Unveiling the dimer/monomer propensities of Smad MH1-DNA complexes
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (Review Commons)
Abstract
Article activity feed
-
-
-
Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Reply to the reviewers
__Review comments Rebuttal __
Reviewer #1 (Evidence, reproducibility and clarity (Required)):
__**Summary ** __ This manuscript describes the X-ray structure determination of two SMAD-DNA complexes confirm that SMAD family proteins bind at least two DNA sequences in a similar fashion, and explores dimer versus monomer formation of the non-DNA bounds forms of the proteins which could influence whether the proteins bind as monomers and dimers. This includes identifying a loop which appears to make a major contribution to this process. There is a lot of experimental work and analysis included.
__**Major comments: ** __The overall conclusions of the manuscript …
Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Reply to the reviewers
__Review comments Rebuttal __
Reviewer #1 (Evidence, reproducibility and clarity (Required)):
__**Summary ** __ This manuscript describes the X-ray structure determination of two SMAD-DNA complexes confirm that SMAD family proteins bind at least two DNA sequences in a similar fashion, and explores dimer versus monomer formation of the non-DNA bounds forms of the proteins which could influence whether the proteins bind as monomers and dimers. This includes identifying a loop which appears to make a major contribution to this process. There is a lot of experimental work and analysis included.
__**Major comments: ** __The overall conclusions of the manuscript are convincing, but some of the detailed analysis is not clear. The structures look good, the experiments look to be generally well controlled, although some details could be provided in the main text to be clear about what methodology is being used or how analysis was carried out and stepwise conclusions obtained.
In particular the analysis of SAXS data is not clear. I'd like to see initial data analysis presented as per the guidelines of Trewhella et al 2017 (PMID: 28876235). There is some mention of data in the SASREF database, but it should be in the supplemental data.
We have prepared a table following this recommendation.
I can't see any evidence for the conclusions about open versus closed monomer state (how good were the fits obtained) - just a graph and a statement. If this can't be better justified please remove the conclusions about these states (they don't really add to the overall conclusions about monomer/dimer which are much less specific), but even the simple analysis supports mostly monomer and small amounts of dimer or higher aggregates. I would also like to see a clear explanation provided about why the MS data supports dimer over other oligomers
We have revised and simplified the SAXS section to clarify the main points. We have re-analyzed the conformations in solution, and the values are presented in new Table S4 and new Figure 3D. We have also included new panels (Figure 3E) and explanations with respect to the IM-MS data (pages 8-9).
State what thermal unfolding experiments are were carried out in the text (and why is the data biphasic?)
The biphasic graphics were interpreted as the presence of dimers and monomers in equilibrium. As suggested by the other reviewer, we have removed these sections as they do not contribute to clarify the main points of our work.
The concept of long versus short loops re domain swapping have been studied in the past but there isn't much reference to this.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2373619/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6349918/
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0039305
https://www.ncbi.nlm.nih.gov/pubmed/22411444
We now mention some related examples in page 8.
**Minor comments: **
__ __The last couple of paragraphs of the introduction are a fairly comprehensive summary of the study overall and the conclusions of the paper. While presaging the key findings and conclusions is fairly common in an introduction this seems to be way too much detail. Unless it is a requirement of the journal, reduce these sections to a couple of sentences and use any other word count to explain your analysis better.
Thank you for this recommendation. We have rephrased and reduced this part of the introduction.
Figures are quite small and hard to see detail at 1X magnification (in both the main and Supplemental figures).
We have removed some panels that were not necessary and increased the size of the figures and labels.
NB.The difference in Tm of SMAD 5 over 8 doesn't seem particularly high as it’s only a couple of degrees (especially when SMAD4 is quite different). The explanation for the Ile>Cys mutation might be about competition of zinc ligation (except that it doesn't seem to cause issues for many zinc finger proteins) but more likely that you've replaced a reasonably bulky hydrophobic sidechain and therefore have lost a bunch of hydrophobic contacts.
We have removed this section entirely.
With respect to the Ile>Cys difference, the residue is located in a loop, and it does not participate in hydrophobic contacts. We still believe that its negative role in the stability of the domains arises from the competition for Zn coordination but we agree with the reviewer that quantifying its specific role is not obvious.
Reviewer #1 (Significance (Required)):
-This paper clarifies concepts about the state of isolated SMAD proteins (thought be largely monomeric in the absence of DNA) and DNA-binding preferences of these proteins.
