CKIP-1 couples Smurf1 ubiquitin ligase with Rpt6 subunit of proteasome to promote substrate degradation

Transcription

1 Manuscript EMBO CKIP-1 couples Smurf1 ubiquitin ligase with Rpt6 subunit of proteasome to promote substrate degradation Yifang Wang, Jing Nie, Yiwu Wang, Luo Zhang, Guichun Xing, Ping Xie, Fuchu He and Lingqiang Zhang Corresponding author: Lingqiang Zhang, Beijing Institute of Radiation Medicine Review timeline: Submission date: 21 January 2012 Editorial Decision: 29 February 2012 Resubmission: 25 June 2012 Editorial Decision: 16 August 2012 Revision received: 24 August 2012 Accepted: 07 September 2012 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 29 February 2012 Thank you for your patience while your study has been under peer-review at EMBO reports. Please accept my apologies for the unusually long time that we have needed to make a decision regarding your study, as we received one of the referee reports only today. As you will see from the enclosed three reports, although the referees acknowledge the potential interest of the findings, they find the study is far from acceptable in our journal at this point. There are numerous technical concerns, and many issues regarding the conclusiveness of the data as well as its physiological relevance. In all, the referees have rated the work as of poor technical quality and referees 1 and 2 the general interest as medium in the summary evaluation sheets returned with their reports Given these opinions and the fact that, due to pressure for space, EMBO reports can only invite revision of papers that receive enthusiastic support from a majority of referees, I am afraid that we cannot offer to publish your manuscript. We would, however, be willing to consider a new manuscript on the same topic that successfully addressed all of the referee concerns in full, if the outcome of such effort were support your current model. To be completely clear, I would like to stress that such a manuscript would be treated as a new submission and would be reviewed afresh. It would only be sent back to the referees if the findings were still novel at the time of submission and all referee concerns were fully and satisfactorily addressed through experimentation, so I would advise against this option unless your are willing to thoroughly rework your study. European Molecular Biology Organization 1

2 I am sorry to bear these negative news, especially after such a protracted process, and hope that the referee comments are helpful in your continued work in this area. Yours sincerely, Editor EMBO Reports Referee #1: The authors previously identified CKIP1 as a positive regulator of the Smurf1 E3 ligase. In this current manuscript, the authors used both biochemical and cellular approaches to demonstrate that CKIP1 could interact with the Rpt6 ATPase subunit of 19S regulatory particle of the proteasome, through which CKIP1 could govern the stability of Smurf1 as well as Smurf1 downstream substrates. The paper is clearly written, and for most part of the manuscript, the experiments are well thought-out and performed to address the points. However, there are several concerns that prevent this manuscript from being published in EMBO Report at its present form. Most importantly, the authors did not provide the biological significance for Rpt6-mediated regulation of the Smurf1 signaling pathway. In their previous paper, they showed that depletion of CKIP1 affects osteoblast differentiation process primarily by affecting Smurf1 activity. Thus it is critical to further show whether manipulating Rpt6 could also affect Smurf1 pathway to influence osteoblast differentiation. Furthermore, in some cases, critical controls are necessary to clarify their statements and further experimental evidences are required to support the proposed model. Below please find some specific comments. 1. Figure 1C, it will be nice to illustrate the interaction between endogenous CKIP1 and endogenous Rpt6. 2. Figure S1 and Table 1. Through mass spect analysis of CKIP-associated proteins, the authors identified a long list of 26S proteasome subunits including Rpt2, Rpt3 and Rpn10. However, later co-immunoprecipitation expression experiment showed that CKIP1 only interacts with Rpt1 and Rpt6 but not other components. Although the authors claimed that CKIP1 may only interacts directly with Rpt1 and Rpt6, but indirectly with other components, experimental evidence is required to support this notion. GST-pull down assays should be performed to show that CKIP1 only directly interacts with Rpt1 and Rpt6. Furthermore, it is interesting to note that later on the authors showed that Smurf1 could interact with Rpt6 in co-ip but not in GST-pull down assays, and CKIP serves as a bridging adaptor to mediate Smurf1/Rpt6 interaction. Thus, it is difficult to understand why under co-ip conditions, endogenous Rpt6 cannot serve as bridging module to mediate the interaction between CKIP and other 26S component Rpt2 and Rpt3? If not, whether overexpression of Rpt6 could enhance the interaction between CKIP and Rpt2 and/or Rpt3. 3. Figure 1, a related question, what is the important role of CKIP1 interaction with Rpt1 in addition to Rpt6? 4. Figure 3B-C, the brackets for these two figure panels are mislabeled. 5. Figure 3C, as the GST-pull down results are different from the co-ip experiments in assessing which domain of CKIP is critical to mediate the interaction with CKIP1, it is critical to repeat the interaction mapping between Rpt6 and CKIP1 truncation in the co-ip setting. Furthermore, it is also important to map the critical domain within CKIP1 to interact with Rpt1. 6. Figure 4A, as Rpt6 is a critical component of the 19S regulatory particle of the proteasome, it is reasonable to assume that depletion of Rpt6 will lead to stabilization of various unstable proteins. Therefore, in addition to Smurf1, other proteins such as p27, cyclin E and cyclin B should be included as negative controls to show whether depletion of Rpt6 only affects Smurf1, or affects a wide spectrum of protein half-lives. In addition, to build up a causal relationship between Rpt6 and Smurf1, it is critical to re-introduce WT-Rpt6 and Rpt6 ( ) that does not interact with CKIP1, into Rpt6-depleted cells. It is anticipated that WT, but not mutant Rpt6 should restore Smurf1 levels. 7. Figure 4B, as this work primarily addresses the importance of Rpt6 in regulating Smurf1 downstream substrates, in addition to monitoring the steady state levels of various Smurf1 substrates, pulse chase experiments should be performed to assess the changes in their half-lives as read-outs of Smurf1 activity changes. 8. Figure 4, sirpt1 and SiRpt1/siRpt6 should be performed to see whether there are synergistic roles between Rpt1 and Rpt6 in regulating Smurf1, and more importantly, in governing the downstream biological outputs including osteoblast differentiation. European Molecular Biology Organization 2

