On 2021-04-30 13:44:00, user NYUPeerReview wrote:
NOTE: This paper was selected for discussion and critique in “Peer Review in the Life Sciences”, a course for PhD students at the New York University School of Medicine. This course aims to build skills in the critical reading of the scientific literature, and provide formal training in the process of peer review. Following discussion as a class, students wrote this peer review, and received responses from the authors.
Summary<br />
In their recent preprint Kiani et al. report observations of a contact-dependent killing mechanism in Proteus mirabilis unrelated to the only known contact-dependent killing system in P. mirabilis, the type VI secretion system (T6SS). These observations lay the groundwork for further investigation into a yet undiscovered mechanism of contact-dependent bacterial killing.
Initially, the authors of this study observed that fetal mice bred from parents hosting either P. mirabilis or E. coli were only found to host P. mirabilis, as opposed to hosting a combination of the two bacteria, suggesting a P. mirabilis survival advantage or killing mechanism to outcompete E. coli. Attempting to recapitulate this observation in vitro, the authors co-cultured P. mirabilis and E. coli and found that E. coli demonstrated reduced viability in liquid culture and cell death on solid surfaces. This finding was extended to other Gram-negative bacteria. To investigate the nature of the survival mechanism, the authors co-cultured cells and found that cell-cell contact is required for P. mirabilis-mediated killing. Using a P. mirabilis T6SS knockout strain, the authors found that the P. mirabilis-mediated killing persisted in the absence of T6SS effector molecules, supporting the authors’ hypothesis that this killing mechanism is separate from T6SS. P. mirabilis did not kill E. coli during its exponential growth phase, only during stationary phase; however, E. coli was susceptible to this killing at any stage of its growth. Fluorescent microscopy visualized interactions between P. mirabilis and E. coli during the course of their killing and found killing did not compromise membrane integrity while halting metabolic activity. Killing required protein synthesis and was enabled by a heat-sensitive component culture supernatant. Osmotic perturbations in swarm limiting media attenuated killing efficiency, suggesting environmental factors and osmolarity are agents in this P. mirabilis killing mechanism.
General Critiques<br />
In a clear and effective manner, the authors of this study present a series of compelling observations that suggest P. mirabilis employs a contact-dependent killing system other than T6SS to compete against Gram-negative bacteria. The various growth assays reported in this study support the discovery of a new mode of interbacterial killing; however, questions remain regarding the nature and mechanism of this new mode of killing. The authors’ conclusions were based on powerful, yet broad, experiments that we believe could be enhanced by some further studies into the mechanism of P. mirabilis killing.
Upon discussion of the manuscript, a few points were brought up that we highlight below:
Figure 1: The authors may want to consider adding a control that introduces fresh LB to the E. coli and P. mirabilis co-culture to rule out reduced E. coli viability was indeed caused by P. mirabilis and not due to nutrient deprivation and starvation.
Authors’ reply: Thank you very much for your comment. We were initially concerned about this as well. We will perform an experiment specifically addressing this in the context of Figure 1 by adding LB during stationary phase of the co-culture. However, based on results in Fig. 2 and 5, we are very confident that starvation is not the reason for E. coli death: We do not see loss of viability in stationary phase P. mirabilis supernatant (Fig. 2A+B), in split-well assays where E. coli cells are in the same medium as P. mirabilis (Fig. 2D), or in E. coli single cultures (Fig. S1). In Fig. 5, E. coli remained viable after supernatant was exchanged to heat-treated stationary phase P. mirabilis supernatant. Contrary to this, E. coli lost viability in untreated stationary phase P. mirabilis supernatant even when the supernatant was mixed 1:1 with fresh LB.
Figure 2E: The authors use wild-type P. mirabilis in their killing assays on swarming permissive LB agar to show that the same process occurs on plates as they describe in liquid media. To rule out the possibility that P. mirabilis is using T6SS killing in combination with the novel mechanism, perhaps the P. mirabilis ∆T6SS strain would be appropriate to include in this experiment.
Authors’ reply: We thank the reviewers for this interesting suggestion. We will perform a killing assay on solid surface with the T6SS mutant as suggested.
Figure 3A: The authors conclude that the formation of Dienes lines is indicative of P. mirabilis utilizing the T6SS machinery. Can the authors clarify the relationship between this observation and their data (Fig. 3B,C) that show no observable reduction viability in P. mirabilis co-cultures?
