Systematic analysis of microtubule plus-end networks defines EB-cargo complexes critical for mitosis in budding yeast

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

Comprehensive analysis of EB-dependent microtubule plus-end networks in budding yeast defines a Bim1-Bik1-Cik1-Kar3 complex that acts during metaphase spindle assembly and supports tension establishment and sister chromatid biorientation.

Article activity feed

  1. Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons


    Reply to the reviewers

    We thank both reviewers for their constructive criticism and the insightful comments on our manuscript. Reviewer 1 states that:

    „The strength of this manuscript lies in its comprehensive analysis of Bim1 function, the quality of the results and that the experiments are generally well controlled and interpreted. „

    And „the findings of this comprehensive analysis are of great value to the microtubule field, especially for people working in budding yeast. „

    While Reviewer 2 adds:

    „The current study is indeed rich with new insights into the mechanisms by which these molecules function, and will no doubt prove valuable to a number of people in the microtubule/motor/yeast mitosis fields. As someone who is interested in and studies mitosis in budding yeast, I found the study to be interesting.

    Both reviewers conclude that:

    “…there are useful data in the manuscript that make this an important contribution and that it should definitely be published”

    Both reviewers raised two major areas of concern: 1. A confusing overall structure makes the study hard to follow. 2. A clearer distinction needs to made between what has already been reported in the literature, and what are new insights provided in this study. In this regard, the appropriate citations need to be made at various positions throughout the manuscript.

    In this full revision, we have addressed these major points of criticism of the reviewers as follows:

    We have re-organized and re-focused the manuscript to make it more accessible and easier to follow for the reader. We have followed a suggestion from reviewer 1 and now present all experiments characterizing mitotic spindle phenotypes and how they can be suppressed consecutively in Figures 2-5 and then finish the manuscript with the characterization of the spindle orientation phenotype. This way of ordering by biological pathway allows for a better flow of the manuscript.

    Throughout the text, we have added citations to better indicate the previous state of knowledge and how the presented experiments either confirm or extend the previous findings in the field. This helps to put our current study better into and overall perspective.

    In addition, we have addressed the specific points raised by both reviewers in full. Please see below our point-by-point answer.

    Reviewer1

    There is already a huge body of published information on mitotic spindle positioning via the Kar9 and dynein pathways that grew since the late 1990s. The genetic relationships and molecular interactions between the components of these 2

    pathways are well studied (many studies, including Liakopoulos et al. 2003, are not cited by the authors). The authors

    should make sure to cite and compare to the relevant primary literature when they report findings that have been

    described before. This will help to distinguish novel findings from validation of previous results.

    We have added relevant citations throughout the manuscript, please see below.

    "The strict dependence of Kar9 and Cik1-Kar3 on the presence of Bim1, as well as the different effects of bim1Δ on

    nuclear and cytoplasmic Bik1, may reflect the formation of stable complexes between Bim1 and these binding partners in

    cells." I believe this has already been shown (Kumar et al., 2021 and Manatschal et al., 2016). There are several other

    instances as well where additional literature should be cited, for example Gardner et al., 2008 and Gardner et al. 2014.

    We have now cited the Manatschal and Kumar papers in this section of the revised manuscript. We have also cited the mentioned Gardner papers later in the manuscript.

    The selection of targets to study in figure 1 doesn't seem to follow the listed criteria. Many proteins included in the

    study were not found by IP-MS, but some perfect targets according to the listed criteria like Duo1 were not included in the

    study. In addition, there are more sophisticated ways of finding Bim1 binding motifs in the literature

    (https://doi.org/10.1016/j.cub.2012.07.047). I suggest, the authors declare that they rationally chose to study 21 proteins

    of interest but remove the claim that their approach was systematic.

    We have changed the wording accordingly and removed the claim of systematic target selection.

    Much of the microscopy data was acquired after release from alpha factor arrest. What is the reason for this

    perturbation? An exponentially growing culture should mostly consist of mitotic cells anyway. Since this treatment affects

    cell size and potentially protein levels/concentrations, testing its influence on spindle position as well as levels on MTs for

    the most relevant proteins of interest would be important to exclude introduction of artifacts.

    In principle that’s correct, but using synchronized cultures has the great advantage that mitotic timing and all the parameters associated related to it (spindle length etc.) can be quantified much better and we obtain larger N and thus get better statistics using this approach. In a typical log culture only one third of the cells are in mitosis and this entails very different states of mitosis. Observations times are limited due to fluorescent bleaching and low signal intensity. We therefore feel the benefits of alpha-factor release outweigh the problems and we compare all mutants under the same conditions.

    Some of the results obtained from bim1Δ cells are a challenge to interpret due to the wide range of processes that

    involve Bim1 and therefor the potential for many off-target effects- including a global change in microtubule dynamical

    behavior in both the cytoplasm and the nucleus that will influence the length distributions and microtubule lifetime (and

    thus number). The authors must carefully consider these caveats.

    We agree in principle and have therefore not only characterized the bim1 deletion, but also more specific bim1 mutants. We also show that some aspects of the bim1 delta phenotypes, but not others, can be rescued by different strategies.

    The results section on page 12 refers to phenotypes of kar9 delete cells with respect to Bim1-GFP on cytoplasmic

    microtubules. In the figure 3D,F I only found data for Kar9-AID, though. The authors should refer to supplementary figure

    5A or even better include quantification similar to figure 3F.

    We have corrected this in the revised text. We refer to the Kar9-AID, for which we have the quantification.

    The observation that cytoplasmic Bim1 localization depends on interaction with its cargo Kar9 (figure 3 + 7) fits into the

    model that Kumar et al (https://doi.org/10.1016/j.str.2021.06.012) proposed in which Kar9 oligomerization is required for

    its Bim1 dependent localization to microtubules. It would be valuable to point that out.

