Structure and flexibility of the yeast NuA4 histone acetyltransferase complex

  1. Stefan A Zukin
  2. Matthew R Marunde
  3. Irina K Popova
  4. Katarzyna M Soczek
  5. Eva Nogales
  6. Avinash B Patel

  • Curated by eLife

    eLife logo

    eLife assessment

    This manuscript provides insights into the architecture of the yeast histone acetyltransferase complex NuA4 and is of broad interest to those studying transcription and chromatin modification. The cryo-EM data are of very high quality, and enable the authors to devise a structural model that is in much better agreement with biochemical data than previously reported models. This structure represents an important puzzle piece towards a molecular understanding of chromatin modification.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The NuA4 protein complex acetylates histones H4 and H2A to activate both transcription and DNA repair. We report the 3.1-Å resolution cryo-electron microscopy structure of the central hub of NuA4, which flexibly tethers the histone acetyltransferase (HAT) and Trimer Independent of NuA4 involved in Transcription Interactions with Nucleosomes (TINTIN) modules. The hub contains the large Tra1 subunit and a core that includes Swc4, Arp4, Act1, Eaf1, and the C-terminal region of Epl1. Eaf1 stands out as the primary scaffolding factor that interacts with the Tra1, Swc4, and Epl1 subunits and contributes the conserved HSA helix to the Arp module. Using nucleosome-binding assays, we find that the HAT module, which is anchored to the core through Epl1, recognizes H3K4me3 nucleosomes with hyperacetylated H3 tails, while the TINTIN module, anchored to the core via Eaf1, recognizes nucleosomes that have hyperacetylated H2A and H4 tails. Together with the known interaction of Tra1 with site-specific transcription factors, our data suggest a model in which Tra1 recruits NuA4 to specific genomic sites then allowing the flexible HAT and TINTIN modules to select nearby nucleosomes for acetylation.

Article activity feed

  1. Version published to 10.7554/elife.81400 on eLife
  2. eLife

    Author Response

    Reviewer #2 (Public Review):

    Zhukin et al., present the structure of the central scaffold component of the NuA4 complex. They hypothesise how the nucleosome interacting modules not present in the structure could be arranged, based on Alphafold modelling, and comparison of their structure to other complexes that use the same subunits. They show some interesting -albeit fairly preliminary - biochemistry on the binding of the flexible modules, suggesting a role for acetylation affecting H3K4me3 reading.

    While the work builds upon previous structural studies on the Tra1 subunit in isolation and a previous 4.7A resolution structure from another group, there are clear differences and novel findings in this study. The data is presented beautifully and nicely annotated figures make following the many subunits and interactions therein simple. What could have been a very complex manuscript is easy to digest. Some of the figures could do with a couple of additional labels and detailed figure legends to make things a little clearer.

    Overall, a nice study and a wonderfully detailed structure of a large multi-subunit assembly but we would recommend some further experimentation validation to bolster their findings.

    Major comments

    1. All 13 subunits of NuA4 are present by mass spec, however, based on the SDS-page gel (Fig1-1) components of the TINTIN sub-complex seem less than stoichiometric, with Eaf7 and Eaf3 certainly much weaker stained. This is particularly important with reference to Figure 3 and the discussion in the text which assumes the nucleosome interacting modules are all present equally, but too flexible to be observed in the structure.

    Simple peptide numbers from mass spec cannot be used as a measure of protein abundance as this is sensitive to multiple confounding factors.

    We did not identify the locations of individual modules (HAT, TINTIN and Yaf9) within the diffuse density, we merely indicate that this is a likely location for their presents based on the location of connections points and presence of crosslinks in previously published data. We did perform mass photometry analysis of the purified NuA4 sample to better determine the composition of the purified complex (Figure 1-1). We find that the major species peak is center at 1037 kDa, which is very close to the theoretical mass of 1043 kDa. There are a few other minor peaks but none of this would indicate a NuA4 complex lacking TINTIN (Eaf3,5,7) or any other distinct subcomplex.

