Epsilon tubulin is an essential determinant of microtubule-based structures in male germ cells

  1. G. Gemma Stathatos
  2. D. Jo Merriner
  3. Anne E. O’Connor
  4. Jennifer Zenker
  5. Jessica E.M. Dunleavy
  6. Moira K. O’Bryan

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Alpha, beta and gamma tubulins are essential building blocks for all eukaryotic cells. The functions of the non-canonical tubulins, delta, epsilon and zeta, however, remain poorly understood and their requirement in mammalian development untested. Herein we have used a spermatogenesis model to define epsilon tubulin (TUBE1) function in mice. We show that TUBE1 is essential for the function of multiple complex microtubule arrays, including the meiotic spindle, axoneme and manchette and in its absence, there is a dramatic loss of germ cells and male sterility. Through examining axoneme structure, we identify differences in TUBE1 function between somatic and germ cells and potentially between species. Moreover, we provide evidence for the interplay between TUBE1 and katanin-mediated microtubule severing, and for the sub-specialization of individual katanin paralogs in the regulation of specific microtubule arrays.

Article activity feed

  1. Review Commons

    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

    Manuscript number: RC-2023-02111

    Corresponding author(s): Moira O’Bryan

    1. General Statements

    We thank the Review Commons editor and the three reviewers for their overall positive responses in assessing this manuscript. Further, we appreciate and would like to reiterate the similarities across our three reviewers’ comments regarding the significance of this work, where our examination of epsilon tubulin (TUBE1) during mammalian spermatogenesis will be valuable for both microtubule/cytoskeletal and developmental/ reproductive fields. Below, we have made point-by-point responses to the reviewers’ comments, and outlined by the revisions we plan to make, or have made. All line numbers refer to the transferred manuscript file with tracked changes.

    2. Description of the planned revisions

    Reviewer 1:* The authors claim that because the TUBE1 knockout mouse have abnormal centrosome numbers during meiosis, there is a role for TUBE1 in suppressing supernumerary centriole formation. While this is one possibility, it's also possible that abnormal centrosome numbers arose as a result of cell division defects, especially because binucleate cells are present in mutants. The authors should edit the text to state that abnormal centrosome numbers may arise from either supernumerary centriole formation (by the templated or de novo pathways) or from failure to complete cell division. *

    *OPTIONAL: to test these possibilities, the authors may choose to 1) count the number of centrioles in meiosis with two different centriole markers 2) stain for markers of mature centrioles, such as Cep164, to determine the number of parental centrioles. *

    Response: This is a good point. Published data indicates that the Stra8-cre is active within a subset of undifferentiated spermatogonia, and in differentiated spermatogonia through to pre-leptotene spermatocytes (Sadate-Ngatchou et al., 2008). This raises the possibility that the increase in centriole numbers could be due to a failure to complete cell division if cre is active in mitotically active spermatogonia populations. The text has been appropriately modified in lines 207-209 and 352 to reflect these insights. We appreciate the Reviewer’s optional suggestion to perform additional immunolabeling experiments and intend to examine the number of parental centrioles in spermatocytes during meiotic division using a marker of the distal or subdistal appendages. This data will be included in the final revised document.

    Reviewer 2:* Considering the suggested non-canonical function of Epsilon tubulin outside the centriole in mice sperm, it is critical to know the localization of the protein in spermatocytes during meiosis and spermatids during differentiation. *

    Response: We agree with Reviewer 2 that determining the localization of TUBE1 in spermatocytes and spermatids would be desirable. However, we are yet to find an appropriate available antibody for this. We have previously assessed the specificity of a TUBE1 antibody (PA5-56917, Invitrogen), however, this antibody was not suitable for use in our mouse model. This aside, we have recently acquired a new TUBE1 antibody which we plan to evaluate its specificity during this revision period. If it appears to bind specifically to TUBE1, we will perform the requested localization experiments.

    For clarification we have previously defined the location of TUBE1 in spermatids to the manchette and basal body in elongating spermatids (lines 72-74) (Dunleavy et al., 2017). Unfortunately, the antibody used in this study is now discontinued. The phenotypes observed as a consequence of TUBE1 loss of function in this study are, however, consistent with these patterns of localization.