-I don't have specific expertise in the structure/function of SMAD proteins, but the study appears to include sufficient background to place the study in context.
-Audience will mostly be those interested in structure/function of SMAD proteins, with some protein engineers interested in the manipulation of monomeric versus dimer.
-I am a protein chemist and structural biologist with an interest in protein dimerization/oligomerization. I am familiar with most techniques presented, but don't have first-hand experience with IM-MS.
Reviewer #2 (Evidence, reproducibility and clarity (Required)):
**Summary**
The manuscript by Ruiz et al examines how the receptor-activated SMAD (R-SMAD) transcription factors bind DNA, and specifically how the MH1 (DNA-binding domain) of the different classes contributes determining whether they bind as monomers or dimers.
In the context of the full length SMAD proteins, it is thought that hetero-trimers with one SMAD4 plus two R-SMADs are the functional unit. In general, the SMAD1/5/8 R-SMADs respond to BMPs whereas SMAD2/3 respond to TGF beta. However, what is less clear is how the specificity of the gene responses is determined, since all SMADS are able to bind to each of the two sequence classes of response element (GTCT and GC-rich, or 5GC).
Previous structural studies suggest that the major contacts between SMAD MH1 and DNA are very similar, irrespective of the particular SMAD or of whether they bind a TGTC or 5GC element. On DNA, MH1 domains have been observed as dimers, but there has been some concern as to whether this (at least in part) is a crystal artefact, or is perhaps forced by the specific DNA sequences use in these studies. For the BMP R-SMADS this may be less likely, since the amino-terminal helix 1 of one dimer is seen to be dislodged from its own intramolecular interactions allowing it to make contacts with the second MH1 domain in the dimer.
Here the authors test this question of MH1 dimerization and address differences between the BMP responsive and other SMADs. They first show by crystallography that SMAD5 and SMAD8 MH1 domains adopt similar dimeric conformations with the displaced helix 1, and bind to a single 5GC element via one of the MH1 domains. To get at whether these MH1 domains form dimers in solution, they use small angle X-ray scattering, NMR and mass spectrometry, to suggest that the SMAD5 and SMAD8 MH1 domains in solution do not fit with a single conformation, but are better modeled by a mixture of dimer and open monomer. Ion mobility MS also suggested a mix of dimer and open monomer for the BMP SMADs, whereas SMAD3 appeared to be primarily monomer. To test if the MH1 domains themselves encode this potential difference between SMAD5 and SMAD3, they swap loop 1 (6 versus 4 amino acids, between helices 1 and 2) from SMAD3 to SMAD5, and now in solution this chimera appears monomeric, and forms monomers when crystalized with DNA.
____Major comments____
- Adding the SMAD3 loop to SMAD5 prevents the open dimer - does the reverse also work? Can you make SMAD3 form SMAD5-like open-dimers by adding the loop 1 sequence from SMAD5?
We have prepared new Smad3 chimeric constructs and we are currently screening crystallization conditions in order to obtain diffracting crystals (if possible). Unfortunately, due to the COVID-19 pandemic, access to our laboratory is highly restricted, while access to synchrotron and mass spectrometry facilities is not available), therefore this work has been postponed until the end of April/May. For this reason, the revised version of the manuscript does not refer to this question. We hope that we will be able to address it in the future.
- Can the authors include similar schematic models for how the site spacing would be for SMAD2/3-SMAD4 complexes - adding the SMAD2/3/4 model to Figure 5C?
We have incorporated new panels to Figure 5 (Figure 5E,F).
- The authors comment on the possibility that the dimer conformation dictates the spacing of the sites that will be bound in vivo. In this context, they refer to a previous paper (PMID: 29234012) to suggest differences in site clustering between BMP SMAD and TGF beta SMAD regions of the genome (from ChIP-seq) that fit with the spacing they imply here. However, the major difference shown in this work seems to be between the clustering of GC sites and GTCT sites irrespective of the pathway. Can the authors analyze existing ChIP-seq data to more specifically test the question they raise - ie that SMAD4 bound regions of the genome have different site clustering/spacing depending on whether they are BMP or TGF beta responsive?
Thank you very much for this recommendation. We agree with the reviewer that this information is very valuable and can help support our hypothesis on the different binding preferences of monomers and dimers of MH1 domains. We have performed this analysis and is now included as two new sections. The results are displayed as new Figure 5A,B.