    We have now included a sentence that our findings support this model and added the respective citation.

    I don't fully understand the model proposed in Figure 5H and discussion page 26. Based on figure 5E, it does not look

    like there is a higher concentration of Bik1 along the lattice in bim1 delete. So how would Bik1 increase Kip2 processivity

    if its levels are only increased due to a MT length change? If Kip2 was not fully processive, you would rather expect to

    see less of it at the tip of a longer microtubule in bim1 delete. The model suggested by Chen et al

    (https://doi.org/10.7554/eLife.48627.001) suggests that Kip2 only gets loaded at the minus-end and processively walks

    towards the +end without falling off. Are the authors suggesting that bim1 deletion changes this behavior?

    We have rephrased this section in results and discussion and more clearly state that there is no increase in Bik1 per MT length unit in the bim1 deletion. We have amended the discussion and grant that we currently cannot explain by which molecular mechanisms Bik1 may contribute to the observed increase in Kip2 plus-end localization under conditions of a bim1 deletion.

    I don't see evidence for independent pools of Bik1 in the cytoplasm and nucleus as claimed on top of page 21. Total

    Bik1 levels on cytoplasmic microtubules seem to be well explained by their length. Please explain better or remove the

    statement.

    We have removed the respective statement from the revised manuscript.

    The experiments in supplementary figure 7B are difficult to interpret. The localization on cytoplasmic microtubules is

    different, but probably explained by the formation of Bim1 heterodimers. Therefore this experiment is difficult to interpret

    and should be removed.

    As requested, we have removed this experiment from the revised manuscript.

    top of page 24: Kar9 localization in metaphase depends exclusively on SxIP, not on LxxPTPh (Manatschal 2016). The

    paragraph should be removed as it is not supported by published data or sufficiently by the authors to merit the

    conclusion.

    We have reformulated this to avoid a misunderstanding. We merely show that in the context of the artificial GCN4 construct a fragment just including the LxxPTPh motif is sufficient for Bim1-dependent localization to microtubules in nucleus and cytoplasm. This makes no statement about localization determinants of the authentic Kar9 protein.

    Top of page 26: The genetic interactions between the Kar9 pathway and the dynein pathway were already well known

    before this work. Please reformulate accordingly.

    We have re-written this section and introduce the two pathways with the respective citations in the very beginning of the section before describing the experiments.

    page 27 second paragraph: There is no selective pressure to evolve compensation mechanisms for gene deletions. I

    suggest the authors consider that Kar9 and dynein partially redundant, with Kar9 acting to position the spindle prior to

    metaphase and dynein to maintain spindle position in the mother and bud compartments in late metaphase and

    anaphase. The authors should consider the quantitative analysis of Kar9 and dynein dependent spindle positioning

    reported in Shulist et al. 2017 and the method for analysis of spindle length and position in 3D in Meziane et al. 2021.

    We have rephrased the section on the partially redundant Kar9 and Dynein pathways. See below our answer for measuring spindle length.

    In addition, it is not clear to me which results suggest that the relocalization of Bik1 is required in the bim1 delete. Why

    would wild type levels not be sufficient for dynein pathway function? The authors have not conclusively shown that

    nuclear migration relies on upregulating the dynein pathway in bim1Δ cells. If there is no supporting data, the paragraph

    should be removed.

    In this revised manuscript we have phrased our observations more carefully and acknowledge the limitations regarding molecular insights. We present indications for increased levels of Dynein-Dynactin pathway components at plus-ends in the bim1 deletion cells, but it is indeed unclear, whether an increased Bik1 level in the cytoplasm is required to achieve this.

    Please provide more details about intensity quantification on page 35. Were these measured on sum or max

    projected stacks? What was the method of background subtraction?

    Analysed images are optical axis integration scans over 3 μm taken on a Deltavision microscope. This procedure gives a sum projection. Local background was determined for every cell by drawing a line under a signal curve derived by line scan. The background line connects regions that are still within the cell but are outside of spindle (or microtubule). We added a sentence in the materials and methods section under point 2.

    Are the spindle lengths in Figure 2E measured in 2D or 3D? Bim1 deletion might lead to more misalignment of the

    spindles in z due to inactivation of the Kar9 pathway and thus partially explain the shorter spindles. The measurements

    should therefore be performed in 3D.

    As we have used optical axis integration (OAIs) on the Deltavision microscope and obtained a sum projection of this virtual stack, the spindles were measured in 2D and we don’t have the information to measure in 3D (this would require a regular stack). We show that there are different ways to restore different aspects of spindle length with alternative strategies. These are unlikely to influence just spindle orientation. In addition, we see that Bim1 deletion has an effect on the size of a nascent bipolar spindle when spindle orientation is similar to wild-type cells. We agree that z-misalignment may affect absolute values of spindle size of Bim1 deletion in late metaphase and it would be better to measure in 3D. However, we think in this case it is unlikely to affect our conclusions in this study.

    The authors should try to shorten the text. There is a lot of redundancy between results and discussion sections.

    We have to shortened the text to avoid redundancy (before >43000 characters, now around 41000 characters, and we have decreased the number of main figures from 9 to 8.

    Data is shown that leads to conclusions that are already supported by the literature should be moved to the

    supplementary material.

    In the course of re-organizing the manuscript we have tried to do this.

    Reviewer 2:

    "Robustness of Ndc80 loading might be achieved by the coexistence of multiple kinetochore assembly pathways or

    alternatively determined by intrinsic Ndc80 properties." Wouldn't Ndc80 levels be determined by Ndc80 kinetochore

    loading, and not by end-binding proteins? This seems to be the more likely means to regulate Ndc80 levels.

