Centrosome age breaks spindle size symmetry even in “symmetrically” dividing cells

  1. Alexandre Thomas
  2. Patrick Meraldi

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Centrosomes are the main microtubule organizing center in animal cells. Due to the semi-conservative nature of centrosome duplication, the two centrosomes differ in age. In asymmetric stem cell divisions, centrosome age can induce an asymmetry in half-spindle lengths. However, whether centrosome age affects the symmetry of the two half-spindles in tissue culture cells thought to divide symmetrically, is unknown. Here, we show that in human epithelial and fibroblastic cell lines centrosome age imposes a subtle spindle asymmetry that leads to asymmetric cell daughter sizes. At the mechanistic level, we show that this asymmetry depends on the preferential accumulation on old centrosomes of the microtubule nucleation-organizing proteins pericentrin, γ-tubulin, Cdk5Rap2, and TPX2, under the control of a cenexin-bound pool of the mitotic kinase Plk1. Moreover, we find that old centrosomes have a higher microtubule nucleation capacity. We therefore postulate that centrosome age breaks spindle size symmetry via microtubule nucleation even in cells thought to divide symmetrically.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    1) List of the detailed experiments we plan to perform (including aforementioned experiments):

    • Careful analysis of the daughter cell size by measuring the real volume.

    • Quantifications of PCM (pericentrin and γ-tubulin) proteins and Plk1 with respect to centrosome age in G2 and metaphase (for Plk1) cells.

    • Analysis of the amount of Plk1 of metaphase cells when cenexin protein is absent (siControl vs siCenexin), and measurements of Plk1 in WT-cenexin vs. cenexinS796A mutant to test if Cenexin controls a subpool of Plk1 at centrosomes.

    • Careful analysis of Ctrl and TPX2 depletion experiment data in 1:1 cells. We plan to repeat the experiment to confirm or infirm on the contribution of TPX2 in spindle asymmetry.

    • Measurement of the PCM volume/intensity in 2:2 and 1:1 metaphase cells, to highlight on the contribution of the daughter centrioles in recruiting PCM proteins.

    • Live cell imaging of 2:2 cells and measurements of different parameters; cortex-to-centrosome and spindle pole to metaphase plate (half-spindle (a)symmetry) distances.

    • Long-term live cell imaging of 2:2 cells to investigate whether the asymmetry in centrosome-age dependent daughter cell size also affects the duration of the ensuing cell cycle. While we have carried out such long-term movies in the past, we are aware that they can be challenging due to high cell mobility over longer time courses.

    • Investigation of the microtubule nucleation capacity under different conditions of PCM protein depletion (depletion of Cdk5rap2 and/or pericentrin).

    • Analysis of the effect of the over-expression of PCM protein (Cdk5Rap2) on the (a)symmetry of the mitotic spindle size

    2) detailed answers (in green) to the reviewers’ comments:

    __Reviewer #1: __

    __(Major points) __

    The discovery of differences in half-spindle size during symmetric division is intriguing. However, the methodology for quantification of the data remains unclear. Key questions, such as how the center of the metaphase plate is determined from the image data, the definition of exact pole position when centrioles are located at spindle poles, the objective determination of daughter cell diameter and width from the image data, and the referential position of the cortex, need more detailed explanation in the manuscript. Additionally, it's crucial to elucidate the specific index used to quantify differences from the image data, especially when dealing with data that only varies by a few percent. Providing clarity on these aspects and, in some cases, re-quantifying the data should be necessary.

    We have already included clearer explanations in the method parts and results part about our methodology and will include a supplementary figure on how precisely we defined and measured the half-spindle sizes, as well as the index used for the asymmetry (using a methodology that we previously used in Dudka et al., Nature Comm., 2018). In addition, we will use a second method to measure the real daughter cell volume.

    The mechanism behind the difference in half-spindle size, related to the subdistal appendage (SDA), raises questions, especially considering that SDA is believed to disassemble during mitosis. Exploring whether differences in the localization of PCM components and half-spindle size result from disparities in Plk1 and PCM loading during G2/early mitosis, prior to SDA disassembly, necessitates experimental verification.

    As suggested by the reviewer we will quantify the amounts of PCM proteins on the old and young centrosome in G2 cells (and therefore prior SDA reorganization). This will also allow us to test whether the asymmetry depends on the SDA themselves, or the corresponding SDA proteins, which still accumulate specifically on the oldest centrosomes during mitosis

    For investigating the mechanism of half-spindle size asymmetry, many perturbation experiments employ knock-down techniques. To directly address the cause of asymmetry, it might be valuable to artificially localize Plk1 and PCM factors to one spindle pole using optogenetic tools or similar approaches and then quantify half-spindle and daughter cell sizes.

