Regional promoter-independent gene silencing is critical in the establishment of cellular

identity in Saccharomyces. Domains of transcriptionally silent regions in the genome are

associated with certain heritable modifications made to chromatin, such as histone

hypoacetylation and methylation. In Saccharomyces cerevisiae, this type of gene repression

occurs through the activity of the four Silent Information Regulator, or SIR genes (SIR1-4). From

an evolutionary perspective, the SIR genes are unique: except for SIR2, all are specific to

budding yeasts. Many other organisms, from Schizosaccharomyces pombe to human, utilize the

RNA interference (RNAi) pathway, whereas most budding yeasts lack this pathway entirely.

Interestingly, SIR1, SIR3, and SIR4 are also rapidly evolving among Saccharomyces yeasts,

providing a model by which to examine the essential principles governing successful silencing

across various species and the relationship between rapid sequence evolution and evolution of

function.

To examine the relationship between gene duplication, extreme sequence divergence, and

functional evolution, I studied the SIR1 gene in S. cerevisiae and its most ancestral paralog,

KOS3, in the pre-whole-genome-duplication budding yeast, Torulaspora delbrueckii. T.

delbrueckii also possesses genes for RNAi, AGO1 and DCR1, allowing us the possibility of

exploring how the evolutionary divergence of RNAi and SIR silencing occurred. In the process, I

developed genetic tools for T. delbrueckii. To fully characterize SIR1 function in S. cerevisiae

and SIR gene function in T. delbrueckii, I utilized chromatin immunoprecipitation followed by

deep-sequencing (ChIP-Seq) of tagged Sir proteins in both species. This strategy allowed for the

discovery of potential novel functions, as well, revealing functions that may have been gained or

lost throughout SIR1’s evolution. To identify loci that were directly repressed by Sir proteins, I

also generated whole-transcriptome data by performing mRNA-Seq on wild-type and sir mutants

in both species.

Collectively, these data revealed that though SIR1 in both species is still involved in

silencing, its role in that process has dramatically shifted. Previous data suggested that SIR1 is

primarily associated with the establishment or nucleation phase of silencing and not involved in

telomeric silencing. The Sir1 ChIP data in S. cerevisiae corroborated this assessment. In T.

delbrueckii, however, KOS3 was essential for silencing, and was also found at telomeres. Thus,

Sir1 in its early evolution had a more essential role in silencing; this role may have changed due

to the duplication and diversification of the other Sir complex members. This diversification may be contributing to the continual change in interactions between Sir1 and other Sir complex

members across budding yeasts, leading to different mutant phenotypes in each species. Assays

of silencer function in T. delbrueckii answered critical questions about when in the phylogeny

important shifts in transcription factor binding sites took place. My work showed that the arrival

of the Rap1, ORC, and Abf1 binding sites in the silencers of budding yeasts took place prior to

the whole-genome duplication event. Analysis of silencer structure also revealed the diversity of

chromatin architecture in budding yeasts: S. cerevisiae silent mating type loci have two silencers

on either side of each locus, whereas in T. delbrueckii, there appears to be a single silencer on

one side of each mating type locus. Transcriptome analysis of RNAi mutants revealed that this

pathway in T. delbrueckii does not function in heterochromatic gene silencing, suggesting that

this pathway has already been repurposed for some other biological process.

The examination of whole-transcriptome data in S. cerevisiae in conjunction with the

enrichment patterns of the Sir proteins at telomeres allowed us to evaluate widely accepted

models regarding the molecular architecture of heterochromatin and expression at S. cerevisiae

telomeres. I established that repression of gene expression at native telomeres is not as

widespread as previously thought, and that many genes in proximity to regions of Sir protein

enrichment were, in fact, expressed just as equally in wild type as they were in sir mutant genetic

backgrounds. However, twenty-one genes were convincingly repressed by Sir proteins,

highlighting the complex and individual nature of native telomeres and subtelomeric genes. The

sensitivity of RNA-Seq also uncovered a previously under-appreciated class of haploid-regulated

genes: genes that were not fully repressed or de-repressed in the diploid a/α-cell type, but rather

weakly repressed or de-repressed. Thus, my work has expanded the set of known a/α-regulated

genes in S. cerevisiae. In conclusion, this dissertation has broadened our understanding of the

functional constraints dictating silencing gene evolution across species that diverged prior to and

after the whole-genome-duplication event. My data speaks to the actual chromatin architecture

and expression state of native S. cerevisiae telomeres, leading to the refinement of existing

models and an appreciation for how heterogeneous these regions of the genome can be.

