Kinetochore-independent chromosome segregation driven by lateral microtubule bundles

Reviewer #1:

1) I'm confused why KLP-19 is not included in the model, while dynein is. Though it is part of the “ring” complex, it would be helpful if this were shown, or at least mentioned in the figure legend.

We originally did not label KLP-19 in our diagram because we did not want to give the incorrect impression that the ring was comprised only of KLP-19 or that we had evidence that KLP-19 is the only plus-end directed force acting on chromosomes. However, we have taken the reviewer’s helpful suggestion and have now updated the figure legend to include mention of KLP-19, which avoids this problem.

2) A color coded legend would make the model much easier to interpret.

We have now added a legend.

Reviewer #2:

From the data presented, however, it is not clear how the meiotic spindle in C. elegans is organized. Although not within the scope of this light microcopy study, I would like the authors to comment on the following issues related to spindle architecture:

1) The presented model is based on the idea that chromosomes/chromatids move laterally along microtubule bundles. What about the polarity of the microtubules in the system? Isn't it a requirement of the model that all microtubules are oriented in a way that the minus ends of the microtubules are facing towards the acentrosomal poles?

We do not have direct evidence by EM that microtubules within the bundles are organized in the way we depict, and we have now acknowledged this point and brought up the issue of microtubule polarity in the Discussion. As the reviewer suggests, we favor the idea that microtubules in the laterally-associated bundles are organized primarily (if not completely) with the polarity depicted in Figure 7; there may also be non-bundle associated microtubules of mixed polarity within the spindle as well, but our model focuses only on bundle-associated microtubules. Although our experiments don’t definitively show the polarity of microtubules within the bundles, we think that this interpretation is most consistent with our data. In monopolar spindles (where it is easier to distinguish between poleward and anti-poleward forces because there is no zone where bundles from opposing poles overlap), rings consistently stretch in one direction during metaphase arrest and chromosomes are subjected to an opposite force (Figure 4A). The simplest explanation for this behavior is that microtubules within the bundles are all oriented the same way; if microtubules were of mixed polarity, we would predict that the rings would not display such consistent behavior (always stretching in the same direction). Since we have identified a plus-end directed force (KLP-19) in the ring, and dynein inhibition causes chromosomes to be found further from the central pole, our data are most consistent with an organization in which the minus ends are in the center of the aster and the plus ends extend outwards. In the bipolar spindle, this would mean that the minus ends are focused at the two poles. Although we think that performing a detailed analysis of anaphase spindle morphology by EM is a worthy thing to do and is an important test of our model, we agree with the reviewer’s later comment that it is beyond the scope of this study.

2) The model does also imply that the number of microtubules from metaphase to early anaphase is not changing? The authors have presented a model in which an elongation of the microtubules is the major aspect of spindle elongation. This is in contrast to the Dumont model of a regrowth of microtubules between the segregating chromosomes/chromatids. Are the Dumont and Muscat models mutually exclusive?

Some aspects of our model and the Dumont model are mutually exclusive and some are not; we have now added a section to the Discussion discussing the previous model in more detail to try to clarify these points. One discrepancy relates to the idea presented in the Dumont model that the metaphase spindle disassembles from the poles so that a new microtubule array can form between separating chromosomes; this is not consistent with our observations that microtubules extend alongside or beyond chromosomes throughout anaphase and that pole proteins also remain associated with spindles, suggesting that the poles don’t disassemble. Another major difference relates to the proposed force separating chromosomes in the two models. Dumont, et al. suggest that the most likely candidate for the force driving chromosome segregation is microtubule polymerization between the separating chromosomes; microtubules coupled to the inside surfaces of separating chromosomes would serve to push them apart. This idea is not consistent with our observation of open channels within the spindles (new polymerization associated with the inside surfaces of chromosomes would serve to close the channels). Therefore, some aspects of the two models are mutually exclusive. However, the idea put forward by Dumont, et al. that microtubule polymerization contributes to segregation is certainly possible and we think quite likely; the spindle elongates significantly during anaphase and we favor the idea that microtubule polymerization contributes to this process. Our data do not directly address the mechanisms promoting spindle elongation or whether microtubule polymerization is required for segregation, as our assays are designed to focus on the forces acting on the chromosomes themselves. Therefore, we do not know whether microtubule number within the spindle changes from metaphase to anaphase and then subsequently throughout anaphase progression. We expect that future studies addressing this issue will result in a more complete understanding of how these two aspects of anaphase spindle behavior (chromosome to pole movement along lateral bundles, reported here, and spindle elongation) together facilitate segregation.

