Structures of p53/BCL-2 complex suggest a mechanism for p53 to antagonize BCL-2 activity

Mitochondrial apoptosis is strictly controlled by BCL-2 family proteins through a subtle network of protein interactions. The tumor suppressor protein p53 triggers transcription-independent apoptosis through direct interactions with BCL-2 family proteins, but the molecular mechanism is not well understood. In this study, we present three crystal structures of p53-DBD in complex with the anti-apoptotic protein BCL-2 at resolutions of 2.3–2.7 Å. The structures show that two loops of p53-DBD penetrate directly into the BH3-binding pocket of BCL-2. Structure-based mutations at the interface impair the p53/BCL-2 interaction. Specifically, the binding sites for p53 and the pro-apoptotic protein Bax in the BCL-2 pocket are mostly identical. In addition, formation of the p53/BCL-2 complex is negatively correlated with the formation of BCL-2 complexes with pro-apoptotic BCL-2 family members. Defects in the p53/BCL-2 interaction attenuate p53-mediated cell apoptosis. Overall, our study provides a structural basis for the interaction between p53 and BCL-2, and suggests a molecular mechanism by which p53 regulates transcription-independent apoptosis by antagonizing the interaction of BCL-2 with pro-apoptotic BCL-2 family members.

substitution can be made without disrupting p53-DBD binding to Bcl-2, the basis of this paper seems flawed.
-Are there any flaws in the data analysis, interpretation and conclusions? -Do these prohibit publication or require revision?
The analysis, interpretation, and conclusions may be correct for the Bcl-2#3 interaction with p53-DBD, but due to the uncertainty of the validity of using Bcl-2#3 as a surrogate for Bcl-2, it is unclear if this represents a true Bcl-2/p53 interaction.
While for crystallography the Bcl-2#3 construct "had to be used" but for cell based assays, why wasn't full length p53 also tested? Is there a loss of affinity of Bcl-2 to full length p53 vs. p53-DBD? Paper requires revision.
-Is the methodology sound? Does the work meet the expected standards in your field?
Most methodology is sound, well-described, and within the standards of the field. The authors are experts in the field of crystallography.
-Is there enough detail provided in the methods for the work to be reproduced?

Yes
Line 81-trial not trail Line 249-mimetic not mimetics; also grammar issue with that sentence Line 252-has not have Line 258-grammar Line 333-lowercase p53 Line 572-hot pink not hotpink Reviewer #2 (Remarks to the Author): Wei et al describe crystal structures of a complex between p53 and Bcl-2, together with functional data characterizing the interaction. Overall this is a very significant report that sheds light on an important interaction controlling apoptosis. The reported experiments are of high quality, and in their totality make for an exciting study. Data largely support conclusions.
Major issues: 1.The affinity measurements are based largely on ELISAs. The authors should conduct either SPR (Biacore) or ITC measurements to fully quantitate the interactions between p53 and Bcl-xL (and their respective mutants).
2. The data on apoptosis are all based on western blots. For these, full blots need to be shown. More importantly, the blots need to be supported by life cell analysis to show that there is a demonstrable effect on live cell apoptosis via FACS analysis and detection of an appropriate marker (PI release, PS via annexin etc). This is important to properly quantify the effect of mutations, and also allow crosscorrelation to the significance of the p53-Bcl2 interaction vs p53-BclXL.
3. Considering the existence of the p53-BclXL complex, a detailed comparison of the two needs to be included.

Competition with
Bax. This is an exciting finding. However, Bax is only one of the major pro-apoptotic Bcl-2 members that are able to interact with Bcl-2. In the absence of any other data on the remaining key pro-apoptotic regulators one cannot gauge as to how significant and relevant the Bax component is, and whether or not this is really driven by another pro-apoptotic protein that is being competed with. Another limitation is the reliance on Bax BH3 peptide (I note full length recombinant Bax is also used, but in the presence of triton -see comment further below on this). Whilst it is a very useful surrogate for Bax, to fully show that Bax is activated due to the p53 activity as per model and crystal structures this should be shown in a life cell system (ideally with and without Bax present). As is the suggestion of a significant role of Bax is insufficiently supported.
