Binding to nucleosome poises human SIRT6 for histone H3 deacetylation

Reviewer #1 (Public Review):

Smirnova et al. present a cryo-EM structure of a nucleosome-SIRT6 complex to understand how the histone deacetylase SIRT6 deacetylates the N-terminal tail of histone H3. The authors obtained the structure at sub-4 Å resolution and can visualize how interactions between the nucleosome and SIRT6 position SIRT6 to allow for H3 tail deacetylation. Through additional conformational analysis of their cryo-EM data, they reveal that SIRT6 positioning is flexible on the nucleosome surface, and this could accommodate the targeting of certain H3 tail residues. This work is significant as it represents the visualization of a histone deacetylase on its native nucleosomal target and reveals how substrate specificity is achieved. Importantly, it should be noted that recently two additional structures of the nucleosome-SIRT6 complex were already published. Therefore, Smirnova et al. confirm and complement these previous findings. Additionally, Smirnova et al. expand our understanding of the structural flexibility of SIRT6 on the nucleosome and clarify that SIRT6 also shows histone deacetylase activity on H3K27Ac.

Reviewer #2 (Public Review):

Smirnova et al. present a cryo-EM structure of human SIRT6 bound to a nucleosome as well as the results from molecular dynamics simulations. The results show that the combined conformational flexibilities of SIRT6 and the N-terminal tail of histone H3 limit the residues with access to the active site, partially explaining the substrate specificity of this sirtuin-class histone deacetylase. Two other groups have recently published cryo-EM structures of SIRT6:nucleosome complexes; this manuscript confirms and complements these previous findings, with the addition of some novel insights into the role of structural flexibility in substrate selection.

Author Response

The following is the authors’ response to the original reviews.

We are grateful to the reviewers for their remarks, which significantly improved the paper. We repeated the biochemical assay concerning SIRT6 activity on H3-K27Ac and quantified the results as requested. Please find our detailed answers bellow each recommendation of the reviewers.

Major recommendations:

1. Grammatical errors are still common; the authors may need to consider an external editing service if they intend to fix the problems as they indicate that they believe the errors have been removed. The Results section is relatively clean, but parts of the Abstract, Introduction, and Discussion are more difficult to understand, and errors are especially common in the Methods section and those parts of the manuscript that are new in this revision.

We corrected the grammatical errors.

1. The introduction doesn't mention the other structures published; this is considered to be a serious deficiency as it prevents the reader from understanding the context for the contributions described here. Withholding the comparison with (or mention of) the previously published work to the last sentence of the Discussion seems misleading and does not give the reader adequate ability to judge the novelty of the results presented in this manuscript.

A paragraph comparing our paper to the other structures published appear at the end of the discussion. We feel this is still the right place for such a paragraph.

1. The addition of the assay for deacetylation is a significant improvement over the initial submission. This is important both for validating the importance of the acidic patch contacts and for helping to resolve the conflicting reports regarding activity on H3-K27Ac. Given the importance of this assay for the impact of the manuscript, it is not clear why the authors chose to 1) put the data in the supplement instead of in the main manuscript, and 2) provide only single samples without quantitation. These both seem to be significant limitations.

We repeated the experiment and provided quantification of the results. We placed the figure in the main manuscript.

1. The authors should add text or a table to the Methods section explaining which maps were used for each figure. By our count, there are 8 maps and 5 models (plus MD models) based on two datasets, but the relationships among them are not clearly stated, and the names of the maps (such as "Zn-finger focused" and "Rossman-Fold-Focused") might be changed to be more helpful to the reader (for example, the latter includes more than the Rossman fold and might be renamed "Sirt6-focused"). The authors should also explain how the maps were validated, which data were deposited in public repositories, and why some data were not deposited. For example, no statistics or methods regarding how particles were separated into integrated vs. non-integrated motion are provided for the CryoDRGN models. Further, the "two principle movements" described are depicted in 4 maps from two CryoDRGN runs using two separate sets of particles, but the relationships among them are not defined clearly. Finally, the connectivity of densities in Fig 8 are not obvious in the submitted maps. Until these points are addressed, the work is considered incomplete.