    We thank the reviewers for this suggestion, as it could indeed, be of great interest and provide a direct proof of principle. Unfortunately, based on our experience in establishing such a cell line we know that just the generation of such a light-manipulated stable cell line that contains markers for centrosomes and chromosomes or kinetochores takes 6-9 months, in the best-case scenario. This experiment is therefore not possible within a normal revision round (even if extended to 6 months).

    The asymmetry in Plk1 sub-population recruitment by SDA triggers the observed effects, but the evidence for this is relatively weak, given the small difference in spindle asymmetry. Quantifying the amount of Plk1 in its activated form, particularly in the context of SDA dismantling during metaphase, could strengthen this aspect of the study.

    While the commercial antibodies against the activated form of Plk1 (phospho-T210) work very well by immunoblotting, we have not been able to get it to work by immunofluorescence. We will nevertheless, test whether variation in the fixation methods can solve this issue. Alternatively, we will test to which extend depletion of Cenexin, or the presence of Cenexin WT vs the non-phosphorylatable Cenexin mutant affects the overall population of Plk1 on both spindle poles.

    While the focus on half-spindle size asymmetry during symmetric division is intriguing, it's important to address the broader physiological significance. The primary outcome of this asymmetry is differences in daughter cell size, which limits the broader significance of the study. Furthermore, the quantification method for daughter cell size warrants scrutiny and clarification.

    As mentioned above, we will use different method to measure and investigate daughter cell size (a)symmetry. Moreover, we will attempt with long-term live cell movies to test whether the variation in centrosome-age dependent daughter cell size also affects the duration of the ensuing cell cycle.

    (Minor points)

    Table 1 lists factors with asymmetric localization not analyzed in detail in this paper. It would be beneficial to discuss whether these factors play a role in spindle asymmetry, and the authors should address the completeness of the data in Table 1 in terms of selecting factors for analysis.

    We agree with this comment that other factors may participate in the regulation of spindle asymmetry. However, we performed this screening to identify key drivers of spindle (a)symmetry based on an investigation of the Pearson’s correlation coefficient and the value of slope.

    In addition, some of these proteins are known to control spindle size in acting in a same pathway (TPX2/Kif2A/Katanin) and (Pericentrin/CDK5RAP2/ϒ-tubulin). We will incorporate these points and the reasons for our selection in the discussion

    In Figure 1H, the impact of centriolin knock-out on the distribution of unaligned polar chromosomes is different from the effect of cenexin S796A in Figure 6H. This difference should be explained to provide clarity on the observed discrepancies.

    We will better explain this difference.

    In Figure 2A, there is no correlation data presented between daughter cell asymmetry and the presence or absence of cenexin signal. This relationship should be elucidated for a more comprehensive understanding.

    We will clarify this point. Specifically, we plotted the daughter cell symmetry index for 2:2 and 1:1 cells with respect to centrosome age. All the daughter cells display the presence of a cenexin signal at both grandmother and mother centrioles with a difference in fluorescence intensity that enables us to assign them to “old” vs “young centrosomes. We found a significant result indicating that there is a relationship between centrosome age and the formation of daughter cell with different sizes.

    In Figure 4G and H, the mean value of spindle asymmetry increases with siRNA treatment of Cdk5Rap2 or PCNT compared to the control. The possible interpretation of this finding should be discussed.

    This is an interesting observation that needs to be discussed in our revision.

    Figure 4K shows that the asymmetry of PCNT distribution is not eliminated by centriolin knock-down. This observation requires clarification and discussion.

    It has been shown that pericentrin is directly recruited by Plk1 at centriole (Soung et al., 2009). In addition, pericentrin has a PACT-domain that directly targets pericentrin to the centriole (Gillingham and Munro., 2000). Moreover, it has been demonstrated that the grandmother centriole is slightly longer than the mother one (Kong et al., 2020). Altogether, this suggests that the old and young centrosomes, based on this intrinsic property, may recruit different amount of pericentrin.

    We will add this explanation in the discussion.

    It appears that the difference in spindle asymmetry of the control group in Figure 5A is smaller than in other data. This discrepancy should be addressed. Additionally, the influence of TPX2 depletion on spindle formation, and any corresponding spindle staining data, should be included.

    This point will be discussed in the revised version of the manuscript.

    Claiming that the daughter centriole recruits PCM based on Figure 6A data alone may require additional supporting evidence. It is essential to investigate whether there is a clear PCM signal when the daughter centriole disengages in late mitosis and maintain consistency in the interpretation.

    As suggested by the reviewer 2, we will measure PCM volume/intensity in both 2:2 and 1:1 cells to demonstrate that daughter centrioles directly recruit PCM proteins.

