Disengaging the Smc3/kleisin interface releases cohesin from Drosophila chromosomes during interphase and mitosis

Transcription

1 Manuscript EMBO Disengaging the Smc3/kleisin interface releases cohesin from Drosophila chromosomes during interphase and mitosis Christian S. Eichinger, Alexander Kurze, Raquel A. Oliveira, Kim A. Nasmyth Corresponding author: Kim A. Nasmyth, University of Oxford Review timeline: Submission date: 30 August 2012 Editorial Decision: 03 October 2012 Revision received: 05 December 2012 Acceptance letter: 12 December 2012 Accepted: 13 December 2012 Editor: Hartmut Vodermaier Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) 1st Editorial Decision 03 October 2012 After some delay associated with the evaluation of back-to-back submissions, we have now received the feedback of two referees, copied below for your information. As you will see, both referees consider the demonstration of a conserved cohesin exit gate and its role in prophase cohesin release in metazoans important and therefore in principle suited for publication in a broad general journal. Both of them nevertheless demand strengthening of certain aspects of this work, in order to provide the required strong support for the main conclusions. Most of these points seem to be well-taken and easily addressable, however the one major concern of referee 1 may require some more substantial follow-up work to rule out potential alternative explanations for the results in Figure 5. I would therefore like to invite you to respond to the referees' comments through the form of a revised version of the manuscript. Given your recent publication on the cohesin exit gate in yeast, it is my hope that you will be able to resubmit your manuscript in a timely manner, nevertheless I feel it will be primarily important to diligently and thoroughly address the raised concerns during this revision. As per our EMBO Journal editorial policies, related or competing manuscripts published during this revision period will have no negative impact on our final assessment of your revised study. Finally, when revising the manuscript text and organization, I feel it would be important to touch on the Chan et al yeast paper and its key findings already in the introduction section, while at the same time stressing the rationale and conceptual importance of the present, parallel in vivo efforts in a metazoan organism (as you have done in the cover letter). Thank you for the opportunity to consider this work for publication, and please do get back to me should you have any comments or require further clarifications regarding the referee reports and this decision. I look forward to your revision! European Molecular Biology Organization 1

2 REFEREE REPORTS: Referee #1 (Remarks to the Author): In the manuscript by Eichinger et al, the authors investigate whether disengagement of the Smc3- Rad21 interface is necessary for cohesin release during interphase and mitosis in Drosophila. The requirement of this process has been recently demonstrated for the turnover of cohesin at pericentic chromatin during mitosis in S. cerevisiae by the same group (Chan 2012). I believe that the current study provides further proof of the existence of an exit gate for cohesin in a metazoan organism and, moreover, shows its importance for cohesin dynamics during interphase and for the prophase pathway. I have enjoyed reading this paper and watching the very nice movies. Thus, if the authors can address the one major concern that I explain below, I would support its publication in Embo J. My major concern refers to the result in Figure 5. Cleavage of the peptide linking Smc3-Rad21-GFP leads to extensive release of cohesin complexes. This should not be the case if these complexes were behaving as the endogenous. My concern is then that the over expressed Smc3-Rad21-GFP heterodimer binds to chromatin through the Smc3 subunit but does not form a bonafide cohesin complex and thus does not turn over (figure 4). In this scenario, when TEV is injected, Rad21-GFP is released (figure 5). Similarly, the heterodimer would persist on chromatin though prophasemetaphase (figure 6D) and a Rad21-GFP fragment would be released upon separase cleavage in anaphase. To rule out this possibility, I would like to see: 1. Immunoprecipitation with anti-gfp from salivary gland extracts (after induction of Smc3-Rad21- GFP) brings down not only Rad21 and Smc3, but also Smc1, Scc3, Pds5 and Wapl. 2. Repeat the experiment with the GFP tag in Smc3 instead of Rad Demonstrate by an alternative method that cohesin complexes containing Smc3-Rad21-GFP remain on chromatin after TEV injection (isolation of salivary gland chromosomes after TEV injection and analysis of chromatin by western blot). 4. Show that dissociation of GFP labelled complexes from salivary gland chromosomes upon TEV injection depends on Wapl. Minor points: -page 7, "In the salivary glands, we found comparable levels of the fusion protein..." It is not clear what levels are being compared. Please clarify. In the salivary glands there seems to be a huge overexpression of the transgene with respect to endogenous protein. - In the last experiment regarding the prophase pathway (Figure 6D), the authors indicate "cohesin containing an intact Smc3-Rad21-GFP fusion protein persisted...until the onset of anaphase." Why is there not a movie showing this? European Molecular Biology Organization 2