3 9. Figure 4, the role of CKIP1 as an adaptor to recruit ubiquitinated Smurf1 and its substrates to 26S proteasome should be further evaluated by double depletion experiment. It will be important to examine whether there is no further stabilization of Smurf1 and its substrates when Rpt6 is depleted in CKIP1-/- MEFs or CKIP1 knocking down cells. 10. Figure 4, according to the author's model, CKIP1 facilitates Smurf1 auto-ubiquitination and its substrates' turnover through two possible mechanisms, either by inducing Smurf1-substrate interaction or by recruiting them to the 26S proteasome via Rpt6 subunit. Thus it will be nice to dissect these two distinct mechanisms by re-introducing different CKIP1 mutants to CKIP1-depleted cells. The author will expect to see the re-introducing of N-terminal CKIP1 will only partially rescue the CKIP1 depletion-caused Smurf1 stabilization but the re-introducing of LZ domain of CKIP1, which cannot bind Smurf1, will not affect Smurf1 levels. Referee #2: The manuscript titled "CKIP-1 couples Smurf1..." by Wang et al details primarily binding experiments that map domains of interaction between Rpt6 and CKIP-1. Rpt6 was identified as a binding partner of CKIP-1 by two hybrid, and mass spec analysis. Prior studies by the authors had identified the Smurf1 ubiquitin ligase as a CKIP-1 interacting protein. In the current study, the authors determine that Rpt6 does not interact with Smurf1 in vitro, but in vivo the interaction was shown to be mediated in part by CKIP-1. Functional consequences of perturbing these interactions were explored in Figure 4. Major Comments: 1. It is unclear at present if Rpt6 is interacting with CKIP1 in vivo as part of an intact 26S proteasome, or a 19S or Base sub-particle. The mass spec data identifies proteasomal subunits, but immunoblotting data limits interaction of CKIP-1 to Rpt6 and Rpt1. Intact 26S proteasome assemblies can be immunoprecipitated from lysates by using Mg-ATP in the buffers; from the very limited Methods section, there is nothing to suggest that nucleotides were included. 2. Additionally, the proteasome is abundant and good antibodies to subunits are available from researchers, as well as commercially. The authors should immunoblot for endogenous proteasome subunits in their CoIP experiments, rather than transfecting, and overproducing single subunits. 3. In Fig. 4F, the degradation of endogenous Smurf1 is assayed. When Rpt6 is expressed ectopically, the half-life of Smurf1 is further decreased. This suggests that either 26S proteolytic activity is limiting in these cells, or that Rpt6 functions as an adaptor, separate from the 26S entity. The latter hypothesis would be predicted because overproducing one subunit of a hexamer is unlikely to cause an overall increase in proteolytic capacity. This issue can be addressed by measuring peptidase activity from un-transfected, and Rpt6-transfected cells. Or by immunoblotting for Ub conjugate levels. This would shed further mechanistic insight. 4. As mentioned above, Method descriptions are minimal. Figure Legends are equally cryptic. For example, in Figures 4A, 4C, 4D, increasing concentrations of CKIP-1 or Rpt6 are being transfected. At the very least, the concentation range of the plasmids should be given for reproducibility. In Fig. 4A, Lanes 1 and 4 are contrasted for Smurf1 levels. However, the levels of CKIP1 are nearly identical between Lanes 3 and 4. It would be good to state in the figure legend whether this experiment is representative of x # of experiments. Minor points: 1. The Abstract states "Depletion of CKIP-1 significantly reduces the affinity of Smurf1 and Rpt6". This statement implies that binding constants were measured. Instead, the authors are referring to their immuno-precipitation data. The abstract should be reworded to include some of their functional data from Fig The text is unclear due to grammatical errors and improper sentence construction. Referee #3: The goal of the manuscript under review is to demonstrate a novel mechanism of targeting of a set of proteins for degradation by the proteasome. In this mechanism CKIP-1 serves as a bridging adaptor that binds to the proteasome and to the ubiquitin protein ligase Smurf1, which in turn binds to its substrates. These interactions lead to the degradation of Smurf1 and its substrates. European Molecular Biology Organization 3

