred to as ‘relative detuning slope’, for each condition and session type. This analysis allowed us to illustrate the effect of relative detuning on accuracy for each participant. Figure 2B now shows each participant’s best-fit lines from each detuning direction condition and session.

Since we now had relative detuning slopes for each individual (which we did not before), we took advantage of this to assess the relationship between oscillator flexibility and the oscillator’s behavior in different detuning situations (how strongly leaving the preferred rate hurt accuracy, as a proxy for the width of the Arnold tongue slice). Theoretically, flexible oscillators should be able to synchronize to wide range of rates, not suffering in conditions where detuning is large (Pikovsky et al., 2003). Conversely, synchronization of inflexible oscillators should depend strongly on detuning. To test whether our flexibility measure predicted this dependence on detuning, which is a different angle on oscillator flexibility, we first averaged each participant’s detuning slopes across detuning directions (after sign-flipping one of them). Then, we assessed the correlation between the average detuning slopes and flexibility estimates, separately from conditions where |-𝚫IOI| or |+𝚫IOI| predicted accuracy. The results revealed significant negative correlations (Fig. 2F), suggesting that performance of individuals with less flexible oscillators suffered more as detuning increased. Note that flexibility estimates quantified how much accuracy decreased as a function of trial-to-trial changes in stimulus rate (±𝚫IOI). Thus, these results show that oscillators that were robust to changes in stimulus rate were also less dependent on detuning to be able to synchronize across a wide range of stimulus rates. We are excited to be able to provide this extra validation of predictions made by entrainment models. 

To revise the manuscript with the updated analysis on detuning:

• We added the descriptions of the analyses to the Experiment 1 Methods section.

Calculation of detuning slopes and their averaging procedure are in Preferred rate estimates:

“In addition to estimating the preferred rate as stimulus rates with peak performance, we investigated whether accuracy increased as a function of detuning, namely, the difference between stimulus rate and preferred rate, as predicted by the entrainment models (Large, 1994; McAuley, 1995; Jones, 2018). We tested this prediction by assessing the slopes of mixed-effects logistic regression models, where accuracy was regressed on the IOI condition, separately for stimulus rates that were faster or slower than an individual’s preferred rate estimate. To do so, we first z-scored IOIs that were faster and slower than the participant’s preferred rate estimates, separately to render IOI scales comparable across participants. The detuning direction (i.e., whether stimulus IOI was faster or slower than the preferred rate estimate) was coded categorically. Accuracy (binary) was predicted by these variables (zscored IOI, detuning direction), and their interaction. The model was fitted separately to datasets from random-order and linear-order sessions, using the fitglme function in MATLAB. Fixed effects were z-scored IOI and detuning direction and random effect was their interaction. We expected a systematic increase in performance toward the preferred rate, which would result in a significant interaction between stimulus rate and detuning direction. To decompose the significant interaction and to visualize the effects of detuning, we fitted separate models to each participant’s single-session datasets, and obtained slopes from each direction condition, hereafter denoted as the ‘relative-detuning slope’. We treated relative-detuning slope as an index of the magnitude of relative detuning effects on accuracy. We then evaluated these models, using the glmval function in MATLAB to obtain predicted accuracy values for each participant and session. To visualize the relative-detuning curves, we averaged the predicted accuracies across participants within each session, separately for each direction condition (faster or slower than the preferred rate). To obtain a single value of relative-detuning magnitude for each participant, we averaged relative detuning slopes across direction conditions. However, since slopes from IOI > preferred rate conditions quantified an accuracy decrease as a function of detuning, we sign-flipped these slopes before averaging. The resulting average relative detuning slopes, obtained from each participant’s single-session datasets, quantified how much the accuracy increase towards preferred rate was dependent on, in other words, sensitive to, relative detuning.” (p. 7-8)

• We added the information on the correlation analyses between average detuning slopes in Flexibility estimates.

“We further tested the relationship between the flexibility estimates (𝛽 from models where |𝚫IOI| or |+𝚫IOI| predicted accuracy) and average detuning slopes (see Preferred rate estimates) from random-order sessions. We predicted that flexible oscillators (larger 𝛽) would be less severely affected by detuning, and thus have smaller detuning slopes. Conversely, inflexible oscillators (smaller 𝛽) should have more difficulty in adapting to a large range of stimulus rates, and their adaptive abilities should be constrained around the preferred rate, as indexed by steeper relative detuning slopes.” (p. 8)

• We provided the results in Experiment 1 Results section.

“Logistic models assessing a systematic increase in accuracy toward the preferred rate estimate in each session type revealed significant main effects of IOI (linear-order session: 𝛽 = 0.264, p < .001; random-order session: 𝛽 = 0.175, p < .001), and significant interactions between IOI and direction (linear-order session: 𝛽 = -0.444, p < .001; random-order session: 𝛽 = -0.364, p < .001), indicating that accuracy increased as fast rates slowed toward the preferred rate (positive slopes) and decreased again as slow rates slowed further past the preferred rate (negative slopes), regardless of the session type. Fig. 2B illustrates the preferred rate estimation method for an example participant’s dataset and shows the predicted accuracy values from models fitted to each participant’s single-session datasets. Note that the main effect and interaction were obtained from mixed effects models that included aggregated datasets from all participants, whereas the slopes quantifying the accuracy increase as a function of detuning (i.e., relative detuning slopes) were from models fitted to single-participant datasets.” (p. 9-10)

“We tested the relationship between the flexibility estimates and single-participant relative detuning slopes from random-order sessions (Fig. 2B). The results revealed negative correlations between the relative detuning slopes and flexibility estimates, both with 𝛽 (r(23) =0.529, p = 0.007) from models where |-𝚫IOI| predicted accuracy (adapting to speeding-up trials), and 𝛽 (r(23) =-0.580, p = 0.002) from models where |+𝚫IOI| predicted accuracy (adapting to slowing-down trials). That is, the performance of individuals with less flexible oscillators suffered more as detuning increased. These results are shown in Fig. 2F.” (p. 10)

• We modified Figure 2. In Figure 2B, there are now separate subfigures with the z-scored IOI faster (left) or slower (right) than the preferred rate predicting accuracy. We illustrated the correlations between average relative detuning slopes and flexibility estimates in Figure 2F. 

Author response image 1.

Main findings of Experiment 1. A Left: Each circle represents a single participant’s preferred rate estimate from the random-order session (x axis) and linear-order session (y axis). The histograms along the top and right of the plot show the distributions of estimates for each session type. The dotted and dashed lines respectively represent 1:2 and 2:1 ratio between the axes, and the solid line represents one-to-one correspondence. Right: permutation test results. The distribution of summed residuals (distance of data points to the closest y=x, y=2*x and y=x/2 lines) of shuffled data over 1000 iterations, and the summed residual from original data (dashed line) that fell below .008 of the permutation distribution. B Top: Illustration of the preferred rate estimation method from an example participant’s linear-order session dataset. Estimates were the stimulus rates (IOI) where smoothed accuracy (orange line) was maximum (arrow). The dotted lines originating from the IOI axis delineate the stimulus rates that were faster (left, IOI < preferred rate) and slower (right, IOI > preferred rate) than the preferred rate estimate and expand those separate axes, the values of which were Z-scored for the relative-detuning analysis. Bottom: Predicted accuracy, calculated from single-participant models where accuracy in random-order (purple) and linear-order (orange) sessions was predicted by z-scored IOIs that were faster than a participant’s preferred rate estimate (left), and by those that were slower (right). Thin lines show predicted accuracy from single-participant models, solid lines show the averages across participants and the shaded areas represent standard error of the mean. Predicted accuracy is maximal at the preferred rate and decreases as a function of detuning. C Average accuracy from random-order (left, purple) and linear-order (right, orange) sessions. Each circle represents a participant’s average accuracy. D Flexibility estimates. Each circle represents an individuals’ slope (𝛽) obtained from logistic models, fitted separately to conditions where |𝚫IOI| (left, green) or |+𝚫IOI| (right blue) predicted accuracy, with greater values (arrow’s direction) indicating better oscillator flexibility. The means of the distributions of 𝛽 from both conditions were smaller than zero (dashed line), indicating a negative effect of between-trial absolute rate change on accuracy. E Participants’ average bias from |𝚫IOI| (green), and |+𝚫IOI| (blue) conditions in random-order (left) and linear-order (right) sessions. Negative bias indicates underestimation of the comparison intervals, positive bias indicates the opposite. Box plots in C-E show median (black vertical line), 25th and 75th percentiles (box edges) and extreme datapoints (whiskers). In C and E, empty circles show outlier values that remained after data cleaning procedures. F Correlations between participants’ average relative detuning slopes, indexing the steepness of the increase in accuracy towards the preferred rate estimate (from panel B), and flexibility estimates from |-𝚫IOI| (top, green), and |+𝚫IOI| (bottom, blue) conditions (from panel C). Solid black lines represent the best-fit line, dashed lines represent 95% confidence intervals.