A core assumption in chromatin biology is that nucleosomes store and transmit epigenetic memory of gene regulation through cell division. A requirement of this model is that nucleosomes – modified in place in one generation – continue to occupy the same DNA locus in the subsequent generation. In my graduate work, I asked whether nucleosomes remember their position through cell division. To execute my experiments, I developed a synthetic chromatin modifying enzyme that deposited high affinity covalent chromatin modification at a defined locus. The chromatin-modifying enzyme could be delocalized within minutes by the addition of a small molecule antibiotic, and the synthetic chromatin modification could be tracked for several hours after preventing new label deposition. Using this synthetic enzyme, I labeled nucleosomes at the GAL10 locus in S cerevisiae and tracked their position through DNA replication. I found that nucleosomes remained localized through DNA replication, suggesting that nucleosomes can transmit epigenetic. Furthermore, I found that in the absence of the fork associated nucleosome chaperones Mcm2 and Dpb3, nucleosomes did not remain localized, suggesting that mcm2 dpb3 mutants cannot transmit epigenetic memory through chromatin. In addition to exploring the fate of nucleosomes during DNA replication, I tracked nucleosomes through transcription and discovered that contrary to published models, nucleosomes are not translated linearly along an open reading frame during transcription.

In the thesis that follows, I discuss the context for our collective intuition about the role of chromatin in transmitting epigenetic memory, and I identify the points at which our collective understanding rests on thin or contradictory data. I then discuss my attempts to develop and apply technology to reach a satisfactory resolution of one of the central unanswered questions in chromatin biology: do nucleosomes remember their position? Lastly I discuss side projects adjacent to chromatin biology that occupied my attention during the slower moments of my core thesis research.

Transcriptional silencing in the budding yeast Saccharomyces cerevisiae occurs at the cryptic mating-type loci HML and HMR on chromosome III and at all 32 telomeres. Transcriptional silencing occurs through the formation of a repressive chromatin structure, featuring nucleosome compaction and removal of active chromatin marks. These features are the result of the activity of the Silent Information Regulator (SIR) complex, made up of Sir2, Sir3, and Sir4. Sir2, founding member of the wide spread class of protein deacetylases known as sirtuins, deacetylates histone tails, whereas Sir3 and Sir4 serve structural roles, binding to histones and compacting chromatin.

The formation of silent chromatin by the SIR complex conceptually involves two steps: recruitment and spreading. Recruitment, also referred to as nucleation, occurs at DNA sequence elements called silencers that are present at telomeres and flank HML and HMR. The E and I silencers that flank both HML and HMR contain different combinations of binding sites for the Origin Recognition Complex (ORC) and the general transcription factors Rap1 and Abf1. Silencers at telomeres are less well characterized, but include an array of Rap1 binding sites in telomere repeats and possibly ORC/Abf1 sites further into the chromosome. Fully silent chromatin displays SIR complex binding beyond recruitment sites for multiple kilobases. The difference in SIR complex occupancy between nucleation and full silencing occurs through an ill-defined process referred to as ‘spreading’.

To date, most studies on spreading of the SIR complex have focused on telomeres, sometimes only one, and relied on low-resolution ChIP-PCR, inducible systems that resulted in massive overexpression of Sir3, or both. These studies defined to a limited resolution positions of SIR complex binding and occupancy, and provided a foundation for genome-wide studies with higher temporal and spatial resolution. My work further characterized and distinguished the two processes of recruitment and spread at telomeres as well as HML and HMR. I accomplished this by developing a new method for tracing the history and trajectory of SIR complex binding: measuring DNA methylation by long-read nanopore sequencing of DNA from cells expressing a fusion protein between Sir3 and an N6-methyladenosine methyltransferase, M.EcoGII.

The fusion protein Sir3-M.EcoGII strongly and specifically methylated HML, HMR, and telomeres and was able to detect transient or low-affinity Sir3-chromatin interactions better than ChIP-seq. This new method allowed me to characterize the occupancy of a SIR3 allele encoding a protein deficient in binding to nucleosomes, sir3-bah∆. The sir3-bah∆-M.EcoGII fusion protein methylated DNA only at recruitment sites, clearly providing evidence that recruitment and spread of Sir3 were separable processes and that the interaction between Sir3 and nucleosomes was not required for recruitment but was required for spread. I also tested prior claims that overexpression of SIR3 results in binding of the protein at even longer distances from recruitment sites. The overexpression of SIR3-M.ECOGII, with few exceptions, did not extend regions of DNA methylation.

