The CUL4-DDB1-DCAF1 E3 ubiquitin ligase complex regulates PLK4 protein levels to prevent premature centriole duplication

  1. Josina Grossmann
  2. Anne-Sophie Kratz
  3. Alina Kordonsky
  4. Gali Prag
  5. Ingrid Hoffmann

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Centrioles play important roles in the assembly of centrosomes and cilia. Centriole duplication occurs once per cell cycle and is dependent on polo-like kinase 4 (PLK4). To prevent centriole amplification, which is a hallmark of cancer, PLK4 protein levels need to be tightly regulated. Here, we show that the Cullin(CUL)4A/B-DDB1-DCAF1, CRL4 DCAF1 , E3 ligase targets PLK4 for degradation in human cells. DCAF1 binds and ubiquitylates PLK4 in G2 phase to prevent premature centriole duplication in mitosis. In contrast to the regulation of PLK4 by SCF β-TrCP , the interaction between PLK4 and DCAF1 is independent of PLK4 kinase activity and mediated by polo-boxes 1 and 2 of PLK4, suggesting that DCAF1 promotes PLK4 ubiquitylation independently of β-TrCP. Thus, the SCF Slimb/β-TrCP pathway, targeting PLK4 for ubiquitylation based on its phosphorylation state and CRL4 DCAF1 , which ubiquitylates PLK4 by binding to the conserved PB1-PB2 domain, appear to be complementary ways to control PLK4 abundance to prevent centriole overduplication.

Article activity feed

  5. 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

    'The authors do not wish to provide a response at this time.'

  6. 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 manuscript, Grossmann et al. present a new potential pathway that regulates PLK4 levels in cells mediated by the CRL4^DCAF1 E3 ubiquitin ligase complex (CUL4A/B-DDB1-DCAF1). PLK4 plays a crucial role in centriole assembly acting as a master regulator of the centriole biogenesis and thus contributes to centriole number control. Centriole numbers need to be tightly regulated as deviations could lead to aneuploidy and potentially cancer. At the onset of centriole assembly in G1/S, PLK4 is focusing into a single point on each parental centriole together with STIL and SAS6 defining the site of procentriole formation. For this process to happen, PLK4 trans-phosphorylates itself creating a binding site for SCF^β-TrCP E3 ubiquitin ligase that targets PLK4 for ubiquitination and degradation by the proteasome. The authors identified by co-IP and mass spectrometry the CRL4DCAF1 E3 ubiquitin ligase complex as a potential regulator of PLK4. They show that CRL4^DCAF1 E3 ubiquitin ligase complex binds to PLK4 and targets it for degradation. Furthermore, the authors present data where knockdown of DCAF1 leads to increased levels of PLK4 and centriole amplification. Using AlphaFold and followed by IPs with PLK4 point-mutants, they propose that DCAF1 binds to the dimer of PLK4 at PB1-PB2 at a similar site where Cep192/Cep152 bind. Then, they move on to show that CRL4^DCAF1 E3 ubiquitin ligase complex ubiquitylates PLK4 predominantly in G2 phase. Lastly, they propose that DCAF1 regulates the interaction of PLK4 with STIL and that it is required to prevent premature centriole disengagement in G2 phase. The manuscript is written in a clear and concise manner while the experimental approaches are sound and well described. The experimental data are well presented with a good number of replicates in most cases. However, some of the conclusions are drawn from marginal differences in the data and without statistical tests (cases indicated in detail below). I believe that this work is of interest to the scientific community, but it would require revisions to address the following major and minor comments.

    Major comments:

    • The key finding of the paper is that PLK4 co-IPs with DCAF1 and DDB1 that are core components of CUL4A/B E3 ubiquitin ligase. However, the only evidence that this interaction between PLK4 and DCAF1 is direct relies on the ubiquitylation assay performed in E. coli. This experiment was performed only once, and no quantitation is performed (Fig 4A). Given that the components are overexpressed in this heterologous system, it is very plausible to have a non-specific interaction between the DCAF1-Acidic domain and PLK4-PB1-PB2 considering that native binders to this region (Cep192/Cep152) are absent. The PLK4-DCAF1 model that was generated with AlphaFold suggests that this interaction is plausible but stronger verification that the interaction is direct are necessary in this reviewer's point of view. This could be performed for instance by purifying proteins (or fragments of them) to test binding in vitro, or through IPs of full-length proteins from bacterial extracts. If the interaction between PLK4 and DCAF1 is indeed direct, then the authors would need to provide an explanation of the feasibility of this interaction given that the binding site is occupied by Cep192 or Cep152 at the centrioles. Based on the current knowledge, PLK4 is loaded to the centriole through the interaction with Cep192 which is then switched to interaction with Cep152. For the DCAF1 to be able to bind to PLK4 it would need to outcompete Cep152. Thus, in order to prove that DCAF1 can control PLK4 at the centrosome, evidence would need to be provided that this interaction is possible. If that is not the case, then the most likely alternative is that DCAF1 interactions with cytoplasmic pool of PLK4, thus only indirectly controlling the PLK4 levels at centrioles. A plausible alternative interpretation of the data provided would be that DCAF1-Acidic domain could bind weakly and perhaps non-specifically to PLK4 but in human cells the interaction is mediated through another component such as Cep152 or Cep192 (which are also present in the MS data). Based on the AlphaFold model, the authors introduced point mutations that abolish the PLK4-DCAF1 interaction, but this effect could just as easily be an indirect effect due to abolishing of the PLK4-Cep152/Cep192 interaction.
    • The authors state that DCAF1 depletion with siRNA or shRNA leads to increased level of PLK4 which triggers centriole overduplication. However, this statement is not entirely supported by the data provided. Firstly, in the western blots shown (Fig2A, 2D) the increase in PLK4 levels is hardly visible. Given that this is a key finding, stronger evidence would need to be provided. Furthermore, the quantification of the PLK4 levels upon siRNA mediated DCAF1 depletion are confusing as siDCAF1#2 leads to higher PLK4 levels than siDCAF1#1 despite being less effective in DCAF1 depletion (Fig 2A). More importantly, the quantification on the HeLa tet-on shDCAF1, that are used in many experiments, is missing an important statistical test (Fig 2D). Similarly, no statistical test is performed on the quantification of centriole numbers (Fig 2F) which puts to question the conclusion that "CRL4DCAF1 might function to keep PLK4 protein levels low, thus preventing centriole overduplication". Moreover, GFP-PLK4 levels shown in Fig 6A seem unaltered (if not lowered) upon DCAF1 depletion. Lastly, DCAF1 overexpression does not seem to decrease PLK4 levels as shown in Fig 6B. In that experiment, though, PLK4 is also overexpressed. In order to support the proposed function that CRL4DCAF1 keeps PLK4 levels low, it would be useful to also investigate whether overexpression of DCAF1 would lead to further decrease of PLK4 levels.
    • In page 7, the authors mention: "A premature onset of centriole duplication in the absence of DCAF1 should also result in increased numbers of already disengaged centrioles in G2 phase." This premise is not correct as it is inverted to the current knowledge. It is the premature disengagement that licences for premature centriole duplication (or as often stated as re-duplication) rather than the premature onset of centriole duplication that causes disengagement. This is also what the authors correctly state in the discussion. In the data presented (Fig 6C) the authors observe centriole disengagement upon DCAF1 depletion using expansion microscopy, but no re-duplication is visible in the images provided. This is contrary to the overduplication claim made earlier on (Fig 2F). As such, the data presented do not fully support the drawn conclusion that DCAF1 controls PLK4 levels in G2 to prevent unscheduled centriole duplication. The authors would also need to investigate whether the prolonged use of Cdk1 inhibitor RO-3306 to synchronise the cells in G2 in addition to DCAF1 depletion contributes to the centriole disengagement that is observed, considering that Cdk1-Cyclin B acts also on PLK4-STIL complex.
    • The mechanism proposed by the authors is that DCAF1 maintains PLK4 at low levels throughout G2 which prevents premature disengagement. Subsequently, low PLK4 levels prevent binding and activation of STIL impeding premature initiation of centriole duplication. However, this would not happen since centrioles remain engaged at this stage. Overall, some of the aspects of the proposed mechanism are not fully supported by the data presented. In addition, the proposed mechanism does not offer a suitable mechanistic explanation of how lower PLK4 levels by CRL4^DCAF1 mediated ubiquitylation and degradation prevent centriole disengagement.