• We discussed the results in General Discussion and emphasized that only entrainment models, compared to timekeeper models, predict a relationship between detuning and accuracy that is amplified by oscillator’s inflexibility: “we observed systematic increases in task accuracy (Experiment 1) toward the best-performance rates (i.e., preferred rate estimates), with the steepness of this increase being closely related to the effects of rate change (i.e., oscillator flexibility). Two interdependent properties of an underlying system together modulating an individual’s timing responses show strong support for the entrainment approach” (p. 24)

“As a side note, colours blue/green have a different meaning in 2B than 2D and E, which might be confusing.” 

Upon the reviewer’s recommendation, we changed the color scale across Figure 2, such that colors refer to the same set of conditions across all panels. 

(2) Second, I don't understand the additional harmonic relationship analyses in the appendix, and I suspect other readers will not either. As with the previous point, it is not my view that the analyses are faulty or inadequate, it is rather that the lack of clarity makes it challenging to evaluate whether they support an entrainment model or not. 

We decided to remove the analysis that was based on a circular approach, and we have clarified the analysis that was based on a modular approach by giving example cases: 

“We first calculated how much the slower estimate (larger IOI value) diverts, proportionally from the faster estimate (smaller IOI value) or its multiples (i.e., harmonics) by normalizing the estimates from both sessions by the faster estimate. The outcome measure was the modulus of the slower, with respect to the faster estimate, divided by the faster estimate, described as mod(max(X), min(X))/min(X) where X = [session1_estimate session2_estimate]. An example case would be a preferred rate estimate of IOI = 603 ms from the linear-order session and an estimate of IOI = 295 ms from the random-order session. In this case, the slower estimate (603 ms) diverts from the multiple of the faster estimate (295*2 = 590 ms) by 13 ms, a proportional deviation of 4% of the faster estimate (295 ms). The outcome measure in this example is calculated as mod(603,295)/295 = 0.04.” (Supplementary Information, p. 2)

Crucially, the ability of oscillators to respond to harmonically-related stimulus rates is a main distinction between entrainment and interval (timekeeper) models. In the current study, we found that each participant’s best-performance rates, the preferred rate estimates, had harmonic relationships. The additional analyses further showed that these harmonic relationships were not due to chance. This finding speaks against the interval (timekeeper) approaches and is maximally compatible with the entrainment framework. 

Here are a number of questions I would like to list to sketch my confusion: 

• The authors write: "We first normalized each participant's estimates by rescaling the slower estimate with respect to the faster one and converting the values to radians". Does slower estimate mean: "task accuracy in those trials in which IOI was slower than a participant's preferred frequency"? 

Preferred rate estimates were stimulus rates (IOI) with best performance, as described in Experiment 1 Methods section. 

“We conceptualized individuals' preferred rates as the stimulus rates where durationdiscrimination accuracy was highest. To estimate preferred rate on an individual basis, we smoothed response accuracy across the stimulus-rate (IOI) dimension for each session type, using the smoothdata function in Matlab. Estimates of preferred rate were taken as the smoothed IOI that yielded maximum accuracy” (p. 7). 

The estimation method and the resulting estimate for an example participant was provided in Figure 2B. The updated figure in the current revision has this illustration only for linear-order session. 

“Estimates were the stimulus rates (IOI) where smoothed accuracy (orange line) was maximum (arrow)” (Figure caption, p. 9).

• "We reasoned that values with integer-ratio relationships should correspond to the same phase on a unit circle". What is values here; IOI, or accuracy values for certain IOIs? And why should this correspond to the same phase? 

We removed the analysis on integer-ratio relationships that was based on a circular approach that the reviewer is referring to here. We clarified the analysis that was based on a modular approach and avoided using the term ‘values’ without specifying what values corresponded to.

• Des "integer-ratio relationships" have to do with the y=x, y=x*2 and y=x/2 relationships of the other analyses?  

Integer-ratio relationships indeed refer to y=x, y=x*2 and y=x/2 relationships. For example, if a number y is double of another number x (y = x*2), these values have an integer-ratio relationship, since 2 is an integer. This holds true also for the case where y = x/2 since x = y*2. 

• Supplementary Figure S2c shows a distribution of median divergences resulting from the modular approach. The p-value is 0.004 but the dashed line appears to be at a much higher percentile of the distribution. I find this hard to understand. 

We thank the reviewer for a detailed inspection of all figures and information in the manuscript. The reviewer’s comment led us to realize that this figure had an error. We updated the figure in Supplementary Information (Supplementary Figure S2). 

Reviewer #2 (Public Review):

To get a better understanding of the mechanisms underlying the behavioral observations, it would have been useful to compare the observed pattern of results with simulations done with existing biophysical models. However, this point is addressed if the current study is read along with this other publication of the same research group: Kaya, E., & Henry, M. J. (2024, February 5). Modeling rhythm perception and temporal adaptation: top-down influences on a gradually decaying oscillator.       https://doi.org/10.31234/osf.io/q9uvr 

We agree with the reviewer that the mechanisms underlying behavioral responses can be better understood by modeling approaches. We thank the reviewer for acknowledging our computational modeling study that addressed this concern. 

Reviewer #2 (Recommendations):

I very much appreciate the thorough work done by the authors in assessing all reviewers' concerns. In this new version they clearly state the assumptions to be tested by their experiments, added extra analyses further strengthening the conclusions and point the reader to a neurocomputational model compatible with the current observations. 

I only regret that the authors misunderstood the take home message of our Essay (Doelling & Assaneo 2021). Despite this being obviously out of the scope of the current work, I would like to take this opportunity to clarify this point. In that paper, we adopted a Stuart-Landau model not to determine how an oscillator should behave, but as an example to show that some behaviors usually used to prove or refute an underlying "oscillator like" mechanism can be falsified. We obviously acknowledge that some of the examples presented in that work are attainable by specific biophysical models, as explicitly stated in the essay: "There may well be certain conditions, equations, or parameters under which some of these commonly held beliefs are true. In that case, the authors who put forth these claims must clearly state what these conditions are to clarify exactly what hypotheses are being tested." 

This work did not mean to delineate what oscillator is (or in not), but to stress the importance of explicitly introducing biophysical models to be tested instead of relying on vague definitions sometimes reflecting the researchers' own beliefs. The take home message that we wanted to deliver to the reader appears explicitly in the last paragraph of that essay: "We believe that rather than concerning ourselves with supporting or refuting neural oscillators, a more useful framework would be to focus our attention on the specific neural dynamics we hope to explain and to develop candidate quantitative models that are constrained by these dynamics. Furthermore, such models should be able to predict future recordings or be falsified by them. That is to say that it should no longer be sufficient to claim that a particular mechanism is or is not an oscillator but instead to choose specific dynamical systems to test. In so doing, we expect to overcome our looping debate and to ultimately develop-by means of testing many model types in many different experimental conditions-a fundamental understanding of cognitive processes and the general organization of neural behavior." 

We appreciate the reviewer’s clarification of the take-home message from Doelling and Assaneo (2021). We concur with the assertions made in this essay, particularly regarding the benefits of employing computational modeling approaches. Such methodologies provide a nuanced and wellstructured foundation for theoretical predictions, thereby minimizing the potential for reductionist interpretations of behavioral or neural data.

In addition, we would like to underscore the significance of delineating the level of analysis when investigating the mechanisms underlying behavioral or neural observations. The current study or Kaya & Henry (2024) involved no electrophysiological measures. Thus, we would argue that the appropriate level of analysis across our studies concerns the theoretical mechanisms rather than how these mechanisms are implemented on the neural (physical) level. In both studies, we aimed to explore or approximate the theoretical oscillator that guides dynamic attention rather than the neural dynamics underlying these theoretical processes. That is, theoretical (attentional) entrainment may not necessarily correspond to neural entrainment, and differentiating these levels could be informative about the parallels and differences between these levels. 

References

Doelling, K. B., & Assaneo, M. F. (2021). Neural oscillations are a start toward understanding brain activity rather than the end. PLoS Biol, 19(5), e3001234. https://doi.org/10.1371/journal.pbio.3001234  Jones, M. R. (2018). Time will tell: A theory of dynamic attending. Oxford University Press. 