4 I think that the proposed mechanism is quite interesting but the manuscript does not yet provide convincing experimental evidence in its support. The manuscript provides good evidence for the interaction between the proteasome, CKIP-1, and Smurf1 and maps the interactions to specific domains within the proteins. As the authors report, they had previously defined different domains for the interaction of the three components but they provide a plausible explanation why the current manuscript comes to different conclusions. One concern here is that in some experiments the interaction between CKIP-1 and the proteasome does not map to a single subunit as would be expected for a specific adaptor protein. This observation should be discussed. Most importantly, the relevance of the interactions to degradation is poorly supported and this part of the manuscript would have to be very significantly improved before publication can be considered. My biggest concern regards the design of the experiments purported to show that the interaction CKIP-1 and the proteasome subunit Rpt6 is required for degradation of Smurf1. In these experiments the expression of Rpt6 is up and down regulated and corresponding effects on Smurf1 degradation are determined. Since Rpt6 is a key stoichiometric component of the proteasome, it is not surprising that reducing its expression will compromise proteasome function. This observation cannot be taken as evidence for a specific role of Rpt6 in the process under investigation here. Furthermore, I am surprised that overexpression of Rpt6 enhances degradation of Smurf1. If anything, I would have expected it to decrease degradation of Smurf1 as excess Rpt6 that cannot be integrated into proteasome particles would compete CKIP-1 and Smurf1 away from the proteasome. Other suggestions: I recommend explaining the motivation behind the proposed mechanism better. Several times, the manuscript raises the question as to how Smurf1 and its substrates are degraded by the proteasome and then goes on to discuss how all the substrates under discussion are ubiquitinated by Smurf1 and that there are at least five different ubiquitin receptors. Why additional targeting factors are required? I recommend providing fuller background and to cite some of the literature on proteasome targeting adaptors. Many of the figures are difficult to follow and the labeling could be improved. Resubmission 25 June 2012 Responses to the Editor: Comments: Thank you for your patience while your study has been under peer-review at EMBO reports. Please accept my apologies for the unusually long time that we have needed to make a decision regarding your study, as we received one of the referee reports only today. As you will see from the enclosed three reports, although the referees acknowledge the potential interest of the findings, they find the study is far from acceptable in our journal at this point. There are numerous technical concerns, and many issues regarding the conclusiveness of the data as well as its physiological relevance. In all, the referees have rated the work as of poor technical quality and referees 1 and 2 the general interest as medium in the summary evaluation sheets returned with their reports. Given these opinions and the fact that, due to pressure for space, EMBO reports can only invite revision of papers that receive enthusiastic support from a majority of referees, I am afraid that we cannot offer to publish your manuscript. We would, however, be willing to consider a new manuscript on the same topic that successfully addressed all of the referee concerns in full, if the outcome of such effort were support your current model. To be completely clear, I would like to European Molecular Biology Organization 4

5 stress that such a manuscript would be treated as a new submission and would be reviewed afresh. It would only be sent back to the referees if the findings were still novel at the time of submission and all referee concerns were fully and satisfactorily addressed through experimentation, so I would advise against this option unless your are willing to thoroughly rework your study. Response (R): We received the reviewers comments at Feb 29, During the last nearly four months, we have performed a lot of experiments according to the reviewers suggestions and yours. In the resubmitted manuscript, we have provided new evidence to indicate that CKIP-1 specifically interacts with Rpt6 among 19S proteasome regulatory particle; Rpt6 overexpression significantly enhances the proteasome activity; CKIP-1 cooperates with Rpt6 to promote the turnover of Smurf1 and its substrates; CKIP-1 utilizes its C-terminal leucine zipper to interact with Rpt6 while utilizes N-terminal PH domain plus the extreme C-terminus to interact with Smurf1; CKIP-1 functions as an adaptor to couple the ubiquitin ligase (Smurf1) and the proteasome (Rpt6), thus establishes a direct linkage between the ubiquitylation and the degradation machines. Summarizing our and other s findings, we propose a model that on the one hand, CKIP-1 functions as an auxiliary factor of Smurf1 to enhance its interaction with the substrate [6,18]; on the other hand, CKIP-1 functions as an adaptor coupling Smurf1 with Rpt6 to promote the degradation of Smurf1 and its substrates (this study). These findings provide novel insight into the recognition mechanism by proteasome. We hope the modified manuscript will meet the standard for publication in the EMBO Reports. Responses to Reviewer #1: Comments: The authors previously identified CKIP1 as a positive regulator of the Smurf1 E3 ligase. In this current manuscript, the authors used both biochemical and cellular approaches to demonstrate that CKIP1 could interact with the Rpt6 ATPase subunit of 19S regulatory particle of the proteasome, through which CKIP1 could govern the stability of Smurf1 as well as Smurf1 downstream substrates. The paper is clearly written, and for most part of the manuscript, the experiments are well thought-out and performed to address the points. However, there are several concerns that prevent this manuscript from being published in EMBO Report at its present form. Most importantly, the authors did not provide the biological significance for Rpt6-mediated regulation of the Smurf1 signaling pathway. In their previous paper, they showed that depletion of CKIP1 affects osteoblast differentiation process primarily by affecting Smurf1 activity. Thus it is critical to further show whether manipulating Rpt6 could also affect Smurf1 pathway to influence osteoblast differentiation. Furthermore, in some cases, critical controls are necessary to clarify their statements and further experimental evidences are required to support the proposed model. Below please find some specific comments. R: Many thanks for your kind support regarding the novelty of our manuscript. In view of your concerns, we have performed a number of experiments and provided new data. For example, we showed that overexpression of BMP receptor and Smad5 led to a dramatic European Molecular Biology Organization 5