Minor issues: 1. For the reported apoptosis assays, is expression of Ti-p53 sufficient on its own to trigger apoptosis or was this supplemented by another stimulus? 2. Western blots need to be shown in full and loading controls need to be shown. 4. What concentration of protein was used for crystallization trials and to obtain the final crystals? 5. Triton X100 was present in the GST pull down buffer. Triton and certain other detergents are known to induce structural changes in Bax. Can the authors comment on the rational for using Triton here in light of this issue, and also comment on the possibility that triton may induce experimental artefacts in such a context and obfuscate the result. In article 405778 entitled "Structures of p53/BCL-2 complex suggest a mechanism for p53 to 1 antagonize BCL-2 activity" authors Wei, Wang et al. have determined three crystal structures of fusion proteins BCL-2-deltaTM-G-linker-p53DBD where they vary the linker size. All structures reveal that p53DBD binds to the BH3-binding groove of BCL-2, this mode of binding being different from the outside of the groove binding mode observed for BCL-xL-deltaTM-G-linker-p53DBD, which these authors previously characterized, and another study also resolved by NMR. The authors than produce mutants at the binding interface in both partners and characterize their binding to respective WT partner to validate the interactions site. Additionally, they selected a triple RRR mutant in p53DBD, which exhibited altered activity in one of their binding assays and reconstituted it in cells KO for p53 in the context of a transcriptionally inactive mutations previously described showing that it did not induce the typical markers of apoptosis. The direct regulation of the mitochondrial pathway at the BCL-2 family protein by cytosolic p53 has been a controversial area of investigation in the field that has been difficult to prove with existing knowledge and available reagents. Studies in this area therefore need to be carefully conducted.
While the structures reveal how p53 DBD may interact with BCL-2 and seem to have adequate statistics and resolution, I find the functional probing preliminary and not rigorously executed.
Here are some specific examples of problem area: 1. The side chain of D186 does not participate in the water bridge interaction as shown in the electron density in SF3b, instead the D186 CO being implicated. Why would the authors select to replace this residue with D186R, which abolished the pull-down interactions with GST-BCL-2, when presumably the side chain does not participate in the interaction.
2. For such high EC50 values observed in ELISA assays and high IC50 values observed in competitive FP assays, why did the authors not use a direct binding assay to characterize the interactions as they have done in their previous paper?
3. The assays in Figure 2 and Figure 6 should be combined as it deals with characterizing the same interaction. It is unclear why the authors have not included the RRR p53DBD mutant in the latter Figure  6b. Figure 6c and 6d it is remarkable that BAX peptide cannot fully displace p53DBD from BCL-2 suggesting possibly that other interactions may hold the complex in place. I cannot find the concentrations in the Methods section used in this assay which only show the stochiometric considerations. The authors need to include this and do a proper titration to dislodge the BCL-2-p53 complex with excess BAX BH3 or BAX protein. It does not seem that the interaction is titratable. Figure 4b is very preliminary and needs to be better developed. Cell death assay should show extent to cell death. Additionally, the authors need to at least show that the constructs do not induced apoptosis in p53-/-bak-/-bax-/-cell lines to exclude an off target effect especially as others have shown cytosolic p53-/-mediated cell death in the absence of BAK and BAX (https://doi.org/10.1016/j.cell.2012.05.014) 6. The authors need to characterize the homogeneity of the fusions used for crystallography on gel filtration chromatography (are these fusions forming oligomers?). They also need to acknowledge that crystallography could result in crystal contact artifacts, which may be non-native. For instance, if the fusions were p53DBD-G-linker-BCL-2-deltaTM they may lead to totally different crystal contacts. Therefore, the rigor of the investigation needs to be unquestionable. A crystal structure of the DNA binding domain (DBD) of p53 and a modified fragment of Bcl-2 has been reported. Key residues have been identified, purportedly revealing the p53-Bcl-2 inteface (interacting residues).

The apoptosis data in
-Will the work be of significance to the field and related fields? How does it compare to the established literature? If the work is not original, please provide relevant references.