Kaya, E., & Henry, M. J. (2024). Modeling rhythm perception and temporal adaptation: top-down influences on a gradually decaying oscillator. PsyArxiv. https://doi.org/https://doi.org/10.31234/osf.io/q9uvr 

Large, E. W. (1994). Dynamic representation of musical structure. The Ohio State University. 

McAuley, J. D. (1995). Perception of time as phase: Toward an adaptive-oscillator model of rhythmic pattern processing Indiana University Bloomington]. 

Pikovsky, A., Rosenblum, M., & Kurths, J. (2003). Synchronization: A Universal Concept in Nonlinear Sciences. Cambridge University Press.

eLife assessment

This useful study investigates two secreted Mycobacterium tuberculosis proteins, ESAT-6 and CFP10, using biochemical assays, including a Biolayer Interferometry assay. Solid experimental evidence demonstrates that ESAT-6 forms a tight interaction with CFP10 as a heterodimer at neutral pH and that ESAT-6 also forms a homodimer at acidic pH. Additional, more definitive evidence is required to describe how these proteins disrupt the phagosomal membrane. While improved compared to a previous version, the revised manuscript did not address these concerns adequately.

Reviewer #1 (Public Review):

Summary:

The authors sought to establish a biochemical strategy to study ESAT-6 and CFP-10 biochemistry. They established recombinant reagents to study these protein associations in vitro revealing an unexpected relationship at low pH. They next develop much needed reagents to study these proteins in an infection context and reveal that treatment with an ESAT-6 nanobody enhances Mtb control.

Strengths:

The biochemical conclusions are supported by multiple configurations of the experiments. They combine multiple approaches to study a complex problem.

Weaknesses:

It would be valuable to understand if the nanobody is disrupting the formation of the ESAT6-CFP10 complex. It is unclear how the nanobody is functioning to enhance control in the infection context. More detail or speculation in the discussion would have been valuable. Where is the nanobody in the cell during infection?

Reviewer #2 (Public Review):

Summary:

Bates TA. et al. studied the biochemical characteristics of ESAT-6, a major virulence factor of Mycobacterium tuberculosis (Mtb), as part of the heterodimer with CFP10, a molecular chaperon of ESAT-6, as in homodimer and in homotetramer using recombinant ESAT-6 and CFP10 expressed in E. coli by applying several biochemical assays including Biolayer Interferometry (BLI) assay. The main findings show that ESAT-6 forms a tight interaction with CFP10 as a heterodimer at neutral pH, and ESAT-6 forms homodimer and even tetramer based larger molecular aggregates at acidic pH. Although the discussion of the potential problems associated with the contamination of ESAT-6 preparations with ASB-14 during the LPS removal step is interesting, but this research does not test the potential impact of residual ASB-14 contaminant on the biochemical behavior ESAT-6-CFP10 heterodimer and ESAT-6 homodimer or tetramer and their hemolytic activity in comparison with the ones without ASB-14. The main strength of this study is the generation of ESAT-6 specific nanobodies and demonstration of its anti-tuberculosis efficiency in THP-1 cell line infected with Mtb strains with reporter genes.

Strengths:

Generation and demonstration of the anti-ESAT-6 nanobodies against tuberculosis infection in cell line based Mtb infection model. Probably identifying potential anti-ESAT-6 nanobody interacting amino acid residues of ESAT-6 is critical in understanding their effects on ESAT-6 mediated membrane lytic activity.

Weaknesses:

Although the biochemistry studies provide quantitative data about the interactions of ESAT-6 with its molecular chaperon CFP10 and the interaction of ESAT-6 homodimer and tetramers, the novel information from these studies are minimal.

Reviewer #3 (Public Review):

Summary:

This manuscript describes some biochemical experiments on the crucial virulence factor EsxA (ESAT-6) of Mycobacterium tuberculosis. EsxA is secreted via the ESX-1 secretion system. Although this system is recognized to be crucial for virulence the actual mechanisms employed by the ESX-1 substrates are still mostly unknown. The EsxA substrate is attracting most attention as the central player in virulence, especially phagosomal membrane disruption. EsxA is secreted as a dimer together with EsxB. The authors show that EsxA is also able to form homodimers and even tetramers, albeit at very low pH (below 5). Furthermore addition of a nanobody that specifically binds EsxA is blocking intracellular survival, also if the nanobody is produced in the cytosol of the infected macrophages.

Strengths:

Decent biochemical characterization of EsxA and identification of a new and interesting tool to study the function of EsxA (nanobody). Well written.

Weaknesses:

The findings are not critically evaluated using extra experiments or controls.<br /> For instance, tetrameric EsxA in itself is interesting and could reveal how EsxA works. But one would say that this is a starting point to make small point mutations that specifically affect tetramer formation and then evaluate what the effect is on phagosomal membrane lysis. Also one would like to see experiments to indicate whether these structures can be produced under in vitro conditions, especially because it seems that this mainly happens when the pH is lower than 5, which is not normally happening in phagosomes that are loaded with M. tuberculosis.<br /> Also the fact that the addition of the nanobody, either directly to the bacteria or produced in the cytosol of macrophages is interesting, but again the starting point for further experimentation. As a control one would like to se the effect on an Esx-1 secretion mutant. Furthermore, does cytososlic production or direct addition of the nanobody affect phagosomal escape? What happens if an EsxA mutant is produced that does not bind the nanobody?<br /> Finally, it is a bit strange that the authors use a non-native version of esxA that has not only an additional His-tag but also an additional 12 amino acids, which makes the protein in total almost 20% bigger. Of course these additions do not have to alter the characteristics, but they might. On the other hand they easily discard the natural acetylation of EsxA by mycobacteria itself (proven for M. marinum) as not relevant for the function because it might not happen in (the close homologue) M. tuberculosis.

Author response:

The following is the authors’ response to the original reviews.

Reviewer #1 (Recommendations For The Authors):

Will the nanobody be available to the TB research community?

Yes, we will make E11rv available upon request. Please see our materials availability statement.

Reviewer #2 (Recommendations For The Authors):

(1) It would be interesting to test the potential impact of residual ASB-14 contaminant on the biochemical behavior of ESAT-6-CFP10 heterodimer and ESAT-6 homodimer or tetramer and their hemolytic activity in comparison with the ones without ASB-14.

We agree that this is an interesting line of questioning. Based on the study by Refai et al. that we cite in the text, ESAT-6 treated with nonionic detergents ASB-14 or LDAO, but not other common detergents, undergoes a conformational change that increases its cytotoxicity in cell assays, hemolytic activity, and ability to dimerize with CFP-10. What is not known at this point, is how similar the ASB-bound conformation is to anything seen physiologically.

(2) Building on the progress in making anti-ESAT-6 nanobodies and their anti-Mtb effects in the cells, it could have been tested in human or mouse primary macrophages infected with Mtb and a mouse model of Mtb infection for its anti-Mtb efficiency.

We thank the reviewer for this suggestion, and we agree that these would be very informative next steps for determining the therapeutic potential of anti-ESAT-6 nanobodies.

Reviewer #3 (Recommendations For The Authors):

Minor comments:

Line 133: "It is well established that Mm-induced hemolysis is ESX-1 dependent, but our results suggest that Mtb must lack one or more factors necessary for efficient hemolysis.". I would tone this down a bit, as it is also known that M. tuberculosis escapes much later than M. marinum from the phagosome, which could indicate different kinetics.

We thank the reviewer for their insightful comments. We agree that the kinetics of Mtb and Mm infection are quite different and that this may impact the hemolysis assay. As described by Augenstreich et al. some hemolysis by Mtb is observed at 48 hours, though the method of normalization makes it impossible to determine absolute amount of hemolysis that occurred in their experiment. Our findings just show that the absolute amount of Mtb hemolysis in 2 hours is negligible, setting it apart from Mm. We have edited the wording of this statement in the manuscript to avoid any confusion.

Line 155: "Because Mtb often exists in an acidified compartment". First of all, the reference used here does not discuss anything about Mtb, secondly, papers that do measure the acidification of Mtb-loaded phagosomes indicate that this acidification is very mild (typically to pH 6.2).

We agree that this point should be articulated more precisely. We have added additional clarification that the pH of Mtb-containing compartments in macrophages can fall in a broad range depending on the activation state of the macrophages, and that non-activated macrophages are typically only mildly acidic. We have updated our references to better describe the current state of knowledge on this topic.