    The lack of difference in TPX2 distribution in Figure 7E should be explained, along with a discussion of how this observation aligns with the spindle asymmetry data and any inconsistencies.

    We will discuss this point in the revised manuscript.

    The differing N numbers between samples in all the figures may affect the validity of comparisons. The authors should discuss whether it is necessary to have consistent N numbers in each experiment for more robust conclusions.

    Indeed, this is an important point that must be discussed.

    Reviewer #2____:

    Major comments:

    1. It is not completely clear how the authors determined whether a spindle was asymmetric or not. In the methods, they say that statistical tests are described in the legends. In Figure 1 legend they say: "Each condition was compared to a theoretical distribution centered at 0 (dashed line)". How did they generate this theoretical distribution?

    As explained under point 1 of reviewer 1, we will provide a more thorough explanation of our methodology and how we decide whether a spindle is symmetric or not. In brief, a perfectly symmetric spindle would yield an asymmetry index of 0, as there is no difference between the two half-spindle sizes.

    1. The authors claim that TPX2 depletion results in loss of spindle asymmetry in 1:1 cells, but the difference is very small (1.7% in control vs 1.3% in TPX2 depletion, Fig 5B) and the data is more variable in TPX2 depletion, which makes it less likely that a statistically significant difference from 0 would be found. Firstly, perhaps the authors could check the standard error of the mean, which provides a measure of how accurate the mean is with regard to N and variation. If a dataset is more spread (such as in TPX2 depletion) a higher N is required to attain the same accuracy in the mean value. This is normally not so important when directly comparing two datasets, but in this case the authors are comparing each dataset to 0. So, are the authors measuring enough cells in the TPX2 depletion to be sure that a 1.3% value is not significantly different from 0? Secondly, I don't understand why the control cells have such a low asymmetry index (1.7%), when previous data in the paper shows an asymmetry index of 4.1% (Fig 1D) and 3.4% (Fig 4E) in control 1:1 cells. This suggests that something about the way this experiment was carried out dampens the asymmetry, which could therefore lead the authors to conclude that TPX2 is more important than it really is.

    We agree with this comment, the mean of the control condition is smaller compared to others controls. As mentioned above, we will carefully look at the data (SD vs SEM) and in case add a new replicate to confirm or infirm the involvement of TPX2 in the formation of asymmetric spindles.

    1. The authors claim that daughter centrioles are associated with some Pericentrin and suggest that this may be why 2:2 centrosomes have less of an asymmetry than 1:1 centrosomes (Fig 6A). It is unclear whether the authors consider these daughter centrioles as being prematurely disengaged (they make reference to the fact that they previously showed how disengaged daughters recruit γ-tubulin, but it's unclear if this is related to their current observations). In Figure 6A, the Centrin spots look too far apart for engaged centrioles (~750nm). I appreciate that this may be the only way to dectect Pericentrin around the daughter at this resolution, but it may also force the authors to select cells where the centrioles have prematurely disengaged. For the asymmetry measurements, the authors presumably did not select cells where they could distinguish mother and daughter centrioles. One way to address this issue would be to compare PCM size at centrosomes in 2:2 cells with centrosomes in 1:1 cells. The expectation would be that centrosomes in 2:2 cells would have more PCM, due to the contribution of the daughter centrioles.

    We agree that on those high-resolution images the daughter centrioles seem to be far from the mother ones. The metaphase cells presented in this figure, are wild-type non-treated cells for which the daughter centrioles are engaged. Indeed, our own investigation of the centriole engagement status by expansion microscopy, indicates that over 98% of centriole pairs in metaphase RPE1 cells are engaged.

    Nevertheless, as suggested by the reviewer and to validate that daughter centrioles participate in this process, we will compare PCM size in 2:2 and 1:1 metaphase cells.

    1. The authors show that Plk1 recruitment by Cenexin (via S796 phosphorylation), which happens only at mother centrosomes, is important for asymmetry. Nevertheless, they show that Plk1 is symmetrically distributed between mother and daughter centrosomes (Table 1). This does not really fit, unless daughter centrosomes recruit more cenexin-independent Plk1 than mother centrosomes or if the cenexin-bound pool of Plk1 is only a minor fraction of total Plk1. If so, do the authors think that the Cenexin-bound pool of Plk1 is more potent than the rest of centrosomal Plk1?

    As indicated in point 4 of reviewer 1 we will test which proportion of the Plk1 pool at spindle poles depends on the presence of Cenexin, as we suspect that this Plk1 population is only a subpopulation.

    1. The circles drawn to measure cell size in Figures 2A,E and 7C do not look like a good representation of cell area (as the cells are not perfectly round). The authors use a formular for circle area with an approximation of the radius (based on mean length/width of an oval. It would be much better to use ImageJ to draw a freehand line around the perimeter of the cell and use the in-built tool to measure the area.

