Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #2
Evidence, reproducibility and clarity
This paper focusses on the mechanisms underlying chromosome biorientation in mitosis, an essential process that warrants equal chromosome segregation to the dividing cells. Correction of improper kinetochore-microtubule attachments relies on two conserved protein kinases, Aurora B and Mps1, that detach kinetochores that are not under tension in order to provide them with a second opportunity to establish bipolar connections. In vivo, Aurora B and Mps1 have intertwined functions and share some common targets. For this reason, despite the large body of literature on the subject, their precise roles in chromosome biorientation have been difficult to tease apart.
The authors take advantage of an in vitro reconstitution assay that they previously published (Akyioshi et al., 2010) to identify the critical target(s) of Mps1 in weakening kinetochore-microtubule connections. The assay uses kinetochore particles purified from budding yeast cells that bear Mps1 but are notably deprived of Aurora B. Upon addition of ATP to activate the co-purified kinases (e.g. Mps1), kinetochores are added to coverslip-anchored microtubules to which they attach laterally. Through a laser trap, kinetochores are brought to the microtubule plus-end and pulled with increasing force until the kinetochore detaches, which allows measurements of the average rupture forces that reflect the strength of the attachments. The approach is straightforward and potentially very powerful, first because it provides a simplified experimental set-up in comparison to the cellular context, and second because it directly measures the impact of protein phosphorylation on the strength of attachments.
The authors convincingly show that Mps1-dependent phosphorylation of the N-terminal part of Ndc80 significantly weakens the strength of kinetochore-microtubule attachments in vitro, while phosphorylation of other known Mps1 targets, such as Spc105, does not seem to have an effect. Eight phosphorylation sites in Ndc80, which were previously identified as Mps1-dependent phosphorylation sites (Kemmler et al., 2009), are shown to be critical to destabilise kinetochore-microtubule attachments in the in vitro reconstitution assays. The authors also present evidence for a moderate involvement of Ndc80 phosphorylation by Mps1 in correcting improper attachments in vivo, suggesting that additional mechanisms are physiologically relevant for error correction.
The experiments are mostly well designed, the data are solid and support the main conclusions. However, to my opinion additional experiments could be performed, as outlined below, to strengthen the physiological relevance of the main findings and corroborate some of the conclusions.
Major points:
- Given the partially overlapping function of Mps1 and Ipl1 (Aurora B) in error correction, the ndc80-8A mutant should display synthetic growth and chromosome mis-segregation defects with ipl1 temperature-sensitive alleles. Conversely, the ndc80-8D mutant should suppress the lethality at high temperatures of mps1-3 mutant cells, which were recently shown to be defective in chromosome biorientation (Benzi et al., 2020). Finally, chromosome mono-orientation could become apparent in ndc80-8A cells upon a transient treatment with microtubule-depolymerising drugs, which should amplify the cellular need for error correction.
- The authors show that Mps1-dependent phosphorylation of Ndc80 is not involved in the spindle assembly checkpoint, a conclusion that contradicts a previous report (Kemmler et al., 2009). They also find, in contrast with the same report, that the lethal phenotype of the ndc80-14D phospho-mimetic mutant cannot be rescued by disabling the spindle checkpoint. In my opinion, Kemmler et al. convincingly showed, through a number of different experimental approaches, that ndc80-14D cells die because of spindle checkpoint hyperactivation. Not only deletion of checkpoint genes was shown to rescue the lethality, but re-introduction of a wild type copy of the deleted checkpoint gene reinstated lethality. Thus, the explanation invoked here that spontaneous suppressing mutations could underlie the viability of ndc80-14D SAC-deficient mutants is not consistent with the published observations. A thorough examination by the authors of the phenotype of ndc80-14D cells in their hands should be carried out to support these conflicting conclusions. If authors find that ndc80-14D cells actually die because of chromosome mono-orientation, then this would highlight an important function for some or all the six additional phosphorylation sites, relative to the ndc80-8D mutant, for chromosome biorientation in vivo.
- The conclusion that Spc105 phosphorylation by Mps1 is not required for the Mps1-mediated weakening of kinetochore attachments in vitro is based on the comparison between kinetochore particles bearing wild type, untagged Spc105 and particles bearing non-phosphorylatable Spc105-6A tagged at the C-terminus with twelve myc epitopes. Thus, the presence of the tag could obliterate the effects of the mutations in the phosphorylation sites by destabilising kinetochore-microtubule attachments in the presence of ATP. Consistent with this conclusion, Spc105-6A-12myc-bearing kinetochores withstand lower rupture forces than Spc105-bearing kinetochores upon ATP addition. Furthermore, Spc105-6A-12myc kinetochore particles show an interacting protein at MW above 150 KD that is not present in wild type particles (Fig. S2A), suggesting that either the tag or the mutations might affect kinetochore composition. Thus, this set of experiments should be repeated using Spc105-6A kinetochore particles lacking the tag.
- In general, it would have been informative to complement the data presented here with a mass spec analysis of the composition of kinetochore particles, at least for the experiments that are most relevant to the conclusions. For instance, the composition of the Ndc80-8A kinetochore particles is assumed to be similar to that of wild type kinetochores based on gel silver staining (Fig. S4A; note also that ndc80-8A particles are compared to ndc80-8D particles and not to wild type particles). However, the authors previously showed that kinetochore particles purified from dad1-1 mutant cells (affecting the Dam1 complex) have an apparently identical composition to particles purified from wild type cells by silver staining, yet they display significantly lower resistance to the rupture strength in vitro (Akyioshi et al., 2010). What is the status of the Dam1 complex (or other kinetochore subunits) in kinetochores purified from ndc80-8A/-8D or spc105-6A cells relative to wild type kinetochore particles?
Minor comment:
I believe that the right reference for the sentence in the Discussion "If Aurora B is defective, for example, the opposing phosphatase PP1 prematurely localizes to kinetochores" is Liu et al. 2010.
Significance
Although the experiments are well designed and the conclusions are mainly supported by the data, the question arises as to what extent the in vitro assays recapitulate, at least partly, what happens in vivo. An emblematic example is the involvement of Spc105 in the error correction pathway. The Biggins lab previously showed that Spc105 phosphorylation by Mps1 and subsequent Bub1 recruitment is not only essential for the spindle assembly checkpoint, but is also crucial for chromosome segregation in vivo, as shown by slow-growth phenotype and aneuploidy of the spc105-6A non-phosphorylatable mutant (London et al., 2012). Additionally, a recent paper showed that Spc105 is a crucial Mps1 target in chromosome biorientation (Benzi et al., 2020).
In sharp contrast, the ndc80-8A mutant, which in vitro completely erases the ability of Mps1 to destabilise kinetochore-microtubule attachments, displays no growth defects in otherwise wild type cells and only modestly enhances chromosome mis-segregation in a mutant affecting an intrinsic correction pathway (stu2ccΔ). The N-terminal part of Ndc80 (aa 1-116) containing the aforementioned eight phosphorylation sites can even be deleted altogether without any consequence on cell viability (Kemmler et al., 2009). Thus, although the in vitro assays presented here produced clear-cut and reproducible results, their physiological relevance in vivo remains unclear.
Left apart this criticism, the manuscript has several merits outlined above and will be of interest for people working in the fields of chromosome segregation, kinetochore assembly, spindle assembly checkpoint, etc.
Expertise of this reviewer: mitosis and related checkpoints