Line 339: "Whereas most of these functions rely only on the secretion of ESAT-6 into the cytoplasm, the ability of E11rv to access Mtb suggests that this communication is likely two-way." No, not necessary, there are many processes in which ESX-1 substrates affect the macrophage. This nanobody could affect EsxA functioning only once the bacteria reach the cytoplasm. I think checking phagosomal escape in these cells is therefore crucial.

We agree that phagosomal escape and subsequent direct secretion of ESAT-6 into the cytoplasm is a reasonable alternative hypothesis. We have added this point to our discussion, and we agree that looking directly at phagosomal escape is an important next step.

Figure 7 is not mentioned in the text (mistake for Fig 6).

This has been corrected.

Author response:

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

The study is highly interesting and the applied methods are target-oriented. The biophysical characterization of viable N-protein species and several representative N-protein mutants is supported by the data, including polarity, hydrophobicity, thermodynamic stability, CD spectra, particle size, and especially protein self-association. The physicochemical parameters for viable N-protein and related coronavirus are described for comparison in detail. However, the conclusion becomes less convincing that the interaction of peptides or motifs was judged by different biophysical results, with no more direct data about peptide interaction. Additionally, the manuscript could benefit from more results involving peptide interaction to support the author's opinions or make expression more accurate when concerning the interaction of motifs. Although the authors put a lot of effort into the study, there are still some questions to answer.

We thank the Reviewer for this assessment and wholeheartedly agree that there are still many questions. The main thrust of the present work was not intended to unravel the detailed mechanistic origin of all observations, but rather to juxtapose the different observations made with different viable N-protein species across the mutant spectrum, in order to get a sense of how narrowly the biophysical phenotype is confined to ensure virus viability. Such a study has become possible for the first time with the unprecedented genomic database of SARS-CoV-2. This has led to observations of non-local effects of individual mutations that are not independent and non-additive relative to the effects of other mutations, and in that sense we have inferred ‘interactions’. These might be mediated by direct contacts or indirectly through altered chain configurations. In the revised manuscript we have clarified this point.

Meanwhile, a number of documented direct physical intra-molecular and intra-dimer interactions provide a context to our study of mutation effects. The flexibility of the IDRs provides a rich variety of contacts that have been observed in molecular dynamics and single-molecule fluorescence studies (Rozycki & Boura, Biophys Chem. 2022 and Cubuk et al, Nat Communs 2021). We have previously carried out detailed hydrodynamic studies of self-association interfaces located in the leucine-rich region. More recently, NMR data just published by the Blackledge laboratory (Botova et al., bioRxiv 2024) extend the list of intra-molecular contacts with the observation of long-range intra-molecular interactions between the NTD and the CTD, NTD and the phosphorylated SR-rich region, and NTD and the previously studied leucine-rich region. The latter contacts require the C-terminal region of the linker to loop back onto the NTD, which may well introduce susceptibility to any of the linker mutations. However, detailed linker configurations are beyond the scope of the present work.

With regard to the effects of the Omicron mutations in the N-arm IDR, we have shown hydrodynamic data directly demonstrating peptide self-association, and we are currently working on a more detailed functional follow-up study which we hope to communicate soon.

Reviewer #2 (Public Review):

Summary: This work focuses on the biochemical features of the SARS-CoV-2 Nucleocapsid (N)protein, which condenses the large viral RNA genome inside the virus and also plays other roles in the infected cell. The N protein of SARS-CoV-2 and other coronaviruses is known to contain two globular RNA-binding domains, the NTD and CTD, flanked by disordered regions. The central disordered linker is particularly well understood: it contains a long SR-rich region that is extensively phosphorylated in infected cells, followed by a leucine-rich helical segment that was shown previously by these authors to promote N protein oligomerization.

In the current work, the authors analyze 5 million viral sequence variants to assess the conservation of specific amino acids and general sequence features in the major regions of the N protein. This analysis shows that disordered regions are particularly variable but that the general hydrophobic and charge character of these regions are conserved, particularly in the SR and leucine-rich regions of the central linker. The authors then construct a series of N proteins bearing the most prevalent mutations seen in the Delta and Omicron variants, and they subject these mutant proteins to a comprehensive array of biophysical analyses (temperature sensitivity, circular dichroism, oligomerization, RNA binding, and phase separation).

Strengths:

The results include a number of novel findings that are worthy of further exploration. Most notable are the analyses of the previously unstudied P31L mutation of the Omicron variant. The authors use ColabFold and sedimentation analysis to suggest that this mutation promotes the self-association of the disordered N-terminal region and stimulates the formation of N protein condensates. Although the affinity of this interaction is low, it seems likely that this mutation enhances viral fitness by promoting N-terminal interactions. The work also addresses the impact of another unstudied mutation, D63G, that is located on the surface of the globular NTD and has no significant effect on the properties analyzed here, raising interesting questions about how this mutation enhances viral fitness. Finally, the paper ends with studies showing that another common mutant, R203K/G204R,disrupts phase separation and might thereby alter N protein function in a way that enhances viral fitness.

Thank you for highlighting the strengths of our paper.

Weaknesses:

In general, the results in the paper confirm previous ideas about the role of N protein regions. The key novelty of the paper lies in the identification of point mutations, notablyP13L, that suggest previously unsuspected functions of the N-terminal disordered region in protein oligomerization. The paper would benefit from further exploration of these possibilities.

We agree that the bioinformatic results confirm previous ideas about the role of the N protein regions. However, we believe our results go beyond the previous thinking in a crucial aspect, which is that we examine the full (so far known) mutant spectrum of N-protein. Properties previously inferred from the inspection of single consensus sequences can be misleading because of the quasispecies nature of RNA viruses. By considering the mutant spectrum we can obtain a sense for how significant differences in the physicochemical properties of the different regions are, and how much variation is possible without jeopardizing essential protein functions.

With regard to the N-arm IDR mutations we believe this deserves a separate study focusing on the apparent N-arm function. Our rationale for presenting some initial N-arm results in the current paper was to highlight how the variability of N-protein species in the mutant spectrum can even include differences in the type and number of protein self-association interfaces.

Reviewer #3 (Public Review):

Nguyen, Zhao, et al. used bioinformatic analysis of mutational variants of SARS-CoV-2Nucleocapsid (N) protein from the large genomic database of SARS-CoV-2 sequences to identify domains and regions of N where mutations are more highly represented and computationally determined the effects of these mutations on the physicochemical properties of the protein. They found that the intrinsically disordered regions (IDRs) of N protein are more highly mutated than structured regions and that these mutations can lead to higher variability in the physical properties of these domains. These computational predictions are compared to in vitro biophysical experiments to assess the effects of identified mutations on the thermodynamic stability, oligomeric state, particle formation, and liquid-liquid phase separation of a few exemplary mutants.

The paper is well-written and easy to follow, and the conclusions drawn are supported by the evidence presented. The analyses and conclusions are interesting and will be of value to virologists, cell biologists, and biophysicists studying SARS-CoV-2 function and assembly. It would be nice if some further extrapolation or comments could be made regarding the effects of the observed mutations on the in vivo behavior and properties of the virus, but I appreciate that this is much higher-order than could be addressed with the approaches employed here.

We thank the Reviewer for this positive assessment. With regard to the possible in vivo behavior of mutant species, we agree that this would require additional data beyond the scope of the present work.

However, for the N:G215C mutant we can point to a very recent preprint by Kubinski et al. (bioRxiv 2024) that describes reverse genetics experiments where the isolated N:G215C mutation caused altered in vivo pathology, enhanced viral replication, and altered virion morphology. We have cited this work in the revised manuscript.

As mentioned above, for the P13L mutation we hope to communicate a more detailed follow-up study that will allow us to extrapolate on its in vivo behavior.

Recommendations For The Authors:

Reviewer #1:

(1) Given the structure organization of N-protein in Figure 1, the authors should explain why linker region 180-247 is different from linker (175-247) mentioned in the first result.

We thank the reviewer for bringing up this point, which we agree deserves clarification. While often the NTD has been assigned a C-terminal limit of 180 (e.g., in the NMR structure by Dinesh et al, Plos Pathogens 2020), the last several residues in the NTD are already disordered and contain the S176/R177 pair and therefore may be ascribed to the beginning of the SR-rich portion of the linker. In order not to artificially truncate functional sequences of either NTD or linker, we have decided to allow the designations of the NTD and linker regions to overlap. We believe this is conservative in that possible NTD or linker properties extending into this transition region will be preserved. In order to explain this in the manuscript, we have modified Figure 1 and inserted a brief sentence “(Due to ambiguity in delineation between NTD and linker, designations overlapping in 175-180 were used to avoid artificial truncation and permit conservative evaluation of the properties of each domain.)”.