6 increase in BMP-Luc activity; Smurf1, together with CKIP-1, inhibited this reporter activity. The further ectopic expression of Rpt6 significantly strengthened this inhibitory effect (Fig. 4F). The detailed responses are described below point-by-point. Question 1 (Q1): Figure 1C, it will be nice to illustrate the interaction between endogenous CKIP1 and endogenous Rpt6. R: According to your suggestion, endogenous co-immunoprecipitation assays in human cells were performed. Fig. 1F of the revised manuscript shows that endogenous CKIP-1 interacted with endogenous Rpt6 but not other Rpt subunits. Q2: Figure S1 and Table 1. Through mass spect analysis of CKIP-associated proteins, the authors identified a long list of 26S proteasome subunits including Rpt2, Rpt3 and Rpn10. However, later co-immunoprecipitation expression experiment showed that CKIP1 only interacts with Rpt1 and Rpt6 but not other components. Although the authors claimed that CKIP1 may only interacts directly with Rpt1 and Rpt6, but indirectly with other components, experimental evidence is required to support this notion. GST-pull down assays should be performed to show that CKIP1 only directly interacts with Rpt1 and Rpt6. Furthermore, it is interesting to note that later on the authors showed that Smurf1 could interact with Rpt6 in co-ip but not in GST-pull down assays, and CKIP serves as a bridging adaptor to mediate Smurf1/Rpt6 interaction. Thus, it is difficult to understand why under co-ip conditions, endogenous Rpt6 cannot serve as bridging module to mediate the interaction between CKIP and other 26S component Rpt2 and Rpt3? If not, whether overexpression of Rpt6 could enhance the interaction between CKIP and Rpt2 and/or Rpt3. R: Thanks for the reviewer s valuable suggestions. We constructed His-tagged each Rpt subunit and tested their possible interaction with GST-CKIP-1 (GST alone as the negative control). GST pull-down assays showed clearly that only Rpt6 interacted with CKIP-1 but all other five Rpt proteins did not (Fig. 1C). This result suggested that the previous data ectopic Rpt1 could interact with ectopic CKIP-1 might be mediated by Rpt6 but not direct binding. We then tested this possibility. As shown in Fig. 1G, overexpression of ectopic Rpt6 can enhance the interaction between endogenous CKIP-1 and endogenous Rpt1 (but not other subunits including Rpt2, Rpt3, Rpt5), although this effect is still not strong. In addition, endogenous Rpt6 was detectable in the immunoprecipitate of endogenous CKIP-1, whereas other Rpt subunits were not detectable (Fig. 1F). No interaction was detectable between CKIP- 1 and non-atpase ubiquitin receptor Rpn10 or Rpn13 (Fig. 1D). We then made the conclusion that CKIP-1 specifically recognized Rpt6 among 19S regulatory particle. The previous Supplementary Fig. S1 has been removed off in this new version. The appearance of Rpt2, Rpt3 and Rpn10 in CKIP-1 associated proteins in mass-spec analysis might be caused by the high sensitivity of this technique and these interactions seem to be indirect ones. Q3: Figure 1, a related question, what is the important role of CKIP1 interaction with Rpt1 in addition to Rpt6? European Molecular Biology Organization 6

7 R: As described above, our new data point to the conclusion that CKIP-1 specifically interact with Rpt6 in a direct manner. By contrast, the weak interaction between CKIP-1 and other Rpt subunits may be in an indirect manner. Q4: Figure 3B-C, the brackets for these two figure panels are mislabeled. R: Previous Figures 3B-C are now Fig. 3B and 3D. We carefully checked the labeling of these two panels and have made sure they are correct. Q5: Figure 3C, as the GST-pull down results are different from the co-ip experiments in assessing which domain of CKIP is critical to mediate the interaction with CKIP1, it is critical to repeat the interaction mapping between Rpt6 and CKIP1 truncation in the co-ip setting. Furthermore, it is also important to map the critical domain within CKIP1 to interact with Rpt1. R: We did the Co-IP assays as the reviewer requested. As shown in newly-added Fig. 3C, we compared the possible binding of CKIP-1 WT (1-409), ΔC (1-337), or ΔPH ( ) to Rpt6 in cultured cells. As indicated, both CKIP-1 WT and ΔPH interacted with Rpt6 but the ΔC mutant did not, confirming that the C-terminal region of CKIP-1 is required for Rpt6 binding. For the case of Rpt1, as we described above, CKIP-1 did not interact with Rpt1 directly, so we did not perform further mapping analysis. Q6: Figure 4A, as Rpt6 is a critical component of the 19S regulatory particle of the proteasome, it is reasonable to assume that depletion of Rpt6 will lead to stabilization of various unstable proteins. Therefore, in addition to Smurf1, other proteins such as p27, cyclin E and cyclin B should be included as negative controls to show whether depletion of Rpt6 only affects Smurf1, or affects a wide spectrum of protein half-lives. In addition, to build up a causal relationship between Rpt6 and Smurf1, it is critical to re-introduce WT-Rpt6 and Rpt6 ( ) that does not interact with CKIP1, into Rpt6-depleted cells. It is anticipated that WT, but not mutant Rpt6 should restore Smurf1 levels. R: As shown in Fig. 5B, depletion of Rpt6 affected the protein levels of Smurf1, Cyclin E and Cyclin B, suggesting its role with a wide spectrum. We also performed the rescue experiment. Depletion of Rpt6 resulted in an increase of Smurf1 level, and re-introduction of wild-type Rpt6 promoted the degradation of Smurf1. By contrast, re-introduction of Rpt mutant lacking of the CKIP-1-binding coiled-coil region had no such degradation effect. Instead, this mutant even displayed somehow the increasing effect (Fig. 5F). Q7: Figure 4B, as this work primarily addresses the importance of Rpt6 in regulating Smurf1 downstream substrates, in addition to monitoring the steady state levels of various Smurf1 substrates, pulse chase experiments should be performed to assess the changes in their half-lives as read-outs of Smurf1 activity changes. R: Pulse chase experiments of Smurf1 substrate (Smad5) have been performed as the reviewer recommended. As shown in Fig. 4E, Rpt6 expression significantly shortened the half-life of Smad5 whereas the treatment of proteasome inhibitor MG132 blocked such effect. European Molecular Biology Organization 7