Reviewer #3:

Suggested experiments/presentation of data:

1) The authors state that “chromosomes associate with lateral bundles of microtubules during anaphase” and that this is “inconsistent with a model in which a new array of microtubules polymerizes between separating chromosomes”. This is initially based on the observation that open “channels” exist in the MT spindle network between separating chromosomes in early-late anaphase. However, a pushing force could still come from polymerizing MTs within the lateral MT bundles. Furthermore, the fact that chromosomes move beyond the pole region at the end of anaphase suggests that a pushing force from within the midzone is involved. A simple quantification of MT levels (total fluorescence of the entire spindle) during different stages of meiosis would help address the issue of whether more MTs are generated during anaphase. This would help to verify the authors' claims that the spindle does not disassemble and reassemble MTs, rather, it reorganizes existing MTs during the transition from metaphase to anaphase.

We did not intend to give the impression that we think that microtubule polymerization does not contribute to segregation on bipolar spindles; our experiments neither support nor rule out this idea. As discussed above in response to reviewer 2, the idea that microtubule polymerization may contribute to chromosome separation during the process of spindle elongation is entirely compatible with our model (e.g., by elongation of the microtubule bundles), and microtubule polymerization in this context could be promoted by CLASP, as Dumont et al. suggest. We had tried to account for this possibility in the original manuscript with the sentence: “However, the major length changes that occur in the C. elegans oocyte spindle (with it shrinking and then elongating as anaphase progresses) make it likely that spindle elongation also contributes to segregation, potentially driven by microtubule polymerization as this previous study proposed.”

However, we have now tried to more clearly articulate this point, with an expanded discussion of the differences between the two models and which features are compatible/incompatible, which will hopefully prevent further confusion. We have also revised the quoted sentence to read “…are inconsistent with a model in which a new array of microtubules polymerized between separating chromosomes provides a pushing force on their inside surfaces to drive separation”, to be more precise about the feature of the previous model that is most incompatible with our data.

In addition to these changes to the text, we also took the reviewer’s suggestion and attempted to quantify our fluorescence images, to see if we could find evidence to either support or disprove the idea of microtubule polymerization occurring during anaphase. We did a number of new imaging experiments, where we fixed large numbers of embryos and collected as many anaphase images as we could per experiment, using uniform parameters (e.g., exposure, image size, images from an individual experiment taken on the same day, etc.). We then processed each undeconvolved image using the analysis software Imaris v 8.0.2. We used the Surfaces module to calculate the size of the surface representing the spindle, keeping parameters applied to each image the same, and we then recorded the sum intensity of tubulin fluorescence from each surface. For each image we also calculated chromosome distance by generating a surface model for each cluster of separating chromosomes, allowing the software to call the center of each surface, and measuring the distance between the two center points. Using this strategy, we obtained the following intensity values for a set of anaphase spindles from one of our experiments, arranged from shortest chromosome distance (i.e., early anaphase) to largest (late anaphase):

From this data, we did not see a consistent trend that would allow us to argue either for or against microtubule polymerization occurring during anaphase. We think that the variations we observe are more likely the result of experimental variation (e.g., fixation differences that lead to differences in the efficiency of staining) rather than reflecting actual changes in microtubule number. We tried extensively to optimize this procedure, imaging both metaphase (not shown) and anaphase spindles, but we saw a similar level of variation in each experiment. We also used a number of different parameters in Imaris to calculate spindle volume, in case that was the source of variability, and we also performed a set of calculations where we manually set a region of interest around each spindle by eye, to rule out the possibility that the program was not accurately calculating spindle volume. However, we observed a similar level of variation no matter what method we used. We therefore think that definitively answering this question will not be possible using fixed imaging, and would be better determined by high resolution live imaging, where timepoints within the same spindle can be compared. Because we have not yet optimized live imaging conditions at a resolution that would be sufficient to accurately calculate microtubule density in the entire spindle volume, and also because neither outcome of this experiment would negatively impact our model (as our model focuses more on the forces acting on chromosomes and not on the molecular mechanisms that mediate spindle reorganization) we request that this experiment be viewed as the subject of future investigation, beyond the scope of this study.

Other points:

1) It looks like MTs do not form bundles in the monopolar spindles. If bundles do not form, is it fair to compare the movement of chromosomes in this system to a bipolar array?