(2) Please specify the "physicochemical requirements" in the fourth paragraph of the first result, and its physicochemical meaning and references.

Thank you for pointing this out; we agree this was not well expressed. We have rephrased this (including new references) to “…we find that hydrophobicity is uniformly high and polarity correspondingly low in the folded NTD and CTD domains, which is consistent with the expectation that folded structures are stabilized by buried hydrophobic residues (Eisenberg and McLachlan, 1986; Kauzmann, 1959)”.

(3) The authors should clarify the biological meaning of the net charge and phosphorylation charge in the first result, just like the description in the results of polarity and hydrophobicity.

We agree this will improve readability, and have inserted an introductory sentence to the study of charges in the mutant spectrum: “Charges in proteins can control multiple properties related to electrostatic interactions, from functions of active sites to protein solubility, protein interactions, and conformational ensembles in IDRs (Garcia-Viloca et al., 2004; Gerstein and Chothia, 1996; Gitlin et al., 2006; Mao et al., 2010).”.

(4) The authors should clarify the calculation method and meaning of the column "occurs in % of all genomes" in Table 2.

We have inserted a footnote specifying that this is the “Percentage of all sequenced genomes carrying the specific mutation.”.

(5) Please specify what information or conclusion we can get for the shift of the intrinsic fluorescent spectrum of N: D63G in the third result paragraph 2.

We have rephrased the second sentence of this paragraph to “The presence of the N:D63G mutation in the NTD is highlighted in the shift of the intrinsic fluorescence quantum yield of this mutant in comparison to Nref ”. It confirms the structural prediction, which positions D63G at the protein surface near the NA binding site, and sets up the question whether this obligatory mutation of Delta-variant N-protein affects NA binding and thereby possibly assembly. Unexpectedly, we did not find any impact of the D63G mutation on NA binding, although we observed a modest impact on temperature-dependent particle formation by DLS.

(6) The conclusion, "some epistatic interaction between mutation of the linker and N-arm" in the third result paragraph 4, is over-interpreted from the result of the CD spectra because they didn't detect peptide interaction between mutation of the linker and N-arm.

Thank you for raising this point. We did not mean to make a strong conclusion here, and have now deleted this statement.

(7) The parallel assay for N: G215C and Nδ in SV-AUC experiments is recommended to be conducted with other groups to avoid experimental error.

I believe this may be a misunderstanding: Indeed we had carried out SV-AUC experiments for all the mutants, as shown in Figure 5A. However, since all but the N:G215C and Nδ formed only dimers as the reference protein, we did not comment on these in the results text. We have rectified this omission in the revision by inserting the sentence: “…The same behavior is observed for N:D63G, No, N:R203K/G204R, as well as N:P13L/Δ31-33 at low micromolar concentrations (Figure 5A). By contrast, the G215C mutation promotes the formation of higher oligomers…”

With regard to experimental error, SV-AUC is an absolute method based on first principles and we have maintained our instruments by performing regular calibrations, using methods developed by us and colleagues at NIST, as described in the literature (Anal Biochem 2013, PLOS ONE 2018, Eur. Biophys. J. 2021). Previously we have critically examined the accuracy of s-values by SV-AUC before and after calibration in a large multi-laboratory study (PLOS ONE 2015), and found that the accuracy of s-values is ~1%. This allows detailed comparisons of results from different runs and different points in time. To alleviate any concerns we have now mentioned our calibration methods in the methods section.

(8) The authors did not test the function of Nδ R203M mutation, so they should not mention about it like in the third result paragraph 5, which is over-interpreted from result 5A.

We accept the criticism that we have not yet examined the R203M mutation in isolation. However, we believe some speculation is in order: Nδ consists of D63G, R203M, G215C, and D377Y, of which D63G is unlikely to impact oligomeric state based on our data of N:D63G. It is therefore reasonable to assume that R203M and/or D377Y interfere with the observed promotion of oligomerization that we have observed with N:G215C. In previous work, we have traced the 215C-incuded oligomerization to the transient helix in the leucine-rich region of the linker 215-235 (Science Advances, 2023), Since 377Y is quite far away, the more proximal 203M appears to be the most plausible origin of the modulation of dimerization.

In the revision we have more clearly outlined this speculation: “ Of the three additional mutations of Nδ relative to N:G215C, we speculate that D63G does not impact dimerization (as in N:D63G, Figure 5A), and that therefore either the distant D377Y and/or R203M might cause this reduction of helicity and oligomerization relative to N:G215C, noting that R203M is proximal to the L-rich region (215-235) reshaped by 215C. ”. Later we refer to this as “any potential inhibitory role suspected of the R203M mutation on self-association…”.

(9) The description of LLPS formation lacks reference in the third result paragraph 6.

Thank you. To improve the transition to this new paragraph in the results, we have inserted “As outlined in the introduction, …” and repeated the 8 references to the fact that N-protein undergoes LLPS. The two additional, separate references refer to just those published studies that examined the temperature-dependence of LLPS, which I believe is now clearer.

(10) The authors did not test the interaction between the N-arm IDR mutation and linker IDR, it is not exponible that interaction promoted particle formation of No in the third result paragraph 8, which is over-interpreted from result 5B.

We thank the Reviewer for raising this point. In fact, we did not want to imply a direct physical interaction (in terms of binding) between the N-arm IDR mutation and that in the linker. But clearly there are non-additive effects in particle formation since P13L/Δ31-33 inhibits slightly and R203K/G204R inhibits almost completely, whereas the combination of the two (constituting No) promotes particle formation. We have rephrased this to “alter the effect of”, avoiding the term “interact with” not to suggest a picture of direct binding and invoke instead the idea of epistatic interactions.

(11) In the third result paragraph 9, why did the authors choose to examine the role of the N-arm mutations of the Omicron variants in greater detail? This reason should be added to the manuscript.

Thank you for this suggestion. Naturally, we were curious how the defining N-arm mutations of Omicron variants could impact particle formation. Even though no obvious enhancement of self-association by either Omicron N-arm or linker mutations was observed at low micromolar concentrations in SV-AUC (Figure 5A), we knew from experience with the study of the leucine-rich transient helix in the linker IDR that even weak interfaces with mM Kd can be highly relevant in the context of multivalent assemblies (Science Advances, 2023). Therefore we followed the same roadmap and focused on IDR peptides with the goal to study them at higher concentrations that might reveal weak interactions.

We have described this motivation as follows: “We were curious whether IDR mutations might alter particle formation through modulation of existing or introduction of new protein-protein interfaces. We focused on Omicron mutations as these are obligatory an all currently circulating strains, and specifically on N-arm mutations, which have recently been implicated in altered intramolecular interactions with NA-occupied NTD (Cubuk et al., 2023). Even though SV-AUC showed no indication of self-association of N:P13L/Δ31-33 at low micromolar concentrations, weak interactions with Kd > mM would not be detectable under these conditions yet could be highly relevant in the context of multi-valent complexes (Zhao et al., 2024). Following the roadmap used previously for the study of the weak self-association of the leucine-rich linker IDR (Zhao et al., 2023), we restricted the protein to the N-arm peptide such that it can be studied at much higher concentrations. To this end, we …”

(12) Why were different proteins dissolved in either high-salt buffer or low-salt buffer for biophysical experiments? Did this affect the experimental results? Explanations and evidence are required.

We appreciate this is an important point. Unfortunately, for practical reasons of available sample concentrations and quantities, it was not always possible to dialyze protein into both buffers. For example, the DSF data in Figure 4B show all proteins in low-salt buffer except N:R203K/G204R, which is in high-salt buffer. We had previously reported the absence of changes in Ti in DSF for Nref in the two buffers, which we have documented better in the revised manuscript by providing an additional Supplementary Figure S7: “As a buffer control, the difference in Ti for Nref in LS and HS buffer was measured and found to be within error of data acquisition (Supplementary Figure S7A).” This new Supplementary Figure provides an overlay of low-salt and high-salt DSF data for Nref, N:D63G, and No, which have variations in the Ti values for different buffers on the order of 0.1 °C. This is comparable to the precision of the measurement, and significantly smaller than the changes in Ti values between the different mutant protein species. Finally, we note that the one species for which we were unable to collect DSF data in low-salt buffer, N:R203K/G204R, was unremarkable relative to Nref, No, and N:P13L/Δ31-33.

In the case of CD, the only species for which we could not collect spectra in low-salt buffer was No. Again, this spectrum was similar to the group including Nref, along with N:P13L/Δ31-33, and N:D63G. In the results we interpreted significant differences from Nref for N:G215C and N:R203K/G204R.