Solving this structure of the interaction between p53-DBD and Bcl-2 is potentially highly significant, due to p53's role in blocking Bcl-2's anti-apoptotic activity. Solving the structure gives key information on the mechanism of action of p53's mitochondrial apoptosis pathway and how it disrupts Bcl-2 action by binding. This work appears to be original.

-Does the work support the conclusions and claims, or is additional evidence needed?
The critical issue with this structure is the use of a modified Bcl-2 (called Bcl-2#3) which lacks the flexible loop region of Bcl-2 (aa 32-68) which has been reported to interact with the DBD of p53 (Deng et al., Mol Cell Bio, 2006). Based on the reference to this Bcl-2#3 construct within this paper (and references therein), Bcl-2#3, which is used in this paper, instead has the first 34 aa's of Bcl-2, followed by the flexible domain from another protein, Bcl-XL (aas 35-50), then the Bcr-2 structure continues starting at residue 92. As shown in the diagram below (pasted from Petros et al., PNAS 2001, https://doi.org/10.1073/pnas.041619798) the flexible linker aa sequence (from around 32-68) from Bcl-2 (any of the 3 isoforms shown below) is vastly different from the flexible linker sequence from Bcl-XL.
Understandably, the authors used a modified Bcl-2 sequence due to the inability to crystallize native or truncated Bcl-2. Using Bcl-2#3 led to an "ability to achieve a high yield and was suitable for crystallization." Surprisingly, the details on this critical construct were not clearly explained in the paper.
Modifications in a protein to make it crystallize should ensure that the interacting domains between 2 proteins in question are intact. The substitution of the Bcl-XL flexible domain for the Bcl-2 flexible domain may not seem reasonable, given that p53-DBD binds to Bcl-2 via Bcl-2's original flexible domain, according to prior literature, which is no longer present in Bcl-2#3. The substituted flexible domain has a different length compared to Bcl-2's flexible domain, which may change the overall structure of Bcl-2 (and subsequently impact crystal packing) Unless the authors can rationalize/justify how this substitution can be made without disrupting p53-DBD binding to Bcl-2, the basis of this paper seems flawed.
-Are there any flaws in the data analysis, interpretation and conclusions? -Do these prohibit publication or require revision?
The analysis, interpretation, and conclusions may be correct for the Bcl-2#3 interaction with p53-DBD, but due to the uncertainty of the validity of using Bcl-2#3 as a surrogate for Bcl-2, it is unclear if this represents a true Bcl-2/p53 interaction.
While for crystallography the Bcl-2#3 construct "had to be used" but for cell based assays, why wasn't full length p53 also tested? Is there a loss of affinity of Bcl-2 to full length p53 vs. p53-DBD? Paper requires revision.
-Is the methodology sound? Does the work meet the expected standards in your field?
Most methodology is sound, well-described, and within the standards of the field. The authors are experts in the field of crystallography.
-Is there enough detail provided in the methods for the work to be reproduced?

Yes
Line 81-trial not trail Line 249-mimetic not mimetics; also grammar issue with that sentence Line 252-has not have Line 258-grammar Line 333-lowercase p53 Line 572-hot pink not hotpink We would like to express our gratitude to all the reviewers for their valuable and constructive comments on our manuscript. We are submitting a revised version of the manuscript, in which we have carefully addressed all the comments, incorporated additional experiments, and clarified the text. Below is a point-by-point response to the reviewers' comments.

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): 1. The critical issue with this structure is the use of a modified Bcl-2 (called Bcl-2#3) which lacks the flexible loop region of Bcl-2 (aa 32-68) which has been reported to interact with the DBD of p53 (Deng et al., Mol Cell Bio, 2006). Based on the reference to this Bcl-2#3 construct within this paper (and references therein), Bcl-2#3, which is used in this paper, instead has the first 34 aa's of Bcl-2, followed by the flexible domain from another protein, Bcl-XL (aas 35-50), then the Bcr-2 structure continues starting at residue 92. As shown in the diagram below (pasted from Petros et al., PNAS 2001, https://doi.org/10.1073/pnas.041619798)-see supplemental file-the flexible linker aa sequence (from around 32-68) from Bcl-2 (any of the 3 isoforms shown below) is vastly different from the flexible linker sequence from Bcl-XL.