    Minor comments

    • In Fig S2A authors need to indicate the expected size of the expressed protein. In its current form blot is difficult to be assessed. More specifically, it is unclear what is the result on the IP with the PLK4 fragment (1-879) since the more intense band in the input in not the same as in in the IP with Flag.
    • In Fig 1C, S2B, S3B, it would be helpful to have a summary of the interactions observed next to each construct. This is commonly represented with (-, +, ++, +++) depending on the amounts present in the IP.
    • In Fig 1D, even though not statistically significant, there seems to be a reduction in the IP of AA and PEST. Do the authors have some suggestion why that might be?
    • Authors used two different cell lines in the experiments presented in Fig2A and Fig 2B. Given that depletion of siDCAF5 is provided as a control of having no effect in the PLK4 levels I would expect to have the experiment performed on same cell line.
    • No statistical test is provided in the comparison on PLK4 levels upon siRNA treatment coupled with CHX (Fig 2C).
    • In the quantification of the PLK4 levels at the centrosomes (Fig 2E), it is not specified whether a background subtraction step was performed prior to the normalisation to the untreated control.
    • In the blot shown in Fig S3, no input is visible in the lane with expression of the Acidic domain.
    • Authors claim that both WD40 and acidic domain contribute to binding of PLK4 because WD40-Acidic is more efficient in binding PLK4 that Acidic domain alone. However, in the blot provided, WD40 alone does not interact with PLK4. Thus, the most likely explanation would be that Acidic domain is the major interactor and WD40 has only minor contribution or it offers a stabilisation role to the acidic domain.
    • Regarding the AlphaFold model provided, and in addition to the comments above, some further clarifications and controls would need to be provided. AlphaFold is a powerful tool but not without its caveats and needs to be used with caution. The authors need to provide a description on how they used AlphaFold to generate the model presented. Typically, AlphaFold produces 5 output models. At which site was DCAF1-Acidic domain positioned in the other output models? Based on what criteria the model shown was selected? Also, a confidence score for the model should be provided.
    • The authors compare their PLK4-DCAF1 AlphaFold model with the structure of PLK4-CEP192 complex but not with the PLK4-Cep152. What is the explanation for this? Given that Cep152 is reported to have higher affinity than Cep192 (Park SY et al., 2014) it would be important to be included in the comparisons performed.
    • The phrase "An overlay between the two structures revealed that ..." is not accurate as one is a merely a model. There are also other instances in the text that the model is referred to as 'structure' which is not correct.
    • Please provide a citation for "Poisson-Boltzmann solver (APBS)".
    • In Fig 3A ribbon representations are too small to see DCAF1 in the printout.
    • The mutations designed might affect the folding of the PBs and thus no interaction is observed. Authors could test how the mutations would affect PB1-PB2 and also design one or two mutants that are in the vicinity but not in the interaction interface to serve as true negative controls in addition to the PLK4-WT. Do these mutants localise to centrioles or also the interaction with Cep192/Cep152 is affected?
    • There is no statistical test in the quantification in Fig 3D, but it is not critical as the difference is very clear and certainly statistically significant.
    • Authors state that DCAF1 strongly interacts with PLK4 during interphase but only weakly in mitosis with quantification in Fig 5A but there is no statistical test.
    • Based on the data shown in Fig 5B, authors state that PLK4 is predominantly ubiquitylated by CRL4DCAF1 in G2 phase. However, in the blot shown, PLK4 seems to be in more abundance in G2 that might explain the apparent higher ubiquitylation. Furthermore, the experiment was performed once and no quantification of the ubiquitylation is performed. Lastly, there are no evidence that this apparent higher ubiquitylation in G2 is mediated by CRL4DCAF1.
    • In Fig 6A, STIL levels upon DCAF1 depletion seem to be lower, is there any potential explanation for that? No statistical test is performed for the STIL/GFP-PLK4 levels difference in siGL2 versus siDCAF1. The authors should provide a justification for over-expressing PLK4 in this experiment. Similarly, in Fig 6B, the authors use overexpression of both PLK4 and DCAF1 and no statistical test is performed.
    • Authors report in Fig 6C disengaged centrioles. How are disengaged centriole defined, is it based on a distance cut-off or loss of orthogonality? In the images provided, this reviewer's impression is that in the (+) Dox condition, there are two parental centrioles that have separated rather disengaged procentriole. Do the images come from the same cell?
    • Based on the data presented, would overexpression of PLK4 in G2 would result in centriole disengagement? This is something that the authors would optionally check.
    • The quantification of rootletin as an additional confirmation of centriole disengagement is puzzling to me as I would expect an increase rather than decrease of its levels. As centrioles disengage, a new link would need to form and thus the expected increase in its levels. However, new rootlet might form only later in mitosis. Also, given that the cells are synchronised in G2 the quantification is more complex. In late G2, centrioles separate in order to move to opposite poles to form the mitotic spindle. This would result in removal of the rootlet that might reflect the reduction the authors report. Ideally the quantification should be limited to cells in late G2 (that centrioles have separated) stained with Centrin 2 to allow for a quantification per centriole pair.
    • In the discussion, authors state "It is conceivable that increasing amounts of PLK4 during mitosis, when the interaction between CRL4DCAF1 and PLK4 is weak, might capture STIL from binding to CDK1 initiating the interaction between PLK4 and STIL". In mitosis CDK1-Cyclin B binds to STIL and prevents formation of the PLK4-STIL complex, thus inhibiting untimely onset of centriole biogenesis (Zitouni et al., 2016). In addition, the authors show that total PLK4 levels are low in mitosis (Fig 5A). The conclusion drawn are not in line with the current literature.
    • The addition of a graphical representation of the proposed mechanism would be beneficial to the readers.
    • A reference for the Ac.Tubulin antibody used is missing.
    • Please provide a citation for FiJi.