Similarly, SV-AUC experiments were carried out in high-salt buffer, except Nref, Nδ , and N:G215C. In this case, we could observe a ≈ 5% difference in s-value for the same protein in different buffers, but the magnitude of this change is negligible compared to the ≈ 60-90% increase observed for altered oligomeric states. To clarify this we have inserted a sentence “Proteins for self-association studies were in buffer HS, except Nref, Nδ , and N:G215C were in LS, the latter causing a ≈5% increase in s-value (Supplementary Figure S7B).”, with the new Supplementary Figure S7B showing a comparison of sedimentation coefficient distributions of Nref and N:D63G in low- and high-salt buffers. Whether the small differences in s-values are indeed significant and reflective of salt-dependent conformational ensembles of IDRs will require a more detailed follow-up study, but is outside the scope of the present work.

All other experiments were carried out with uniform buffer conditions for all protein species.

(13) DLS data of N from other research suggests oligomers beyond dimer. Please address this discrepancy.

Unfortunately several previous studies in the literature did not recognize the importance of eliminating nucleic acid contaminations in the N-protein preparations, and/or did not succeed in completely removing nucleic acid from the protein. We and others have repeatedly commented on this issue. For example, Tarczewska et al (IJBM 188 (2021) 391-403) clearly demonstrate this in much detail in a study dedicated to this problem.

The clarify this point we have included a sentence in the paragraph describing the protein preparation “…the ratio of absorbance at 260 nm and 280 nm of ~0.50-0.55 confirmed absence of nucleic acid. The latter is important to eliminate higher order N-protein oligomers induced by nucleic acid binding (Carlson et al., 2020; Tarczewska et al., 2021; Zhao et al., 2021)” .

In order to strengthen the statement in the Results that the ancestral N-protein is dimeric we have added additional references from other labs that have carried out detailed biophysical analyses: “As reported previously, the ancestral N-protein at micromolar concentrations in NA-free form is a tightly linked dimer sedimenting at ≈4 S , without significant populations of higher oligomers (Forsythe et al., 2021; Ribeiro-Filho et al., 2022; Tarczewska et al., 2021; Zhao et al., 2022, 2021).”

Reviewer #2:

The key novel finding of the work lies in the evidence that P31L promotes N-terminal interactions. The paper would be strengthened by additional studies of the impact of P31Lon the oligomerization of full-length N protein. The sedimentation analysis in Fig 6 shows that high concentrations of the N arm alone self-associate, while the analysis in Fig 5 argues that P31L does not have an effect on the oligomerization of the full-length protein. Perhaps there are specific conditions or mutation combinations that would provide evidence that P31L has an effect on protein behavior that might explain the prevalence of this mutation.

We agree that the finding of P13L promoting N-terminal interactions is of great interest, and we thank the Reviewer for the suggestion to examine cross-correlations of N-arm mutations with other mutations as a tool to study its function and relevance.

The observation of self-association in Figure 6 at high concentrations is not necessarily at odds with the absence of self-association at 100fold lower concentrations. Rather, it seems to show that the interaction mediated by the N-terminal mutation P13L is weak with an effective Kd in the mM range. It will likely not be possible to reach sufficiently high protein concentrations with the full-length protein to visualize the oligomerization of N-terminal IDR. But even if it was possible to concentrate the protein enough, very likely other assembly processes would take place, including LLPS, obscuring potential P13L interfaces. Nonetheless we believe the protein-protein interface created by the N-arm IDR is highly relevant in the context of multi-valent complexes, where entropic co-localization enhances the effective N-arm IDR concentration that then can provide additional binding energy and strengthen the assembly of multi-protein complexes.

We are currently pursuing further experiments examining the properties and relevance of the N-arm mutations and intend to publish this in a separate study, not to distract from the thrust of the current work exploring of the extent of the biophysical phenotype space.

The R203K/G204R mutations have a surprising impact on LLPS in Figure 7: it is not clear how such limited mutations would alter the many nonspecific, multivalent interactions that presumably lead to phase separation. The paper would benefit from a more extensive analysis of LLPS in this mutant and in the P31L mutant, perhaps by performing the analysis at various protein concentrations and times.

Following this recommendation we have expanded the study of LLPS of Figure 7 by comparison of two different time points for Nref, N:R203K/G204R, and N:P13L in a new Supplementary Figure S6. We have also quantified the droplet distributions as shown in the new Supplementary Figure S5. Both clearly confirm the strong inhibitory effect of the R203K/G204R mutation on LLPS under our experimental conditions. What this shows is not that this protein could not undergo LLPS per se, but that the phase boundaries have shifted such that under the experimental conditions we applied LLPS does not occur yet. (In this context it is interesting to note that ≈50,000 genomes in the GISAID database have R203K/G204R as the sole N-protein mutation, without impact on viral viability.)

That individual point-mutations in IDRs can have significant impact on LLPS has been observed previously for several other proteins. Examples include SPOP [Bouchard et al., Mol Cell 72 (2018) 19-36.e8], SHP2 [Zhu et al., Cell 183 (2020) 490-502.e18], FUS [Niaki et al., Mol Cell 77 (2020) 82-94.e4], and CAPRIN1 [Kim et al., PNAS 118 (2021) 1-11]. The latter work applies NMR and reveals that promotion of LLPS is not uniform but centered in hot-spot residues of CAPRIN1.

While the precise molecular mechanism for LLPS of the N-protein is unclear, we can speculate how the effect of 203K/204R might be amplified. As shown by the coarse-grained MD simulations from Rozycki & Boura (Biophys. Chem. 2022), the linker IDR is highly flexible and the 203/204 residues make transient contacts to other residues throughout the linker as well as to distinct sites on the NTD. Furthermore, recent NMR data from the Blackledge lab (Botova et al., bioRxiv 2024, doi:10.1101/2024.02.22.579423) have revealed intra-molecular interactions, including a state where the L-rich (C-terminal) portion of the linker IDR interacts with a site on the distant NTD. (We have included a reference to this preprint in the discussion.) This intra-molecular contact observed in NMR must cause significant chain compaction and may thereby modulate the accessibility of portions of the linker IDR available to inter-molecular interactions contributing to LLPS. The residues 203/204 are in the middle between the SR-rich and L-rich region where bending of the chain must occur to allow for the intra-molecular contacts. The 203K/204R mutation may alter the dynamics or population of this intra-molecular bound state, especially considering the introduction of a bulky positively charged R replacing G204.

In summary, considering the dynamics of intra-molecular contacts and considering precedent of several other disordered proteins, we believe it is not unreasonable that the local mutation in the IDR R203K/G204R may cause a significant shift in LLPS phase boundaries. We note that this mutant also shows a very distinct behavior in the temperature-dependent DLS, entirely lacking particle formation below 70 °C. This observation seems consistent with altered inter-molecular interactions.

Reviewer #3:

I have only a few minor specific comments:

(1) Page 4, last paragraph - typo: "The large number of structural and non-structural N-protein functions poses the question of how they are conserved...". This either needs a colon or to be changed to "... poses the question of how they are conserved...".

Thank you – we have changed this sentence accordingly.

(2) Page 7, 2nd and 3rd paragraphs of "Physicochemical properties" section: why is Figure2B discussed before Figure 2A?

Initially when we present the results of polarity and hydrophobicity we refer more generally to Figure 2, as the two properties are so closely related. Later, in the section on related coronaviruses we do refer once more to Figure 2. Here we begin this section by discussing Figure 2B since in this plot the symbols for the different viruses are most recognizable.

(3) Page 11, lines 1-2: "Since this is a tell-tale of weak protein..." -> "tell-tale sign of ...".

We thank the reviewer for pointing this out and have fixed this sentence.

(4) Further down in the same paragraph, the meaning of "SV-AUC" should be spelled out at its first use.

We have double checked that SV-AUC is spelled out at its first use.

(5) Figures 1 and 2. Is there a good reason that the color scheme for the IDRs (magenta and cyan) is so close to the color scheme for the identifying mutations of Omicron and Delta (magenta and blue)? This initially led me to try to search for some connection, and it remains unclear to me if there is.

We apologize for this confusion. This was indeed a poor color choice, and we have rectified this in the revised manuscript by changing the colors of the identifying mutations of Omicron and Delta to dashed green and dotted red, respectively, so that there is no connection to the shading of the IDRs. Thank you very much for pointing this out!

(6) Figure 1: The physical limits of the subdomains, e.g. SR-rich, L-rich, C-arm1, and N3 could be more clearly delineated with lines, or some other visual representation.