    Reviewer 2:* Localization of Epsilon tubulin is needed to distinguish between mutant sperm cells and those that are not Epsilon tubulin mutants in the Tube1GCKO/GCKO mice. E.g., are the 28.07% of Tube1GCKO/GCKO tubules that showed a Sertoli cell only (SCO) phenotype the one where all the cells are mutants? *

    Response: As per our response to Reviewer 2’s comment above, we plan to test a new TUBE1 antibody to determine TUBE1 localization in this model. Outlined in our response to Reviewer 2 below, we also plan to sequence DNA from mature epididymal sperm from our mutant mice to further confirm the deletion of *Tube1 *exon 3.

    Reviewer 2:* The generated conditional germ cell-specific mutants are demonstrated by mRNA expression spermatocytes. It would help if DNA sequencing, western, and Immunohistochemical staining were used to show the gene and protein are affected. *

    Response: We thank Reviewer 2 for their suggestions. Should we successfully validate an appropriate TUBE1 antibody for use in our model, we will perform immunohistochemical staining during the revision process. Our qPCR results from purified spermatocytes however, strongly suggest that the Tube1 gene is deleted in our model, noting that such purifications are on average 81% pure with the major contaminants being Sertoli cells and spermatids (Dunleavy et al., 2019). To further confirm the deletion of *Tube1 *exon 3, we plan to sequence DNA from mature epididymal sperm from our mutant mice.

    Reviewer 2:* "Suggesting a core TUBE1 function that can be supplemented by either z-tubulin or TUBD1." Can you test what happens to mice Z and D tubulin isoforms in the mutant? Did their level increase in the centrioles? This is informative since there is no clear centriolar phenotype (other than centriole number that may be due to cell division failure) in mice spermatogenesis and the paper's central hypothesis in the introduction. *

    Response: We appreciate this question by Reviewer 2. Zeta tubulin is not present in the mouse genome as outlined in our introduction (lines 38-39). We do acknowledge that exploring *Tubd1 *will be informative in our mutant and thereby plan to examine its expression in round spermatids.

    Reviewer 2:* The authors looked at the Metaphase stage cells to assess meiosis. It would be more interesting to look at the meiosis prophase I. Since the Stra8 acts very early leptotene stage, it would be interesting to see if meiosis is defective from the very beginning. Also, some suggest that the manchette is nucleated at the pachytene stage. Is the manchette defective from the very early stage of nucleation?*

    Response: We thank Reviewer 2 for this suggestion. To this end, we plan to examine juvenile mouse testes at days 10 and 17 post-partum where leptotene and pachytene spermatocytes are the most mature germ cells respectively.

    In regard to the Reviewer’s comment of the manchette being nucleated in pachytene stage spermatocytes, we acknowledge that the precise mechanism of manchette nucleation has not been confirmed. We are aware of the alternative hypothesis introduced by Moreno and Schatten (2000), which postulates manchette microtubules may be nucleated prior to pachytene period, through their examination of bovine male germ cells. This hasn’t, however, been supported by evidence and with more recent data, others have suggested that the manchette is nucleated at the centrosomal adjunct (Lehti and Sironen, 2016). Indeed, our unpublished data suggests this is the case (another study). Regardless, the origin of the microtubule seeds that ultimately extend to form the manchette is not relevant to the hypothesis we have proposed. As we note that in our manuscript and mouse model, manchettes appear to assemble normally in step 8 spermatids. Rather, their movement and disassembly is abnormal i.e. TUBE1 serves critical roles more manchette movement and disassembly rather than manchette formation.

    Reviewer 2:* Is the acetylation of manchette microtubules affected in the absence of TUBE1? *

    Response: Reviewer 2 raises an interesting question, which we plan to answer through immunolabeling of testis sections for acetylated tubulin in our control and mutant groups.