    As mentioned in point 1 of reviewer 1 we will use another method to measure daughter cell size.

    Minor comments:

    1. Asymmetry in centrosome size that correlates with centrosome age in apparently symmetrically dividing "cells" has been observed previously in Drosophila syncytial embryos (Conduit et al., 2010a, Curr. Bio.). I think this should be mentioned somewhere given the topic of the study.

    We thank the reviewer for this information. This paper will be discussed in the revised version.

    1. A full description of statistical tests and n numbers for each experiment should be provided in the methods, even if this duplicates information in the Figure legends.

    We will add this information in the method.

    OPTIONAL EXPERIMENTS:

    1. Given that chTOG is very important for microtubule nucleation, it seems strange that this protein was not analysed for a potential asymmetry.

    As suggested by the reviewers we will test for a potential chTOG asymmetry and its impact on spindle size asymmetry.

    1. Cooling-warming experiments could be done using higher concentration of formaldehyde, as it's likely that microtubule nucleation is not immediately halted when using 4% formaldehyde.

    The fixation solution was chilled at 4°C, which should halt any further depolymerization. We will specify this point in the Material and Methods section.

    Reviewer ____#____3:

    Major points:

    1. The evaluation of spindle and cell size asymmetry related to centrosome age only relies on fixed sample preparation. Cells should be followed by time-lapse microscopy as the metaphase plate position relative to the spindle poles and/or the cell cortex may fluctuate over time and as the observed differences remain in a very subtle range. This is an important possibility to consider for 1:1, 1:0 or 0:0 spindle pole configurations where centrosome integrity is impaired.

    We agree with the reviewer that this is a drawback of our approach, but the experiments the reviewer suggests is not possible for 1:0 or 0:0 or only in an approximate manner. Indeed, we do not have a centriole-independent spindle pole marker that would allow us to mark precisely the position of the spindle pole. In the past we used Sir-tubulin, which gave us an approximate position of the spindle poles, and which allowed to us monitor the spindle asymmetry over time of 1:0 cells (see Dudka et al., 2019), a point that we will discuss. Nevertheless, as suggested by the reviewer we will attempt to monitor these asymmetries in 2:2 and/or 1:1 cells expressing GFP-Centrin1 and GFP-CENPA (kinetochore marker) in WT conditions. Indeed, we cannot expand this approach to all the conditions, as the calculation of the spindle asymmetry index is based on a very high number of cells, and the monitoring of spindle asymmetry can only be achieved by selecting mitotic cells one-by-one and then monitoring them over a short period of them (Tan et al., eLife, 2015), which makes such an approach extremely time-consuming.

    1. Cell size asymmetry was evaluated based on cell area at the equator. Volumes will be a better indicator as daughter cell shapes can be different in telophase if they do not re-adhere at the same speed. This evaluation should also be confirmed with another readout, like the position of the cleavage furrow relative to the spindle poles in late anaphase, as again the observed differences are in a very subtle range.

    As indicated in the similar points of reviewer 1 and 2, we will improve our methodology to take this comment in account

    1. The authors propose that differential microtubule nucleation at the spindle poles underlies spindle size symmetry breaking without providing direct evidence. If the observed spindle symmetry in the 1:1 configuration after pericentrin, CDK5RAP2 or g-tubulin siRNA fuels this interpretation (Fig4C), the differential microtubule nucleation capacity at the spindle poles after microtubule-depolymerisation-repolymerisation assays was not evaluated in these conditions, as compared to the control situation.

    As suggested by the reviewer we will analyze the microtubule nucleation capacity after the downregulation of PCM proteins.

    1. If differential microtubule nucleation at the spindle poles is responsible for spindle asymmetry, overexpression of PCM proteins or g-tubulin should be sufficient for re-establishment of symmetric protein distribution, spindle and cell size symmetry in 2:2 or 1:1 configuration. The authors should evaluate whether this is the case or not.

    This is an interesting suggestion, which we will test, although overexpression of these proteins might also lead to other defects in the spindle, such as multipolar spindles.

    1. The authors describe that the cortex-centrosome distance is not changed according to centrosome age (Fig2C), but centrosome-metaphase plate distance is (Fig1D). These observations are difficult to reconcile if differential microtubule-nucleation capacity is at play. Again, time-lapse microscopy would enable to detect over time whether only metaphase plate position relative to spindle poles is changing or if spindle pole position relative to the cell cortex is also fluctuating.

    We plan to give a try to image WT 2:2 cells by time lapse microscopy and to measure several parameters such as half-spindle size, spindle (a)symmetry and the cortex to centrosome distance over time.