8 Q8: Figure 4, sirpt1 and SiRpt1/siRpt6 should be performed to see whether there are synergistic roles between Rpt1 and Rpt6 in regulating Smurf1, and more importantly, in governing the downstream biological outputs including osteoblast differentiation. R: The double depletion of Rpt1 and Rpt6 had no additional effects on Smurf1 upregulation (Fig 5E), which might stem from the fact that Rpt1 and Rpt6 regulate Smurf1 through the same 26S proteasome mechanism. Smad5 plays a critical role in initiating gene expression during osteoblast differentiation and thus Smurf1-mediated degradation of Smad5 inhibits bone formation. We tested whether Rpt6 had any effect on the Smad5-driven luciferase activity. As seen in Fig. 4F, overexpression of BMP receptor (a constitutively active form) and Smad5 led to a dramatic increase in BMP- Luc activity; Smurf1, together with CKIP-1, inhibited this reporter activity. The further ectopic expression of Rpt6 significantly strengthened this inhibitory effect, suggesting that Rpt6 had an inhibitory role on BMP-Smad signaling. Q9: Figure 4, the role of CKIP1 as an adaptor to recruit ubiquitinated Smurf1 and its substrates to 26S proteasome should be further evaluated by double depletion experiment. It will be important to examine whether there is no further stabilization of Smurf1 and its substrates when Rpt6 is depleted in CKIP1-/- MEFs or CKIP1 knocking down cells. R: Thank you for the useful advice. To examine the relevance between CKIP-1 and Rpt6, we performed the double knockdown assays. As shown in Fig. 5D, single knockdown of CKIP-1, or any subunit of Rpt1, Rpt4, Rpt5 and Rpt6, resulted in a moderate increase of Smurf1 level. Simultaneous knockdown of Rpt1, Rpt4, or Rpt5 together with CKIP-1 further elevated Smurf1 level. By contrast, knockdown of Rpt6 had no such effect, implicating that Rpt6 and CKIP-1 cooperated with each other to promote the degradation of Smurf1. Q10: Figure 4, according to the author's model, CKIP1 facilitates Smurf1 auto-ubiquitination and its substrates' turnover through two possible mechanisms, either by inducing Smurf1-substrate interaction or by recruiting them to the 26S proteasome via Rpt6 subunit. Thus it will be nice to dissect these two distinct mechanisms by re-introducing different CKIP1 mutants to CKIP1-depleted cells. The author will expect to see the re-introducing of N-terminal CKIP1 will only partially rescue the CKIP1 depletion-caused Smurf1 stabilization but the re-introducing of LZ domain of CKIP1, which cannot bind Smurf1, will not affect Smurf1 levels. R: Thanks, this suggestion is indeed valuable for us to improve the manuscript. According to this suggestion, we did the CKIP-1 rescue experiment. CKIP-1 depletion by virus-mediated RNA interference caused Smurf1 stabilization as expected. Re-introduction of wild-type CKIP-1 into the CKIP-1-depleted cells resulted in the decrease of Smurf1 level. By contrast, such effect was not detectable with the CKIP-1 ΔLZ mutant which is lack of the Rpt6-binding leucine zipper (Fig. 5G). Re-introduction of CKIP-1 leucine zipper ( ), which cannot bind to Smurf1, also had no effect on Smurf1 levels (Fig. 5G). Since CKIP-1 utilizes the leucine European Molecular Biology Organization 8

9 zipper ( ) to interact with Rpt6 and the extreme C-terminus ( ) plus N-terminal PH domain (1-135) to interact with Smurf1, respectively (Fig 3), we proposed that both the interaction between CKIP-1 and Rpt6, and the interaction between CKIP-1 and Smurf1, were required for CKIP-1 to promote Smurf1 turnover. Thus, both mechanisms (by inducing Smurf1-substrate interaction or by recruiting them to the 26S proteasome via Rpt6 subunit) are bona fide and critical for CKIP-1 function. Responses to Reviewer #2: Major Comments: Q1: It is unclear at present if Rpt6 is interacting with CKIP1 in vivo as part of an intact 26S proteasome, or a 19S or Base sub-particle. The mass spec data identifies proteasomal subunits, but immunoblotting data limits interaction of CKIP-1 to Rpt6 and Rpt1. Intact 26S proteasome assemblies can be immunoprecipitated from lysates by using Mg-ATP in the buffers; from the very limited Methods section, there is nothing to suggest that nucleotides were included. R: Thanks for these very valuable suggestions. We carefully performed the CKIP-1 interaction with Rpt6 or other subunits of proteasome. We proposed that Rpt6 likely interacted with CKIP-1 in vivo mainly in its free isoform and secondly as part of an intact proteasome. This conclusion was based on the following evidence. First, in vitro binding assays showed that CKIP-1 uniquely recognized Rpt6 but not Rpt1-5 (Fig. 1C). Second, Co-IP assays showed that the interaction between endogenous Rpt6 and endogenous CKIP-1 was detectable whereas the interaction between Rpt1, Rpt2, or Rpt3 with CKIP-1 was undetectable (Fig. 1F). Even in the presence of ectopic Rpt6, the interaction between CKIP-1 and Rpt1, Rpt2, Rpt3, Rpt5 was either very weak (for Rpt1) or undetectable (for Rpt2, 3, 5) (Fig. 1G). Third, an in vitro proteasome activity assay in which the cell lysate expressed ectopic Rpt6 or other Rpt subunits clearly showed that Rpt6 expression led to a three-fold increase in proteasome activity, whereas the expression of other Rpt subunits caused moderate or weak effects (Fig. 5A). This result suggests that in addition to the component of the intact proteasome, free Rpt6 also plays a regulatory role on the assembly and activation of the proteasome. Therefore, we proposed that CKIP-1 interacted with free Rpt6 and enhanced the activity of Rpt6 to elevate the whole activity of 26S proteasome. Q2: Additionally, the proteasome is abundant and good antibodies to subunits are available from researchers, as well as commercially. The authors should immunoblot for endogenous proteasome subunits in their CoIP experiments, rather than transfecting, and overproducing single subunits. R: Thanks, we bought five other antibodies each recognizes Rpt1-5. They were used in the experiments shown in Fig. 1F, 1G, 5D, 5E. The results of Co-IP experiments (Fig. 1F, 1G) are described above. (P.S. The Rpt4 antibody did not work well in the immunoblotting, so we showed the results of Rpt1, 2, 3, 5 and 6 in our manuscript) European Molecular Biology Organization 9