    Referees cross-commenting

    I find the comments by the other two reviewers to be valid, clear, insightful, and complementary to those made by this reviewer. There is a good convergence between the reviewers on the critical aspects in this manuscript that require attention. Following revisions this study will contribute to the understanding of regulatory mechanisms acting at the centrioles.

    Significance

    Centriole number control is an important aspect that is relevant not only to the centrosome research field but is also related to cilia, cells signaling, and cancer research. This work presents a novel pathway involved in the regulation of PLK4 levels in cells mediated by the CRL4^DCAF1 E3 ubiquitin ligase complex (CUL4A/B-DDB1-DCAF1). The authors present extensive data to characterise when and how DCAF1 interactions with PLK4 to lowers its levels through ubiquitination and subsequent degradation by the proteasome. However, the effects from various treatments are often minor. The study from Grossmann et al. comes to complement already known pathways of controlling centriole numbers, at G1/S through SCFβ-TrCP E3 ubiquitin ligase mediated PLK4 degradation, and in mitosis by CDK1-Cyclin B through STIL 'capturing' to block centriole reduplication. Given that certain aspects of the manuscript are revised, and an updated and more thorough mechanism is proposed and supported, it will contribute to the conceptual advancement or our understanding of centriole number control across the cell cycle. It could potentially also contribute to the ubiquitin research field of research, but it is hard for me to assess this as it is not my field of expertise.

  7. 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

    Polo-like kinase 4 (Plk4) is the master regulator of centriole assembly and previous studies have shown that its level must be tightly regulated to ensure the precise duplication of centrioles during each cell cycle. It is well documented that the abundance of Plk4 is regulated by E3 ubiquitin ligases and in particular the SCF-TrCp ubiquitin ligase. However, in the absence of SCF-TrCp mediated regulation, PLk4 is still ubiquitylated suggesting that other ubiquitin ligases function to regulate Plk4 levels. Here Grossman and colleagues identify the CUL4-DDB1-DCAF1 (CRL4DCAF1) E3 ubiquitin ligase as a regulator of Plk4 levels and show that it functions predominantly in G2 phase to prevent centriole assembly in M phase. They propose a model whereby SCF-TrCp and CRL4DCAF1 cooperate to control the levels of PLK4 at different points in the cell cycle.

