Modeling the regulatory network of histone acetylation in Saccharomyces cerevisiae

Abstract

Acetylation of histones plays an important role in regulating transcription. Histone acetylation is mediated partly by the recruitment of specific histone acetyltransferases (HATs) and deacetylases (HDACs) to genomic loci by transcription factors, resulting in modulation of gene expression. Although several specific interactions between transcription factors and HATs and HDACs have been elaborated in Saccharomyces cerevisiae , the full regulatory network remains uncharacterized. We have utilized a linear regression of optimized sigmoidal functions to correlate transcription factor binding patterns to the acetylation profiles of 11 lysines in the four core histones measured at all S. cerevisiae promoters. The resulting associations are combined with large‐scale protein–protein interaction data sets to generate a comprehensive model that relates recruitment of specific HDACs and HATs to transcription factors and their target genes and the resulting effects on individual lysines. This model provides a broad and detailed view of the regulatory network, describing which transcription factors are most significant in regulating acetylation of specific lysines at defined promoters. We validate the model, both computationally and experimentally, to demonstrate that it yields accurate predictions of these regulatory mechanisms. ### Synopsis Acetylation of histone N‐termini plays an important role in regulating transcription ([Kurdistani and Grunstein, 2003][1]). The modification of these termini is mediated by the recruitment of histone acetyltransferases (HATs) and deacetylases (HDACs) to specific genomic loci by transcription factors (TFs), resulting in either suppression or enhancement of gene expression. Although many specific interactions between TFs and HATs and HDACs have been elaborated in Saccharomyces cerevisiae , the full regulatory network remains uncharacterized. Here we propose a new approach to reconstruct this network from genome‐wide data sets. By analyzing genome‐wide acetylation profiles of 11 lysines and the binding profiles of 204 TFs using linear regression of optimized sigmoidal functions, we identify TFs that are correlated with the acetylation profiles of specific lysines. These associations allow us to hypothesize that the correlation might be due to the recruitment by the TF of a HAT or HDAC. We then utilize protein–protein interaction data to infer which HAT or HDAC is likely recruited by the factor in the few cases where these supporting data are available. The validity of the resulting network, shown in [Figure 1A and B][2], is supported by the observation that TFs that are correlated with hypoacetylation tend to recruit deacetylases while TFs that are correlated with hyperacetylation tend to recruit acetylases. When we compare the real network to random ones, we find that the enrichment of correct associations is very statistically significant. Furthermore, we have performed experimental tests of the hypotheses generated from the network. In particular, we have measured the changes in acetylation in three deletion mutants of TFs that were significant in our model, Sum1, Hir3 and Yml081W, along with three control factors, shown in [Figure 4][3]. We observe increased acetylation of two lysines in two promoters where Sum1 binds, but little or no effect on another lysine and a control promoter, as predicted by our model. We note a marked decrease in the acetylation of three lysines in the hir3 mutant at a histone gene promoter where Hir3 binds strongly. Finally we also test an uncharacterized factor that appears in our model and show that its deletion has a small but reproducible effect on acetylation, supporting our regulatory network. The combined computational and experimental validation of the network suggests that it accurately captures the regulation of acetylation levels of histone promoters in S. cerevisiae on a genome‐wide scale. Based on this model, we see that yeast cells have evolved a complex regulatory network to regulate this important epigenetic mechanism. Furthermore, we are now capable of generating multiple testable hypotheses about the regulation of acetylation levels of specific lysines at individual promoters. Finally, we expect that as chromatin immunoprecipitation techniques improve, we will be able to reconstruct the analogous network in human cells, thus generating new hypotheses regarding the connection between these and human disease. Mol Syst Biol. 3: 153 [1]: #ref-11 [2]: #F1 [3]: #F4