    We have removed this statement from the revised manuscript.

    "Upon analyzing the associations in the cytoplasm, we found that Kar9-3xGFP foci on bud-directed cytoplasmic

    microtubules were abolished in the bim1Δ strain, consistent with earlier reports." It would be helpful if the authors

    commented on the how the localization of some of these proteins are affected by bim1Δ on the mother-directed plus

    ends. Although I understand the need to account for one class of plus end for the sake of consistency (and the distinct

    behaviors of the mother vs bud-directed plus end), the text as written leaves me wondering about the other plus end.

    We have noticed that the bim1 deletion led to the loss of asymmetric distribution on cytoplasmic microtubules for a number of components. Most prominent are Bik1, Kip2 and proteins of dynein-dynactin complex. We felt that further analysing this phenotype was beyond the scope of this study.

    "The CAP-Gly domain construct, expressed from a BIM1 promoter, almost exclusively localized to the spindle of yeast

    cells." For clarity, the authors should explicitly state that the CAP-Gly domain in question is from Bik1. Although this can

    be deduced, this was not abundantly clear.

    We have clarified this in the text and in the figure.

    "In addition to Ase1, we followed the kinetochore proteins Ndc80-GFP and Sgo1-GFP which specifically marks

    kinetochores that lack tension." This sentence should add "the latter of which..." to clarify that SgoI, but not Ndc80

    exhibits this behavior.

    We have added the phrase “the latter of which” to clarify this point.

    "We observed that bim1Δ cells had mispositioned kinetochores with a bright Sgo1-GFP signal that was much stronger

    than in wild-type cells." I don't see the mispositioned kinetochores described here. Are the authors referring to the fact

    that Sgo1 is brighter, which suggests tension-free KTs? If so, this should be clearly stated as such, since the authors are

    not explicitly assessed kinetochore "positioning".

    We have rephrased the sentence to clarify. We refer to a lack of bi-lobed Ndc80 signal and a bright Sgo1-GFP signal as two aspects of the phenotype.

    "We speculate that Bim1-Bik1 in a complex with its cargo Cik1-Kar3 is active after bi-polar spindle formation but before

    late metaphase and Ase1 can partially substitute for nuclear Bim1 functions." I struggled to grasp the reasoning for these

    conclusions. I assume the former point (the timing for Bim1-Bik1-Cik-Kar3) is due to the localization dynamics of Bim1

    and Bik1, while the latter (Ase1 can substitute for Bim1) is due to the synthetic interaction between Bim1 and Ase1 (I

    needed to look this latter point up myself). Or is this latter point due to the brighter spindle Ase1-GFP intensity? In either

    case, the authors should more clearly state their reasoning.

    We have clarified this statement in the revised discussion.

    The error bars in Figures 3A and 6E (shown as 95% CI) and elsewhere seem very small for the parameters that are

    being plotted. Spindle length values as shown in Figure 2E cover a broad range (as would be expected for a biological

    process), and it would be more accurate if the error bars in Fig 3A and 6E reflect this, even if it means they start

    overlapping each other. I find the error as shown to be misleading to your readers, and unless the authors have very

    good reason to use 95% CI (which is not as meaningful as standard deviation), then I would encourage them to use

    standard deviation.

    We prefer to use CI for the spindle length plots over time for consistency reason and to avoid overlap, which would make the graphs difficult to read. We have changed the text to provide the standard deviation instead of the standard error of the mean for spindle length and metaphase duration, see point below.

    The same is true for the values stated throughout the text (e.g., for mitotic timing "47{plus minus}2 min" for metaphase

    duration; for distance between SPB and bud neck {plus minus} 0.1 μm, etc). I am highly skeptical that metaphase

    duration (for example) ranged from only 46-48 minutes. Please use standard deviation to describe a more accurate

    description of the range of values for these parameters.

    In the revised manuscript, we now give the mean values plus/minus standard deviation, instead of the standard error of the mean, as requested. In addition, the range of values is directly visible from the individual data points in the plots.

    "Unexpectedly, the kar9 deletion mutant displayed a slightly accelerated metaphase progression relative to wild-type

    cells (26{plus minus}1 min) (Figure 3C). This could be attributed to an increased level of Bim1 on the metaphase spindle

    of kar9Δ (or Kar9-AID) cells." The authors should give us more rationale to explain the "attributing the increased levels of

    Bim1" point here. Do they think that the levels of spindle-associated Bim1 impact metaphase duration somehow? If so,

    how?

    We have added a sentence, speculating about how this could be accomplished.

    "Overall, our cell biology data suggested that major nuclear Bim1 functions are conducted in a complex with Cik1-

    Kar3, while Bik1 and Kar9 have a smaller impact, probably affecting the nuclear- cytoplasmic distribution of Bim1."

    Although I understand and agree with the former conclusion (that Bim1 functions are conducted via Cik1-Kar3"), the latter

    was confusing to me. Did the authors mean that "Bim1 impacts Bik1 and Kar9 to a lesser extent", rather than vice versa?

    The authors are discussing Bim1 functioning via Cik1, but then switch to discussing how Bik1 and Kar9 affect Bim1.

    We have removed the second part of the sentence from the revised manuscript.

    "Next, we compared the comparing genetic interaction profile of a bim1 deletion to that of various other factors by reanalyzing the synthetic genetic interaction data..." Remove "comparing".

    Thanks for pointing out this typo, we have removed it in the revised manuscript.

    As someone who is unfamiliar with the analysis shown in Figure 3H, I think it would be useful to list a Pearson

    correlation value for two genes that are not functionally related. This would help define a lower limit for this analysis.