    *Reviewer 3: **Minor points, a substantial percentage of sperm produced had a normal head shape in the KO (Figure 1I), which undermine the function of tube1 in nuclear shaping, the author should address this point in their manuscript. It is also curious whether there are phenotype in other tissues, can the authors comment on that? *

    Response: We thank Reviewer 3 for highlighting this point. As reported in Fig. 1I, 28.5% of sperm from Tube1GCKO/GCKO epididymides have abnormal nuclear shape. This is a 4.4-fold increase over that seen in wild type sperm. These data clearly highlight the role of TUBE1 in defining nuclear morphology. Variations between cells does not undermine this conclusion. It appears that prior to sperm release from the testis, the majority of TUBE1 null spermatids heads are abnormally shaped. However, in the epididymis there appears to be an increase in the proportion of normally shaped heads. We thus hypothesize that the high rates of spermiation failure in the TUBE1 null mice reflect the preferential removal of abnormally shaped sperm by Sertoli cells, thus enriching for normally shaped heads that are released. During the revision process, we will quantify the percentage of spermatids with normal versus abnormally shaped heads prior to spermiation in testis sections. All Tube1 null mice were sterile.

    To Reviewer 3’s second point - we have not examined other tissues in this conditional male germ cell knockout mouse model, as the cre used in this manuscript is only expressed in the testis (Sadate-Ngatchou et al., 2008). Consistent with the specificity of the deletion, null male mice are overtly healthy, with the exception of male fertility, and exhibit normal body weight as detailed on line 123 and in Fig S1D.

    3. Description of the revisions that have already been incorporated in the transferred manuscript

    Reviewer 1:* In figure 5, based on quantification of fluorescence intensity, the authors conclude that loss of epsilon-tubulin results in an increase in the levels of KATNAL1, KATNAL2, and KATNB1. Given the inherent variability in immunofluorescence staining, the authors should at a minimum normalize their intensity measurements to those of an unrelated control protein stained in the same cell (ex: alpha-tubulin). It would be more convincing to quantify the levels of these proteins by Western blot (again, normalized to a control protein or to total cellular protein), which should be feasible given that the authors can isolate elongating spermatids. *

    Response: We thank Reviewer 1 for this suggestion to better account for any potential variability between immunofluorescence staining in cells. In this instance, alpha-tubulin would be a related protein in our model, making it unsuitable for normalization - the longer manchette phenotypes in our mutant spermatids indicate more tubulin present in mutant cells. We have therefore normalized the fluorescence intensity in our cells to DNA content (DAPI staining). This has provided comparable results to our initial analysis, and we have edited our text accordingly at lines 303, 307-310, 563-564, 845, 850 and Fig. 5. We respectfully disagree that western blotting would be informative, as the point is that katanin proteins are accumulating abnormally on the elongating sperm manchette. This does not necessarily mean that overall katanin levels will be increased. This aside, given the low numbers of elongating spermatids in the Tube1GCKO/GCKO mice, obtaining sufficient materials of western blotting is prohibitive. With the severity of germ cell loss indicated by our daily sperm production calculations, we predict the isolated spermatids of up to 5 Tube1GCKO/GCKO animals would be required to make up one biological replicate. It would not be feasible to collect the large number of animals required for at least three biological replicates in the revision timeframe.

    Reviewer 1:* A major claim of the paper is that epsilon-tubulin plays a different role within mammalian germ cells (abstract, line 22; p9, lines 167-168; p15 lines 315-316), because the Tube1GCKO/GCKO mice can form some sperm with relatively normal ciliary ultrastructure, whereas ciliates lacking epsilon-tubulin fail to form cilia. However, it's unclear whether the centrioles that templated these normal cilia were formed before or after epsilon-tubulin loss. Given that centrioles are inherited from one generation to the next, it's possible that the few normal cilia may be templated by relatively normal parental centrioles. These parental centrioles would have been present in spermatogonia prior to Cre expression/epsilon-tubulin deletion, and inherited by a fraction of sperm after the mitotic and meiotic divisions, resulting in sperm with normal ciliary ultrastructure. Other spermatocytes may have inherited centrioles formed in the absence of epsilon-tubulin, resulting in aberrant centrioles similar to those reported in human somatic cells, but these would not form any sperm flagella due to a loss of cell viability, as has been reported for acentriolar cells in a p53+ background. Underscoring this point, Chlamydomonas and human somatic mutant cells constitutively lack epsilon-tubulin. In these systems, the parental centrioles were diluted from the population over many cell divisions, and phenotypic analysis would only include the centrioles that formed in the absence of epsilon-tubulin. To make their major claim, the authors need to demonstrate that the basal bodies of sperm flagella with normal ultrastructure were formed in the absence of epsilon-tubulin, and were not normal parental centrioles. Given the difficulty of this experiment, the authors may instead choose to remove their claim that epsilon-tubulin plays a different role within mammalian germ cells. *

    Response: The authors thank Reviewer 1 for their detailed input regarding TUBE1’s centriolar importance across species. From their feedback, we recognize the need to modulate our interpretation of this result. We have also added a line to our manuscript highlighting that the normal axonemal structure observed may be due to the inheritance of normal centrioles (lines 328-329). We note however, that sperm produced within the null animals were immotile and that motility could not be recovered by the addition of exogenous ATP thus revealing that TUBE1 is required to form functional sperm tails.

