UC Berkeley

Many cells possess a remarkable ability to pass on non-genetic information to daughter cells. In some cases, the transmission of epigenetic information is clearly linked to processes such as DNA methylation, RNAi, or transcription factor feedback loops. In other cases, a mechanism for epigenetic memory clearly exists but has evaded discovery. One such example is the inheritance of heterochromatin. Heterochromatic gene silencing often requires silencing machinery that can both deposit and bind to specific histone modifications, sometimes referred to as writers and readers of epigenetic information, coupled with inheritance of histones during DNA replication. Therefore, a feedback loop between silencing factors and modified histones could allow a silenced state to be “remembered” through DNA replication. However, multiple studies argue that this intuitive model is either incorrect or incomplete, as discussed in Chapter 1 of this manuscript. During my thesis work, I investigated the histone-based memory model by studying epigenetic states in S. cerevisiae. In Chapter 2, I tested two predictions of the histone-based memory model. One prediction builds on the observation that histone H3-H4 tetramers are randomly distributed between daughter chromatids during DNA replication. This phenomenon suggests that rare events of asymmetric H3-H4 tetramer inheritance could cause one daughter chromatid to receive an insufficient number of modified parental tetramers, and thereby lose the silenced state. In this case, reductions in heterochromatin domain size would be predicted to increase the frequency of asymmetric inheritance and silencing loss. However, I found that severe reductions in heterochromatin domain size had no effects on the rate of silencing inheritance. Additionally, two mutations that led to severe reductions in H3-H4 tetramer inheritance had minimal effects on the inheritance of silencing. Finally, combining these tetramer-inheritance defects with reductions in heterochromatin domain size still permitted robust inheritance of the silenced state. Therefore, reducing the number of inherited H3-H4 tetramers and reducing the number of H3-H4 tetramers available for inheritance had surprisingly minor effects on the inheritance of silencing. These findings argue that the histone-based memory model cannot explain epigenetic memory. In Chapter 3, I asked whether robust nucleosome positioning can contribute to the efficiency of silencing. To test this, I made a series of deletions between two nucleosome-depleted regions (NDRs) in a silenced domain, which would be predicted to affect nucleosome packing. These deletions revealed a striking oscillatory relationship between inter-NDR distance and silencing stability. I found that instances of robust silencing corresponded to efficient nucleosome packing, and weak silencing corresponded to poorly positioned nucleosomes. Additionally, organized nucleosome packing correlated with an enhanced ability to both establish and inherit a silenced epigenetic state. These findings argue that proper NDR placement in heterochromatin is important for silencing and suggest that well-organized nucleosome arrays provide a better substrate to produce a silenced structure. Under normal conditions, mating-type loci in S. cerevisiae are constitutively silenced. In Chapter 4, I worked with a talented undergraduate/research technician, Delaney Farris, to perform a screen for mutations that cause constitutive silencing to become metastable. This screen uncovered many mutations, as expected, in SIR1. Surprisingly, this screen also identified a point mutation, sir2-G436D, that caused cells to exhibit bistable silencing at the single-cell level. Interestingly, whereas bistable silencing normally manifests as a mix of cells that are either fully silenced or fully expressed, sir2-G436D exhibited a mix of cells that were either fully silenced or partially silenced. This is the first time that a heritable intermediate silenced state has been documented, to our knowledge. Additionally, given that Sir2 is a histone deacetylase, this finding suggests that histone modifiers are important for silencing to mature from an intermediate silenced state to a fully silenced state. This idea is consistent with previous studies, and the sir2-G436D mutant provides an exciting avenue to examine the structure of chromatin that can achieve intermediate levels of repression and how that structure might be inherited through DNA replication.