    For functionally unrelated genes the Pearson correlation between genetic interaction (GI) profiles is very close to zero. The graph below depicts Pearson correlation between GI profile of Bim1 and GIs of every yeast gene (data used for graph is taken from thecellmap.org).

    The axes for the plots in Figure 5E and 5I are very confusing to me. I don't understand what I'm looking at. Why does

    it go from 0 to 1, and then back to 0-1 again? I don't see how this can account for MTs of different lengths. Normalizing all MT length values to 1 would do this, no?

    We have clarified the labelling in the revised manuscript. The x-axis gives the relative position from either the plus-end, or the Spindle pole body (both set to position 0) in micrometres. This allowed us to compare fluorescent intensities on cytoplasmic microtubules of different lengths in wild-type and bim1 delete.

    "These observations are consistent with the idea that Bik1 acts as a processivity factor for Kip2: If more Bik1 is

    present on the lattice, then more Kip2 molecules are able to reach plus-ends without detachment." Perhaps I'm

    misunderstanding the plot shown in Figure 5E, but it seems to indicate that the levels of lattice-bound Bik1 are the same

    in BIM1 and bim1Δ cells (higher SPB-localized levels, though). There are also lower levels of Bik1 at the plus ends in

    bim1Δ cells. So, if Bik1 were a processivity factor for Kip2, this would suggest that they would remain bound at plus ends

    as well, which these data suggest is not the case…

    We have added a section to the discussion that deals with this point and we speculate about the reasons why Kip2 is increased at plus-ends, while Bik1 is not.

    "The data on the CH-Cik1 fusion is very compelling, and indeed supports their hypothesis that Bim1's main role in the

    nucleus is to target Cik1 to the spindle MT plus ends. That being said, it would be a simple, but important task to ensure

    that this fusion behaves as suggested (restores Cik1 plus end binding in cells). Otherwise, it can't' be ruled out that this

    fusion is rescuing bim1Δ functions by some other means. However, as stated above, it's unclear how much was already

    known about this fusion from the lab's previous work.

    In our previous study (Kornakov et al., 2020) we have shown that the CH-Cik1Delta74 fusion indeed is sufficient to enrich Kar3 at plus ends. We expect the same to be true for this slightly different fusion construct. We have added a respective sentence to the results section.

    Regarding the p1-p6 promoter data: p6 is missing from Figure S6A, in spite of it being referenced in the text and the

    figure.

    Thanks for pointing this out, we have corrected that in the revised manuscript and do not refer to p6 anymore.

    "Exogenously expressed Ase1 displayed a similar level and kinetics of localization compared to the endogenous

    protein, indicating that binding sites for microtubule crosslinkers are not a limiting factor on the budding yeast spindle."

    Specifically, the authors show that binding sites for Ase1 may not be limiting (the overlapping 95% CI bars if Fig S6B

    suggest this is not significant), not all crosslinkers. The authors should not use such broad language to describe results

    from one experiment with one crosslinker.

    We have rephrased to make clear that our statement only refers to Ase1.

    "We found that all bim1 mutants were less well recruited to the metaphase spindle compared to the wild-type protein,

    indicating that Bim1-interacting proteins strongly contribute to Bim1 localization." Can the authors rule out the defects in

    localization of these mutants is not compromised MT binding by the Bim1 mutants? Also, regarding this statement: "To

    test that the observed recruitment defects of bim1 mutants are not a result of a compromised spindle or microtubule

    structure, we examined their localization in a situation when GFP-tagged mutants were covered with the unlabeled wildtype

    allele. Indeed, in this situation, the Bim1 mutants displayed very similar localization profiles (Supplementary Figure

    7B)." I wasn't sure what these results were similar to: the wild-type protein, or the mutant without the presence of WT

    Bim1? The lack of quantitation made this difficult to determine.

    At the request of reviewer 1, we have removed the analysis of Bim1-GFP mutants over an unlabelled Bim1 wild-type from the manuscript.

    The zoom crops for many of the images (Fig 1F and C, 3D, 5J, etc) are not labeled. I realize the legends indicated

    what was what, but it would be much easier for the reader if these panels were labeled in the figure.

    We have indicated the channel by a respective frame around the zoom throughout the manuscript. We think this makes orientation easier.

    "While in vitro reconstitution experiments have suggested that Bim1 is required to fully reconstitute the Kip2-

    dependent loading of the Dynein-Dynactin complex to microtubule-plus ends in vitro (Roberts et al., 2014), our

    experiments indicate that it may contribute relatively little to this pathway in cells." Work from other labs have also shown

    Bim1 is dispensable for dynein function in cells. This should be noted by the authors, and the appropriate work cited (see

    work from Lee and Pellman labs. In fact work from the Lee lab showed that Kip2 is dispensable for plus end binding of

    dynein).

    We have re-written this section and now also refer to the Markus 2009 paper (Wei-Lih Lee lab).

    References are missing throughout the text. I have listed a few examples below:

    "We have previously shown that the phenotype of Bim1-binding deficient Cik1 mutants can be rescued by fusing the

    CH-domain to this Cik1 mutant (cik1-Δ74)."

    We have listed the citation of our 2020 paper (Kornakov et al.)

    "We constructed a series of strains expressing an extra copy of Ase1-GFP under different constitutive promoters of

    increasing strength (p1 to p6)"; where did these promoters come from?

    They were selected based on a systematic analysis of promoter strength in Shaw et al., 2019, DOI: 10.1016/j.cell.2019.02.023 . We have added that citation to the methods section.

    "double point mutation exchanging two conserved residues in the EBH domain (bim1 Y220A E228A) is predicted to

    eliminate all EBH-dependent cargo interactions, but does not affect protein dimerization."