Once more, we thank the reviewer for pointing this out. We have revised Figure 1 to indicate the limits between these subdomains.

(7) Figures 4, 5, and 6: are there any kind of error bars or confidence intervals on these measurements?

We appreciate this concern and have addressed it in different ways for the different methods.

For the spectra of intrinsic fluorescence in Figure 4A, we have now plotted an overlay of three acquired spectra, from which the experimental error as a function of wavelength may be assessed. It is clear that the differences between Nref and N:D63G are far greater than the measurement error.

With regard to DSF, we have provide an error estimate of 0.3 °C for the Ti-values, a value that we have revised from the previously reported errors of sequential replicates to now include Ti variation observed with different preparations of the same protein over long time periods.

For CD spectra we have included a new Supplementary Figure S3 that shows standard deviations of triplicate measurements as a function of wavelength. Since an overlay including errors for all species would be too crowded, we have created separate plots for all species in comparison with Nref. (On this occasion we discovered a 3% error in the magnitude of the Nref spectrum due to previously incorrect conversion to MRE, which we have now fixed.)

In SV-AUC, for data with typical signal-noise ratio, the statistical error is very small due to the large number (> 104 ) of raw data points included in the calculation of each c(s) trace, which each data point carrying a statistical error that is usually better than 1%. Therefore, the dominant error is systematic. In the past we have carried out large studies quantifying the accuracy of the major peaks of the sedimentation coefficient distributions, and found they are typically ≈1% in s-value and 1-2% for relative peak areas. In the AUC methods section we have now included the sentence “Typical accuracy of c(s) peaks are on the order of ≈1% for peak s-values and ≈1-2% for relative peak areas (Zhao et al., 2015).”

Finally, for the temperature-dependent DLS data we have to resort to the scatter in the temperature-dependent Rh-values. The calculated Rh-values can exhibit fluctuations once particles start to form and the distribution becomes highly polydisperse. As is characteristic for DLS under those conditions, individual Rh-values can be dominated by adventitious diffusion of few large particles into the laser focal spot. Although customarily autocorrelation functions can be filtered out through software filters (e.g., setting baseline and amplitude thresholds), this still presents the largest source of error in the Rh-values. These are systematic for the individual autocorrelation functions. We believe that the variation of Rh-values at similar temperatures outside the transition region provides a reasonable estimate for the experimental error.

(8) Figure 7: My most major comment. It would be good to somehow quantify the differences between these images. The claim is made that the LLPS droplets are different sizes, or for the P13L/\Delta31-33 variant that droplets are coalescing or changing shape over time. It would be good to quantify this rather than rely on eyeballing the pictures.

We are grateful to the Reviewer for this suggestion. As mentioned above, to improve the LLPS analysis we have now carried out segmentation of the images in Figure 7 to quantify the droplet numbers and areas. Histograms and statistical analyses are now provided in the new Supplementary Figure S5. In addition, we have added a comparison of the droplet numbers and sizes at two time-points for Nref, N:R203K/G204R, in addition to the previously shown N:P13L/Δ31-33, provided in the new Supplementary Figure S6. The results corroborate the previous conclusions, and depict how droplets in the N:P13L/Δ31-33 merge and grow in area more strongly than those from Nref.

eLife assessment

This important manuscript provides new insights into the biophysics of the SARS-CoV-2 nucleocapsid. The evidence, which relies on a convincing combination of genetic and biophysical data, nicely supports the conclusions.

Reviewer #2 (Public Review):

This work focuses on the biochemical features of the SARS-CoV-2 Nucleocapsid (N) protein, which condenses the large viral RNA genome inside the virus and also plays other roles in the infected cell. The N protein of SARS-CoV-2 and other coronaviruses is known to contain two globular RNA-binding domains, the NTD and CTD, flanked by disordered regions. The central disordered linker is particularly well understood: it contains a long SR-rich region that is extensively phosphorylated in infected cells, followed by a leucine-rich helical segment that was shown previously by these authors to promote N protein oligomerization.

In the current work, the authors analyze 5 million viral sequence variants to assess the conservation of specific amino acids and general sequence features in the major regions of the N protein. This analysis shows that disordered regions are particularly variable but that the general hydrophobic and charge character of these regions are conserved, particularly in the SR and leucine-rich regions of the central linker. The authors then construct a series of N proteins bearing the most prevalent mutations seen in the Delta and Omicron variants, and they subject these mutant proteins to a comprehensive array of biophysical analyses (temperature sensitivity, circular dichroism, oligomerization, RNA binding, and phase separation).

The results include a number of novel findings that are worthy of further exploration. Most notable are the analyses of the previously unstudied P31L mutation of the Omicron variant. The authors use ColabFold and sedimentation analysis to suggest that this mutation promotes self-association of the disordered N-terminal region and stimulates the formation of N protein condensates. Although the affinity of this interaction is low, it seems likely that this mutation enhances viral fitness by promoting N-terminal interactions. The work also addresses the impact of another unstudied mutation, D63G, that is located on the surface of the globular NTD and has no significant effect on the properties analyzed here, raising interesting questions about how this mutation enhances viral fitness. Finally, the paper ends with studies showing that another common mutant, R203K/G204R, disrupts phase separation and might thereby alter N protein function in a way that enhances viral fitness. These provocative results set the stage for in-depth analyses of these mutations in future work.

Reviewer #3 (Public Review):

Nguyen, Zhao et al. used bioinformatic analysis of mutational variants of SARS-CoV-2 Nucleocapsid (N) protein from the large genomic database of SARS-CoV-2 sequences to identify domains and regions of N where mutations are more highly represented, and computationally determined the effects of these mutations on the physicochemical properties of the protein. They found that the intrinsically disordered regions (IDRs) of N protein are more highly mutated than structured regions, and that these mutations can lead to higher variability in the physical properties of these domains. These computational predictions are compared to in vitro biophysical experiments to assess the effects of identified mutations on the thermodynamic stability, oligomeric state, particle formation, and liquid-liquid phase separation of a few exemplary mutants.

The paper is well written, easy to follow and the conclusions drawn are supported by the evidence presented. The analyses and conclusions are interesting and will be of value to virologists, cell biologists, and biophysicists studying SARS-CoV-2 function and assembly.

eLife assessment

This study represents a fundamental contribution to our understanding of how gene expression levels are controlled in bacteria. Through a series of compelling and careful experiments, relying on a mutant that blocks DNA replication but permits growth, and using various methods, the authors reveal how genome concentration rapidly becomes limiting for growth when replication is inhibited. This work contributes to our understanding of the contributions and limiting roles of DNA, mRNA, and ribosomes for growth in bacteria, and will be of considerable interest within both systems biology and microbial physiology.

Reviewer #1 (Public Review):

Summary:

The manuscript by Mäkelä et al. presents compelling experimental evidence that the amount of chromosomal DNA can become limiting for the total rate of mRNA transcription and consequently protein production in the model bacterium Escherichia coli. Specifically, the authors demonstrate that upon inhibition of DNA replication the single-cell growth rate continuously decreases, in direct proportion to the concentration of active ribosomes, as measured indirectly by single-particle tracking. The decrease of ribosomal activity with filamentation, in turn, is likely caused by a decrease of the concentration of mRNAs, as suggested by an observed plateau of the total number of active RNA polymerases. These observations are compatible with the hypothesis that DNA limits the total rate of transcription and thus translation. The authors also demonstrate that the decrease of RNAp activity is independent of two candidate stress response pathways, the SOS stress response and the stringent response, as well as an anti-sigma factor previously implicated in variations of RNAp activity upon variations of nutrient sources.

Remarkably, the reduction of growth rate is observed soon after the inhibition of DNA replication, suggesting that the amount of DNA in wild-type cells is tuned to provide just as much substrate for RNA polymerase as needed to saturate most ribosomes with mRNAs. While previous studies of bacterial growth have most often focused on ribosomes and metabolic proteins, this study provides important evidence that chromosomal DNA has a previously underestimated important and potentially rate-limiting role for growth.

Strengths:

This article links the growth of single cells to the amount of DNA, the number of active ribosomes and to the number of RNA polymerases, combining quantitative experiments with theory. The correlations observed during depletion of DNA, notably in M9gluCAA medium, are compelling and point towards a limiting role of DNA for transcription and subsequently for protein production soon after reduction of the amount of DNA in the cell. The article also contains a theoretical model of transcription-translation that contains a Michaelis-Menten type dependency of transcription on DNA availability and is fit to the data. While the model fits well with the continuous reduction of relative growth rate in rich medium (M9gluCAA), the behavior in minimal media without casamino acids is a bit less clear (see comments below).