In a previous study, we demonstrated that microtubules in monopolar spindles do form bundles that form lateral associations with bivalents, just as they do in bipolar spindles. This data is reported in Wignall and Villeneuve (2009), Nature Cell Biology. Bundles are shown in Figures 2A and 2D (monopolar spindles) and Figures 2B and 2C (bipolar). These bundles are apparent in images that are partial projections of z-stacks immediately surrounding individual bivalents; in full projections other microtubules in the spindle often obscure the bundles (which is the case in most of our images in the current manuscript). Therefore, although we did not present our images in a way that highlights this particular feature of monopolar spindle organization, we think that the presence of lateral bundles surrounding bivalents in both monopolar and bipolar spindles makes it possible to compare the movement of chromosomes in these two contexts. To make this point more clear to readers, we have added a sentence to the text and cited our previous work.

2) In Figure 3, it is difficult to determine whether the chromosomes are moving along the sides of microtubules in these panels (and the supplemental movie) or whether they are moving in conjunction with MT depolymerization. Could MT depolymerization be the main driver of minus-end movement?

We were also concerned about this possibility. However, we think that two lines of evidence support the interpretation that microtubule depolymerization is not the main driver of poleward movement. First, spindles are much larger following MCAK depletion (with clear microtubules extending beyond the chromosomes in anaphase) and separated chromosomes still move to the center of the monopolar spindle (Figure 3D). Second, during metaphase arrest we observe a significant number of monopolar spindles where chromosomes exhibit anaphase-like poleward movement after ring removal (Figure 4A). However, these monopolar spindles remain large since they are still in metaphase, making it unlikely that spindle shrinkage and microtubule depolymerization drives these poleward movements.

3) Some aspects of the model are unclear. Muscat et al. suggest that MT polymerization is not required for chromosome segregation, so how do the authors reconcile data from the Desai lab that showed that loss of CLASP abolished segregation? Could polymerization along the lateral bundles contribute to the segregation, or do the authors think that CLASP is doing something other than promoting MT polymerization?

This issue is addressed in response to major point #1, above.

The authors suggest that, during congression, the plus-end forces are active but that the minus-end forces start to accumulate (Discussion, second paragraph). But in the next sentence, it is the overlapping bundles of MTs with mixed polarity that causes the movement of chromosomes to stop in the middle of the spindle.

First, how can the congression work if dynein is also present? Figure 7 suggests that plus-end forces are active during congression, so is dynein accumulating but not active until after congression? What is a “weak” minus-end force, and what changes to make it stronger?

We have expanded our description of our dynein data in the manuscript to clarify these questions. We have done extensive imaging of dynein during meiosis using a variety of reagents/antibodies, and our results are consistent with the idea that dynein is present at low levels on chromosomes during prometaphase/metaphase; we can see faint dynein cupping chromosomes but it is usually difficult to distinguish from spindle-localized dynein. The images we show of dynein localization on monopolar spindles (Figure 5A) are clearer (probably because the chromosomes are no longer located in a dense zone of overlapping microtubules that would obscure the chromosomal pool), but the staining is still faint and the images we show are not deconvolved, which was necessary to preserve the signal. However, it is much easier to visualize chromosomal dynein in anaphase, even in deconvolved images (Figure 5B), suggesting that the amount of dynein on the chromosomes increases prior to or during anaphase. We therefore suggest that the amount of dynein on the chromosomes contributes to the strength of the force. Early in congression the minus-end force would be weaker and the plus-end forces would be dominant, but then dynein would accumulate to help drive segregation. This idea is now more clearly articulated in the text and depicted in Figure 7B.

Second, it is not obvious how the chromosomes escape the overlapping MT zone as depicted in the last two panels of Figure 7B (ring removal alone cannot explain it, unless the overlapping MT zone becomes extremely small). Is it possible that something else (e.g. CLASP-dependent MT polymerization?) is required to push the chromosomes beyond the region of mixed MT-polarity, to allow the minus-end forces to take over? Or could CLASP simply provide many “loose”, non-spindle-pole MTs that reduce the opposing plus-end resistance simply by competition? Or is it that MT plus-end directed forces acting between anti-parallel MTs within the bundles eventually shrinks the overlap region to the size depicted in the cartoon? Please provide some suggestion(s) for how this transition might occur.

We have now added a paragraph to the Discussion to address this issue and propose a number of suggestions.