    Minor points:

    1. Main PCM and MT nucleation protein "depletion" do not appear to impact spindle assembly, but only spindle symmetry in 1:1 and 1:0 configurations (Fig4A and 4F-H). Can it be explained by the fact that their depletion is not always total (for pericentrin, Fig5F versus FigS2A or Fig7G)? Can they comment on this point?

    Spindles displaying abnormal centriole number at spindle poles (1:1 and 1:0) can still assemble bipolar spindle in absence of the main PCM proteins (Chinen et al., JCB, 2021, and Watanabe et al., JCB, 2020).

    In our study, the depletion of PCM protein is almost total (97% for pericentrin, 98% for Cdk5Rap2).

    1. If centrosome age dictates spindle and cell size asymmetry through differential MT-nucleation capacity at the spindle poles, how can this process be modulated? Indeed, centrosome age is common to all cell types, but cell size asymmetry is more or less pronounced. The authors should further discuss this point based on the literature.

    We will discuss this point in the discussion.

    __ Description of the revisions that we have already carried out in the revised manuscript__

    The discovery of differences in half-spindle size during symmetric division is intriguing. However, the methodology for quantification of the data remains unclear. Key questions, such as how the center of the metaphase plate is determined from the image data, the definition of exact pole position when centrioles are located at spindle poles, the objective determination of daughter cell diameter and width from the image data, and the referential position of the cortex, need more detailed explanation in the manuscript. Additionally, it's crucial to elucidate the specific index used to quantify differences from the image data, especially when dealing with data that only varies by a few percent. Providing clarity on these aspects and, in some cases, re-quantifying the data should be necessary.

    We have already included clearer explanations in the method parts and results part about our methodology and will include a supplementary figure on how precisely we defined and measured the half-spindle sizes, as well as the index used for the asymmetry (using a methodology that we previously used in Dudka et al., Nature Comm., 2018). In addition, we will use a second method to measure the real daughter cell volume.

    __ Description of the experiments that we prefer not to carry out:__

    Point 3 of reviewer 1 : For investigating the mechanism of half-spindle size asymmetry, many perturbation experiments employ knock-down techniques. To directly address the cause of asymmetry, it might be valuable to artificially localize Plk1 and PCM factors to one spindle pole using optogenetic tools or similar approaches and then quantify half-spindle and daughter cell sizes.

    We thank the reviewers for this suggestion, as it could indeed, be of great interest and provide a direct proof of principle. Unfortunately, based on our experience in establishing such a cell line we know that just the generation of such a light-manipulated stable cell line that contains markers for centrosomes and chromosomes or kinetochores takes 6-9 months, in the best-case scenario. This experiment is therefore not possible within a normal revision round (even if extended to 6 months).

  2. Review Commons

    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 #3

    Evidence, reproducibility and clarity

    The manuscript entitled "centrosome age breaks spindle size symmetry even in "symmetrically" dividing cells" by Thomas and Meraldi reports that centrosome age impacts microtubule-nucleation capacity and is sufficient to tune spindle symmetry and cell size in human culture cell lines. The manuscript is overall clear, well written, illustrated and discussed. Nonetheless, some key experiments are missing as the authors report very subtle differences that need to be confirmed with complementary experiments, including time-lapse microscopy and alternative evaluations of cell sizes. The mechanism by which spindle symmetry breaking is established by centrosome age is not clear, even if the authors have identified some important actors at the spindle poles.

    Major points:

    1. The evaluation of spindle and cell size asymmetry related to centrosome age only relies on fixed sample preparation. Cells should be followed by time-lapse microscopy as the metaphase plate position relative to the spindle poles and/or the cell cortex may fluctuate over time and as the observed differences remain in a very subtle range. This is an important possibility to consider for 1:1, 1:0 or 0:0 spindle pole configurations where centrosome integrity is impaired.
    2. Cell size asymmetry was evaluated based on cell area at the equator. Volumes will be a better indicator as daughter cell shapes can be different in telophase if they do not re-adhere at the same speed. This evaluation should also be confirmed with another readout, like the position of the cleavage furrow relative to the spindle poles in late anaphase, as again the observed differences are in a very subtle range.
    3. The authors propose that differential microtubule nucleation at the spindle poles underlies spindle size symmetry breaking without providing direct evidence. If the observed spindle symmetry in the 1:1 configuration after pericentrin, CDK5RAP2 or -tubulin siRNA fuels this interpretation (Fig4C), the differential microtubule nucleation capacity at the spindle poles after microtubule-depolymerisation-repolymerisation assays was not evaluated in these conditions, as compared to the control situation.
    4. If differential microtubule nucleation at the spindle poles is responsible for spindle asymmetry, overexpression of PCM proteins or -tubulin should be sufficient for re-establishment of symmetric protein distribution, spindle and cell size symmetry in 2:2 or 1:1 configuration. The authors should evaluate whether this is the case or not.
    5. The authors describe that the cortex-centrosome distance is not changed according to centrosome age (Fig2C), but centrosome-metaphase plate distance is (Fig1D). These observations are difficult to reconcile if differential microtubule-nucleation capacity is at play. Again, time-lapse microscopy would enable to detect over time whether only metaphase plate position relative to spindle poles is changing or if spindle pole position relative to the cell cortex is also fluctuating.