    This study has the potential to yield some important and novel insights into the regulation of centriole assembly. However, in its current form the relationship between the increased Plk4 levels and the other effects described by the authors remain unclear. In particular, it is not clear to me that the small increase in Plk4 levels upon CRL4DCAF1 inhibition is responsible for the multipolar spindle phenotype. Nor is it clear how this increase in Plk4 is related to the premature disengagement defect. Finally, some of the experimental results could be made more convincing by including quantitation and/or additional controls. Major and minor issues are listed point-by-point below.

    Major issues

    Figure 2A and E. The authors report that depletion of DCAF-1 results in an increase in Plk4 levels. However, the actual increase is pretty small, about 1.5 fold for total levels (2A) and approximately 1.2 fold at centrosomes. How can the authors be sure this small increase in Plk4 levels is responsible for the multipolar spindle phenotype reported in figure 2F? It seems to me that CRL4DCAF1 could have other relevant substrates that are responsible for this defect. Related to this, can the authors show that the multipolar spindle phenotype is due to an overproduction of centrioles versus some other defect such as cytokinesis failure? Did the authors examine DCAF-1-depleted cells to if there are cell division defects that could explain the multipolar spindle defect?

    Do the authors know if DCAF1 is operating within the context of the CRL4DCAF1 complex to control Plk4 levels? I know they showed that the entire complex is bound to Plk4 in pull down experiments, but have they tried to deplete other components of CRL4DCAF1 to see if they have the same effect on Plk4 levels?

    Page 5 and Figure 3A. Th authors provide a model where the acidic domain of DCAF1 binds to a groove within the PB1-2 domain of Plk4. This is the same groove that binds CEP192, a protein that cooperates with Cep152 to recruit Plk4 to centrioles. Could it be that DCAF-1, at least in part, is competing with Cep192 and possibly cep152 for binding to Plk4? Thus, in the absence of DCAF1, Cep192 (and possibly Cep152) could recruit more Plk4. Can such a model be ruled out?

    Figure 4. I don't find the results of the in vitro ubiquitin assays all that compelling. Here the authors are fusing DCAF1 to the E2 enzyme and show that this synthetic construct can ubiquitylate Plk4. I wonder in such a system if any protein could be ubiquitylated simply by tethering a binding domain for that protein to an E2 enzyme. So, I guess this is a question of specificity. Is there a control the authors can do to demonstrate specificity in this system?

    Figure 5A and S5A. In figure 5A the authors use a flag-tagged Plk4 pulldown to show that DCAF1 strongly interacts with Plk4 during interphase and weakly during mitosis. In figure S5A, they perform the reverse experiment by pulling down endogenous DCAF1 and state that they obtained similar results. Looking at Figure S5A, this doesn't appear to be true. There is not much difference in the amount of Plk4 pulled down from interphase cells versus mitotic cells. The authors also do not indicate if any of the differences are significant.

    Figure 5B. The authors investigate the cell-cycle-dependent pattern of Plk4 ubiquitination by co-expressing Flag-Plk4, HA-ubiquitin, and Myc-DCAF1 in HEk293 cells followed by a series of Flag IPs from cells arrested at different points in the cell cycle. They claim based on the retarded migration of Plk4, that CRL4DCAF1 ubiquitylates Plk4 specifically during G2 phase. It's hard to make any firm conclusions without quantitation. Furthermore, it's impossible to know how much of the ubiquitylation at any given cell cycle stage is dependent on DCAF1. The correct experiment would have been to have a no DCAF1 control for each cell cycle stage and to quantitate the differences. Since ubiquitin is tagged with HA, would it not be possible to probe the immunoprecipitate with an anti-HA antibody followed by quantitation.

    Figures 6A and 6B. Why do the levels of Plk4 not respond to decreased or increased levels of DCAF1? In 6A for instance strong depletion of DCAF1 does not appear to affect the level of Plk4. Also, given that there is no change in Plk4 levels, the amount of STIL that is pulled down with PLK4 still increases upon DCAF1 knockdown. Does this mean that DCAF1 might function by directly inhibiting the Plk4-STIL interaction.

