Embryonic development and differentiation produce organisms with many cell types whose identities are stably maintained over numerous cell divisions. Maintenance of cell identity depends on epigenetic control mechanisms that are linked to the assembly of specialized chromatin structures. Genes that are located in silent heterochromatic DNA domains display variegated or bistable on and off expression states. These states, which are maintained during cell division, are examples of epigenetic states that result from changes in chromatin structure. Research in my laboratory is focused on understanding the mechanisms that are involved in the formation, function, and inheritance of heterochromatin.
We apply a combination of approaches—ranging from biochemistry and cell biology to proteomics and genomics—to study heterochromatin in budding and fission yeasts. In addition, we are beginning to investigate how changes in gene expression that are induced by neuronal activity contribute to the formation of stable memories.
RNAi-Mediated Assembly of Heterochromatin
RNA interference (RNAi), a post-transcriptional gene silencing mechanism, is triggered by double-stranded RNA (dsRNA) which is processed to small interfering RNA (siRNA) by the Dicer ribonuclease. The RNAi machinery participates in heterochromatin formation in plants, fission yeast, Caenorhabditis elegans, and Drosophila, and related small RNA pathways are required for DNA methylation and transposon silencing in germline cells of vertebrates.
From fission yeast Schizosaccharomyces pombe, we have purified the RNA-induced transcriptional silencing (RITS) complex, which directly links the RNAi pathway to heterochromatin assembly. The RITS complex contains the sole S. pombe Argonaute protein, the chromodomain protein Chp1, and the GW-repeat protein Tas3. In addition, the RITS complex is loaded with Dicer-generated heterochromatic siRNAs, which are required for the spreading of histone H3 lysine 9 (H3K9) methylation in the pericentromeric repeat regions. We have identified a second RNAi complex, termed the RNA-dependent RNA polymerase complex (RDRC), which contains an RNA-directed RNA polymerase (Rdp1), a putative poly(A) polymerase (Cid12), and a helicase (Hrr1). The RDRC is recruited via interactions with the RITS complex and is responsible for the generation of dsRNA. Components of both the RITS complex and the RDRC associate with noncoding centromeric transcripts as well as with centromeric DNA repeats. These observations suggest that the RITS/RDRC associate with nascent noncoding centromeric transcripts to initiate RNAi and heterochromatin assembly (Figure 1). In support of this model, we have shown that tethering of the RITS complex to the transcript of a normally active gene (S. pombe ura4+) results in de novo siRNA generation and heterochromatin formation. Moreover, consistent with the nascent transcript model, our studies indicate that ability of ectopically produced siRNAs to induce heterochromatin formation depends on 3’ end processing pathways that mediate polyadenylation and RNA export. Thus mRNAs that contain canonical polyadenylation signals are refractory to siRNA-mediated heterochromatin formation, presumably because polyadenylation of nascent pre-mRNA promotes rapid export from the nucleus before the RNA can act as a scaffold for the recruitment and assembly of RITS. In contrast, aberrant or foreign RNAs that lack proper 3’ end processing signals are readily targeted by siRNAs, which mediate their silencing by heterochromatin formation.
Heterochromatic gene silencing also requires RNAi-independent RNA turnover mechanisms. In this regard, we have found that Cid14, the fission yeast homolog of the Trf4/5 non-canonical poly A polymerases, and the exosome also contribute to efficient gene silencing within heterochromatic DNA domains and compete with RNAi for substrates. Our results further suggest that the RNAi and exosome pathways make independent contributions to cotranscriptional degradation of RNAs that are transcribed from heterochromatic regions. We refer to the degradation of heterochromatic transcripts as cotranscriptional gene silencing (CTGS), which provides a layer of regulation that augments transcriptional gene silencing within heterochromatic domains.