3 Additional suggestions: -In the Introduction: -The prophase pathway was initially identified in Xenopus (Losada 1998) and the involvement of mitotic kinases was also first described by Sumara 2002 and Losada In page 3 of Introduction, we read "Wapl is recruited to cohesin by binding...pds5, which..." and Chan KL, personal communication is cited. I do not think this is adequate for the Introduction. Maybe for the first part of the sentence Chan et al (2012) should be cited instead, although it should be also mentioned that the requirement of Pds5 for Wapl recruitment does not exist in Xenopus (Shintomi 2009). - Results By the end of the first paragraph in page 9, Kueng 2006 should be cited: In that study it was already shown by live cell imaging in HeLa cells that despite failure of the prophase pathway in Wapl sirna cells, all cohesin was released in anaphase. -Figure Legends. Although I am no fan of long Figure legends, I would encourage the authors to provide some more info in Figures 5 and 6 to allow the reader to understand the figure independently of the main text. Also, magnification bars should be added. Referee #2 (Remarks to the Author): Experiments in yeast have suggested that cohesin undergoes DNA entrapment and release dynamically, through transient opening of Smc1/3 interface and Smc3/alpha-kleisin (Scc1) interface, respectively. It is well known that DNA is released from cohesin by the proteolytic cleavage of kleisin upon anaphase onset, but the proteolysis-independent dissociation of cohesin in prophase/prometaphase through what it is called prophase pathway is not well understood. This paper eloquently addresses the long-standing question of how cohesin is released nonproteolytically. The authors made use of live cell imaging analyses in Drosophila non-dividing salivary gland cells and showed that cohesin turnover in these interphase cells, which required proficient Wapl. Thus, in Wapl mutant cells, cohesin over-enriched at chromosomal loci revealed characteristic structures, but artificial cleavage of kleisin Rad21 caused immediate dissociation of cohesin from chromatin. This finding significantly implied that cohesins that are not conferring cohesion during interphase associates to chromosomes in a topological manner. To my knowledge this is the first demonstration that cohesin association is basically topological. The authors then tested the idea that DNA might escape from the cohesin ring through the transient disconnection of the Smc3-Rad21 interaction, in a manner depending on Wapl, the hypothesis based on yeast works. To address this the authors tethered Smc3 and Rad21 heads by short polypeptide links, which can be artificially cleaved by TEV protease. Cohesin ring with Smc3-Rad21 fusion caused over-enrichment of the complex on chromatin, as seen in Wapl mutant, but this was released by cleaving the tethering peptide. Crucially, the authors described that not only on polytene chromosomes in salivary gland cells but also in prophase in neuroblasts, Wapl-promoting cohesin release is blocked by fusing Smc3 and kleisin. Based on these results the authors concluded that proteolysis-independent release of cohesin from chromatin is universally mediated by escape of DNA through Smc3/Rad21 gate. I found the paper provide important results indicating how cohesin complexes are associated with chromatin and how most of them are released before anaphase. It is true that similar lines of conclusions are already drawn in yeast, nevertheless I do see that the current study goes beyond the yeast studies, as explicitly described in the discussion. It is a significant step toward our understanding of cohesin regulation and clearly deserves for the publication. That said, I have several concerns that might taken into consideration, which should be addressed without much difficulty. European Molecular Biology Organization 3