    We have cited the Honnapa 2009 paper here.

    "A deletion of the terminal five amino acids is predicted to prevent binding of the CAP-Gly domain of Bik1 to Bim1. The

    combination of both mutations is expected to simultaneously prevent both types of interaction."

    We have cited the Stangier 2018 paper here.

    "Spindle positioning in budding yeast is achieved via two pathways, one relying on the protein Kar9 which interacts

    with the actin-based motor Myo2." Yin et al 2000 should be added (in addition to Hwang et al).

    We have now included the Yin et al. 2000 citation.

    "For nuclear migration to occur efficiently, the Dynein-Dynactin complex must be enriched at the plus-ends of

    cytoplasmic microtubules..." Should cite work from the Lee lab here.

    We now cite Markus and Lee, 2011 as an example.

    "These long microtubules can interact with the bud cortex and initiate pulling events to move the nucleus (Omer et al.,

    2018)." Many papers pre-dating the Omer study found this to the case, including work from the Cooper lab (see Adames

    et al). These studies should be cited either in place of the Omer study, or in addition.

    We have cited additional studies besides the Omer paper.

  2. 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

    The authors of the study performed a systematic assessment of the role of Bim1 in the MT-binding activity and function of a large number of nuclear and cytoplasmic MT-associated proteins (MAPs), as well as their role during mitosis and spindle positioning. For example, they find that the reliance of MT-binding activity of several MAPs varies from complete reliance on Bim1, to almost no role (in some cases, loss of Bim1 even increases MAP-MT binding). The density and quality of the data, and the large number of players analyzed by the study, are certainly impressive, and there is no doubt a lot of valuable information contained within that will be of use to many people in the MAP/mitosis/yeast cell biology community. However, I feel the manuscript can be greatly improved following some significant revisions. In particular, although some of their findings are indeed interesting and useful, and can be used to reliably draw conclusions, it is difficult to parse out what is novel, and what is a rehashing of old data. For instance, the role of Bim1 in Bik1/Kip2 targeting was described years (Carvalho et al), and I was surprised to see that the CH-Cik1 fusion was previously described by the authors' lab a couple years ago (see note below regarding lack of appropriate citations and lack of description of previous knowledge). Also, how much did we already know about the Bim1 truncations shown in Figure 7 and S7, and how they might disrupt binding to partners? Finally, regarding this statement in the Discussion: "Our analysis indicates that Bim1 contributes to both of these processes as part of two key protein complexes (Figure 9A): Bim1-Kar9-Myo2 in the cytoplasm and Bim1- Bik1-Cik1-Kar3 in the nucleus." As far as I know, these things have been known for many years; their work might help to support these findings, but the statement as written misleads the readers in to believing the present work proves these old concepts.

    One of the main issues with reading a manuscript with so much data about so many different players and pathways is that this leads to a situation in which each story is only superficially covered, with only minimal depth or detail. This made the paper somewhat difficult for me to follow (and I am a fan of budding yeast mitosis!), especially given the frequent switching from one pathway to another (e.g., the Cik1 section started on page 12 appears to be continued on page.17, only after talking about the spindle orientation story in between the two Cik1 sections). I'm not sure what to suggest, but the manuscript can be improved if the authors try to refocus some of the sections to make it easier to follow one story at a time, for a particular molecule (e.g., Cik1) or pathway (spindle orientation). In addition to explicitly describing what is already known about a particular molecule/pathway, the writing can be greatly improved by introducing their reasoning for the experiments in question. Some of the sections lack sufficient rationale for me to understand the justification for their experiments (e.g., why try to overexpress Ase1 to rescue bim1∆ phenotypes, as described on page 19?).

    Although there is likely much to learn from this study, I felt that some conclusions were a little bold (see below), while alternative hypotheses were not addressed (perhaps Bim1 simply competes for MT binding with some of these factors, thus accounting for them increasing their spindle-binding behavior?). For example, the authors make a point that loss of Bim1 enhances dynein-dynactin function. However, it is important to note that mutations in tubulin (tub2-430∆) and other MAPs (Kar9 or Ase1, the latter of which the authors point out) also lead to increased dynein activity (see work by Yeh et al., 2000, and work from the Moore lab). It is unknown whether mutations to these genes affect dynein targeting in cells similar to what the authors describe here. Thus, a direct causal relationship between their bim1∆ phenotypes and enhanced dynein activity is unclear, and at best is speculative. It's also worth noting that overexpression of Bik1 has been shown to actually reduce Dhc1 localization to plus ends in cells (see Markus et al 2011), which would argues against a simple mechanism of increasing Bik1 correlating with increasing dynein localization and activity.

    Below are some specific points.