10 Q3: In Fig. 4F, the degradation of endogenous Smurf1 is assayed. When Rpt6 is expressed ectopically, the half-life of Smurf1 is further decreased. This suggests that either 26S proteolytic activity is limiting in these cells, or that Rpt6 functions as an adaptor, separate from the 26S entity. The latter hypothesis would be predicted because overproducing one subunit of a hexamer is unlikely to cause an overall increase in proteolytic capacity. This issue can be addressed by measuring peptidase activity from un-transfected, and Rpt6-transfected cells. Or by immunoblotting for Ub conjugate levels. This would shed further mechanistic insight. R: We performed the proteasome peptidase activity assay as the reviewer recommended. The peptidase activity of the un-transfected cell lysate can be easily detected whereas treatment of MG132, the proteasome inhibitor, resulted in the lost of the peptidase activity (Fig. 5A, columns 2-3). When Rpt6 was overexpressed, the peptidase activity increased to three folds of the un-transfected group (Fig. 5A, column 9). Also, the MG132 treatment blocked the peptidase activity (column 10). Expression of Rpt1-Rpt5 resulted in relatively moderate effects (columns 4-8), compared to that of Rpt6, suggesting that Rpt6 might have a special regulatory effect on the intact proteasome activity. These data support the latter hypothesis that Rpt6 functions as an adaptor, separate from the 26S entity, as the reviewer predicted. Q4: As mentioned above, Method descriptions are minimal. Figure Legends are equally cryptic. For example, in Figures 4A, 4C, 4D, increasing concentrations of CKIP-1 or Rpt6 are being transfected. At the very least, the concentation range of the plasmids should be given for reproducibility. In Fig. 4A, Lanes 1 and 4 are contrasted for Smurf1 levels. However, the levels of CKIP1 are nearly identical between Lanes 3 and 4. It would be good to state in the figure legend whether this experiment is representative of x # of experiments. R: Thanks. In the revised manuscript, we carefully modified the description of the figure legends, adding the concentrations of CKIP-1 and Rpt6 in the legends of Fig. 4A, 4C. New and better data are provided in the revised Fig. 4A to make the results clear and clean. Also we added the description of statistical analysis, for example, Each point represents the mean±s.d. of triplicate experiments. Minor points: Q1: The Abstract states "Depletion of CKIP-1 significantly reduces the affinity of Smurf1 and Rpt6". This statement implies that binding constants were measured. Instead, the authors are referring to their immuno-precipitation data. The abstract should be reworded to include some of their functional data from Fig. 4. R: Thanks for the suggestion. The statement has been replaced by Depletion of CKIP-1 reduces the degradation of Smurf1 and its substrates by Rpt6. Also, the abstract writing has been further modified in the revised manuscript. Q2: The text is unclear due to grammatical errors and improper sentence construction. European Molecular Biology Organization 10

11 R: We apologize for the grammatical errors and improper sentence construction in the manuscript. The current revised manuscript has been edited by an expert company, American Journal Experts ( and we therefore believe the grammatical errors have been avoided. Response to Reviewer #3: Q1: The manuscript provides good evidence for the interaction between the proteasome, CKIP-1, and Smurf1 and maps the interactions to specific domains within the proteins. As the authors report, they had previously defined different domains for the interaction of the three components but they provide a plausible explanation why the current manuscript comes to different conclusions. One concern here is that in some experiments the interaction between CKIP-1 and the proteasome does not map to a single subunit as would be expected for a specific adaptor protein. This observation should be discussed. R: As we described in the response to Reviewer #1, we provided more evidence to make the conclusion that CKIP-1 specifically recognizes Rpt6 among the proteasome subunits. We performed in vitro GST pull-down assays to test the possible interaction of CKIP-1 with each Rpt protein of the base sub-particle. The data showed clearly that only Rpt6 interacted with CKIP-1 but all other five Rpt proteins did not (Fig. 1C). Co-IP assays showed that endogenous CKIP-1 was co-immunoprecipitated with Rpt6, but was not with Rpt1, Rpt2, and Rpt3 (Fig. 1F). These results suggested that the previous data ectopic Rpt1 could interact with ectopic CKIP-1 might be mediated by Rpt6 but not direct binding. We then tested this possibility. As shown in Fig. 1G, overexpression of ectopic Rpt6 can enhance the interaction between endogenous CKIP-1 and endogenous Rpt1 (but not other subunits including Rpt2, Rpt3, Rpt5), although this effect is still weak. No interaction was detectable between CKIP-1 and non-atpase Rpn10 or Rpn13 (Fig. 1D). We then made the conclusion that CKIP-1 specifically recognized Rpt6 among 19S regulatory particle. Q2: Most importantly, the relevance of the interactions to degradation is poorly supported and this part of the manuscript would have to be very significantly improved before publication can be considered. My biggest concern regards the design of the experiments purported to show that the interaction CKIP-1 and the proteasome subunit Rpt6 is required for degradation of Smurf1. In these experiments the expression of Rpt6 is up and down regulated and corresponding effects on Smurf1 degradation are determined. Since Rpt6 is a key stoichiometric component of the proteasome, it is not surprising that reducing its expression will compromise proteasome function. This observation cannot be taken as evidence for a specific role of Rpt6 in the process under investigation here. Furthermore, I am surprised that overexpression of Rpt6 enhances degradation of Smurf1. If anything, I would have expected it to decrease degradation of Smurf1 as excess Rpt6 that cannot be integrated into proteasome particles would compete CKIP-1 and Smurf1 away from the proteasome. R: Thank you for these very valuable comments. We provided new data to show that the interaction between CKIP-1 and Rpt6 is required for degradation of Smurf1. As seen in Fig. European Molecular Biology Organization 11