4 1. Provided that Wapl-promoted release of chromatin involves DNA's escape through the Smc3- kleisin gate, I am curious to know if and to which extent the TEV-induced release of Smc3-Rad21- GFP from polytene chromosomes depends on Wapl in this experimental setting (Figure 5). If the release is primarily driven by Wapl, what one would expect to see in a Wapl mutant background is a marked enrichment of Smc3-Rad21-GFP on chromosomes, which do not grossly affected after microinjecting the active TEV. Is it the case? These results would be in sharp contrast to the experiment cleaving Rad21 (Figure 2E). 2. Smc3-Rad21-GFP fusion protein is found to accumulate at chromosomal loci that result in emergence of bands or circular structure appearance (Figure 3D). To estimate the net effect of the tethering of Smc3 and Rad21, it is informative to show side-by-side the picture of Smc3-Rad21-GFP when TEV was co-expressed. As for the Western blots in Figure 3, antibodies used for the analyses are missing. In the blots in Figure 3C,, is it possible to explain why endogenous Rad21 band is not detectable in Smc-Rad21-GFP lanes? 3. FRAP analysis in Figure 4 shows the turnover rate of Smc3-Rad21-GFP with our without the Smc3/Rad21 tethering. The authors' interpretation of the data for +TEV kinetics is "similar to wild type", but the fluorescence recovery seems to be consistently lower (60-70% of RFI) than that of wild type which is shown in Figure 2C. It seems to imply that there is more stably bound fraction of non-tethered Smc3-Rad21-GFP than Rad21-GFP. What could be a possible explanation for this? My suggestion is to compare the half-recovery time here instead, and say it is similar to wild-type, if that is the case. 4. The supplementary movie data in neuroblast provide a unique opportunity to show that the dissociation of cohesin from chromosome arms in prophase and from centromeres in anaphase, and that in Wapl mutants considerable amount of cohesin remains throughout the chromosome lengths until anaphase (Movie S10 and S11). It will be helpful to additionally provide the movie data for Smc3-Rad21-GFP with or without TEV, because it is difficult to tell that cohesin remains on arms solely from a still image (Figure 6D). For more comprehensive presentation for Figure 6D, drawings of schematic illustration may help. 5. The paper provides mechanistic explanation for the first time for how prophase pathway might work to release cohesin from chromosomes. It has long been known that the prophase pathway involves activity of mitotic kinases such as Plk1 and phosphorylation of Scc3/SA2. Therefore a perspective view for how these signals might contribute to promote Wapl-mediated cohesin release would be interesting to discuss. 1st Revision - authors' response 05 December 2012 Regarding editor s remarks: In the introduction of our new manuscript, we describe briefly the parallel work of Chan et al. and the rational and conceptual importance of the present manuscript. Also in the discussion, the conclusions of the parallel yeast study is put in relation to our work. Referee #1 (Remarks to the Author): In the manuscript by Eichinger et al, the authors investigate whether disengagement of the Smc3- Rad21 interface is necessary for cohesin release during interphase and mitosis in Drosophila. The requirement of this process has been recently demonstrated for the turnover of cohesin at pericentic chromatin during mitosis in S. cerevisiae by the same group (Chan 2012). I believe that the current study provides further proof of the existence of an exit gate for cohesin in a metazoan organism and, moreover, shows its importance for cohesin dynamics during interphase and for the prophase pathway. I have enjoyed reading this paper and watching the very nice movies. Thus, if the authors can address the one major concern that I explain below, I would support its publication in Embo J. European Molecular Biology Organization 4

5 My major concern refers to the result in Figure 5. Cleavage of the peptide linking Smc3-Rad21- GFP leads to extensive release of cohesin complexes. This should not be the case if these complexes were behaving as the endogenous. My concern is then that the over expressed Smc3-Rad21-GFP heterodimer binds to chromatin through the Smc3 subunit but does not form a bonafide cohesin complex and thus does not turn over (figure 4). In this scenario, when TEV is injected, Rad21-GFP is released (figure 5). Similarly, the heterodimer would persist on chromatin though prophasemetaphase (figure 6D) and a Rad21-GFP fragment would be released upon separase cleavage in anaphase. We agree that referee #1 s major concern whether or not the Smc3-Rad21-GFP fusion protein is incorporated into a bona fide cohesin ring needs to be clarified. For this, we carried out two experiments as suggested by referee 1: 1) First and foremost, we set up a fly cross that generates larvae, which express the Smc3-Rad21- GFP fusion protein in a wapl C204 mutant background and injected TEV protease after the fusion protein has been loaded onto chromatin. This experiment clearly demonstrated that Rad21-GFP is not released from Smc3 merely by cleaving the linker between the two proteins. Crucially, release also depends on the cohesin-associated protein Wapl. This experiment is now described by the revised version as Figure 5C and Supplementary Movie S10 and S11 (Movies S10 and S11 from our first submission are now named Movies S12 and S13, respectively). 