UC Berkeley

In eukaryotes, the physical packaging of DNA into chromatin plays a major role in modulating gene expression. In the budding yeast Saccharomyces cerevisiae, silent chromatin, a repressive chromatin structure analogous to the heterochromatin found in animals, is required for cell mating-type identity. Silent chromatin forms at the cryptic mating type loci HML and HMR and depends on the interactions between the Silent Information Regulator SIR proteins and nucleosomes. A longstanding model for how Sir proteins build a repressive chromatin structure posits a stepwise process of histone tail deacetylation by Sir2, followed by recruitment of Sir3 and Sir4 to those deacetylated tails. However, this model is incomplete, as it does not account for the observed dependence of silencing establishment on progression through S phase of the cell cycle, which has been known for decades, but remained unexplained when I began these studies. I worked to resolve this mystery first by characterizing the establishment of silencing using modern molecular techniques and then by performing genetic analysis to identify the molecular determinants of silencing establishment.

Using the inducible allele SIR3-EBD, I demonstrated that silencing establishment has similar cell-cycle requirements at HML and HMR, which corrected an incorrect claim in the literature. This finding simplified the possible models that could explain an S-phase dependence for silencing establishment, as the models did not have to account for locus-specific effects. Using chromatin immunoprecipitation followed by sequencing (ChIP-seq), I confirmed the earlier finding that Sir protein recruitment and spread across HML and HMR does not require cell-cycle progression. However, I also showed that S phase promotes Sir protein binding beyond the level that can be obtained without cell-cycle progression. Using single-molecule RNA fluorescence in situ hybridization (smRNA-FISH) to study the establishment of silencing, I found that transcriptional repression occurs gradually in individual cells, not via discrete transitions between the expressed and repressed states. Thus, the model for silencing must account for intermediate levels of repression occurring during establishment. I found that cells lacking DOT1, SAS2, or RTT109, which code for anti-silencing histone modifying enzymes, could partially establish silencing without S phase passage. Dot1 was particularly interesting, because the anti-silencing mark it deposits, methylation of H3K79, lacks a known demethylase, and thus the mark can only be removed through histone turnover, which occurs mainly during S phase. Consistent with this expectation, I found that Sir proteins were unable to deplete H3K79 methylation from HML and HMR without passage through S phase. Together, these results suggest that removal of methylation from H3K79 is a major cell-cycle-dependent step in the establishment of silent chromatin.

While silencing establishment relies on incorporation of new silencing-competent histones during S phase, a long-standing hypothesis in the field of heterochromatin is that the epigenetic inheritance of the silent state depends on inheritance of modified parental histones during S phase. Recent work from the Rine lab has demonstrated that parental histones are locally redeposited after DNA replication at the non-Sir-bound GAL10 locus. Furthermore, mutations in the replisome components DPB3 and MCM2 severely reduce local histone inheritance at GAL10, but do not cause complete loss of epigenetic memory at silent loci, calling into question the model that local histone inheritance is required for epigenetic inheritance. A major open question is whether the mechanisms of histone position memory at GAL10 function similarly at the silent locus HML.

I found that histones are locally inherited at HML in wild-type cells and that this inheritance seems to be weakened in sir mutant cells. In the course of these experiments, I observed a synthetic sickness that occurs when cells lack both the histone inheritance factor Dpb3 and the histone deacetylases Hst3 and/or Hst4, which remove a histone modification that marks newly-synthesized histones. A genetic screen for suppressors of this synthetic phenotype yielded mutations in many proteasome subunits, suggesting that aberrant degradation of some protein, potentially H3 itself, is responsible for the sickness in these cells. The work on both local histone inheritance at HML and the interaction between DPB3 and HST3/4 is ongoing.