    Reviewer 2:* It will help if the introduction summarizes the knowledge on Epsilon tubulin in spermatogenesis with emesis on its localization and the method used to find the localization. *

    Response: We have modified the introduction accordingly in lines 72-73.

    Reviewer 2:* How many independent mutant animals were studied, and what was the elfishness of generating mutants with a complete mutant testis? From Fig s1c, it appears all mutants generated were total mutations in almost all cells - is this correct? *

    Response: We have updated the number of animals studied as per the comment below. Regarding the mutant status of our mouse model, we used Stra8-Cre which is active between early (postnatal day 3) spermatogonia to pre-leptotene spermatocytes (Sadate-Ngatchou et al., 2008) thus all spermatocytes, spermatids, and sperm will carry the deletion. As shown in Fig. S1C we measured a 90.1% reduction in *Tube1 *mRNA expression from purified spermatocytes. As mentioned above, we note that the purified germ cells always contain a low percentage of contaminating cells. Using our optimized Staput method we obtain isolated germ cell populations of high purity, where in spermatocyte populations we calculate 19% contamination with other testicular cell types (e.g. somatic Sertoli/interstitial cells, spermatogonia, spermatids) (Dunleavy et al., 2019). We therefore believe the 9.9% Tube1 mRNA expression detected in our Tube1GCKO/GCKO group are the origin of that residual mRNA. We have included this information in the materials and methods section (lines 491-493).

    Reviewer 2:* Add a definition to "ZED-tubulins." *

    Response: A definition to the ZED-tubulins can be found on line 32.

    Reviewer 2:* From the paper, it is unclear if Epsilon tubulin is dispensable for centriole function only in sperm cells or if the same is true in mice somatic cells in vivo. *

    Response: In this study we have used a conditional male germ cell knockout mouse model to examine TUBE1’s function specifically in male germ cells. As mentioned in our introduction, the function of TUBE1 has not been examined in murine somatic cells in vivo (lines 68-70). To avoid confusion, we have reiterated this point in lines 356-358 of our discussion.

    Reviewer 2:* Fig. S1 and other figures: "n {greater than or equal to} 3 samples/genotype" - this is unclear - please indicate the number of independent animals tested. *

    Response: We have modified the figure legends accordingly in lines 11-13 and 33-35 of the transferred supplementary information file and lines 787-788 and 810-811 of the transferred manuscript file.

    Reviewer 2:* "suppressing supernumerary centriole formation" is this due to access centriole formation or failed mitosis? *

    Response: We acknowledge Reviewer 2’s comment is similar to the comment made by Reviewer 1 above and note we have modified the associated text in lines 207-209 in response to the above comment.

    Reviewer 2:* The KATNAL1, KATNAL2, and KATNB1 staining in Fig 5 show multiple foci in the nucleus. Are these foci-specific staining or nonspecific? It is surprising to see such a large complex. *

    Response: As outlined in the materials and methods and the Fig. 5 legend, Fig. 5 displays three-dimensional (3D) z-stack images of whole elongating spermatids presented as 2D maximum intensity projections. The katanin subunit staining is around the nucleus rather than inside of it, however the flattening of the image from 3D to 2D make the foci appear inside the nucleus. To clarify this, we have modified the Fig. 5 legend in lines 845 and 848.

    Reviewer 2:* How the staging of spermatids was performed needs to be explained in the method. *

    Response: We have included additional explanation the materials and methods section (lines 513-514).

    Reviewer 3:* The experimental part is of the highest quality and the manuscript is very well written.* My only reservation with the manuscript is concerning the model proposed for manchette migration in the Discussion section (Figure 6). I find the proposed model highly speculative and pre-mature, not supported enough by data, as even admitted by the authors (lines 415-427). Having it as a figure and concluding remark gives it too match weight, my suggestion would be to remove figure 6 and tone down the discussion.