2) Second, we carried out a Co-Immunoprecipitation experiment showing that the Smc3-Rad21- GFP fusion protein does interact with endogenous Smc1 but not with endogenous Rad21, thus showing that the Smc3-Rad21-GFP fusion protein is incorporated into a bona fide cohesin complex. The result of this experiment is included in the revised manuscript as a new Supplementary Figure S1. Further details on each of referee 1 s comments are described in the following: To rule out this possibility, I would like to see: 1. Immunoprecipitation with anti-gfp from salivary gland extracts (after induction of Smc3-Rad21- GFP) brings down not only Rad21 and Smc3, but also Smc1, Scc3, Pds5 and Wapl. We carried out an immunoprecipitation experiment (using anti-gfp beads) of chromatin-bound proteins from larval extracts after induction of Smc3-Rad21-GFP and tested binding for endogenous Smc1 and endogenous Rad21 using available antibodies against Drosophila Rad21 and Smc1. In our attempts to use salivary gland extracts we did not obtain enough material to purify chromatin-bound proteins and to carry out subsequent co-immunoprecipitation. However, using third instar larval extracts, we found that Smc3-Rad21-GFP interacts with endogenous Smc1, but not endogenous Rad21, indicating that chromatin-bound Smc3-Rad21-GFP forms an intact cohesin ring with endogenous Smc1 and does not incorporate another endogenous Rad21. The data of this experiment are shown in the new Supplementary Figure Repeat the experiment with the GFP tag in Smc3 instead of Rad21. Unfortunately, we do not have transgenic flies where an Smc3-Rad21 fusion protein is tagged at the N-terminus of Smc3 with GFP. Thus, in order to do this experiment, the generation of new transgenic flies would be necessary. This is unfortunately not doable in a reasonable amount of time. We believe however that our new experimental data described in referee 1 s points 1 and 4 show strong evidence against his/her major concern and indicates that Smc3-Rad21-GFP forms bona-fide cohesin complexes and that the release of Rad21-GFP from chromatin after cleaving the linkage between Smc3 and Rad21 depends on Wapl. 3. Demonstrate by an alternative method that cohesin complexes containing Smc3-Rad21-GFP remain on chromatin after TEV injection (isolation of salivary gland chromosomes after TEV injection and analysis of chromatin by western blot). Although we agree that this experiment would be very informative, it is unfortunately technically European Molecular Biology Organization 5

6 impossible to do for two reasons: 1) Proteins or mrnas are injected directly into the cytoplasm of only a single or of very few cells of a salivary gland in order to keep the tissue and its physiology intact. For this experiment however, one would need to inject every single cell in the tissue (one salivary gland contains around 120 cells). 2) Although live imaging after injection is possible without perturbing the physiology of the gland, it is technically very difficult to recover an injected intact gland reliably from the injection chamber for Western Blot analysis. 4. Show that dissociation of GFP labelled complexes from salivary gland chromosomes upon TEV injection depends on Wapl. We agree that this is a very important biological question that can be addressed using our expertise. We therefore set up a fly cross that generates progeny, which expresses the Smc3-Rad21-GFP protein upon heat-shock in a wapl C204 mutant background. We injected salivary gland cells with TEV protease in exactly the same manner as in a wildtype background. In a wildtype background, the fusion protein is released from chromatin within minutes (Figure 5A and B; Supplementary Movies S8 and S9). In sharp contrast, our new experiments clearly show that there is no release or turnover (measured by FRAP) of Smc3-Rad21-GFP seen in a wapl C204 mutant upon TEV protease injection (Figure 5C; Supplementary Movies S10 and S11). This indicates that separaseindependent cohesin release happens via opening of the Smc3/kleisin interface and that this process is mediated by the protein Wapl. This experiment also proves that Rad21-GFP is not released from Smc3 only by cleaving the linker between the two proteins, but that the process also depends on the cohesin-associated protein Wapl. Minor