At a technical level, single-cell growth experiments and single-particle tracking experiments are well described, suggesting that different diffusive states of molecules represent different states of RNAp/ribosome activities, which reflect the reduction of growth. However, I still have a few points about the interpretation of the data and the measured fractions of active ribosomes (see below).

Apart from correlations in DNA-deplete cells, the article also investigates the role of candidate stress response pathways for reduced transcription, demonstrating that neither the SOS nor the stringent response are responsible for the reduced rate of growth. Equally, the anti-sigma factor Rsd recently described for its role in controlling RNA polymerase activity in nutrient-poor growth media, seems also not involved according to mass-spec data. While other (unknown) pathways might still be involved in reducing the number of active RNA polymerases, the proposed hypothesis of the DNA substrate itself being limiting for the total rate of transcription is appealing.

Finally, the authors confirm the reduction of growth in the distant Caulobacter crescentus, which lacks overlapping rounds of replication and could thus have shown a different dependency on DNA concentration.

Weaknesses:

There are a range of points that should be clarified or addressed, either by additional experiments/analyses or by explanations or clear disclaimers.

First, the continuous reduction of growth rate upon arrest of DNA replication initiation observed in rich growth medium (M9gluCAA) is not equally observed in poor media. Instead, the relative growth rate is immediately/quickly reduced by about 10-20% and then maintained for long times, as if the arrest of replication initiation had an immediate effect but would then not lead to saturation of the DNA substrate. In particular, the long plateau of a constant relative growth rate in M9ala is difficult to reconcile with the model fit in Fig 4S2. Is it possible that DNA is not limiting in poor media (at least not for the cell sizes studied here) while replication arrest still elicits a reduction of growth rate in a different way? Might this have something to do with the naturally much higher oscillations of DNA concentration in minimal medium?

The authors argue that DNA becomes limiting in the range of physiological cell sizes, in particular for M9glCAA (Fig. 1BC). It would be helpful to know by how much (fold-change) the DNA concentration is reduced below wild-type (or multi-N) levels at t=0 in Fig 1B and how DNA concentration decays with time or cell area, to get a sense by how many-fold DNA is essentially 'overexpressed/overprovided' in wild-type cells.

Fig. 2: The distribution of diffusion coefficients of RpsB is fit to Gaussians on the log scale. Is this based on a model or on previous work or simply an empirical fit to the data? An exact analytical model for the distribution of diffusion constants can be found in the tool anaDDA by Vink, ..., Hohlbein Biophys J 2020. Alternatively, distributions of displacements are expressed analytically in other tools (e.g., in SpotOn).

The estimated fraction of active ribosomes in wild-type cells shows a very strong reduction with decreasing growth rate (down from 75% to 30%), twice as strong as measured in bulk experiments (Dai et al Nat Microbiology 2016; decrease from 90% to 60% for the same growth rate range) and probably incompatible with measurements of growth rate, ribosome concentrations, and almost constant translation elongation rate in this regime of growth rates. Might the different diffusive fractions of RpsB not represent active/inactive ribosomes? See also the problem of quantification above. The authors should explain and compare their results to previous work.

To measure the reduction of mRNA transcripts in the cell, the authors rely on the fluorescent dye SYTO RNAselect. They argue that 70% of the dye signal represents mRNA. The argument is based on the previously observed reduction of the total signal by 70% upon treatment with rifampicin, an RNA polymerase inhibitor (Bakshi et al 2014). The idea here is presumably that mRNA should undergo rapid degradation upon rif treatment while rRNA or tRNA are stable. However, work from Hamouche et al. RNA (2021) 27:946 demonstrates that rifampicin treatment also leads to a rapid degradation of rRNA. Furthermore, the timescale of fluorescent-signal decay in the paper by Bakshi et al. (half life about 10min) is not compatible with the previously reported rapid decay of mRNA (2-4min) but rather compatible with the slower, still somewhat rapid, decay of rRNA reported by Hamouche et al.. A bulk method to measure total mRNA as in the cited Balakrishnan et al. (Science 2022) would thus be a preferred method to quantify mRNA. Alternatively, the authors could also test whether the mass contribution of total RNA remains constant, which would suggest that rRNA decay does not contribute to signal loss. However, since rRNA dominates total RNA, this measurement requires high accuracy. The authors might thus tone down their conclusions on mRNA concentration changes while still highlighting the compelling data on RNAp diffusion.

The proteomics experiments are a great addition to the single-cell studies, and the correlations between distance from ori and protein abundance is compelling. However, I was missing a different test, the authors might have already done but not put in the manuscript: If DNA is indeed limiting the initiation of transcription, genes that are already highly transcribed in non-perturbed conditions might saturate fastest upon replication inhibition, while genes rarely transcribed should have no problem to accommodate additional RNA polymerases. One might thus want to test, whether the (unperturbed) transcription initiation rate is a predictor of changes in protein composition. This is just a suggestion the authors may also ignore, but since it is an easy analysis, I chose to mention it here.

Related to the proteomics, in l. 380 the authors write that the reduced expression close to the ori might reflect a gene-dosage compensatory mechanism. I don't understand this argument. Can the authors add a sentence to explain their hypothesis?

In Fig. 1E the authors show evidence that growth rate increases with cell length/area. While this is not a main point of the paper it might be cited by others in the future. There are two possible artifacts that could influence this experiment: a) segmentation: an overestimation of the physical length of the cell based on phase-contrast images (e.g., 200 nm would cause a 10% error in the relative rate of 2 um cells, but not of longer cells). b) time-dependent changes of growth rate, e.g., due to change from liquid to solid or other perturbations. To test for the latter, one could measure growth rate as a function of time, restricting the analysis to short or long cells, or measuring growth rate for short/long cells at selected time points. For the former, I recommend comparison of phase-contrast segmentation with FM4-64-stained cell boundaries.

Reviewer #2 (Public Review):

In this work, the authors uncovered the effects of DNA dilution on E. coli, including a decrease in growth rate and a significant change in proteome composition. The authors demonstrated that the decline in growth rate is due to the reduction of active ribosomes and active RNA polymerases because of the limited DNA copy numbers. They further showed that the change in the DNA-to-volume ratio leads to concentration changes in almost 60% of proteins, and these changes mainly stem from the change in the mRNA levels.

Reviewer #3 (Public Review):

Summary:

Mäkelä et al. here investigate genome concentration as a limiting factor on growth. Previous work has identified key roles for transcription (RNA polymerase) and translation (ribosomes) as limiting factors on growth, which enable an exponential increase in cell mass. While a potential limiting role of genome concentration under certain conditions has been explored theoretically, Mäkelä et al. here present direct evidence that when replication is inhibited, genome concentration emerges as a limiting factor.

Strengths:

A major strength of this paper is the diligent and compelling combination of experiment and modeling used to address this core question. The use of origin- and ftsZ-targeted CRISPRi is a very nice approach that enables dissection of the specific effects of limiting genome dosage in the context of a growing cytoplasm. While it might be expected that genome concentration eventually becomes a limiting factor, what is surprising and novel here is that this happens very rapidly, with growth transitioning even for cells within the normal length distribution for E. coli. Fundamentally, it demonstrates the fine balance of bacterial physiology, where the concentration of the genome itself (at least under rapid growth conditions) is no higher than it needs to be.

Weaknesses:

One limitation of the study is that genome concentration is largely treated as a single commodity. While this facilitates their modeling approach, one would expect that the growth phenotypes observed arise due to copy number limitation in a relatively small number of rate-limiting genes. The authors do report shifts in the composition of both the proteome and the transcriptome in response to replication inhibition, but while they report a positional effect of distance from the replication origin (reflecting loss of high-copy, origin-proximal genes), other factors shaping compositional shifts and their functional effects on growth are not extensively explored. This is particularly true for ribosomal RNA itself, which the authors assume to grow proportionately with protein. More generally, understanding which genes exert the greatest copy number-dependent influence on growth may aid both efforts to enhance (biotechnology) and inhibit (infection) bacterial growth.

Overall, this study provides a fundamental contribution to bacterial physiology by illuminating the relationship between DNA, mRNA, and protein in determining growth rate. While coarse-grained, the work invites exciting questions about how the composition of major cellular components is fine-tuned to a cell's needs and which specific gene products mediate this connection. This work has implications not only for biotechnology, as the authors discuss, but potentially also for our understanding of how DNA-targeted antibiotics limit bacterial growth.