Both the RITS complex and the RDRC are required for the spreading of H3K9 methylation, which is mediated by the conserved Clr4 methyltransferase. Conversely, Clr4 is required for the association of the RITS complex and the RDRC with chromatin. We have recently purified a Clr4-containing complex that also contains the heterochromatin protein Rik1, the Cullin 4 E3 ubiquitin ligase, and two previously uncharacterized proteins. We have shown that this complex physically interacts with the RNAi machinery, indicating that RNAi can directly recruit histone H3K9 methylation to specific chromosome regions. Remarkably, in S. pombe, siRNA generation requires Clr4 and the heterochromatin protein 1 (HP1) homolog Swi6, indicating a role for heterochromatin in RNAi. Our findings suggest that primary small RNAs (siRNAs and primal small RNAs) are generated by both Dicer-dependent and -independent mechanisms and recruit the RDRC to chromatin-associated centromeric RNAs to initiate dsRNA synthesis and siRNA amplification (Figure 1).
Epigenetic Inheritance of Silent Chromatin
Post-translational modifications (PTMs) of histones are generally believed to carry epigenetic information that can be maintained during cell division. However, it had remained unclear whether epigenetic information could be transmitted independently of the underlying DNA sequences, DNA methylation, or RNAi. We recently used a system of inducible heterochromatin to demonstrate that histone H3K9 methylated domains and transcriptionally silent states in S. pombe could be inherited through both mitotic and meiotic cell divisions independently of sequence-specific initiators. A putative histone demethylase promotes erasure of H3K9 methylation and masks ready detection of the inheritance process in wildtype cells, while the ability of the chromodomain of Clr4 H3K9 methyltransferase to recognize its own mark is critical for transmission of epigenetic states (Figure 2).
SIR-Mediated Assembly of Silent Chromatin
In budding yeast, heterochromatin plays important roles in both gene regulation and maintenance of chromosome stability. The silent information regulator (SIR) complex, containing the Sir2, Sir3, and Sir4 proteins, mediates heterochromatin formation at the mating-type loci and telomeres. Sir2 is an NAD-dependent deacetylase, and the Sir3 and Sir4 proteins are histone-binding proteins. Deacetylation by Sir2 is coupled to NAD hydrolysis and synthesis of a metabolite, O-acetyl-ADP-ribose (acetoxy-ADP-ribose, AAR, also called OAADPr). Deacetylation is critical for silencing, and AAR synthesis may also contribute to SIR-complex assembly and silencing. Deacetylation of histone H4 lysine 16 (H4K16) creates a high-affinity binding site for the SIR complex, while AAR induces a structural rearrangement in the complex that is accompanied by the binding of multiple copies of Sir3 to Sir2/Sir4. The AAR-induced structural change in the SIR complex may contribute to the polymerization of the SIR complex along the chromatin fiber.
We identified the conserved bromo-adjacent homology (BAH) domain of Sir3 as the main histone- and nucleosome-binding domain in the SIR complex. The association of the BAH domain with the nucleosome is regulated by acetylation of H4K16 and methylation of histone H3 lysine 79, suggesting that the exposed surface of the histone octamer in the nucleosome mediates the association of the SIR complex with nucleosomes. Furthermore, our structural studies indicate that the association of the BAH domain with the nucleosome promotes the formation of salt bridges between arginine 17 and 19 (R17 and R19) in H4 tail and specific phosphates in the DNA backbone (Figure 3). R17 and R19 are essential for silencing and may act as clamps that promote silencing by making the nucleosome more stable. Recent studies by others indicate that BAH domains in plant and mammalian chromatin proteins have nucleosome binding activities that are regulated by other histone modifications. The BAH domain has therefore evolved to recognize a diversity of histone modifications.
Epigenetic Regulation of Learning and Memory
Previous studies have shown that the formation of stable memories induces new gene expression, which is required for remodeling of synaptic connections. We are interested in understanding how changes in transcription mediate stable changes in synaptic connections during learning and memory. Toward this end, we are using a fear conditioning paradigm to map changes in transcription-associated chromatin modifications and microRNA (miRNA) levels in the hippocampus of trained versus naïve rats.
Our studies have identified several miRNAs, which are upregulated in trained animals and regulate mRNAs that encode synaptic proteins. We are interested in understanding how miRNA-mediated regulation contributes to stable changes in synapses that underlie the formation of long-term memories. We hope to help uncover the principles that underlie neuronal memory, which in contrast to developmental epigenetic memory, operates in non-mitotic cells and involves stable changes in synapses located far away from the the cell body and nucleus.
This work was supported in part by grants from the National Institutes of Health.
As of April 24, 2016