points: - page 7, "In the salivary glands, we found comparable levels of the fusion protein..." It is not clear what levels are being compared. Please clarify. In the salivary glands there seems to be a huge overexpression of the transgene with respect to endogenous protein. Yes, the Smc3-Rad21-GFP protein is indeed overexpressed in salivary glands as compared to the endogenous level at the time and condition at which the experiment was carried out and we therefore corrected the text accordingly. - In the last experiment regarding the prophase pathway (Figure 6D), the authors indicate "cohesin containing an intact Smc3-Rad21-GFP fusion protein persisted...until the onset of anaphase." Why is there not a movie showing this? We had many attempts to analyse the behavior of the fusion in live brains, however getting highquality time-lapse movies is technically very challenging due to low expression or incorporation of Smc3-Rad21-GFP in addition to the endogenous cohesin complex. This difficulty may result from the fact that there is not a constant turnover of cohesin in neuroblasts as compared to salivary glands where the Smc3-Rad21-GFP fusion protein is highly expressed, rapidly loaded onto chromatin and presumably replaces most of the dynamic endogenous cohesin. In order to get solid expression and chromatin binding of the fusion protein, we were therefore using conditions which do not allow time-lapse/live imaging analysis. Namely, we carried out heat-shock induction and recovery in live animals, thus ensuring best culture conditions. Afterwards, we dissected and analyzed brain tissues and imaged neuroblasts, which have undergone nuclear envelope breakdown but not metaphase-toanaphase transition. We also used poly-lysine-coated slides to improve the still imaging quality, which again is of disadvantage for live imaging as it interferes with proper cell divisions. Due to these experimental limitations, we changed the text accordingly to soften our conclusion to: This revealed that cohesin containing an intact Smc3-Rad21-GFP fusion protein persisted on chromosome arms after nuclear envelope breakdown. Additional suggestions: - In the Introduction: - The prophase pathway was initially identified in Xenopus (Losada 1998) and the involvement of mitotic kinases was also first described by Sumara 2002 and Losada The citations were added accordingly in the introduction of our revised manuscript. - In page 3 of Introduction, we read "Wapl is recruited to cohesin by binding...pds5, which..." and European Molecular Biology Organization 6

7 Chan KL, personal communication is cited. I do not think this is adequate for the Introduction. Maybe for the first part of the sentence Chan et al (2012) should be cited instead, although it should be also mentioned that the requirement of Pds5 for Wapl recruitment does not exist in Xenopus (Shintomi 2009). We changed the text accordingly in our revised manuscript. - Results By the end of the first paragraph in page 9, Kueng 2006 should be cited: In that study it was already shown by live cell imaging in HeLa cells that despite failure of the prophase pathway in Wapl sirna cells, all cohesin was released in anaphase. We changed the text accordingly in our revised manuscript. - Figure Legends. Although I am no fan of long Figure legends, I would encourage the authors to provide some more info in Figures 5 and 6 to allow the reader to understand the figure independently of the main text. Also, magnification bars should be added. We changed the figure legends accordingly in our revised manuscript. Referee #2 (Remarks to the Author): Experiments in yeast have suggested that cohesin undergoes DNA entrapment and release dynamically, through transient opening of Smc1/3 interface and Smc3/alpha-kleisin (Scc1) interface, respectively. It is well known that DNA is released from cohesin by the proteolytic cleavage of kleisin upon anaphase onset, but the proteolysis-independent dissociation of cohesin in prophase/prometaphase through what it is called prophase pathway is not well understood. This paper eloquently addresses the long-standing question of how cohesin is released nonproteolytically. The authors made use of live cell imaging analyses in Drosophila non-dividing salivary gland cells and showed that cohesin turnover in these interphase cells, which required proficient Wapl. Thus, in Wapl mutant cells, cohesin over-enriched at chromosomal loci revealed characteristic structures, but artificial cleavage of kleisin Rad21 caused immediate dissociation of cohesin from chromatin. This finding significantly implied that cohesins that are not conferring cohesion during interphase associates to chromosomes in a topological manner. To my knowledge this is the first demonstration that cohesin