Understandably, the authors used a modified Bcl-2 sequence due to the inability to crystallize native or truncated Bcl-2. Using Bcl-2#3 led to an "ability to achieve a high yield and was suitable for crystallization." Surprisingly, the details on this critical construct were not clearly explained in the paper.
Modifications in a protein to make it crystallize should ensure that the interacting domains between 2 proteins in question are intact. The substitution of the Bcl-XL flexible domain for the Bcl-2 flexible domain may not seem reasonable, given that p53-DBD binds to Bcl-2 via Bcl-2's original flexible domain, according to prior literature, which is no longer present in Bcl-2#3. The substituted flexible domain has a different length compared to Bcl-2's flexible domain, which may change the overall structure of Bcl-2 (and subsequently impact crystal packing) Unless the authors can rationalize/justify how this substitution can be made without disrupting p53-DBD binding to Bcl-2, the basis of this paper seems flawed. Response: We thank the reviewer for bringing up the concern. Based on the reviewer's suggestion, the sequences of BCL-2 constructs have been provided in the revised Supplementary Figure 1, and the details on the critical constructs have been explained in the revised manuscript. BCL-2#3 is a truncated BCL-2 used in a previous structural study, featuring a shortened α1-α2 loop and certain surface residues replaced with a BCL-xL sequence (ref 25). BCL-2#3 has been reported to have a similar selectivity profile for BH3-only peptides as wild-type BCL-2 (ref 25). A similar truncated strategy has been employed in many biochemical and structural studies of the BCL-2 family (ref 25-30).
To investigate whether the modifications affect the interaction between BCL-2 and p53-DBD, we carried out MST to determine the binding affinity. The results showed that p53-DBD had similar Kd values for both BCL-2 (3.5 µM) and BCL-2#3 (3.9 µM), suggesting that the modifications of BCL-2 didn't influence its binding affinity to p53-DBD (revised Supplementary Figure 3). In the p53-DBD/BCL-2#3 structures, the flexible loop and modified residues are not located on the contact surface with p53-DBD (revised Supplementary Figure 4c). The overall structure of BCL-2#3 closely resembles the apo solution structure of BCL-2 (PDB: 1GJH) (revised Supplementary Figure 4b) or the BCL-2 in the BCL-2/Bax-BH3 structure (PDB: 2XA0) (revised Figure 4b). This indicates the modification has minimal influence on the overall structure of BCL-2. In addition, mutational analysis based on wild-type recombinant BCL-2 (revised Figure 3) or full-length BCL-2 (revised Figure 6a) supports the interface defined by the p53-DBD/BCL-2#3 structures. Collectively, these data suggest that the modifications of BCL-2 do not disrupt the binding of p53-DBD to BCL-2.
In the study mentioned by the reviewer (Deng et al., Mol Cell Bio, 2006), cell-based coimmunoprecipitation demonstrated that a BCL-2 construct with a deletion of amino acids 32-68 failed to co-precipitated with p53 in H7 cells. Notably, several studies have reported that posttranslational modifications or structural flexibility of the α1-α2 loop (aa 32-92) of BCL-2 or BCL-xL can affect ligand access into the BH3 binding pocket (ref 36, 50-53). It is possible that the α1-α2 loop of BCL-2 also plays a regulatory role in the intracellular p53/BCL-2 interactions. We have addressed the issue in the revised discussion section.
2. The analysis, interpretation, and conclusions may be correct for the Bcl-2#3 interaction with p53-DBD, but due to the uncertainty of the validity of using Bcl-2#3 as a surrogate for Bcl-2, it is unclear if this represents a true Bcl-2/p53 interaction. While for crystallography the Bcl-2#3 construct "had to be used" but for cell based assays, why wasn't full length p53 also tested? Is there a loss of affinity of Bcl-2 to full length p53 vs. p53-DBD? Paper requires revision. Response: Thanks for the comment. In our study, BCL-2#3 was solely used for crystallography, and all the biochemical assays were conducted using recombinant wild-type BCL-2. For the cell-based assays, full-length p53 and full-length BCL-2 were used (revised Figure 6). To better distinguish these constructs, we have classified BCL-2 constructs as BCL-2#3, recombinant BCL-2 and FL-BCL-2, and p53 constructs as p53-DBD, FL-p53 and Ti-p53 (Ti-p53 is generated based on FL-p53) in the revised manuscript.