AND

1. The PDB model provided for review and submitted to the PDB database shows loosely bound DNA at the nucleosomal entry/exit points near the binding site of SIRT6, but the maps provided for review and submitted to the EMDB show stronger density for the canonical location of the DNA expected at these sites. The CryoDRGN maps support a more extended conformation, but these maps were not deposited or provided for review so their validity cannot be assessed.

We added a section to the methods listing the different maps used for the figures. We deposited the map we used to trance the H2A N-terminal tail (EMD-18497). Unfortunately, we couldn’t deposit the cryoDRGN maps as the deposition system either accepts composite maps, where the consensus should be deposited too or experimental maps, where the deposition of half maps are mandatory. Nevertheless, the cryoDRGN maps are available upon request. We also added a supplementary figure (Supplementary Fig 6) to show how the cryoDRGN analyses were performed.

1. The orientation, angle and threshold used in Fig 1 make it difficult to see the multiple DNA orientations that are visible in the deposited consensus map. Examination of the map suggests that the DNA model submitted to PDB corresponds to a weaker DNA conformation than is present in the map where both DNA conformations are visible. The authors should consider modeling both conformations in their deposited model to provide a more complete, accurate representation of the data. It is concerning that a key conclusion of the manuscript is that the DNA conformation changes upon SIRT6 binding, but density for the canonical position is observable in Fig 8a.

Figure 1 is showing the overall representation of the SIRT6 bound nucleosome structure. We show the DNA linker orientations in the subsequent figure. Figure 8 (now Figure 9) shows the rearrangement of the SIRT6 Rossmann fold domain not the DNA linker.

1. Figure 4 needs a more complete legend, indicating that it is a hybrid of the consensus structure (one color) and the MD simulations (another color). In general, the colors used in the figure should be changed to make the main points more accessible.

As there is a color code for the histones, changing colors might be confusing. The figure legend mentions that panels c, d and e are from MD simulations.

Minor recommendations:

1. Figures 2c, e, and f are not referenced in the text.

We now referenced all figure panels in the text.

1. Consider moving Supp. 5C to Fig. 2 as the models in that figure come from the CryoDRGN maps and not the consensus map.

Supplemental Figure 5c show the DNA linker deviation upon SIRT6 binding from another angle. We prefer to keep it there.

1.) Supp Fig 3 is labeled "ZnF-nucleosome" refinement, but this appears to come from Data Set #2 processing. The map might be labeled ZnF-nucleosome but then a mask should be shown that excludes the Rossman Fold. It is not clear if this is a focused refinement or just a 2.9 A map that was merged with the "Rossman-fold" map.

We changed both supplemental figures accordingly.

1. The orientation of Fig 2 b and e do not show the differences in these models as well as panels c and f. Panels b and e could be replaced with the 4 CryoDRGN maps.

The models reflect the cryoDRGN maps and panels c and f were added to clarify the movement.

1. The MD description should emphasize that the H3 tails are moving with respect to the active site, as it currently suggests the active site is moving.

In the results and in the discussion section we mention that we observe new conformations of the H3 tail, not of the active site.

1. The authors refer to the "flexibility of the Rossmann fold domain," but the Rossman Fold domain isn't flexible, the linkage to the ZnF is flexible. Perhaps "observed conformational space" or "dynamic Rossman-fold domain position" are meant.

The text was changed accordingly.

1. The H2A C-terminal tail present in Fig 1 (bottom right) and Figure 3e is not present in the model in Fig 4a,b.

The H2A tails conformation was not resolved in the cryoDRGN maps so we didn’t model it.

1. The crosslinking agent used is not specified.

The crosslinking agent used is specified more clearly in the methods.

1. Supp Table 1 and EM methods do not agree on the magnification for Dataset #1. Verify nominal versus binned magnification and reported pixel size.
   The magnification in the methods was changed.

1. Fig 3F showing the difference between affinity for H2A and H2A.Z-containing nucleosomes would be more convincing with a titration rather than the current comparison of a single concentration.

We agree with this remark however, we find single concentration comparison is convincing enough for the purposes of this paper as it is not a central finding.

1. Fig S1 legend; both the Zn-finger and helix bundle are stated to be shown in green.

Figure S1 legend was changed.