association is basically topological. The authors then tested the idea that DNA might escape from the cohesin ring through the transient disconnection of the Smc3-Rad21 interaction, in a manner depending on Wapl, the hypothesis based on yeast works. To address this the authors tethered Smc3 and Rad21 heads by short polypeptide links, which can be artificially cleaved by TEV protease. Cohesin ring with Smc3-Rad21 fusion caused over-enrichment of the complex on chromatin, as seen in Wapl mutant, but this was released by cleaving the tethering peptide. Crucially, the authors described that not only on polytene chromosomes in salivary gland cells but also in prophase in neuroblasts, Wapl-promoting cohesin release is blocked by fusing Smc3 and kleisin. Based on these results the authors concluded that proteolysis-independent release of cohesin from chromatin is universally mediated by escape of DNA through Smc3/Rad21 gate. I found the paper provide important results indicating how cohesin complexes are associated with chromatin and how most of them are released before anaphase. It is true that similar lines of conclusions are already drawn in yeast, nevertheless I do see that the current study goes beyond the yeast studies, as explicitly described in the discussion. It is a significant step toward our understanding of cohesin regulation and clearly deserves for the publication. That said, I have several concerns that might taken into consideration, which should be addressed without much difficulty. 1. Provided that Wapl-promoted release of chromatin involves DNA's escape through the Smc3- kleisin gate, I am curious to know if and to which extent the TEV-induced release of Smc3-Rad21- GFP from polytene chromosomes depends on Wapl in this experimental setting (Figure 5). If the release is primarily driven by Wapl, what one would expect to see in a Wapl mutant background is a marked enrichment of Smc3-Rad21-GFP on chromosomes, which do not grossly affected after European Molecular Biology Organization 7

8 microinjecting the active TEV. Is it the case? These results would be in sharp contrast to the experiment cleaving Rad21 (Figure 2E). This point raises the same question as referee 1 (see before) and we agree that this is a very important biological question that can be addressed using our expertise. We therefore set up a fly cross that generates progeny, which expresses the Smc3-Rad21-GFP protein upon heat-shock in a wapl C204 mutant background. We injected salivary gland cells with TEV protease in exactly the same manner as in a wildtype background. In a wildtype background, the fusion protein is released from chromatin within minutes (Figure 5A and B; Supplementary Movies S8 and S9). In sharp contrast, our new experiments clearly show that there is no release or turnover (measured by FRAP) of Smc3-Rad21-GFP seen in a wapl C204 mutant upon TEV protease injection (Figure 5C; Supplementary Movies S10 and S11). This indicates that separase-independent cohesin release happens via opening of the Smc3/kleisin interface and that this process is mediated by the protein Wapl. This experiment also proves that Rad21-GFP is not released from Smc3 only by cleaving the linker between the two proteins, but that the process also depends on the cohesin-associated protein Wapl. Moreover, it seem that Wapl-dependent release affects the entire cohesin population in Drosophila salivary glands. 2. Smc3-Rad21-GFP fusion protein is found to accumulate at chromosomal loci that result in emergence of bands or circular structure appearance (Figure 3D). To estimate the net effect of the tethering of Smc3 and Rad21, it is informative to show side-by-side the picture of Smc3-Rad21-GFP when TEV was co-expressed. As for the Western blots in Figure 3, antibodies used for the analyses are missing. In the blots in Figure 3C, is it possible to explain why endogenous Rad21 band is not detectable in Smc-Rad21-GFP lanes? The net effect of tethering Smc3 and Rad21 compared to its co-expression with TEV protease is basically shown in Figure 4A upper and lower panel. The key difference between the two conditions is the observation of less pronounced circular structures regarding the cohesin localization pattern. We added a description of antibodies in Figure legend 3 as well as in the methods part in our revised manuscript. We can only speculate on why the endogenous Rad21 levels may be lower in the Smc3-Rad21-GFP without TEV co-expression as compared to co-expressed TEV. Given our newly added experiments (Cleavage of the linker in the Smc3-Rad21-GFP fusion protein by TEV injection and Coimmunoprecipitation of Smc3-Rad21-GFP), we strongly believe that the Smc3-Rad21-GFP protein occupies most of the possible cohesin binding sites (without turning over) and that endogenous Rad21 remains mainly in the soluble pool where it could be more prone to degradation. 