Line 81-trial not trail Line 249-mimetic not mimetics; also grammar issue with that sentence Line 252-has not have Line 258-grammar Line 333-lowercase p53 Line 572-hot pink not hotpink Response: We appreciate the careful reading by the reviewer. We have made corresponding modifications to the wording and phrases in the revised manuscript.
Reviewer #2 (Remarks to the Author): Wei et al describe crystal structures of a complex between p53 and Bcl-2, together with functional data characterizing the interaction. Overall, this is a very significant report that sheds light on an important interaction controlling apoptosis. The reported experiments are of high quality, and in their totality make for an exciting study. Data largely support conclusions. Major issues: 1. The affinity measurements are based largely on ELISAs. The authors should conduct either SPR (Biacore) or ITC measurements to fully quantitate the interactions between p53 and Bcl-xL (and their respective mutants). Response: We thank the reviewer for the positive comments and the suggestion. We speculate that the issue raised here by the reviewer is about the interaction of p53 with BCL-2. In fact, we attempted Biacore X100 and ITC measurements; however, both recombinant p53-DBD and BCL-2 proteins, when present at high concentrations, tend to precipitate after prolonged exposure to the instruments at room temperature. In the revised manuscript, we employed MST to quantitate the interactions between p53-DBD and BCL-2. GST-p53-DBD was used to help stabilize the p53 protein. The GST tag was tested and did not bind to BCL-2. After the MST experiments, the samples remained transparent. The results are presented in the revised Supplementary 3 and the revised Figure 3c-d. The descriptions and detailed methods have been added to the revised manuscript. The data from both methods have shown that the structure-based mutations decreased the binding ability of p53-DBD to BCL-2, and supported the p53-DBD/BCL-2#3 interface defined by the complex structure. We have also observed that the Kd values of MST were different from the EC50 values of ELISA (revised Supplementary 6), possibly due to the utilization of different methods.
2. The data on apoptosis are all based on western blots. For these, full blots need to be shown. More importantly, the blots need to be supported by life cell analysis to show that there is a demonstrable effect on live cell apoptosis via FACS analysis and detection of an appropriate marker (PI release, PS via annexin etc). This is important to properly quantify the effect of mutations, and also allow cross-correlation to the significance of the p53-Bcl2 interaction vs p53-BclXL. Response: We appreciate the suggestions. Per the reviewer's suggestion, the full blots have been included in the revised Supplementary Figure 12. Additionally, we have used FACS analysis and detection of Annexin-V/7-AAD marker to demonstrate the effect of p53 mutations on live cell apoptosis (revised Figure 6c-d, revised Supplementary Figure 8). Ti-p53 induced approximately 19.6% of apoptotic cells, whereas the ratio of apoptotic cells reduced significantly to 8% with Ti-p53 RRR . The relevant descriptions have been added to the revised Results section.
3. Considering the existence of the p53-BclXL complex, a detailed comparison of the two needs to be included. Response: Thanks for the suggestion. We have provided a detailed comparison of p53/BCL-2 complex with p53/BCL-xL complex in the revised Supplementary Figure 10 -11. 4. Competition with Bax. This is an exciting finding. However, Bax is only one of the major proapoptotic Bcl-2 members that are able to interact with Bcl-2. In the absence of any other data on the remaining key pro-apoptotic regulators one cannot gauge as to how significant and relevant the Bax component is, and whether or not this is really driven by another pro-apoptotic protein that is being competed with. Another limitation is the reliance on Bax BH3 peptide (I note full length recombinant Bax is also used, but in the presence of tritonsee comment further below on this). Whilst it is a very useful surrogate for Bax, to fully show that Bax is activated due to the p53 activity as per model and crystal structures this should be shown in a life cell system (ideally with and without Bax present). As is the suggestion of a significant role of Bax is insufficiently supported. Response: We thank the reviewer for raising these concerns. In the revised manuscript, we have modified the GST pull-down assay by replacing triton with a buffer containing CHAPS, which has been reported not to alter the Bax or Bak conformation (ref 60-61). We utilized GST pull-down to detect the effects of p53-DBD on the interactions of BCL-2 with BH3 peptides (Bid, Bim, Puma, Bax, and Bak) (revised Figure 5c), as well as Bax proteins (revised Figure 5d), and Bak proteins (revised Figure 5e). The results indicate a negative correlation between the complex formation of p53-DBD/BCL-2 and the complex formation of BCL-2 with pro-apoptotic peptides or proteins. The descriptions of these findings have been added to the revised Results section.