Authors’ reply: We will clarify this section in the manuscript. Fig. 3 panels A-C show that T6SS are inactive during conditions when we see E. coli killing and thus make a contribution unlikely, whereas panels D+E show directly that T6SS are not involved in the killing of E. coli. <br />
Panels A-C illustrate the known pattern of T6SS activity in P. mirabilis: killing of other P. mirabilis strains on solid surface, no killing of other P. mirabilis strains in liquid culture. These panels therefore confirm that T6SS are not active during conditions where we see killing of E. coli. Fig. 3A demonstrates that the three P. mirabilis strains all possess a functional T6SS that is different from each of the other strains. Dienes lines on agar plates are the direct result of an active T6SS. Contrary to agar, there is limited extended contact time between the cells in shaking liquid media (Fig 1). T6SS are generally thought to require this extended contact time to be effective. Fig. 3B confirms that the same two species that killed each other on the solid media, as seen with the Dienes lines, do not kill each other in liquid media. The killing of E. coli that we observe thus does not seem to be the result of T6SS. To directly show that the killing is T6SS-independent and not to merely infer, we used a T6SS-deficient strain and a co-culture of P. mirabilis and E. coli in panels D+E.
Figure 5 D-F: The authors conduct several initial investigatory assays to narrow down the potential component(s) of the P. mirabilis killing system. To help further triangulate the nature/ identity of the effector molecule(s), have the authors considered studies such as SDS PAGE, HPLC, or mass spectrometry analysis of the different P. mirabilis conditioned media (heat-inactivated, inducing vs non-inducing supernatant, etc)?
Authors’ reply: These are indeed future directions we are very interested in pursuing for a future manuscript. In order to identify the nature of the communicatory or potential quorum sensing molecule, we will fractionate the supernatant and test fractions in killing induction experiments. The fraction(s) with the highest potency in killing induction can then be analyzed via Mass Spectrometry to identify the molecule(s). <br />
Our data indicate that the killing is contact dependent. We therefore consider it likely that the components that are required for the killing process (e.g. a scaffold that binds effector molecules) are cell-bound and not in the supernatant. In order to identify the genes coding for these molecules, we plan on screening a P. mirabilis library to identify mutants unable to kill E. coli. This screen might also help us to identify the communicatory molecule. Additionally, we are performing RNAseq to identify genes whose transcription is upregulated during treatment with “inducing” supernatant (Fig. 5). As E. coli loses viability in the inducing supernatant, but not the heat-treated inducing supernatant, we expect to find transcription of genes coding for killing system genes upregulated in the inducing compared to the heat-treated inducing supernatant.
While these experiments may not elucidate the mechanism entirely, they would likely help describe what components in P. mirabilis regulate this novel system of contact-dependent interference against competing Enterobacteriaceae.
Authors’ reply: We thank you for carefully reviewing our manuscript and your helpful comments.
We also noted a few other minor points while discussing this manuscript:
Figure 2: The authors delineate conditions required for P. mirabilis-mediated killing of E. coli (contact and live cells). Have the authors considered how anaerobic conditions (such as the gut) affect P. mirabilis killing?<br />
• On a small note, Fig. 2F is referenced in the manuscript, but does not appear in the figure.
Authors’ reply: The reviewers raised a very important question. We are currently performing co-culture experiments in anaerobic and microaerophilic environments. The gut mucosa is known to contain more oxygen than the lumen, especially in the infant gut microbiota. Specifically, P. mirabilis is known to colonize the mucosa more than the lumen. We will address the questions under which conditions the system is active with future in vitro and in vivo experiments.<br />
Thank you for noticing the reference to the non-existing figure. This will be corrected in a future version of this manuscript.
Figure 4B: The observation that E. coli membrane is not compromised during killing raised the question—are we observing true cell death or metabolic inactivity more similar to senescence?<br />
• We would appreciate a comparison of biological loss of membrane integrity (as opposed to ethanol treatment) to gain a better sense of the morphology of compromised bacterial membranes
Authors’ reply: This is a very interesting comment and something that we had wondered as well. We indeed do not know without any doubt whether the E. coli cells are dead or metabolically inactive. However, bacteria generally resume metabolic activity and emerge from senescence once the stressor is removed. This is the case for bacteria that enter a persister state during antibiotic pressure and resume growth when the antibiotic is removed. As E. coli cells are unable to resume growth on rich media agar plates, this strongly suggests that they are indeed dead. There are enzymatic toxins, for example some colicins, that kill cells without affecting membrane integrity. The effector molecules cleave either DNA, rRNA, or tRNA, thus degrading the genome or arresting protein synthesis. The cells maintain membrane integrity but fail to grow as vital parts of the cell were destroyed by the toxin. <br />
Antimicrobial peptide LL-37 results in loss of membrane integrity and we can use this as a control to visualize loss of membrane integrity by an antimicrobial compound. In addition, we plan to use negatively-charged dyes like DiBAC4 (PMID: 31792213) that enter cells that have lost membrane potential but are excluded from metabolically active cells.
We congratulate the authors on their efforts and hope our reviews are helpful in any revisions to this manuscript.
Authors’ reply: We thank the reviewers for taking the time to review and for their positive evaluation of our manuscript.