PLoS Genetics (May 2020)

Multiple mechanisms regulate H3 acetylation of enhancers in response to thyroid hormone.

  • Stine M Præstholm,
  • Majken S Siersbæk,
  • Ronni Nielsen,
  • Xuguang Zhu,
  • Anthony N Hollenberg,
  • Sheue-Yann Cheng,
  • Lars Grøntved

DOI
https://doi.org/10.1371/journal.pgen.1008770
Journal volume & issue
Vol. 16, no. 5
p. e1008770

Abstract

Read online

Hormone-dependent activation of enhancers includes histone hyperacetylation and mediator recruitment. Histone hyperacetylation is mostly explained by a bimodal switch model, where histone deacetylases (HDACs) disassociate from chromatin, and histone acetyl transferases (HATs) are recruited. This model builds on decades of research on steroid receptor regulation of transcription. Yet, the general concept of the bimodal switch model has not been rigorously tested genome wide. We have used a genomics approach to study enhancer hyperacetylation by the thyroid hormone receptor (TR), described to operate as a bimodal switch. H3 acetylation, HAT and HDAC ChIP-seq analyses of livers from hypo- and hyperthyroid wildtype, TR deficient and NCOR1 disrupted mice reveal three types of thyroid hormone (T3)-regulated enhancers. One subset of enhancers is bound by HDAC3-NCOR1 in the absence of hormone and constitutively occupy TR and HATs irrespective of T3 levels, suggesting a poised enhancer state in absence of hormone. In presence of T3, HDAC3-NCOR1 dissociates from these enhancers leading to histone hyperacetylation, suggesting a histone acetylation rheostat function of HDACs at poised enhancers. Another subset of enhancers, not occupied by HDACs, is hyperacetylated in a T3-dependent manner, where TR is recruited to chromatin together with HATs. Lastly, a subset of enhancers, is not occupied directly by TR yet requires TR for histone hyperacetylation. This indirect enhancer activation involves co-association with TR bound enhancers within super-enhancers or topological associated domains. Collectively, this demonstrates various mechanisms controlling hormone-dependent transcription and adds significant details to the otherwise simple bimodal switch model.