To examine whether Bax or Bak is activated by Ti-p53 activity in a cell system, we knocked down Bax or Bak by small interfering RNA (siRNA) in Ti-p53 stably-expressing HCT116 cells, and then used FACS analysis and detection of an Annexin-V/7-AAD marker to detect cell apoptosis (revised Supplementary Figure 9). Compared to Ti-p53 (19.6%), Bax siRNA and Bak siRNA decreased the apoptotic cell to 12.1% and 13.5%, respectively. The simultaneous knockdown of Bax and Bak still led to 9.6% of cell apoptosis. The results indicate that apoptosis induced by Ti-p53 is dependent on both Bax and Bak. The related descriptions have been added to the revised Results section.
Minor issues: 1. For the reported apoptosis assays, is expression of Ti-p53 sufficient on its own to trigger apoptosis or was this supplemented by another stimulus? Response: Yes. The expression of Ti-p53 is sufficient on its own to trigger apoptosis, and no additional stimulus is required. We have provided a detailed description of the experimental condition in the revised Methods section.
2. Western blots need to be shown in full and loading controls need to be shown. Response: In the revised manuscript, western blots have been shown in full (revised Supplementary Figure 12). We have carefully examined all the loading controls. Please kindly bring it to our attention if any information appears to be missing. Fig 3D but not 3C or in Fig  6. Response: Thanks for the suggestion. We have accordingly added the number of experiments and error bars in the revised manuscript. For specific data points, the error bars may be shorter than the height of the symbol, making them less visually prominent.

Number of experiments and error bars only defined in the legend for
4. What concentration of protein was used for crystallization trials and to obtain the final crystals? Response: The proteins were used for initial crystal screening at approximately 10 mg/ml. Through optimization, final crystals were obtained using 13 mg/ml protein concentration. We have added detailed information to the revised Methods section in the revised manuscript.
5. Triton X100 was present in the GST pull down buffer. Triton and certain other detergents are known to induce structural changes in Bax. Can the authors comment on the rational for using Triton here in light of this issue, and also comment on the possibility that triton may induce experimental artefacts in such a context and obfuscate the result. Response: We thank the reviewer for the valuable comments. In the revised manuscript, we have modified the GST pull-down assay by replacing triton with a buffer containing CHAPS, which has been reported not to alter the Bax or Bak conformation (ref 60-61). We then re-performed the experiments and the results are presented in the revised Figure 5c-e. The experimental procedure has been provided in the revised Methods section.
Typographical issues: Line # -comment 43through forming formation of inhibitory complexes 47networks 51source? Nevertheless -> However? 53 -,(called the BH3-binding pocket). 60 -This Therefore, the 85formatting (spacing), To improve the success rate of crystallisation?, 88of all these three 92 -We hereafter henceforth used 93representative model? 94the p53-DBD 98,99wording 104corner regions. 111in vitro or in silico analysis? 123formed much less too informal/innacurate 129proteins 131before a few times followed by multiple 134these 140had an about approximate 5-fold decreased in affinity 147assays was were 148formed the prominent predominant 153at the p53/BCL-2 159 -Taking Taken together, 170 -They would thus clash and prevent each other This could lead to competition and prevent either protein 213 while however the detailed mechanistic interaction 214mechanism 215 -What's more However 233differences 244in cells -> in vitro? 249a wild-type p53 activity contributes 262activity through by directly 278missing tag description? 287sonication/centrifugation details? 296 -Then, 298 -GL. 324 -To carried carry 332,333consistency of values i.e., Fifty microliters. Consider rewording. 370 -Surviving cells are were Response: We appreciate the reviewer's careful reading and valuable feedback. We have addressed the typographical issues and revised the grammar and references accordingly.