    1. "Robustness of Ndc80 loading might be achieved by the coexistence of multiple kinetochore assembly pathways or alternatively determined by intrinsic Ndc80 properties." Wouldn't Ndc80 levels be determined by Ndc80 kinetochore loading, and not by end-binding proteins? This seems to be the more likely means to regulate Ndc80 levels.
    2. "Upon analyzing the associations in the cytoplasm, we found that Kar9-3xGFP foci on bud-directed cytoplasmic microtubules were abolished in the bim1Δ strain, consistent with earlier reports." It would be helpful if the authors commented on the how the localization of some of these proteins are affected by bim1∆ on the mother-directed plus ends. Although I understand the need to account for one class of plus end for the sake of consistency (and the distinct behaviors of the mother vs bud-directed plus end), the text as written leaves me wondering about the other plus end.
    3. "The CAP-Gly domain construct, expressed from a BIM1 promoter, almost exclusively localized to the spindle of yeast cells." For clarity, the authors should explicitly state that the CAP-Gly domain in question is from Bik1. Although this can be deduced, this was not abundantly clear.
    4. "In addition to Ase1, we followed the kinetochore proteins Ndc80-GFP and Sgo1-GFP which specifically marks kinetochores that lack tension." This sentence should add "the latter of which..." to clarify that SgoI, but not Ndc80 exhibits this behavior.
    5. "We observed that bim1Δ cells had mispositioned kinetochores with a bright Sgo1-GFP signal that was much stronger than in wild-type cells." I don't see the mispositioned kinetochores described here. Are the authors referring to the fact that Sgo1 is brighter, which suggests tension-free KTs? If so, this should be clearly stated as such, since the authors are not explicitly assessed kinetochore "positioning".
    6. "We speculate that Bim1-Bik1 in a complex with its cargo Cik1-Kar3 is active after bi-polar spindle formation but before late metaphase and Ase1 can partially substitute for nuclear Bim1 functions." I struggled to grasp the reasoning for these conclusions. I assume the former point (the timing for Bim1-Bik1-Cik-Kar3) is due to the localization dynamics of Bim1 and Bik1, while the latter (Ase1 can substitute for Bim1) is due to the synthetic interaction between Bim1 and Ase1 (I needed to look this latter point up myself). Or is this latter point due to the brighter spindle Ase1-GFP intensity? In either case, the authors should more clearly state their reasoning.
    7. The error bars in Figures 3A and 6E (shown as 95% CI) and elsewhere seem very small for the parameters that are being plotted. Spindle length values as shown in Figure 2E cover a broad range (as would be expected for a biological process), and it would be more accurate if the error bars in Fig 3A and 6E reflect this, even if it means they start overlapping each other. I find the error as shown to be misleading to your readers, and unless the authors have very good reason to use 95% CI (which is not as meaningful as standard deviation), then I would encourage them to use standard deviation.
    8. The same is true for the values stated throughout the text (e.g., for mitotic timing "47{plus minus}2 min" for metaphase duration; for distance between SPB and bud neck {plus minus} 0.1 µm, etc). I am highly skeptical that metaphase duration (for example) ranged from only 46-48 minutes. Please use standard deviation to describe a more accurate description of the range of values for these parameters.
    9. "Unexpectedly, the kar9 deletion mutant displayed a slightly accelerated metaphase progression relative to wild-type cells (26{plus minus}1 min) (Figure 3C). This could be attributed to an increased level of Bim1 on the metaphase spindle of kar9Δ (or Kar9-AID) cells." The authors should give us more rationale to explain the "attributing the increased levels of Bim1" point here. Do they think that the levels of spindle-associated Bim1 impact metaphase duration somehow? If so, how?
    10. "Overall, our cell biology data suggested that major nuclear Bim1 functions are conducted in a complex with Cik1- Kar3, while Bik1 and Kar9 have a smaller impact, probably affecting the nuclear- cytoplasmic distribution of Bim1." Although I understand and agree with the former conclusion (that Bim1 functions are conducted via Cik1-Kar3"), the latter was confusing to me. Did the authors mean that "Bim1 impacts Bik1 and Kar9 to a lesser extent", rather than vice versa? The authors are discussing Bim1 functioning via Cik1, but then switch to discussing how Bik1 and Kar9 affect Bim1.
    11. "Next, we compared the comparing genetic interaction profile of a bim1 deletion to that of various other factors by re-analyzing the synthetic genetic interaction data..." Remove "comparing".
    12. As someone who is unfamiliar with the analysis shown in Figure 3H, I think it would be useful to list a Pearson correlation value for two genes that are not functionally related. This would help define a lower limit for this analysis.
    13. The axes for the plots in Figure 5E and 5I are very confusing to me. I don't understand what I'm looking at. Why does it go from 0 to 1, and then back to 0-1 again? I don't see how this can account for MTs of different lengths. Normalizing all MT length values to 1 would do this, no?
    14. "These observations are consistent with the idea that Bik1 acts as a processivity factor for Kip2: If more Bik1 is present on the lattice, then more Kip2 molecules are able to reach plus-ends without detachment." Perhaps I'm misunderstanding the plot shown in Figure 5E, but it seems to indicate that the levels of lattice-bound Bik1 are the same in BIM1 and bim1∆ cells (higher SPB-localized levels, though). There are also lower levels of Bik1 at the plus ends in bim1∆ cells. So, if Bik1 were a processivity factor for Kip2, this would suggest that they would remain bound at plus ends as well, which these data suggest is not the case.
    15. "The data on the CH-Cik1 fusion is very compelling, and indeed supports their hypothesis that Bim1's main role in the nucleus is to target Cik1 to the spindle MT plus ends. That being said, it would be a simple, but important task to ensure that this fusion behaves as suggested (restores Cik1 plus end binding in cells). Otherwise, it can't' be ruled out that this fusion is rescuing bim1∆ functions by some other means. However, as stated above, it's unclear how much was already known about this fusion from the lab's previous work.
    16. Regarding the p1-p6 promoter data: p6 is missing from Figure S6A, in spite of it being referenced in the text and the figure.
    17. "Exogenously expressed Ase1 displayed a similar level and kinetics of localization compared to the endogenous protein, indicating that binding sites for microtubule crosslinkers are not a limiting factor on the budding yeast spindle." Specifically, the authors show that binding sites for Ase1 may not be limiting (the overlapping 95% CI bars if Fig S6B suggest this is not significant), not all crosslinkers. The authors should not use such broad language to describe results from one experiment with one crosslinker.
    18. "We found that all bim1 mutants were less well recruited to the metaphase spindle compared to the wild-type protein, indicating that Bim1-interacting proteins strongly contribute to Bim1 localization." Can the authors rule out the defects in localization of these mutants is not compromised MT binding by the Bim1 mutants? Also, regarding this statement: "To test that the observed recruitment defects of bim1 mutants are not a result of a compromised spindle or microtubule structure, we examined their localization in a situation when GFP-tagged mutants were covered with the unlabeled wild-type allele. Indeed, in this situation, the Bim1 mutants displayed very similar localization profiles (Supplementary Figure 7B)." I wasn't sure what these results were similar to: the wild-type protein, or the mutant without the presence of WT Bim1? The lack of quantitation made this difficult to determine.
    19. The zoom crops for many of the images (Fig 1F and C, 3D, 5J, etc) are not labeled. I realize the legends indicated what was what, but it would be much easier for the reader if these panels were labeled in the figure.
    20. "While in vitro reconstitution experiments have suggested that Bim1 is required to fully reconstitute the Kip2- dependent loading of the Dynein-Dynactin complex to microtubule-plus ends in vitro (Roberts et al., 2014), our experiments indicate that it may contribute relatively little to this pathway in cells." Work from other labs have also shown Bim1 is dispensable for dynein function in cells. This should be noted by the authors, and the appropriate work cited (see work from Lee and Pellman labs. In fact work from the Lee lab showed that Kip2 is dispensable for plus end binding of dynein).
    21. References are missing throughout the text. I have listed a few examples below:
      • a. "We have previously shown that the phenotype of Bim1-binding deficient Cik1 mutants can be rescued by fusing the CH-domain to this Cik1 mutant (cik1-Δ74)."
      • b. "We constructed a series of strains expressing an extra copy of Ase1-GFP under different constitutive promoters of increasing strength (p1 to p6)"; where did these promoters come from?
      • c. "double point mutation exchanging two conserved residues in the EBH domain (bim1 Y220A E228A) is predicted to eliminate all EBH-dependent cargo interactions, but does not affect protein dimerization."
      • d. "A deletion of the terminal five amino acids is predicted to prevent binding of the CAP-Gly domain of Bik1 to Bim1. The combination of both mutations is expected to simultaneously prevent both types of interaction."
      • e. "Spindle positioning in budding yeast is achieved via two pathways, one relying on the protein Kar9 which interacts with the actin-based motor Myo2." Yin et al 2000 should be added (in addition to Hwang et al).
      • f. "For nuclear migration to occur efficiently, the Dynein-Dynactin complex must be enriched at the plus-ends of cytoplasmic microtubules..." Should cite work from the Lee lab here.
      • g. "These long microtubules can interact with the bud cortex and initiate pulling events to move the nucleus (Omer et al., 2018)." Many papers pre-dating the Omer study found this to the case, including work from the Cooper lab (see Adames et al). These studies should be cited either in place of the Omer study, or in addition.