    Figure 6C The authors find that upon DCAF1 knockdown, centrioles prematurely disengage during G2. They attribute this effect to the increased levels of Plk4. Is there any evidence that increased Plk4 levels lead to premature disengagement? Isn't it possible that this defect is independent of the increase in Plk4 protein? Either the authors should provide evidence of this or offer the possibility that the premature disengagement defect arises independently of the effect on Plk4 levels.

    The authors should also consider exploring the possibility that CRL4DCAF1 functions semi-redundantly with the SCF. It would be interesting to see if there is a synergistic effect of knocking out both E3 ligases on Plk4 levels and centriole number. Such a finding would highlight the importance of the cooperative model the authors propose in this paper.

    Minor issues

    Page 5 typo: "in addition DCAF1 strongly binds to a WD40-acidic motif" I think you meant to say Plk4.

    Figure 4. The terms l.e. and s.e. should be explicitly defined.

    Many figures: Error bars are not defined. Do these represent SD or SE?

    SCF-TrCp is not the only known E3 ligase that controls Plk4 levels. For instance Erich Nigg's group showed some time ago that the E3 ubiquitin ligase Mindbomb (Mib1) also regulates Plk4. (CAjanek et al 2015 J. Cell Sci. 128: 1674-82). This should also be mentioned in the introduction in order to paint a more complete picture of what is known about E3-based regulation of Plk4.

    Significance

    If my criticisms can be successfully addressed, this study has the potential to provide significant new insight into how centriole number is controlled. At least two E3 ligases have already been described that regulate Plk4 levels. This manuscript would provide a third. In an of itself, the discovery of a third E3 involved in the regulation of PLK4 levels would not have a major effect on the field. However if the authors can demonstrate how these two E3s are coordinated to control centriole assembly during the cell cycle that would be a great interest to those studying centriole assembly.

  8. 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 CUL4-DDB1-DCAF1 E3 ubiquitin ligase complex regulates PLK4 protein levels to prevent premature centriole duplication

    Previously, it was thought that PLK4 is mainly regulated by autophosphorylation and degradation by the E3 ligase SCFSlimb/beta-TrCP in a phosphorylation dependent manner. In this manuscript by the Hoffmann group, the authors add an additional layer to the regulation of PLK4 as they identify the CRL4DCAF1 E3 ubiquitin ligase as a regulator of PLK4 that prevents PLK4 accumulation in G2 when beta-TrCP is low and therefore helps to restrict centrosome duplication to one event per cell cycle. More specifically, Grossmann et al. identified CRL4DCAF1 E3 ubiquitin ligase subunits in an immunoprecipitation mass spec approach. Using PLK4 kinase dead and phospho-mutants, they first show that CRL4DCAF1 binding is distinct from the SCFSlimb/beta-TrCP binding site. Depletion of DCAF1 leads to a modest increase in cellular PLK4 levels, PLK4 at centrosomes and cells with supernumerary centrosomes. Based on an IP experiments, they convincingly show that the acid C-terminus of DCAF1 interacts with PLK4 and they provide a model based on AlphaFold and analysis of mutations in the putative interaction interface how PLK4 and DCAF1 interact. They further provide evidences that DCAF1 directly ubiquitinates PLK4 in vitro. The interaction between DCAF1 and PLK4 is cell cycle dependent (peak in G2; Fig. 5) following the ubiquitination of PLK4. In Fig. 6 the authors analyze whether PLK4-STIL interaction is regulated by DCAF1. This is indeed the case and Fig. 6B likely indicates that DCAF1 functions as a competitive inhibitor for PLK4 and in this way blocks PLK4 binding to STIL. Finally, in Fig. 6C the authors analyze centrioles by expansion microscopy. The authors show mother-daughter centriole pair disengagement upon depletion of DCAF1 (on p. 7, bottom: "knockdown of DCAF1 leads to a significant higher number of disengaged centrioles"). This is similar to CEP57 depletion as shown by Kitagawa: JCB 2021 220: e202005153. Instead of analyzing centriole disengagement in further depth, the authors analyze in Fig. 6D centrosome separation, which is mechanistically quite distinct from centriole disengagement. Centrosome separation (mother-daughter pairs) in G2 is triggered by resolution of the rootletin linker through the action of the kinase Nek2A. Thus, Fig. 6 refers to two different events/mechanisms and it will be important to clarify whether DCFA1 depletion causes centriole disengagement or centrosome separation (e.g. by analyzing the centrin pattern and whether daughter centrioles mature). To my knowledge, there is no connection between PLK4/STIL and the centrosome linker. Thus, if DCFA1 regulates centrosome separation, Fig. 6 would be disconnected from the rest of the paper.