    Response: The authors thank Reviewer 3 for their complimentary overview of our manuscript. We agree that some unanswered questions remain in our proposed model of manchette migration. This study has however, added several critical missing pieces. With respect, we prefer to keep Figure 6 in the manuscript as explaining manchette function to non-experts is very difficult without a visual aide. To ensure transparency with the audience that our model is indeed hypothetical, we have edited our discussion and Figure 6 legend to reflect this (lines 406, 417, 428, 435, 463, 860, 863, 869).

    4. Description of analyses that authors prefer not to carry out

    None

    References

    DUNLEAVY, J. E., GRAFFEO, M., WOZNIAK, K., O’CONNOR, A. E., MERRINER, D. J., NGUYEN, J., SCHITTENHELM, R. B., HOUSTON, B. J. & O’BRYAN, M. K. 2022. Male mammalian meiosis and spermiogenesis is critically dependent on the shared functions of the katanins KATNA1 and KATNAL1. bioRxiv, 2022.11.11.516072.

    DUNLEAVY, J. E. M., O’CONNOR, A. E. & O’BRYAN, M. K. 2019. An optimised STAPUT method for the purification of mouse spermatocyte and spermatid populations. Molecular Human Reproduction.

    DUNLEAVY, J. E. M., OKUDA, H., O’CONNOR, A. E., MERRINER, D. J., O’DONNELL, L., JAMSAI, D., BERGMANN, M. & O’BRYAN, M. K. 2017. Katanin-like 2 (KATNAL2) functions in multiple aspects of haploid male germ cell development in the mouse. PLOS Genetics, 13.

    LEHTI, M. S. & SIRONEN, A. 2016. Formation and function of the manchette and flagellum during spermatogenesis. Reproduction, 151__,__ R43-54.

    MORENO, R. D. & SCHATTEN, G. 2000. Microtubule configurations and post-translational alpha-tubulin modifications during mammalian spermatogenesis. Cell Motil Cytoskeleton, 46__,__ 235-46.

    SADATE-NGATCHOU, P. I., PAYNE, C. J., DEARTH, A. T. & BRAUN, R. E. 2008. Cre recombinase activity specific to postnatal, premeiotic male germ cells in transgenic mice. Genesis, 46__,__ 738-42.

  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

    In this study Stathatos et al looked at the function of epsilon tubulin (tube1), specifically in male germ cells. Previous work showed that tube1 is an important member of the tubulin family but its function is more enigmatic compared to alpha, beta and gamma tubulin. The authors produced a mouse KO line of tube1 and the data presented in this manuscript concerns the effects on spermatogenesis. They found that tube1 is essential for multiple microtubule dependent functions, including meiosis, nuclear shaping and sperm motility.

    The experimental part is of the highest quality and the manuscript is very well written. My only reservation with the manuscript is concerning the model proposed for manchette migration in the Discussion section (Figure 6). I find the proposed model highly speculative and pre-mature, not supported enough by data, as even admitted by the authors (lines 415-427). Having it as a figure and concluding remark gives it too match weight, my suggestion would be to remove figure 6 and tone down the discussion. Minor points, a substantial percentage of sperm produced had a normal head shape in the KO (Figure 1I), which undermine the function of tube1 in nuclear shaping, the author should address this point in their manuscript. It is also curious whether there are phenotype in other tissues, can the authors comment on that?

    Significance

    The observations reported are novel and will be highly valuable specifically for the sperm biology field but also very interesting to the microtubule field in general.

  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

    The paper "Epsilon tubulin is an essential determinant of microtubule-based structures in male germ cells" provides the first insight into the essential function of Epsilon tubulin. TUBE1 (epsilon tubulin) is a non-canonical tubulin localized at the pericentriolar material of somatic and germ cell centrosome. TUBE1 has been primarily studied in unicellular organisms and cell lines, and multiple studies have shown its role in ciliogenesis and flagellum formation. However, its role in mammals, specifically in fertility, is unknown. Here, Stathatos et al address the critical question of whether TUBE1 plays a role in mammalian spermatogenesis and fertility. The authors show by germline inactivation of TUBE1 that the mice lacking TUBE1 are sterile, defective in meiosis, form abnormal manchette, and sperms are nonmotile. The authors further correlate that the TUBE1 functions together with KATNAL-1, KATNAL-1, and KATNB1, the microtubule severing protein. As little is known about the role of non-canonical tubulin like TUBE1 in fertility, this manuscript addresses a significant knowledge gap and generates an exciting hypothesis that TUBE1 regulates the KATNAL1-KATNB1 and KATNAL2-KATNB1 dynamic at manchette microtubules and perinuclear ring to control the manchette microtubule severing and migration.