    Minor points:

    1. Main PCM and MT nucleation protein "depletion" do not appear to impact spindle assembly, but only spindle symmetry in 1:1 and 1:0 configurations (Fig4A and 4F-H). Can it be explained by the fact that their depletion is not always total (for pericentrin, Fig5F versus FigS2A or Fig7G)? Can they comment on this point?
    2. If centrosome age dictates spindle and cell size asymmetry through differential MT-nucleation capacity at the spindle poles, how can this process be modulated? Indeed, centrosome age is common to all cell types, but cell size asymmetry is more or less pronounced. The authors should further discuss this point based on the literature.

    Significance

    The question of whether centrosome age is translated into different capacity to nucleate microtubules and related consequences on spindle and cell size symmetry has already been addressed in different model systems. Nonetheless, cell lines were previously described as dividing symmetrically since their spindle is symmetric in size and since they give rise to daughter cells of equivalent sizes. The present manuscript reports a thorough re-evaluation of this question and provides evidence that subtle differences in PCM and spindle pole protein recruitment, microtubule-nucleation capacity and spindle symmetry can be observed as a function of centrosome age. They also identify some key actors whose differential recruitment at the spindle poles can underlie spindle symmetry breaking, even if their involvement seems to differ from one cell line to another one. This manuscript could be submitted after appropriate revisions as a report and will benefit to the basic research cell biology community.

  3. Review Commons

    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

    Summary:

    In this study, the authors show that in two types of human tissue culture cells the half spindles associated with mother centrosomes are slightly longer, on average, than the half spindles associated with daughter centrioles. They show that this correlates with centrosome age, with mother centrosomes tending to be associated with the longer half spindle, and with a correlative asymmetry in the size of daughter cells. They show that spindle asymmetry relates to asymmetries in the amount of certain PCM components at centrosomes, including Pericentrin, CDK5RAP2, TPX2, and γ-tubulin, which preferentially accumulate at mother centrosomes. Pericentrin/CDK5RAP2/TPX2/γ-tubulin are known to be involved, directly or indirectly, with microtubule nucleation, and the authors also show how that microtubule nucleation is more robust at mother centrosomes and that depletion of either Pericentrin, CDK5RAP2, TPX2, or γ-tubulin abolishes (or reduces) spindle asymmetry. The suggestion is that enhanced microtubule nucleation at the mother centrosome leads to longer half spindles and subsequent asymmetric positioning of the division plane and daughter cells of unequal size. Centrosome and spindle asymmetry is partially masked by the apparent equal accumulation of PCM at daughter centrioles, such that cells with centrosomes containing only mother centrioles show higher levels of asymmetry.

    Mechanistically, the authors show that a Cenexin-bound pool of Plk1, a kinase required for PCM assembly, is important for centrosome and spindle asymmetry. Cenexin is an "upstream" sub-distal appendage protein only found at mother centrosomes (due to appendage structures only being present on the grandmother centriole). Nevertheless, depletion of a more downstream sub-distal appendage protein, Centriolin, also abrogated spindle asymmetry, suggesting that multiple proteins of the sub-distal appendages are necessary for asymmetry. Results from some experiments show that spindle asymmetry and a known asymmetry in the distribution of polar centrosomes are mechanistically separable, while other experiments show a link.

    Major comments:

    1. It is not completely clear how the authors determined whether a spindle was asymmetric or not. In the methods, they say that statistical tests are described in the legends. In Figure 1 legend they say: "Each condition was compared to a theoretical distribution centered at 0 (dashed line)". How did they generate this theoretical distribution?
    2. The authors claim that TPX2 depletion results in loss of spindle asymmetry in 1:1 cells, but the difference is very small (1.7% in control vs 1.3% in TPX2 depletion, Fig 5B) and the data is more variable in TPX2 depletion, which makes it less likely that a statistically significant difference from 0 would be found. Firstly, perhaps the authors could check the standard error of the mean, which provides a measure of how accurate the mean is with regard to N and variation. If a dataset is more spread (such as in TPX2 depletion) a higher N is required to attain the same accuracy in the mean value. This is normally not so important when directly comparing two datasets, but in this case the authors are comparing each dataset to 0. So, are the authors measuring enough cells in the TPX2 depletion to be sure that a 1.3% value is not significantly different from 0? Secondly, I don't understand why the control cells have such a low asymmetry index (1.7%), when previous data in the paper shows an asymmetry index of 4.1% (Fig 1D) and 3.4% (Fig 4E) in control 1:1 cells. This suggests that something about the way this experiment was carried out dampens the asymmetry, which could therefore lead the authors to conclude that TPX2 is more important than it really is.
    3. The authors claim that daughter centrioles are associated with some Pericentrin and suggest that this may be why 2:2 centrosomes have less of an asymmetry than 1:1 centrosomes (Fig 6A). It is unclear whether the authors consider these daughter centrioles as being prematurely disengaged (they make reference to the fact that they previously showed how disengaged daughters recruit γ-tubulin, but it's unclear if this is related to their current observations). In Figure 6A, the Centrin spots look too far apart for engaged centrioles (~750nm). I appreciate that this may be the only way to dectect Pericentrin around the daughter at this resolution, but it may also force the authors to select cells where the centrioles have prematurely disengaged. For the asymmetry measurements, the authors presumably did not select cells where they could distinguish mother and daughter centrioles. One way to address this issue would be to compare PCM size at centrosomes in 2:2 cells with centrosomes in 1:1 cells. The expectation would be that centrosomes in 2:2 cells would have more PCM, due to the contribution of the daughter centrioles.
    4. The authors show that Plk1 recruitment by Cenexin (via S796 phosphorylation), which happens only at mother centrosomes, is important for asymmetry. Nevertheless, they show that Plk1 is symmetrically distributed between mother and daughter centrosomes (Table 1). This does not really fit, unless daughter centrosomes recruit more cenexin-independent Plk1 than mother centrosomes or if the cenexin-bound pool of Plk1 is only a minor fraction of total Plk1. If so, do the authors think that the Cenexin-bound pool of Plk1 is more potent than the rest of centrosomal Plk1?
    5. The circles drawn to measure cell size in Figures 2A,E and 7C do not look like a good representation of cell area (as the cells are not perfectly round). The authors use a formular for circle area with an approximation of the radius (based on mean length/width of an oval. It would be much better to use ImageJ to draw a freehand line around the perimeter of the cell and use the in-built tool to measure the area.

    Minor comments:

    1. Asymmetry in centrosome size that correlates with centrosome age in apparently symmetrically dividing "cells" has been observed previously in Drosophila syncytial embryos (Conduit et al., 2010a, Curr. Bio.). I think this should be mentioned somewhere given the topic of the study.
    2. A full description of statistical tests and n numbers for each experiment should be provided in the methods, even if this duplicates information in the Figure legends.

    OPTIONAL EXPERIMENTS:

    1. Given that chTOG is very important for microtubule nucleation, it seems strange that this protein was not analysed for a potential asymmetry.
    2. Cooling-warming experiments could be done using higher concentration of formaldehyde, as it's likely that microtubule nucleation is not immediately halted when using 4% formaldehyde.

    Significance

    This a well-conducted study with results being presented clearly and concisely. The methodology is solid in the main. The study reveals something unexpected - that apparently symmetrically dividing human tissue culture cells divide asymmetrically. While the asymmetry is only slight, it could be important - although the authors do not address its relevance for the cell population. Having analysed only 2 cultured cell types, it remains unclear if this is a widespread phenomenon, and whether this occurs in a more natural setting. Nevertheless, the proposed model (Plk1 at SDA's => increased PCM at mother => increased nucleation => offset division plane), which is supported by the data, would suggest this could be a widespread phenomenon. This study will be of interest to anyone studying cell division, but it would require some degree of insight into the importance of the observations for it to appeal to a very broad audience.

    I am a cell biologist with an interest in cell division and microtubule regulation.

  4. Review Commons

    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

    In this paper, the authors demonstrated that there is asymmetry in mitotic spindles, which are usually considered symmetric. That is, they found that centrosome age causes asymmetry in the size of the half-spindle even when the number of centrioles forming the spindle poles is the usual pair or when there is only one mother centriole. It is also suggested that the difference in half-spindle size is due to the different microtubule-organizing activity of the centrosomes at each spindle pole. Furthermore, they observe that the difference in half-spindle size also results in asymmetries in the size of the daughter cells after cell division. In this study, they mainly analyze the mechanism by employing the condition of 1:1 number of centrioles, in which the difference in half-spindle size is more sharply pronounced. They showed that the subdistal appendage (SDA) of the centriole of the old centrosome is important for the molecular basis of this half-spindle size difference, as the SDA-dependent recruitment of the Plk1 pool mediates an asymmetric localization of pericentrin, Cdk5rap2, gamma-tubulin, TPX2, and other factors at the spindle poles. In addition, knockdown of these factors eliminated the half-spindle size asymmetry. They also confirm these findings using a different human cell line, BJ cells. In conclusion, they propose that, reflecting centrosome age, the old centrosome promotes asymmetric spindle formation by localizing a group of factors that promote microtubule organization, originating from the SDA-Plk1 pathway.