    Main points

    1. Fig. 2E: it would make sense to quantify the PLK4 signal at centrioles according to the cell cycle phase of the cell. G2 is probably the cell cycle phase when PLK4 is regulated by the CRL4DCAF1 E3 ubiquitin ligase.
    2. It is known that PLK4 has a function in cytokinesis (i.e.: https://doi.org/10.1073/pnas.181882011). Thus, there is the possibility that the supernumerary centrosomes observed in Fig. 2F result from a cytokinesis defect and not from centriole over-duplication. To address this, the authors can use procentriole marker Sas6, and show that a newly disengaged centriole should still posseses Sas6. When the premature onset of centriole duplication happens to those newly disengaed centrioles, both mother and daughter centriole in the pair should posses Sas6 since Sas6 removal only happens in upcoming mitosis.
    3. The authors suggest a competitive interaction between proteins DCAF1, PLK4 and STIL in Fig. 6A and Fig. 6B. However, they have not excluded direct binding of DCAF1 to STIL as an alternative explanation. Additionally, is the enhanced PLK4/STIL interaction in Fig. 6A G2 dependent?
    4. The quality of the expansion microscopy in Fig. 6C could be improved.
    5. The authors have to resolve whether Fig. 6C and D relate to centriole disengagement or centrosome separation and how this is connected to DCAF1, STIL and PLK4.

    Minor points

    1. P. 5: WD40-Acidic motif. This fragment needs to be described in the text and not just in Fig. S3.
    2. Fig. 2E: The authors analyze the phenotype by combining all data points from three experiments. It would be better to show the average of the three independent experiments and do the statistics on the three data points.
    3. Is the difference (> 4) in Fig. 2F significant?
    4. Fig. 3A-C is difficult to follow. It is too small and DCAF1 and CEP192 are very difficult to see. I am sure that there are simple ways to improve this figure.
    5. Define BP1 and BP2 in Fig. 3A. Does BP1 = PB1?
    6. P. 5. "the first helix (D1420-E1436) of DCAF1 positioned .... (add DCAF1).
    7. P. 20 Fig. 4B: 200 nm should be 200 nM.
    8. The authors may want to test additional PLK4 mutations that are not localized in the predicted interaction interface with DCAF1 to show that these mutations do not affect binding.
    9. In Fig. 4A the authors could IP GFP-PLK4 and show that a fraction of this protein carries His-Ubi conjugation using His antibodies.
    10. Difference in quantification of Fig. 6D is not significant,
    11. Fig. 4B (also Fig. 5B): Explain "l.e." and "s.e." in figure. Both blots are not the same (at least in case of Fig. 4B comparing the kDa numbers), thus l.e. = low exposure and s.e. = short exposure does not work. How was PLK4 detected in Fig. 4B?

    Referees cross-commenting

    I believe that all three reviewers have very similar concerns. I guess, everybody agrees that this manuscript, although potentialy very intresting, needs a substrainail amout of revision

    Significance

    The manuscript convincingly identifies CRL4DCAF1 E3 ubiquitin ligase as an PLK4 regulator and therefore is a very important contribution to the field. However, the impact of DCAF1 depletion is not too high. I therefore recommend double depletion of DCAF1 and SCFSlimb/beta-TrCP (not absolutely necessary but could increase impact). The interaction analysis of DCAF1 with PLK4 and the ubiquitination of PLK4 by the DCAF1 E3 ligase is convincing. I see a problems with the data in Fig. 6 that need to be revised.

    Thus, key experiments that should be done are Main points 2), 3) and 5). The revision of the manuscript will take 3-6 months.

  9. Version published to 10.1101/2023.09.13.555514v1 on bioRxiv
  10. Version published to 10.1101/2023.09.13.555514v2 on bioRxiv