UC Berkeley

This thesis describes studies exploring the evolution of the genetic circuits regulating yeast mating-type and silencing by Sir (Silent Information Regulator) proteins in the budding yeast Saccharomyces bayanus, a close relative of the laboratory workhorse S. cerevisiae (a.k.a., budding yeast, or brewer's yeast). The two central subjects of these studies, mating type and silencing, are textbook examples of "well understood" mechanisms of eukaryotic gene regulation: the former serves as a model for understanding the genetic control of cell-type differentiation, the latter serves as a model for understanding physically condensed, transcriptionally repressed portions of the genome, often referred to as "heterochromatin". The two subjects are intimately connected in the biology of the budding yeast life cycle, as explained below, and I argue that a deeper appreciation of this connection is necessary for further progress in the study of either subject. My thesis brings a critical evolutionary perspective to certain assumptions underlying current knowledge of mating-type regulation and silencing--in short, an appreciation of organismal biology that has been marginalized in the pursuit of understanding molecular mechanisms. The value of this perspective is in attempting to understand the purpose of a biological process--why is there such a thing as silencing, and why does it require the particular proteins and DNA elements that it does? To ask what silencing does for a yeast cell, we can start by asking how the silencing mechanism is constrained over evolutionary time. One of the surprising findings of my thesis is how unconstrained some elements of the silencing machinery are during evolution.

At least three major findings arise from the comparative genetics studies described here: First, I describe the first new branch of the mating-type control circuit in almost 25 years. Although alpha-specific genes were previously thought to be "off" in MATa cells due to the absence of the alpha1 activator protein (i.e., by default), I show that these genes are, in fact, actively repressed by the Sum1 protein. This novel regulatory branch highlights the sophisticated control mechanisms necessary to coordinate the mating and mating-type switching processes. This finding has additional implications, including questioning the extent to which the "absence of activator" model is sufficient to explain the absence of a particular gene's expression; and that at least one subset of mating genes may be under environmental or metabolic regulation via the Sum1-associated NAD+-dependent histone deacetylase Hst1.

Second, I show that at least two major genetic alterations to the Sir-based silencing machinery occurred in the recent ancestry of S. cerevisiae and its closest relative species. These changes reveal that our understanding of the silencing mechanism has been limited by the relative lack of comparative genetic sampling of the silencing process. That is, our understanding can improve via functional studies of silencing in close relatives of S. cerevisiae with variant silencing machinery, fueling new hypotheses about how silencing works. Although the identities of the major players (Sir1-4) largely remain the same, my discovery that certain silencing proteins are incompatible across closely related Saccharomyces species suggests evolutionary alterations in the genetic network of silencing--variation that could be tapped in future studies to understand better the way that silencing works. Of particular note are the rapid sequence evolution of SIR4, and the changes in copy number and sequence of SIR1, between S. bayanus and S. cerevisiae. SIR4 and SIR1 appear to rapidly evolve for interesting, though not completely overlapping, reasons. SIR4 appears to be under diversifying selection in modern yeast populations, and its coding sequence evolves rapidly across two rather distant clades spanning the Saccharomyces complex--the sensu stricto clade, and the Torulaspora clade.

Third, I show that Sir4 and silencers are engaged in a remarkable pattern of co-evolution in Saccharomyces yeasts. I used a novel combination of classical genetic techniques in S. cerevisiae/S. bayanus hybrids to test cis versus trans contributions to a genetic incompatibility between S. cerevisiae SIR4 and the S. bayanus HMR locus. Comparative ChIP-Seq of Sir4 in these hybrids helped identify the molecular basis for this incompatibility. Critically, I show that the S. bayanus HMR locus, when transferred into S. cerevisiae, can be silenced only by the specific combination of S. bayanus Sir4 and Kos3 proteins, with potential contributions by S. bayanus ORC and the other Sir1 paralogs. A striking asymmetry in cross-species compatibility of S. bayanus versus S. cerevisiae SIR4 genes, and in each species' Sir4 ChIP-Seq profile, suggests that compensatory changes have occurred in SIR4 and in silencers along the S. cerevisiae lineage. Although the initial evolutionary pressure(s) driving these rapid changes remains uncertain, my results point to some pressure driving either the silencers' or Sir4's rapid sequence change, with the other factor subsequently changing to maintain compatibility within a species. From a practical standpoint, these results suggest that molecular studies of silencing using only S. cerevisiae suffer from a previously unrecognized bias. That S. bayanus has four Sir1-like proteins, each important for silencing, suggests additional dimensions (i.e., temporal and/or spatial components) to the interactions occurring at silencers between Sir1, Sir4, ORC, and Rap1.

An interesting consequence of the comparative Sir4 ChIP-Seq experiments was the generation of a high-resolution picture of the architecture of silent chromatin in yeast. The unexpected non-uniform distributions of Sir4 protein across HML and HMR bring into question the standard "spreading" model for yeast silent chromatin formation, and will fuel future experiments to determine how Sir-based chromatin structures determine gene silencing and the epigenetic inheritance of gene expression states. I describe the novel ChIP-Seq picture of Sir protein association with silenced loci in Appendix A.

Finally, in addition to these specific biological insights, my comparative genetic studies provide guidelines for using the genetic variation between S. bayanus and S. cerevisiae as a tool to learn more about conserved genetic circuits and gene regulation mechanisms in general. Two substantial advances in evolutionary genetic techniques are presented in Chapters 3 and 4, which involve the use of yeast hybrids. First, I show that the genetic facility of S. cerevisiae/S. bayanus hybrids can be used to tease apart interspecies genetic variation of functional consequence that resides in cis-regulatory DNA elements from that in trans-acting transcriptional regulatory proteins. Second, in the case of silencing, the very act of re-introducing genetic factors that have been independently evolving for millions of years leads to unexpected, emergent phenotypes in the hybrids that can be used to understand the silencing mechanism itself. Lessons from my work should inform principles of comparative genetics using organisms closely related to classical "model organism" species such as S. cerevisiae.