I also defined the dynamics of Sir3 spreading during silencing establishment and how its occupancy related to transcriptional silencing of HML and HMR. A fusion between sir3-8–a temperature sensitive allele of SIR3–and M.ECOGII allowed for regulated induction of DNA methylation without straying above the endogenous level of SIR3 expression. Over the course of about one cell cycle, methylation appeared only at the E and I silencers and the promoters of HML and HMR, demonstrating recruitment. Despite a lack of Sir3 occupancy between these recruitment sites, repression of transcription occurred early in the time course, suggesting that the early stages of silencing did not require Sir3 occupancy across the entire locus.

Once silent chromatin is established, it must be faithfully inherited through the disruptive process of DNA replication. Certain point mutations in PCNA (POL30), the processivity clamp for DNA polymerase at replication forks, result in loss of transcriptional silencing at HML and HMR. I used classical genetics to study three of these alleles, pol30-6, pol30-8, and pol30-79, in more detail. All three alleles disrupted silencing only in actively-cycling cells, and the disruption in silencing was only transient, suggesting that the inheritance of silent chromatin through cell division was not as robust as in wild-type cells. All three alleles of POL30 destabilized silencing through disrupting the function of histone chaperones, highlighting the importance of histone trafficking at the replication fork for the stability of transcriptional silencing.

Abstract

Intersections of Nutrient Availability, Heterochromatic Silencing, Growth, Aging and DNA Damage Repair in Saccharomyces cerevisiae

By

David Frankel McCleary

Doctor of Philosophy in Molecular and Cell Biology

University of California, Berkeley

Professor Jasper D. Rine, Chair

Calorie restriction extends lifespan in organisms as diverse as yeast and mammals through incompletely understood mechanism(s). The role of NAD+-dependent deacetylases, known as Sirtuins, in this process, particularly in the yeast S. cerevisiae is controversial.

In Chapter 1, I measured chronological lifespan of wild-type and sir2Δ strains over a higher glucose range than typically used for studying yeast calorie restriction. sir2Δ extended lifespan in high-glucose synthetic complete medium, and had little effect in low-glucose medium, revealing a partial role for Sir2 in the calorie restriction response under these conditions. I measured growth, glucose consumption, and pH of these cultures and revealed a mild effect of sir2Δ that inhibited growth, lowered glucose consumption, and increased extracellular pH. Buffering of growth medium eliminated the sir2Δ-dependent extension of chronological lifespan, as did replacing stationary-phase media with water. Replacing medium with water at stationary phase actually led to a sir2Δ-dependent reduction in lifespan in high-glucose. Growth on high levels of other fermentable sugars still resulted in the sir2Δ-dependent extension of lifespan, whereas growth on high levels of glycerol, a non-fermentable carbon source, did not allow for sir2Δ-dependent lifespan extension. When cells were grown in high- or low-glucose yeast peptone medium, large amounts of stationary-phase adaptive regrowth were observed, especially in high-glucose cultures. This regrowth complicated analysis of chronological aging. Interestingly, high-glucose yeast peptone cultures had dramatically decreased initial viability as compared to low-glucose yeast peptone or any other media tested. Also of interest, adaptively regrown cells from high-glucose yeast peptone cultures displayed enhanced growth, a trait that was heritable.

In Chapter 2, I measured chronological lifespan of wild-type and sir2Δ strains in rich yeast peptone medium with a newly-developed genetic strategy. This approach revealed that sir2Δ shortened lifespan in low glucose while having little effect in high glucose, revealing a partial role for Sir2 in calorie restriction-mediated lifespan extension in yeast peptone medium. In complete minimal media like synthetic complete, Sir2 shortened lifespan as glucose levels increased, whereas in rich media, Sir2 extended lifespan as glucose levels decreased. Both of these phenotypes are consistent with roll(s) for Sir2 in the calorie restriction response. Using a second genetic strategy to measure the strength of gene silencing at HML, I determined increasing glucose stabilized Sir2-based silencing during growth on complete minimal media. Conversely, increasing glucose destabilized Sir-based silencing during growth on rich media, specifically during late cell divisions. In rich medium, silencing was far less stable in high glucose than in low glucose during stationary phase. Therefore, Sir2 is involved in a response to nutrient cues including glucose that regulates chronological aging, possibly through Sir2-dependent modification of chromatin or deacetylation of a non-histone protein.