    Referees cross-commenting

    It seems that one of my major concerns is reflected in Reviewer #1's review: that a lot of the findings described in the manuscript have been published elsewhere, and are not novel. In spite of this, I do think there are useful data in this manuscript that make this an important contribution, and that it should definitely be published. However, this would first require a significant re-writing with appropriate description of known vs unknown, and additional citations.

    Significance

    The current study aims to clarify the role of Bim1 (EB1 homolog in budding yeast) in the various pathways in which it has been implicated. To achieve this aim, the authors assess the localization of numerous other microtubule-associated proteins in cells with and without Bim1. In addition to high quality localization data (e.g., intensity values), the authors perform a number of cell biological assessments (e.g., mitotic spindle length values before, during and after anaphase), genetic assessments (synthetic interaction assays), and in vitro binding assays. The current study is indeed rich with new insights into the mechanisms by which these molecules function, and will no doubt prove valuable to a number of people in the microtubule/motor/yeast mitosis fields. As someone who is interested in and studies mitosis in budding yeast, I found the study to be interesting.

  3. 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 #1

    Evidence, reproducibility and clarity

    Kornakov and Westermann provide a comprehensive analysis of the functions of microtubule +end binding proteins (+TIPs) in budding yeast. Bim1 is the EB1 ortholog of budding yeast and serves as a scaffold for other +TIPs. As Bim1 is not essential for cell viability, the authors could use a deletion of Bim1 to evaluate the global loss of function of +TIP proteins on cellular processes involving microtubules in the nucleus and the cytoplasm. During mitosis Bim1 functions almost exclusively in two previously characterized complexes with different functions: 1. The Kar9-Bim1-Bik1 complex which mediates spindle positioning at the bud neck in metaphase and 2. The Kar3-Cik1-Bik1-Bim1 complex which functions in spindle assembly and spindle elongation. Much is already known about Bim1's function in spindle positioning e.g. =TIPs on cytoplasmic microtubules, but the authors provide new insights into Kar9-Bim1 co-dependence in localization to cytoplasmic microtubules and how in turn affects localization of Bik1 and Kip2, both of which act in the dynein dependent spindle positioning pathway. The role of Bim1 in spindle assembly has also been characterized previously. The authors show how Bim1 is required to recruit the kinesin-14 Cik1/Kar3 to mitotic spindles, which interactions are involved, and that spindle elongation is delayed in its absence. Unfortunately, there is considerable overlap between their results and the published literature, and the impact of their finding is therefore reduced. The strength of this manuscript lies in its comprehensive analysis of Bim1 function , the quality of the results and that the experiments are generally well controlled and interpreted.