- I think Figure 2C,D is not really well described in terms of the importance to this work. As it is this data does not really seem to add very much, but perhaps I am missing the importance.
We have entirely removed this section in the new version of the manuscript.
5.Can the authors comment about the compressed GC element or BRE? This seems to be an unfavorable conformation. How might it be bound in vivo, is it an unusual element, or is it relatively widely found? Is it possible that in vitro it binds two MH1 domains, but in vivo might simply act as a normal 5GC, with an additional site nearby?
The BRE domain is less abundant than 5GCs and SBE sites, and in fact, this sequence is not enriched in the ChIP-Seq datasets that we have analyzed. We have included a sentence refering to these findings in page 11.
We have also revised the section comparing the 5GC and BRE-GC site and illustrate this interaction as well as the comparison to our 5GC complex by including two panel in Figure2 that before were displayed as supplementary information. The panels have been edited to clarify the similarities and differences between both complexes. Indeed, the protein-DNA complex made of one BRE motif bound by two MH1 domains as found in the PDB:5X6H crystal structure suffers from several issues, including compactness of the two overlapping Smad-binding motifs that led to distortion of DNA geometry, clashes at protein-protein interface and local cancelation of protein-DNA interactions.
In this new section (page 8) we include the sentence that “we believe that the most probable binding mode in vivo should be that observed in the 5GC and SBE complexes. It seems very unlikely that two MH1 domains would interact with a reduced BRE motif —using half of their protein binding site and causing a high distortion to the DNA structure— if there is the possibility to interact with neighboring sites (Figure 2B,C) using the full protein binding interface and a perfect accommodation to the DNA”.
**Minor comments: **
1.In Figure 1B is one the two DNAs assumed? In the structure was it two MH1 to one DNA or two of each?
One DNA was hidden for clarity. The crystallographic structure is now shown in full. The crystal structure was solved for a complex made of two MH1 domains bound to a dsDNA molecule that included two Smad-binding sites.
2.Figure 2C and page 9: the stabilization of SMADs in the text and figure do not agree. Maybe just state the exact numbers from the figure in the text.
We have entirely removed this section in the new version of the manuscript.
3.In Figure S1C, can the authors label the retarded complexes on the gels?
Done.
4.Figure 4A - explain the asterisk (presumably the SMAD2 insert).
Yes, it corresponds to the Gly rich region present in loop1. It is indicated now.
5.In Figure 4B, C (and maybe D) can they color helix 1, loop 1, and helix 2 three separate colors, it might really emphasize the effect of the loop if it was more immediately visible.
We have improved these figures but we did not change the colors because the figure was getting even more complicated.
6.The legend to Figure 4 is missing F.
Thank you. This has now been corrected.
Reviewer #2 (Significance (Required)):
The authors conclude that the length of the loop between helices 1 and 2 determines the dimer versus monomer state - a shorter loop as in the BMP SMADS hinders the intramolecular interactions needed for the closed monomeric form, whereas the longer loop in the other SMADS allows the flexibility for these interactions so favors a more closed monomeric form. Showing that the dimers are not forced by crystallography or by binding to fixed DNA elements clearly adds to our understanding of the mechanisms of SMAD function, and it is of interest that the BMP and TGF beta SMADS are different in this respect.
They speculate that this may contribute to the specificity of the responses activated by BMP versus TGF beta signaling based on the requirements for different site spacing depending on whether an open (BMP) or closed (TGF beta) dimer of R-SMADS is present. This idea is likely to be of interest to anyone who studies the responses to the TGF beta superfamily of signaling molecules, and should spur additional experimentation to test it.
-
Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #3
Evidence, reproducibility and clarity
Summary:
The manuscript by Ruiz et al examines how the receptor-activated SMAD (R-SMAD) transcription factors bind DNA, and specifically how the MH1 (DNA-binding domain) of the different classes contributes determining whether they bind as monomers or dimers.
In the context of the full length SMAD proteins, it is thought that hetero-trimers with one SMAD4 plus two R-SMADs are the functional unit. In general, the SMAD1/5/8 R-SMADs respond to BMPs whereas SMAD2/3 respond to TGF beta. However, what is less clear is how the specificity of the gene responses is determined, since all SMADS are able to bind to each of the two sequence …
Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #3
Evidence, reproducibility and clarity
Summary:
The manuscript by Ruiz et al examines how the receptor-activated SMAD (R-SMAD) transcription factors bind DNA, and specifically how the MH1 (DNA-binding domain) of the different classes contributes determining whether they bind as monomers or dimers.