Reviewer #3 (Remarks to the Author): In article 405778 entitled "Structures of p53/BCL-2 complex suggest a mechanism for p53 to 1 antagonize BCL-2 activity" authors Wei, Wang et al. have determined three crystal structures of fusion proteins BCL-2-deltaTM-G-linker-p53DBD where they vary the linker size. All structures reveal that p53DBD binds to the BH3-binding groove of BCL-2, this mode of binding being different from the outside of the groove binding mode observed for BCL-xL-deltaTM-G-linker-p53DBD, which these authors previously characterized, and another study also resolved by NMR. The authors than produce mutants at the binding interface in both partners and characterize their binding to respective WT partner to validate the interactions site. Additionally, they selected a triple RRR mutant in p53DBD, which exhibited altered activity in one of their binding assays and reconstituted it in cells KO for p53 in the context of a transcriptionally inactive mutations previously described showing that it did not induce the typical markers of apoptosis. The direct regulation of the mitochondrial pathway at the BCL-2 family protein by cytosolic p53 has been a controversial area of investigation in the field that has been difficult to prove with existing knowledge and available reagents. Studies in this area therefore need to be carefully conducted. While the structures reveal how p53 DBD may interact with BCL-2 and seem to have adequate statistics and resolution, I find the functional probing preliminary and not rigorously executed.
Here are some specific examples of problem area: 1. The side chain of D186 does not participate in the water bridge interaction as shown in the electron density in SF3b, instead the D186 CO being implicated. Why would the authors select to replace this residue with D186R, which abolished the pull-down interactions with GST-BCL-2, when presumably the side chain does not participate in the interaction. Response: We thank the reviewer for pointing out the concern. In the structures determined in our study, main chain of D186 participate in the water-bridged interaction with BCL-2 residues R107 and Y108 in the structure of the 22-residue linked protein (revised Supplementary Figure 5a), while in the structures of the 16-residue or 19-residue linked proteins, the side chain of residue D186 is involved in the direct interactions with BCL-2 residues R107 or Y108 (revised Supplementary  Figure 5b-c). We, therefore, hypothesized that D186 could play an important role for the p53-DBD/BCL-2 interactions, and designed mutation D186R.
2. For such high EC50 values observed in ELISA assays and high IC50 values observed in competitive FP assays, why did the authors not use a direct binding assay to characterize the interactions as they have done in their previous paper? Response: Thanks for the comment. We attempted to characterize the interactions between p53-DBD and BCL-2 using Biacore X100 and ITC measurements. However, both recombinant p53-DBD and BCL-2 proteins, when present at high concentrations, tend to precipitate after prolonged exposure to the instruments at room temperature. In the revised manuscript, we employed MST method to characterize the direct binding ability. GST-p53-DBD was used to help stabilize the p53 proteins. The GST tag was tested and did not bind to the BCL-2 protein. After the MST experiments, the samples remained transparent. The results are presented in the revised Supplementary 3 and the revised Figure 3c-d. The descriptions and detailed methods have been added to the revised manuscript. The data from both methods have shown that these structure-based mutations decreased the binding ability of p53-DBD to BCL-2, and supported the p53-DBD/BCL-2#3 interface defined by the complex structure. We have also observed that the Kd values of MST are different from the EC50 values of ELISA (revised Supplementary 6), possibly due to the utilization of different methods.
3. The assays in Figure 2 and Figure 6 should be combined as it deals with characterizing the same interaction. It is unclear why the authors have not included the RRR p53DBD mutant in the latter Figure 6b. Response: We guess the issue raised here by the reviewer refers to Figure 3 and Figure 6 in the original manuscript. According to the reviewer's suggestion, we have adjusted the figures in the revised manuscript. The revised figures 2-3 show the detailed interactions of p53/BCL-2 complex, and the revised figures 4-5 show the effect of p53/BCL-2 complex on the interaction of BCL-2 with pro-apoptotic proteins. The results of the cell-based assay are shown in the revised Figure 6.