12 5F, the depletion of Rpt6 resulted in an increase of Smurf1 level, and the re-introduction of wild-type Rpt6 promoted the degradation of Smurf1. By contrast, the re-introduction of the Rpt mutant, which lacks the coiled-coil region, had no such degradation effect. By contrast, this mutant somehow displayed an increasing effect. Moreover, the re-introduction of wild-type CKIP-1 into the CKIP-1-depleted cells resulted in the decrease of Smurf1 levels. By contrast, this effect was not detectable with the CKIP-1 ΔLZ mutant, which lacks the Rpt6- binding leucine zipper (Fig 5G). The re-introduction of the CKIP-1 leucine zipper ( ), which cannot bind to Smurf1, also had no effect on Smurf1 levels (Fig. 5G). These data indicated that the interaction between Rpt6 and CKIP-1 is critical for the proteasomal turnover of Smurf1. About the effect of Rpt6 overexpression on Smurf1 degradation, we hypothesized that Rpt6 overexpression might enhance the peptidase activity of the 26S proteasome. To test this possibility, the proteasome activity was determined with the cell lysate from the un-transfected or Rpt-transfected cells. As shown in Fig. 5A, Rpt6 overexpression significantly enhanced the proteasome activity and such effect of Rpt6 was more significant than that of other Rpt subunits. This result might explain why Rpt6 overexpression promoted the degradation of Smurf1 and its substrates (Fig. 4). Q3: I recommend explaining the motivation behind the proposed mechanism better. Several times, the manuscript raises the question as to how Smurf1 and its substrates are degraded by the proteasome and then goes on to discuss how all the substrates under discussion are ubiquitinated by Smurf1 and that there are at least five different ubiquitin receptors. Why additional targeting factors are required? R: We highly appreciate the reviewer s careful reading on our opinion. Indeed, our previous description did not explain clearly what we want to tell the readers. We revised our manuscript and attempted to explain clearly our opinion. Although at least five different ubiquitin receptors (Rpn10, Rpn13, Rad23, Dsk2, Ddi1), all of these proteins recognize the targets through binding the ubiquitin chain (the general mechanism). We noticed that previous study showed a different recognition mechanism (here we called VIP mechanism) by which the proteasome subunit (Rpt5) can interact with the ubiquitin ligase (pvhl) and promote the substrate degradation. In our current study, we show that another ATPase subunit Rpt6 can interact with the co-factor of Smurf1 ubiquitin ligase (i.e. CKIP-1), and promote the substrate degradation (and the ligase degradation). Thus this study provides novel insight into the recognition mechanism used by the proteasome. We revised the writing according to the reviewer s suggestion. In the abstract, the unresolved question is described as The mechanisms involved in the recognition and degradation of these substrates by the proteasome remain unclear ; and, the corresponding part in the Introduction has been also modified. The sentence At present, how Smurf1 and substrates are delivered to or recognized by 26S proteasome still remain unclear in the last of European Molecular Biology Organization 12

13 the 2 nd paragraph of Introduction has been deleted. The 3 rd paragraph of Introduction has been dramatically re-written as follow: The recruitment of proteins attached with polyubiquitin chains to the 26S proteasome is fundamental in the ubiquitin-proteasomal degradation system. Simply, the ubiquitylated proteins can bind to either the presence of intrinsic ubiquitin receptors in 19S RP (i.e. Rpn10 and Rpn13) [7-10] or an extrinsic adaptor protein that bears both polyubiquitin-binding and proteasome-binding domains (Rad23, Dsk2 and Ddi1) [11-14]. All of these ubiquitin receptors recognize the targets through binding the ubiquitin chain, which represents a general mechanism of recognition. Intriguingly, there is evidence that the proteasome also recognizes the target protein itself rather than the ubiquitin chain, which represents a specific VIP mechanism. For example, one ATPase subunit, Rpt5 (also known as TBP-1 or PSMC3), has been reported to bind to pvhl, a core component of a multi-subunit E3 ubiquitin ligase, to promote the degradation of HIF1α [15]. The ability of pvhl to degrade HIF1α depends, in part, on the interaction of pvhl with Rpt5. The delivery of ubiquitylated proteins to the 26S proteasome is a complex and sequential process [16], and it is unknown whether any other subunit of 19S RP recognizes the substrate through the VIP mechanism exhibited by Rpt5. The 4 th paragraph of Introduction has been re-written as: Here, we provide evidence that Rpt6 (also known as PSMC5) interacts directly with CKIP-1 and promotes the turnover of Smurf1 and its substrates. This study establishes an adaptor role of CKIP-1 in coupling the E3 ligase Smurf1 and the 26S proteasome and demonstrates another case of the VIP mechanism of substrate recognition by a proteasome. Q4: I recommend providing fuller background and to cite some of the literature on proteasome targeting adaptors. R: According to your useful suggestion, we provided fuller background of proteasome targeting adaptors (as described in the response to Q3) and updated the new literatures. The references 2, 8, 11, 13, 14, 15 are all updated ones. Q5: Many of the figures are difficult to follow and the labeling could be improved. R: We appreciate the reviewer s criticisms. We carefully re-organized the figures and made the labeling more clearly. The statistic analysis was also included in the figures. In addition, the manuscript has been edited by an expert company, American Journal Experts ( and we therefore believe the grammatical errors have been avoided. We hope the modified manuscript would satisfy the reviewer. We deeply hope that these modifications will satisfy you and all of the referees and that the current manuscript will be considered for publication in your journal. Thank you very much in advance for your attention to our work. We look forward to your response. European Molecular Biology Organization 13