3. FRAP analysis in Figure 4 shows the turnover rate of Smc3-Rad21-GFP with our without the Smc3/Rad21 tethering. The authors' interpretation of the data for +TEV kinetics is "similar to wild type", but the fluorescence recovery seems to be consistently lower (60-70% of RFI) than that of wild type which is shown in Figure 2C. It seems to imply that there is more stably bound fraction of non-tethered Smc3-Rad21-GFP than Rad21-GFP. What could be a possible explanation for this? My suggestion is to compare the half-recovery time here instead, and say it is similar to wild-type, if that is the case. The expression similar to wild-type may indeed have been not accurate and therefore misleading. The only main point we wanted to make here is the observation that co-expression of TEV together with the Smc3-Rad21-GFP fusion leads to a very significant portion of fluorescence recovery (around 70% of RFI). The fact that recovery does not go back to the level seen in a wild-type (Figure 1C) can have several reasons: 1) TEV protease may need time to access and cleave the linker, 2) TEV protease concentration is lower compared to TEV injected cells, 3) Some fusion protein is present in a non-cleaved version which presumably represents a fraction of around 20 % as indicated by Western blot (Figure 3C). 4. The supplementary movie data in neuroblast provide a unique opportunity to show that the dissociation of cohesin from chromosome arms in prophase and from centromeres in anaphase, and that in Wapl mutants considerable amount of cohesin remains throughout the chromosome lengths European Molecular Biology Organization 8

9 until anaphase (Movie S10 and S11). It will be helpful to additionally provide the movie data for Smc3-Rad21-GFP with or without TEV, because it is difficult to tell that cohesin remains on arms solely from a still image (Figure 6D). For more comprehensive presentation for Figure 6D, drawings of schematic illustration may help. A similar point has also been raised by referee 1 (see before). We indeed had many attempts to analyse the behavior of the fusion in live brains, however getting high-quality time-lapse movies is technically very challenging due to low expression or incorporation of Smc3-Rad21-GFP in addition to the endogenous cohesin. This difficulty may result from the fact that there is not a constant turnover of cohesin in neuroblasts as compared to salivary glands where the Smc3-Rad21-GFP fusion protein is highly expressed, rapidly loaded onto chromatin and presumably replaces most of the dynamic endogenous cohesin. In order to get solid expression and chromatin binding of the fusion protein, we were therefore using conditions which do not allow time-lapse/live imaging analysis. Namely, we carried out heat-shock induction and recovery in live animals, thus ensuring best culture conditions. Afterwards, we dissected and analyzed brain tissues and imaged neuroblasts, which have undergone nuclear envelope breakdown but not metaphase-to-anaphase transition. We also used poly-lysine-coated slides to improve the still imaging quality, we again is of disadvantage for live imaging as it interferes with proper cell divisions. Due to these experimental limitations, we changed the text accordingly to soften our conclusion to: This revealed that cohesin containing an intact Smc3-Rad21-GFP fusion protein persisted on chromosome arms after nuclear envelope breakdown. 5. The paper provides mechanistic explanation for the first time for how prophase pathway might work to release cohesin from chromosomes. It has long been known that the prophase pathway involves activity of mitotic kinases such as Plk1 and phosphorylation of Scc3/SA2. Therefore a perspective view for how these signals might contribute to promote Wapl-mediated cohesin release would be interesting to discuss. We have added a comment in the discussion on how SA/Scc3-P and Plk1 may contribute to the Wapl-mediated cohesin release in prophase and changed text accordingly in our revised manuscript. Acceptance letter 12 December 2012 Thank you for submitting your revised manuscript for our consideration. I have now had a chance to look through it and your responses, and referee 1 has also taken another look at the study and found their concerns satisfactorily addressed. I am therefore happy to inform you that we have decided to accept your manuscript for publication in The EMBO Journal at this stage. Referee #1 (Remarks to the Author) I read the revised manuscript and the rebuttal letter. I am satisfied with the revisions made and the additional data included so from my part it is OK to publish it. European Molecular Biology Organization 9