In the context of the full length SMAD proteins, it is thought that hetero-trimers with one SMAD4 plus two R-SMADs are the functional unit. In general, the SMAD1/5/8 R-SMADs respond to BMPs whereas SMAD2/3 respond to TGF beta. However, what is less clear is how the specificity of the gene responses is determined, since all SMADS are able to bind to each of the two sequence classes of response element (GTCT and GC-rich, or 5GC).
Previous structural studies suggest that the major contacts between SMAD MH1 and DNA are very similar, irrespective of the particular SMAD or of whether they bind a TGTC or 5GC element. On DNA, MH1 domains have been observed as dimers, but there has been some concern as to whether this (at least in part) is a crystal artefact, or is perhaps forced by the specific DNA sequences use in these studies. For the BMP R-SMADS this may be less likely, since the amino-terminal helix 1 of one dimer is seen to be dislodged from its own intramolecular interactions allowing it to make contacts with the second MH1 domain in the dimer.
Here the authors test this question of MH1 dimerization and address differences between the BMP responsive and other SMADs. They first show by crystallography that SMAD5 and SMAD8 MH1 domains adopt similar dimeric conformations with the displaced helix 1, and bind to a single 5GC element via one of the MH1 domains. To get at whether these MH1 domains form dimers in solution, they use small angle X-ray scattering, NMR and mass spectrometry, to suggest that the SMAD5 and SMAD8 MH1 domains in solution do not fit with a single conformation, but are better modeled by a mixture of dimer and open monomer. Ion mobility MS also suggested a mix of dimer and open monomer for the BMP SMADs, whereas SMAD3 appeared to be primarily monomer. To test if the MH1 domains themselves encode this potential difference between SMAD5 and SMAD3, they swap loop 1 (6 versus 4 amino acids, between helices 1 and 2) from SMAD3 to SMAD5, and now in solution this chimera appears monomeric, and forms monomers when crystalized with or without DNA.
Major comments:
1.Adding the SMAD3 loop to SMAD5 prevents the open dimer - does the reverse also work? Can you make SMAD3 form SMAD5-like open dimers by adding the loop 1 sequence from SMAD5?
2.Can the authors include similar schematic models for how the site spacing would be for SMAD2/3-SMAD4 complexes - adding the SMAD2/3/4 model to Figure 5C?
3.The authors comment on the possibility that the dimer conformation dictates the spacing of the sites that will be bound in vivo. In this context they refer to a previous paper (PMID: 29234012) to suggest differences in site clustering between BMP SMAD and TGF beta SMAD regions of the genome (from ChIP-seq) that fit with the spacing they imply here. However, the major difference shown in this work seems to be between the clustering of GC sites and GTCT sites irrespective of the pathway. Can the authors analyze existing ChIP-seq data to more specifically test the question they raise - ie that SMAD4 bound regions of the genome have different site clustering/spacing depending on whether they are BMP or TGF beta responsive?
4.I think Figure 2C,D is not really well described in terms of the importance to this work. As it is this data does not really seem to add very much, but perhaps I am missing the importance.
5.Can the authors comment about the compressed GC element or BRE? This seems to be an unfavorable conformation. How might it be bound in vivo, is it an unusual element, or is it relatively widely found? Is it possible that in vitro it binds two MH1 domains, but in vivo might simply act as a normal 5GC, with an additional site nearby?
Minor comments:
1.In Figure 1B is one the two DNAs assumed? In the structure was it two MH1 to one DNA or two of each?
2.Figure 2C and page 9: the stabilization of SMADs in the text and figure do not agree. Maybe just state the exact numbers from the figure in the text.
3.In Figure S1C, can the authors label the retarded complexes on the gels?
4.Figure 4A - explain the asterisk (presumably the SMAD2 insert).
5.In Figure 4B, C (and maybe D) can they color helix 1, loop 1, and helix 2 three separate colors, it might really emphasize the effect of the loop if it was more immediately visible.
6.The legend to Figure 4 is missing F.