    1. A major novel biological finding and conclusion from the abstract concerns the binding to modified nucleosomes. However, this seemed somewhat preliminary, especially considering the discussion around the role of acetylation affecting binding to H3K4me3 nucleosomes based solely on the dCypher screen used.

    The discussion on the role of HAT module binding preferential to acetylated and methylated tails concludes that the acetylation liberates the H3 tail from DNA interaction, making H3K4me3 more available for binding by the PHD domain. This is an interesting hypothesis but is stated as fact with very little evidence to make this assertion.

    Whilst others have seen similar results (cited in the paper), no data is presented to disregard an alternative hypothesis that there is some additional acetyl-binding activity in the complex. Indeed, in one of the references they cite the authors do show a direct reading of acetylation as well as methylation.

    TINTIN binding is subject to high background and a fairly minor effect. The biological relevance to these observations while intriguing needs to be proved further.

    We have changed the language of this section to hopefully better leave open other possibilities. As for the TINTIN dCypher results, we do not try to draw too many conclusions, but the data indicates that there is very little (if any) interaction with the histones tails (at least for the modifications present in the panel). One thing we can say is that the TINTIN module does not seem to have any binding preference for H3K36me3 nucleosomes.

    1. There is a large focus on the cross-linking mass spec study from another group and the previously published structure of the NuA4 complex. The authors are fairly aggressive in suggesting the other structure from Wang et al., is incorrect. It is very nice that their built structure shows a better interpretation of previous XL-MS data, but still many of the crosslinks are outside of the modelled density. One possibility that should be entertained is that the two studies are comparing different structures/states of NuA4. The authors of the Wang et al., paper indeed comment that Swc4 and Yaf9 are missing from their purified complex. It is of course possible that both structures are correct as they appear to be biochemically different, with the crosslinking in the Setiaputra paper better reflecting the complex presented here.

    Response given above.

  3. eLife

    eLife assessment

    This manuscript provides insights into the architecture of the yeast histone acetyltransferase complex NuA4 and is of broad interest to those studying transcription and chromatin modification. The cryo-EM data are of very high quality, and enable the authors to devise a structural model that is in much better agreement with biochemical data than previously reported models. This structure represents an important puzzle piece towards a molecular understanding of chromatin modification.

  4. eLife

    Reviewer #1 (Public Review):

    Zukin and colleagues present a high-resolution cryo-EM structure of the yeast histone acetyltransferase complex NuA4, which acetylates histones H4 and H2A. The structural data is of very high quality and was obtained using state-of-the-art methodology. The resulting structural model comprises the rigid "Hub" of the NuA4 complex, consisting of a core module and the Tra1 subunit, while the functional TINTIN and HAT modules remain unresolved, likely due to high flexibility. Nevertheless, the structure provides detailed insights into the architecture of the NuA4 complex and reveals how the subunits in the Hub interact. The authors supplement the structural data with functional characterization of the binding of reconstituted TINTIN and HAT modules to modified nucleosomes, which reveals different specificities of the two. In combination, these data lead to a model for chromatin binding and modification by the NuA4 complex.

    Notably, the structural model presented by the authors here differs from a previous structure of the NuA4 core in several key details, including the assignment of densities to subunits (Wang et al., Nat Comm 2018). This is supported by two key lines of evidence. First, the structural data presented here is of higher resolution. Second, the new model presented here is in good agreement with available cross-linking data. Therefore, the revised model presented here is very likely to be more accurate than previous structural models.

    One "downside" (if one wishes) of the structural data is the lack of defined density for the HAT and TINTIN modules. However, this is not a shortcoming of the experimental approach employed here but is caused by the inherently flexible nature of this complex. Thus, this is not something that could easily be improved. Indeed, as the authors point out by comparison to the SAGA complex, flexible tethering of the functional modules appears to be common among chromatin-modifying complexes. This issue is elegantly addressed by the authors through a detailed analysis of AlphaFold predicted structures of subcomplexes of the HAT and TINTIN modules, which are in good agreement with previous cross-linking data. This analysis supports the assumption that the poorly defined density observed by the authors originates from these modules.