    Major points

    1. The discovery of differences in half-spindle size during symmetric division is intriguing. However, the methodology for quantification of the data remains unclear. Key questions, such as how the center of the metaphase plate is determined from the image data, the definition of exact pole position when centrioles are located at spindle poles, the objective determination of daughter cell diameter and width from the image data, and the referential position of the cortex, need more detailed explanation in the manuscript. Additionally, it's crucial to elucidate the specific index used to quantify differences from the image data, especially when dealing with data that only varies by a few percent. Providing clarity on these aspects and, in some cases, re-quantifying the data should be necessary.
    2. The mechanism behind the difference in half-spindle size, related to the subdistal appendage (SDA), raises questions, especially considering that SDA is believed to disassemble during mitosis. Exploring whether differences in the localization of PCM components and half-spindle size result from disparities in Plk1 and PCM loading during G2/early mitosis, prior to SDA disassembly, necessitates experimental verification.
    3. For investigating the mechanism of half-spindle size asymmetry, many perturbation experiments employ knock-down techniques. To directly address the cause of asymmetry, it might be valuable to artificially localize Plk1 and PCM factors to one spindle pole using optogenetic tools or similar approaches and then quantify half-spindle and daughter cell sizes.
    4. The asymmetry in Plk1 sub-population recruitment by SDA triggers the observed effects, but the evidence for this is relatively weak, given the small difference in spindle asymmetry. Quantifying the amount of Plk1 in its activated form, particularly in the context of SDA dismantling during metaphase, could strengthen this aspect of the study.
    5. While the focus on half-spindle size asymmetry during symmetric division is intriguing, it's important to address the broader physiological significance. The primary outcome of this asymmetry is differences in daughter cell size, which limits the broader significance of the study. Furthermore, the quantification method for daughter cell size warrants scrutiny and clarification.

    Minor points

    1. Table 1 lists factors with asymmetric localization not analyzed in detail in this paper. It would be beneficial to discuss whether these factors play a role in spindle asymmetry, and the authors should address the completeness of the data in Table 1 in terms of selecting factors for analysis.
    2. In Figure 1H, the impact of centriolin knock-out on the distribution of unaligned polar chromosomes is different from the effect of cenexin S796A in Figure 6H. This difference should be explained to provide clarity on the observed discrepancies.
    3. In Figure 2A, there is no correlation data presented between daughter cell asymmetry and the presence or absence of cenexin signal. This relationship should be elucidated for a more comprehensive understanding.
    4. In Figure 4G and H, the mean value of spindle asymmetry increases with siRNA treatment of Cdk5Rap2 or PCNT compared to the control. The possible interpretation of this finding should be discussed.
    5. Figure 4K shows that the asymmetry of PCNT distribution is not eliminated by centriolin knock-down. This observation requires clarification and discussion.
    6. It appears that the difference in spindle asymmetry of the control group in Figure 5A is smaller than in other data. This discrepancy should be addressed. Additionally, the influence of TPX2 depletion on spindle formation, and any corresponding spindle staining data, should be included.
    7. Claiming that the daughter centriole recruits PCM based on Figure 6A data alone may require additional supporting evidence. It is essential to investigate whether there is a clear PCM signal when the daughter centriole disengages in late mitosis and maintain consistency in the interpretation.
    8. The lack of difference in TPX2 distribution in Figure 7E should be explained, along with a discussion of how this observation aligns with the spindle asymmetry data and any inconsistencies.
    9. The differing N numbers between samples in all the figures may affect the validity of comparisons. The authors should discuss whether it is necessary to have consistent N numbers in each experiment for more robust conclusions.

    Significance

    In summary, while the study is intriguing for its exploration of spindle asymmetry during symmetric division, the major points raised here highlight areas where further clarification and data interpretation are needed. A more rigorous quantification method and additional evidence to support the proposed SDA-Plk1 signal as the initiator of asymmetry would enhance the study's validity. Moreover, addressing concerns about daughter cell size quantification and the physiological relevance of spindle asymmetry is essential for a more comprehensive understanding of the findings. This research presents an interesting challenge for researchers in the centrosome and mitotic spindle field.

  5. Version published to 10.1101/2023.09.15.557935v1 on bioRxiv