In Chapter 3, I investigated the origins of a color change of the growth medium at stationary phase when sir2Δ cells were grown on 3% glycerol synthetic complete medium. sir2Δ cultures eventually turned a yellow-brown color while wild-type cultures remained near-white throughout stationary phase. To determine whether the color change was caused by wild-type and sir2Δ cells having different levels of anabolism or catabolism of any small molecules, I performed gas chromatography and mass spectroscopy on filtered medium from wild-type and sir2Δ stationary-phase cultures. Quinoline and several chemical derivatives were highly enriched in filtered sir2Δ medium relative to wild-type. These molecules in solution are reported to have a yellow-brown color, consistent with my observations. GC-MS analysis of fresh 3% glycerol synthetic complete medium revealed that these quinoline compounds were present before cultures were seeded. Thus, wild-type cultures catabolized these molecules at a higher rate than sir2Δ cultures, removing them from the extracellular environment. The quinoline compounds identified here are biosynthesized from tryptophan in some organisms, and are closely related to known UV-degradation products of tryptophan. Closely related compounds including kynurenine and quinolinic acid are also chemical intermediates of the yeast kynurenine pathway, which generates NAD+ from tryptophan. Since sirtuins including Sir2 are the only enzymes that consume NAD+ in yeast, loss of Sir2 could lead to less kynurenine pathway activity, limiting consumption of quinoline compounds that formed either through UV-degradation or catabolism of tryptophan.

In Chapter 4, I exploited a microfluidics approach to investigate yeast replicative lifespan using the same calorie restriction paradigm used to investigate chronological lifespan in past chapters. I used a microfluidics chip, provided by Dr. Hao Li, which was designed to trap aging mother cells while washing away smaller, newborn daughter cells. By filming multiple positions on the chip over the course of a single experiment, I was able to count the number of offspring given off by dozens of cells at once. Surprisingly, wild-type cells had the same replicative lifespan in high- and low-glucose, defying the logic of calorie restriction and the results of past studies of the effects of calorie restriction on replicative lifespan. Since classical replicative lifespan experiments involve micromanipulating daughter cells away from a stationary mother cell growing on solid media, that mother cell is experiencing a changing environment as it metabolizes essential nutrients locally and generates waste. In the case of my microfluidics approach, mother cells are experiencing a static environment, since fresh media is constantly flowing through the chip. Therefore, a changing extracellular environment is likely necessary for calorie restriction to increase replicative lifespan. Additional experiments will be necessary to determine this with certainty.

Finally, in Chapter 5, we investigate a report claiming Sir2 protein was recruited to dicentric chromosomes under tension, and such chromosomes are reported to be especially vulnerable to breakage in sir2Δ mutants. We found that the loss of viability in such mutants was an indirect effect of the repression of non-

homologous end joining in Sir- mutants, and that the apparent recruitment of Sir2 protein to chromosomes under tension was likely due to methodological weakness in early chromatin immunoprecipitation studies.