    Taken together, this is a very well-executed study that provides important insights into the molecular basis of chromatin modification. The conclusions drawn by the authors are supported by the structural data. The model for the mechanism of histone acetylation derived by the authors is very plausible based on the available data but remains somewhat speculative in the absence of experimental structural data for the HAT and TINTIN domains in complex with their substrates as well as functional data for the complete NuA4 complex. However, these data provide an important milestone towards a mechanistic understanding of chromatin acetylation and will serve as a framework for addressing the open questions in the future.

  5. eLife

    Reviewer #2 (Public Review):

    Zhukin et al., present the structure of the central scaffold component of the NuA4 complex. They hypothesise how the nucleosome interacting modules not present in the structure could be arranged, based on Alphafold modelling, and comparison of their structure to other complexes that use the same subunits. They show some interesting -albeit fairly preliminary - biochemistry on the binding of the flexible modules, suggesting a role for acetylation affecting H3K4me3 reading.

    While the work builds upon previous structural studies on the Tra1 subunit in isolation and a previous 4.7A resolution structure from another group, there are clear differences and novel findings in this study. The data is presented beautifully and nicely annotated figures make following the many subunits and interactions therein simple. What could have been a very complex manuscript is easy to digest. Some of the figures could do with a couple of additional labels and detailed figure legends to make things a little clearer.

    Overall, a nice study and a wonderfully detailed structure of a large multi-subunit assembly but we would recommend some further experimentation validation to bolster their findings.

    Major comments

    1. All 13 subunits of NuA4 are present by mass spec, however, based on the SDS-page gel (Fig1-1) components of the TINTIN sub-complex seem less than stoichiometric, with Eaf7 and Eaf3 certainly much weaker stained. This is particularly important with reference to Figure 3 and the discussion in the text which assumes the nucleosome interacting modules are all present equally, but too flexible to be observed in the structure.

    Simple peptide numbers from mass spec cannot be used as a measure of protein abundance as this is sensitive to multiple confounding factors.

    1. A major novel biological finding and conclusion from the abstract concerns the binding to modified nucleosomes. However, this seemed somewhat preliminary, especially considering the discussion around the role of acetylation affecting binding to H3K4me3 nucleosomes based solely on the dCypher screen used.

    The discussion on the role of HAT module binding preferential to acetylated and methylated tails concludes that the acetylation liberates the H3 tail from DNA interaction, making H3K4me3 more available for binding by the PHD domain. This is an interesting hypothesis but is stated as fact with very little evidence to make this assertion.

    Whilst others have seen similar results (cited in the paper), no data is presented to disregard an alternative hypothesis that there is some additional acetyl-binding activity in the complex. Indeed, in one of the references they cite the authors do show a direct reading of acetylation as well as methylation.

    TINTIN binding is subject to high background and a fairly minor effect. The biological relevance to these observations while intriguing needs to be proved further.

    1. There is a large focus on the cross-linking mass spec study from another group and the previously published structure of the NuA4 complex. The authors are fairly aggressive in suggesting the other structure from Wang et al., is incorrect. It is very nice that their built structure shows a better interpretation of previous XL-MS data, but still many of the crosslinks are outside of the modelled density. One possibility that should be entertained is that the two studies are comparing different structures/states of NuA4. The authors of the Wang et al., paper indeed comment that Swc4 and Yaf9 are missing from their purified complex. It is of course possible that both structures are correct as they appear to be biochemically different, with the crosslinking in the Setiaputra paper better reflecting the complex presented here.
  6. Version published to 10.1101/2022.06.24.497536v1 on bioRxiv