Significance
The authors conclude that the length of the loop between helices 1 and 2 determines the dimer versus monomer state - a shorter loop as in the BMP SMADS hinders the intramolecular interactions needed for the closed monomeric form, whereas the longer loop in the other SMADS allows the flexibility for these interactions so favors a more closed monomeric form. Showing that the dimers are not forced by crystallography or by binding to fixed DNA elements clearly adds to our understanding of the mechanisms of SMAD function, and it is of interest that the BMP and TGF beta SMADS are different in this respect.
They speculate that this may contribute to the specificity of the responses activated by BMP versus TGF beta signaling based on the requirements for different site spacing depending on whether an open (BMP) or closed (TGF beta) dimer of R-SMADS is present. This idea is likely to be of interest to anyone who studies the responses to the TGF beta superfamily of signaling molecules, and should spur additional experimentation to test it.
-
Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #2
Evidence, reproducibility and clarity
Summary:
This manuscript describes the X-ray structure determination of two SMAD-DNA complexes confirm that SMAD family proteins bind at least two DNA sequences in a similar fashion, and explores dimer versus monomer formation of the non-DNA bounds forms of the proteins which could influence whether the proteins bind as monomers and dimers. This includes identifying a loop which appears to make a major contribution to this process. There is a lot of experimental work and analysis included.
Major comments: The overall conclusions of the manuscript are convincing, but some of the detailed analysis is not clear. The structures look …
Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #2
Evidence, reproducibility and clarity
Summary:
This manuscript describes the X-ray structure determination of two SMAD-DNA complexes confirm that SMAD family proteins bind at least two DNA sequences in a similar fashion, and explores dimer versus monomer formation of the non-DNA bounds forms of the proteins which could influence whether the proteins bind as monomers and dimers. This includes identifying a loop which appears to make a major contribution to this process. There is a lot of experimental work and analysis included.
Major comments: The overall conclusions of the manuscript are convincing, but some of the detailed analysis is not clear. The structures look good, the experiments look to be generally well controlled, although some details could be provided in the main text to be clear about what methodology is being used or how analysis was carried out and stepwise conclusions obtained.
In particular the analysis of SAXS data is not clear. I'd like to see initial data analysis presented as per the guidelines of Trewhella et al 2017 (PMID: 28876235). There is some mention of data in the SASREF database, but it should be in the supplemental data. I can't see any evidence for the conclusions about open versus closed monomer state (how good were the fits obtained) - just a graph and a statement. If this can't be better justified please remove the conclusions about these states (they don't really add to the overall conclusions about monomer/dimer which are much less specific), but even the simple analysis supports mostly monomer and small amounts of dimer or higher aggregates. I would also like to see a clear explanation provided about why the MS data supports dimer over other oligomers State what thermal unfolding experiments are were carried out in the text (and why is the data biphasic?) The concept of long versus short loops re domain swapping have been studied in the past but there isn't much reference to this.
Minor comments: The last couple of paragraphs of the introduction are a fairly comprehensive summary of the study overall and the conclusions of the paper. While presaging the key findings and conclusions is fairly common in an introduction this seems to be way too much detail. Unless it is a requirement of the journal reduce these sections to a couple of sentences and use any other word count to explain your analysis better. Figures are quite small and hard to see detail at 1X magnification (in both the main and Supplemental figures). NB The difference in Tm of SMAD 5 over 8 doesn't seem particularly high as its only a couple of degrees (especially when SMAD4 is quite different). The explanation for the Ile>Cys mutation might be about competition of zinc ligation (except that it doesn't seem to cause issues for many zinc finger proteins) but more likely that you've replaced a reasonably bulky hydrophobic sidechain and therefore have lost a bunch of hydrophobic contacts.
Significance
-This paper clarifies concepts about the state of isolated SMAD proteins (thought be largely monomeric in the absence of DNA) and DNA-binding preferences of these proteins.
-I don't have specific expertise in the structure/function of SMAD proteins, but the study appears to include sufficient background to place the study in context.
-Audience will mostly be those interested in structure/function of SMAD proteins, with some protein engineers interested in the manipulation of monomeric versus dimer.
-I am a protein chemist and structural biologist with an interest in protein dimerization/oligomerisation. I am familiar with most techniques presented, but don't have first hand experience with IM-MS.
-
Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.
Learn more at Review Commons
Referee #1
This reviewer did not leave any comments
-
-