Compaction of the eukaryotic genome into regions of open euchromatin and dense heterochromatin serves as a fundamental regulator of gene expression in cells. In the budding yeast Saccharomyces cerevisiae, the Silent information regulator (Sir) protein complex regulates a heterochromatin-like structure at the silent mating type loci, HML and HMR, and at the telomeres. In addition to the Sir proteins, Repressor activator protein 1 (Rap1) functions in establishing and maintaining silent chromatin at HML and HMR by binding to nucleation sites known as silencers (silent enhancers) and recruiting silencing machinery. As its name suggests, Rap1 is simultaneously an essential transcription factor that activates hundreds of genes across the genome, including many ribosomal protein genes and the MATα1 and α2 genes whose expression determine the mating-type of α cells. Sequence identity between the mating-type locus MAT and the auxiliary HML allows a unique opportunity to study the role of Rap1 in both silent and expressed contexts, and how local chromatin state affects function. Chapter two focuses on discerning the context-dependent functions of Rap1. Using ChIP-seq I established that Rap1 accessed its binding site at the promoter of HML, even in the silent context. My findings supported and extended the view that pre-initiation complex machinery is unable to act in silent chromatin, thus narrowing the mechanism of silencing to a step between recruitment of the native activator, Rap1, and occlusion of the pre-initiation complex. Surprisingly, Rap1 enrichment at its three binding sites across the silent locus was enhanced by the presence of Sir proteins, despite binding of this transcription factor preceding Sir protein recruitment. As Rap1 was bound in the silent locus but was not functionally recruiting transcription machinery, I tested whether the presence of this enigmatic protein could instead be enhancing silencing. Utilizing a highly sensitive assay that monitors loss-of-silencing events, I established a novel and specific contribution of promoter-bound Rap1, which was previously seen as having potential only for activation, to silencing. Furthermore, I investigated the mechanism by which Rap1 promotes transcription when acting as an activator and have evidence that it may be aiding in the transition from transcription initiation to elongation. ChIP offers a static view of protein-DNA interactions rather than a dynamic readout. To assess whether in vivo dwell time of Rap1 might contribute to its function as an activator or repressor, I coupled a nuclear depletion strategy with a ChIP-seq time course to measure enrichment at and decay of Rap1 from its binding sites genome wide. The apparent Rap1 dwell time did not differ between silenced HML and expressed MAT. However, genome-wide Rap1 residence time correlated with transcriptional output and nucleosome positioning. These results point toward a model in which the duality of Rap1 function is mediated by local chromatin environment rather than binding-site availability. I have conducted a number of experiments attempting to pinpoint the mechanism by which Rap1 switches function, following up mainly on post-translational modifications. Some of this work is described in the appendix. While study of Sir silencing at the HM loci has been integral in understanding the establishment and maintenance of heterochromatin in eukaryotes, the effects of these same proteins at telomeres has been less clear. Chapter three focuses on an incomplete set of experiments designed to better understand the functional similarities and differences in Sir-mediated silencing at telomeres and the HM loci. Early studies of telomere position effect in yeast led to a hypothesis that distance from telomere dictated the amount of gene repression imparted by Sir proteins in a gradated manner. At the time I began working on the project, there was mounting evidence against this hypothesis. I designed and performed preliminary experiments testing whether silencing was a function of distance from telomere utilizing fluorescent reporters for high throughput analysis. Furthermore, I investigated whether telomere length affected silencing of HML and HMR by titrating available Sir proteins away from these loci. I also pioneered usage of a novel form of single molecule RNA FISH in our lab and I planned to use this technique to assess whether subtelomeric genes, which are enriched for metabolic function, were spatially co-regulated in response to environmental stimuli.

Plants have acquired and maintained an expanded suite of DNA-dependent RNA polymerases (RNAPs) compared to other eukaryotes. Although their exact roles remain unclear, plant-specific RNAPs (Pol IV and Pol V) are involved in epigenetic silencing of transposable elements (TEs). Zea mays (maize) Pol IV is required for proper plant development as well as the establishment and maintenance of paramutations, which are trans-homolog interactions that facilitate heritable gene silencing. Maize has duplications of Pol IV catalytic subunits, which define multiple Pol IV subtypes, and accessory proteins associated with these subtypes define distinct Pol IV complexes. Understanding the roles of Pol IV will require identifying the composition and function of these Pol IV subtypes and complexes. In exploring interactions between the genes encoding a Pol IV catalytic subunit and a putative accessory protein, I identified two new alleles of the Pol IV subunit and a family of potential Pol IV accessory proteins conserved in multicellular plants. Together, these Pol IV complexes are proposed to function through two, potentially overlapping, general mechanisms to control of gene function: direct competition with Pol II and/or generation of 24-nucleotide RNAs that guide de novo cytosine methylation. I analyzed nascent transcriptome data of seedlings lacking Pol IV and their wild-type siblings, which identified a global effect of Pol IV on gene boundary transcription. Pol IV-affected loci serve as molecular and phenotypic models for dissecting the involvement of various Pol IV subtype and complex components. Such analysis implicated a Pol IV-affected allele as being controlled by Pol IV competition with

Pol II. Through expanded sequencing of a paramutation participating allele (Pl1-Rhoades), I identified a new Pol IV target of five tandem repeats that may serve as an enhancer element. I used the Pl1-Rhoades allele in particular to compare and contrast the effects of Pol IV loss with that of Pol IV accessory proteins. Together, the work presented here explores the involvement of an uncharacterized protein family of putative Pol IV accessory proteins in

Pol IV complexes, expands the roles of Pol IV complexes in controlling maize gene function, and identifies new Pol IV-affected loci for future study.

Background

Results

Conclusions