    Overall, the paper suggests that Epsilon tubulin is essential for multiple complex microtubule arrays, including the meiotic spindle, axoneme, and manchette; however, in the absence of Epsilon tubulin localization data, it is unclear which microtubule array is affected directly and which indirectly (e.g., is the axoneme defect is due to Epsilon tubulin in the axoneme or centriole?). In particular, it is interesting that in mice sperm, Epsilon tubulin is dispensable for centriole-mediated axoneme formation, its primary function in single-cell organisms (can this be due to compensation by the other tubulin isoforms?). Once the concerns below are resolved, the paper will be significant for the cytoskeleton and reproductive research fields.

    Major comment

    • Considering the suggested non-canonical function of Epsilon tubulin outside the centriole in mice sperm, it is critical to know the localization of the protein in spermatocytes during meiosis and spermatids during differentiation.
    • Localization of Epsilon tubulin is needed to distinguish between mutant sperm cells and those that are not Epsilon tubulin mutants in the Tube1GCKO/GCKO mice. E.g., are the 28.07% of Tube1GCKO/GCKO tubules that showed a Sertoli cell only (SCO) phenotype the one where all the cells are mutants?

    Minor comment

    • It will help if the introduction summarizes the knowledge on Epsilon tubulin in spermatogenesis with emesis on its localization and the method used to find the localization.
    • The generated conditional germ cell-specific mutants are demonstrated by mRNA expression spermatocytes. It would help if DNA sequencing, western, and Immunohistochemical staining were used to show the gene and protein are affected.
    • How many independent mutant animals were studied, and what was the elfishness of generating mutants with a complete mutant testis? From Fig s1c, it appears all mutants generated were total mutations in almost all cells - is this correct?
    • Add a definition to "ZED-tubulins."
    • "Suggesting a core TUBE1 function that can be supplemented by either z-tubulin or TUBD1." Can you test what happens to mice Z and D tubulin isoforms in the mutant? Did their level increase in the centrioles? This is informative since there is no clear centriolar phenotype (other than centriole number that may be due to cell division failure) in mice spermatogenesis and the paper's central hypothesis in the introduction.
    • From the paper, it is unclear if Epsilon tubulin is dispensable for centriole function only in sperm cells or if the same is true in mice somatic cells in vivo.
    • Fig. S1 and other figures: "n {greater than or equal to} 3 samples/genotype" - this is unclear - please indicate the number of independent animals tested.
    • "suppressing supernumerary centriole formation" is this due to access centriole formation or failed mitosis?
    • The KATNAL1, KATNAL2, and KATNB1 staining in Fig 5 show multiple foci in the nucleus. Are these foci-specific staining or nonspecific? It is surprising to see such a large complex.
    • How the staging of spermatids was performed needs to be explained in the method.
    • The authors looked at the Metaphase stage cells to assess meiosis. It would be more interesting to look at the meiosis prophase I. Since the Stra8 acts very early leptotene stage, it would be interesting to see if meiosis is defective from the very beginning. Also, some suggest that the manchette is nucleated at the pachytene stage. Is the manchette defective from the very early stage of nucleation?
    • Is the acetylation of manchette microtubules affected in the absence of TUBE1?

    Significance

    Overall, the paper suggests that Epsilon tubulin is essential for multiple complex microtubule arrays, including the meiotic spindle, axoneme, and manchette; however, in the absence of Epsilon tubulin localization data, it is unclear which microtubule array is affected directly and which indirectly (e.g., is the axoneme defect is due to Epsilon tubulin in the axoneme or centriole?). In particular, it is interesting that in mice sperm, Epsilon tubulin is dispensable for centriole-mediated axoneme formation, its primary function in single-cell organisms (can this be due to compensation by the other tubulin isoforms?). Once the concerns are resolved, the paper will be significant for the cytoskeleton and reproductive research fields.