    Major points:

    1. There is already a huge body of published information on mitotic spindle positioning via the Kar9 and dynein pathways that grew since the late 1990s. The genetic relationships and molecular interactions between the components of these 2 pathways are well studied (many studies, including Liakopoulos et al. 2003, are not cited by the authors). The authors should make sure to cite and compare to the relevant primary literature when they report findings that have been described before. This will help to distinguish novel findings from validation of previous results.
    2. "The strict dependence of Kar9 and Cik1-Kar3 on the presence of Bim1, as well as the different effects of bim1Δ on nuclear and cytoplasmic Bik1, may reflect the formation of stable complexes between Bim1 and these binding partners in cells." I believe this has already been shown (Kumar et al., 2021 and Manatschal et al., 2016). There are several other instances as well where additional literature should be cited, for example Gardner et al., 2008 and Gardner et al. 2014.
    3. The selection of targets to study in figure 1 doesn't seem to follow the listed criteria. Many proteins included in the study were not found by IP-MS, but some perfect targets according to the listed criteria like Duo1 were not included in the study. In addition, there are more sophisticated ways of finding Bim1 binding motifs in the literature (https://doi.org/10.1016/j.cub.2012.07.047). I suggest, the authors declare that they rationally chose to study 21 proteins of interest but remove the claim that their approach was systematic.
    4. Much of the microscopy data was acquired after release from alpha factor arrest. What is the reason for this perturbation? An exponentially growing culture should mostly consist of mitotic cells anyway. Since this treatment affects cell size and potentially protein levels/concentrations, testing its influence on spindle position as well as levels on MTs for the most relevant proteins of interest would be important to exclude introduction of artifacts.
    5. Some of the results obtained from bim1Δ cells are a challenge to interpret due to the wide range of processes that involve Bim1 and therefor the potential for many off-target effects- including a global change in microtubule dynamical behavior in both the cytoplasm and the nucleus that will influence the length distributions and microtubule lifetime (and thus number). The authors must carefully consider these caveats.

    Minor points:

    1. The results section on page 12 refers to phenotypes of kar9 delete cells with respect to Bim1-GFP on cytoplasmic microtubules. In the figure 3D,F I only found data for Kar9-AID, though. The authors should refer to supplementary figure 5A or even better include quantification similar to figure 3F.
    2. The observation that cytoplasmic Bim1 localization depends on interaction with its cargo Kar9 (figure 3 + 7) fits into the model that Kumar et al (https://doi.org/10.1016/j.str.2021.06.012) proposed in which Kar9 oligomerization is required for its Bim1 dependent localization to microtubules. It would be valuable to point that out.
    3. I don't fully understand the model proposed in Figure 5H and discussion page 26. Based on figure 5E, it does not look like there is a higher concentration of Bik1 along the lattice in bim1 delete. So how would Bik1 increase Kip2 processivity if its levels are only increased due to a MT length change? If Kip2 was not fully processive, you would rather expect to see less of it at the tip of a longer microtubule in bim1 delete. The model suggested by Chen et al (https://doi.org/10.7554/eLife.48627.001) suggests that Kip2 only gets loaded at the minus-end and processively walks towards the +end without falling off. Are the authors suggesting that bim1 deletion changes this behavior?
    4. I don't see evidence for independent pools of Bik1 in the cytoplasm and nucleus as claimed on top of page 21. Total Bik1 levels on cytoplasmic microtubules seem to be well explained by their length. Please explain better or remove the statement.
    5. The experiments in supplementary figure 7B are difficult to interpret. The localization on cytoplasmic microtubules is different, but probably explained by the formation of Bim1 heterodimers. Therefore this experiment is difficult to interpret and should be removed.
    6. top of page 24: Kar9 localization in metaphase depends exclusively on SxIP, not on LxxPTPh (Manatschal 2016). The paragraph should be removed as it is not supported by published data or sufficiently by the authors to merit the conclusion.
    7. Top of page 26: The genetic interactions between the Kar9 pathway and the dynein pathway were already well known before this work. Please reformulate accordingly.
    8. page 27 second paragraph: There is no selective pressure to evolve compensation mechanisms for gene deletions. I suggest the authors consider that Kar9 and dynein partially redundant, with Kar9 acting to position the spindle prior to metaphase and dynein to maintain spindle position in the mother and bud compartments in late metaphase and anaphase. The authors should consider the quantitative analysis of Kar9 and dynein dependent spindle positioning reported in Shulist et al. 2017 and the method for analysis of spindle length and position in 3D in Meziane et al. 2021.
    9. In addition, it is not clear to me which results suggest that the relocalization of Bik1 is required in the bim1 delete. Why would wild type levels not be sufficient for dynein pathway function? The authors have not conclusively shown that nuclear migration relies on upregulating the dynein pathway in bim1Δ cells. If there is no supporting data, the paragraph should be removed.
    10. Please provide more details about intensity quantification on page 35. Were these measured on sum or max projected stacks? What was the method of background subtraction?
    11. Are the spindle lengths in Figure 2E measured in 2D or 3D? Bim1 deletion might lead to more misalignment of the spindles in z due to inactivation of the Kar9 pathway and thus partially explain the shorter spindles. The measurements should therefore be performed in 3D.
    12. The authors should try to shorten the text. There is a lot of redundancy between results and discussion sections.
    13. Data is shown that leads to conclusions that are already supported by the literature should be moved to the supplementary material.

    Referees cross-commenting

    I am in agreement with reviewer 2

    Significance

    The role of Bim1 in the Kar9 spindle positioning pathway and in recruiting Kar3/Cik1 to spindles have been extensively characterized in previous publications, however this manuscript adds mechanistic insight into what interactions are essential for localization, what happens to other proteins that have not previously been studied in the context of Bim1 and what are the exact consequences of Bim1 loss with some explanation for the outcomes. Some data presented here was expected from previous work, but never experimentally confirmed and these findings should be the focus of the manuscript. While the manuscript does not provide a huge conceptual advancement, the findings of this comprehensive analysis are of great value to the microtubule field, especially for people working in budding yeast.