14 2nd Editorial Decision 16 August 2012 Thank you for the resubmission of your manuscript to EMBO Reports. I first must apologize for the great delay in processing your manuscript, which is due to the fact that we have only now received the full set of referee reports that I copy below. As you will see, all referees appreciate your efforts to address their comments and agree that the manuscript has been greatly improved. Referee #2 and Referee #3 both support publication of your work in EMBO Reports, but as Referee #3 suggests with minor modifications in the discussion. Referee #1 however is still concerned about the physiological role of CKIP-1 mediated degradation of Smurf-1 in osteoblast differentiation. Given that the other two referees do not raise this specific concern, we have decided that you do not need to address it for publication of the study in EMBO Reports. Given these evaluations I would therefore like to invite you to revise the manuscript along the lines suggested by referee #3. I assure you this final peer-review process will be faster than before. Please note that acceptance of the manuscript will depend on a positive outcome of a second round of review and that it is EMBO reports policy to allow a single round of revision only. Therefore, acceptance or rejection of the manuscript will depend on the completeness of your responses included in the next, final version of the manuscript. Revised manuscripts should be submitted within three months of a request for revision; they will otherwise be treated as new submissions. Also, the revised manuscript may not exceed 30,000 characters (including spaces, references, and figure legends) and 5 figures plus 5 supplementary figures, which should directly relate to the corresponding main figure. I would also like to remind you that p-values, error bars and the number of experiments performed (n) must be defined in the relevant figure legends. When submitting your revised manuscript, please include: A Microsoft Word file of the manuscript text, editable high resolution TIFF or EPS-formatted figure files, a separate PDF file of any Supplementary information (in its final format) and a letter detailing your responses to the referee comments. Please also include a two sentence-summary of the manuscript that will appear online on our webpage in case of acceptance of the study for publication. We would also welcome the submission of cover suggestions, or motifs to be used by our Graphics Illustrator in designing a cover. As part of the EMBO publication's Transparent Editorial Process, EMBO reports publishes online a Review Process File to accompany accepted manuscripts. This File will be published in conjunction with your paper and will include the referee reports, your point-by-point response and all pertinent correspondence relating to the manuscript. You are able to opt out of this by letting the editorial office know (emboreports@embo.org). If you do opt out, the Review Process File link will point to the following statement: "No Review Process File is available with this article, as the authors have chosen not to make the review process public in this case." I look forward to seeing a revised version of your manuscript when it is ready. Again, please accept our sincerest apologies for the delay in the review of your manuscript. Yours sincerely, Editor EMBO reports European Molecular Biology Organization 14

15 Referee #1 The authors have spent serious efforts in addressing the outstanding concerns raised up in the last round of review. As a result, the manuscript has been significantly improved. I only have some remaining concerns related to the physiological role of CKIP1-mediated degradation of Smurf1, especially for its effects in osteoblast differentiation, which the authors failed to address in this round of revision. These critical points should be fully addressed before publication of this study at EMBO Reports. In the revised manuscript, the authors have incorporated newly generated results to show that overexpression of Rpt6 significantly enhance the ability of CKIP to suppress the BMP-Luc reporter activity (Fig. 4F). It will be nice for the authors to further show whether depletion of endogenous Rpt6 also affects BMP-reporter activity in part through regulation of Smurf1 (which can be addressed by Smurf1/Rpt6 double depletion). More importantly, it will be nice for the authors to carry out more functional assays used in their previous publications to show that depletion of endogenous Rpt6, which stabilizes Smurf1 as well as Smurf1 substrates, could activate osteoblast differentiation. Furthermore, it is critical for the authors to examine whether additional depletion of Smurf1 could restore the obsteobloast differentiation defects in Rpt6-depleted cells. As loss of Rpt6 leads to reduced Smurf1-dependent destruction of its downstream substrates, indicating an inactive form of Smurf1 in Rpt6-depleted cells, it is anticipated that additional depletion of Smurf1 might not have major effects in Rpt6-depleted cells. Conversely, it is also critical for the authors to show that overexpression of Rpt6 could suppress osteoblast differentiation, presumably by activating the Smurf1 E3 ligase activity towards its downstream targets. This reviewer believes that these functions assays will serve to unify these biochemical results with the downstream Smruf1 signaling pathway as well as the subsequent biological effects, thereby providing strong support for the critical physiological importance of CKIP/Rpt6-depednent regulation of Smurf1 stability and its downstream signaling pathways. Referee #2 The authors have satisfactorily answered the questions raised by me. Referee #3 I am impressed with the great effort the authors have put into their response to the referee reports. They have performed difficult and complex experiments and thus some of the data described in the manuscript, by the very nature of the experiments, have to remain ambiguous. However, the body of experimental evidence taken together convincingly shows that CKIP-1 stimulates degradation of Smurf1 and its substrates. Their is good evidence that CKIP-1 does so by functioning as a proteasome adaptor acting through the Rpt6 subunit. I remain somewhat concerned that one aspect of the mechanism invoked in the manuscript is not entirely intuitive to the usual mechanistic thinking about enzyme action. In the simplest mechanism for Rpt6 in Smurf1 degradation, it acts in its established location incorporated into the ATPase ring of the proteasome cap. There is serves as a docking site for CKIP-1, allowing the latter to act as a substrate adaptor for the proteasome. This is a clean and plausible mechanism. The unexpected observation is that significant overexpression of Rpt6 stimulates Smurf1 degradation further. In the simplest mechanism excess free Rpt6 (not incorporated into the proteasome) would compete CKIP- 1 away from the proteasome and therefore inhibit Smurf degradation. The manuscript goes on to suggest that free cellular Rpt6 somehow stimulates proteasome activity broadly against other substrates. The mechanism of this effect is not clear at all and this is unsatisfactory to me. There are other observations that are not entirely straightforward. For example, I am confused that immunoprecipitation from cells with endogenous proteasome subunit levels shows clear and clean specificity of the interaction between CKIP-1 and Rpt6. The ATPase ring in the proteasome is quite stable and I would have expected it to act as one unit in these experiments. (In the over expression experiments, the specificity is expected and strengthens the conclusions). However, biology often does not conform to our simplest models and I am ready to just accept these European Molecular Biology Organization 15