  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

    The ZED (zeta-, epsilon-, and delta-) tubulins are important, yet understudied, members of the tubulin superfamily. Here, Stathatos et al. build upon previously published work and leverage their expertise to uncover the roles of epsilon-tubulin in mouse male germ cells. The authors create a germ cell-specific Tube1 knockout mouse, using Stra8-Cre, which is active in spermatogonia before the meiotic divisions. The authors report that knockout of Tube1 results in a range of defects during spermatogenesis, including: 1) a loss of male germ cells 2) sperm motility defects 3) abnormally shaped sperm heads 4) abnormal meiotic spindle morphology and abnormal centrosome numbers 5) some defects in sperm axoneme ultrastructure, 6) disrupted manchette migration 7) increased levels of katanin subunits at the manchette. Most of the experiments are convincing and well done, and based on this work, the authors propose a novel model for regulation of the manchette. I believe this work is of interest and should be published with revisions addressing the following major and minor comments.

    Major comment:

    1. A major claim of the paper is that epsilon-tubulin plays a different role within mammalian germ cells (abstract, line 22; p9, lines 167-168; p15 lines 315-316), because the Tube1GCKO/GCKO mice can form some sperm with relatively normal ciliary ultrastructure, whereas ciliates lacking epsilon-tubulin fail to form cilia. However, it's unclear whether the centrioles that templated these normal cilia were formed before or after epsilon-tubulin loss. Given that centrioles are inherited from one generation to the next, it's possible that the few normal cilia may be templated by relatively normal parental centrioles. These parental centrioles would have been present in spermatogonia prior to Cre expression/epsilon-tubulin deletion, and inherited by a fraction of sperm after the mitotic and meiotic divisions, resulting in sperm with normal ciliary ultrastructure. Other spermatocytes may have inherited centrioles formed in the absence of epsilon-tubulin, resulting in aberrant centrioles similar to those reported in human somatic cells, but these would not form any sperm flagella due to a loss of cell viability, as has been reported for acentriolar cells in a p53+ background. Underscoring this point, Chlamydomonas and human somatic mutant cells constitutively lack epsilon-tubulin. In these systems, the parental centrioles were diluted from the population over many cell divisions, and phenotypic analysis would only include the centrioles that formed in the absence of epsilon-tubulin. To make their major claim, the authors need to demonstrate that the basal bodies of sperm flagella with normal ultrastructure were formed in the absence of epsilon-tubulin, and were not normal parental centrioles. Given the difficulty of this experiment, the authors may instead choose to remove their claim that epsilon-tubulin plays a different role within mammalian germ cells.

    Minor comments:

    1. The authors claim that because the TUBE1 knockout mouse have abnormal centrosome numbers during meiosis, there is a role for TUBE1 in suppressing supernumerary centriole formation. While this is one possibility, it's also possible that abnormal centrosome numbers arose as a result of cell division defects, especially because binucleate cells are present in mutants. The authors should edit the text to state that abnormal centrosome numbers may arise from either supernumerary centriole formation (by the templated or de novo pathways) or from failure to complete cell division.

    OPTIONAL: to test these possibilities, the authors may choose to 1) count the number of centrioles in meiosis with two different centriole markers 2) stain for markers of mature centrioles, such as Cep164, to determine the number of parental centrioles.

    1. In figure 5, based on quantification of fluorescence intensity, the authors conclude that loss of epsilon-tubulin results in an increase in the levels of KATNAL1, KATNAL2, and KATNB1. Given the inherent variability in immunofluorescence staining, the authors should at a minimum normalize their intensity measurements to those of an unrelated control protein stained in the same cell (ex: alpha-tubulin). It would be more convincing to quantify the levels of these proteins by Western blot (again, normalized to a control protein or to total cellular protein), which should be feasible given that the authors can isolate elongating spermatids.

    Significance

    The strengths of this study lie in the careful phenotypic analysis of loss of epsilon-tubulin, which is well-done and very thorough. The limitations of the study are in interpretation of the results, specifically as relates to centriole formation, but can be addressed as indicated above. This work will be of interest to cell and developmental biologists, especially those interested in centrosomes, cilia, and spermatogenesis.

  5. Version published to 10.1101/2023.06.07.544006v1 on bioRxiv