Regarding the competitive FPA (revised Figure 5b), the p53-DBD mutant RRR at 30 μM failed to cause a 30% change in the fluorescence signal of BCL-2/fBAX. Due to such a low signal-tonoise ratio, we did not proceed with the assay to obtain the IC50 value. Figure 6c and 6d it is remarkable that BAX peptide cannot fully displace p53DBD from BCL-2 suggesting possibly that other interactions may hold the complex in place. I cannot find the concentrations in the Methods section used in this assay which only show the stochiometric considerations. The authors need to include this and do a proper titration to dislodge the BCL-2-p53 complex with excess BAX BH3 or BAX protein. It does not seem that the interaction is titratable. Response: We appreciate the reviewer's comment. In the revised manuscript, we have adjusted the competitive GST pull-down assay according to the reviewers' suggestions. Detailed information on protein concentration has been provided in the revised Methods section. We have repeated the experiments with titration as suggested. A gradient dilution of p53-DBD was used to compete for BCL-2 with Bax (revised Figure 5d) or Bak proteins (revised Figure 5e). When the amount of p53-DBD is about 10-times higher than Bax (Figure 5d, lane 5) or Bak proteins (Figure 5e, lane 5), p53-DBD cannot fully displace Bax or Bak proteins from GST-BCL-2. We acknowledge that the interaction is not fully titratable, possibly due to other interactions that may hold the complex in place. We have included the information in the revised Results section. Figure 4b is very preliminary and needs to be better developed. Cell death assay should show extent to cell death. Additionally, the authors need to at least show that the constructs do not induced apoptosis in p53-/-bak-/-bax-/-cell lines to exclude an off target effect especially as others have shown cytosolic p53-/-mediated cell death in the absence of BAK and BAX (https://doi.org/10.1016/j.cell.2012.05.014). Response: We thank the reviewer for the comment. Based on the reviewer's suggestion, we have further developed the cell apoptosis assays (revised Figure 6, revised Supplementary Figure 8). The extent of cell death has been assessed by FACS analysis after staining with Annexin V-PE/7-AAD kit. To further analyze the target effect of Ti-p53, we knocked down Bax or/and Bak by small interfering RNA (siRNA) in Ti-p53 stably-expressing HCT116 cells, and then used FACS analysis to detect cell apoptosis (revised Supplementary Figure 9). Compared to Ti-p53 (19.6%), Bax siRNA and Bak siRNA decreased the apoptotic cell to 12.1% and 13.5%, respectively. With the simultaneous knockdown of Bax and Bak, there was still a significant level of cell apoptosis at 9.6%. This data suggested that Ti-p53-mediated apoptosis in HCT116 cells depended on both Bax and Bak, and Ti-p53 may also have other targets. We have added the information in the revised Results section.

The apoptosis data in
6. The authors need to characterize the homogeneity of the fusions used for crystallography on gel filtration chromatography (are these fusions forming oligomers?). They also need to acknowledge that crystallography could result in crystal contact artifacts, which may be non-native. For instance, if the fusions were p53DBD-G-linker-BCL-2-deltaTM they may lead to totally different crystal contacts. Therefore, the rigor of the investigation needs to be unquestionable. Response: Gel filtration chromatography to characterize the homogeneity of the fusions used for crystallography has been added in the revised supplementary Figure 2. In the revised Discussion section, we have acknowledged that crystallography could introduce non-native crystal contact artifacts.
Minor: 7. Figure 5 can follow immediately after Figure 2. Response: Thanks for the suggestions. We have adjusted the figures in the revised manuscript. The revised figures 2-3 present the detailed interactions of p53/BCL-2 complex, and the revised figures 4-5 illustrate the effect of p53/BCL-2 complex on the interaction of BCL-2 with pro-apoptotic proteins. Figure 6e is confusing with the p53-BCL-2 complex inhibiting the BAX-BCL-2 complex not making sense Response: We have removed the model based on the reviewer's suggestion and additional data incorporated in the revised manuscript.

The model in
Line 305 -artefacts Line 500 -Protein structure coordinate availability