The Mechanistic Roles for SUMO in Chromatin and Condensates

Abstract

Histone post-translational modifications regulate DNA-templated processes in eukaryotes by shaping chromatin structure and regulating enzyme activity. The unstructured histone termini, or tails, extend outwards from the globular octameric histone core that tightly wraps ~147 bp of DNA to form the nucleosome core particle. The dynamic landscape of histone tail modifications is installed, maintained and remodeled by dedicated nuclear proteins that tune transcriptional outcomes in response to varying cellular needs. One dramatic histone modification that remains poorly studied in cells is conjugation of the histone lysine side-chain with the small ubiquitin like-modifier (SUMO) protein. Prior studies associated the presence of SUMO in chromatin with gene repression, but could not ascribe this effect to specific lysines sites in human histones. Semisynthetic access to site-specifically sumoylated histones, pioneered by our labs, has enabled biochemical investigations of the mechanistic roles for SUMO in chromatin. In this thesis, I examined the mechanistic roles for SUMO in three physiological contexts: its biochemical crosstalk with histone modifications in chromatin, its recruitment of repressive machinery through SUMO interaction motifs (SIMs) in key protein components of gene repressive complexes, and its ability to regulate the biophysical properties of biomolecular condensates in living cells. First, to address mechanisms of biochemical crosstalk, I focused on the role of sumoylation at H4 Lys12, because it is a validated site of sumoylation in several cell lines and lies in the unstructured H4 tail. By generating site-specifically sumoylated H4K12 (H4K12su) using protein semisynthesis and incorporating it into histone octamers and reconstituted mononucleosomes, we showed that sumoylation inhibits Rad6-Bre1-mediated monoubiquitylation of H2BK120. In complementary cell-based assays, I demonstrated that the in vitro biochemical effect of H4K12su could be recapitulated using SUMO-H4 linear fusions that ensure uniform sumoylation on the intact H4 tail in cells. By including both SUMO-H4 and a SUMO-H4(D1-11) construct that positions SUMO closer to Lys12, and hence closely mimics H4K12 sumoylation, I demonstrated that H4 sumoylation is inhibitory toward H2B K120 ubiquitylation, a modification required for subsequent H3K4 methylation by the Set1/COMPASS family of H3K4 methyltransferases that are associated with transcription activation and elongation. Having shown that H4K12su and its isosteres can repress H3K4 methylation by physically hindering Rad6-Bre1 binding to the nucleosome core particle and the enzymatic installation of H2BK120ub, I turned to examining the role of H4K12su in recruiting gene repressive enzyme complexes to sumoylated nucleosomes. Toward this goal, I focused on a non-canonical SIM (ncSIM) within the Co-repressor of REST1 protein (CoREST1) and established its direct binding to SUMO3. Two-dimensional 15N-1H HSQC NMR chemical shift perturbations with short ncSIM peptides were employed to map their binding to the canonical SIM-binding groove in SUMO3. Reciprocal pulldowns with either purified SUMO3 or full-length CoREST1 as bait further confirmed their direct binding, and alanine mutation of key residues in the ncSIM hydrophobic core were found to weaken SUMO3-CoREST binding. Altogether, my results defined the binding interface between the CoREST1 protein and SUMO3 and revealed how SUMO3-CoREST1 binding may position the gene repressive LSD1–CoREST–HDAC1 complex on nucleosomes and influence its demethylase and deacetylase activities on methylated and acetylated nucleosomes, respectively. In the final part of my thesis, I broadened my studies of the SUMO-SIM interaction to investigate the fundamental biophysical properties of SUMO-dependent biomolecular condensates (BMCs), such as Promyelocytic Leukemia bodies. Toward this, I engineered a synthetic polyvalent platform consisting of poly(SUMO3) and poly(SIM) scaffolds and quantified how the SUMO3-SIM binding affinity changes the internal dynamics of BMCs and the activity of enzymes localized in BMCs. I discovered that the phase boundary (csat) for polyvalent constructs correlates well with monovalent SUMO-SIM affinities (Kd). However, condensate number and size do not correlate with Kd. Condensate stability to chaotoropes and internal mobility, as measured by Fluorescence Recovery After Photobleaching (FRAP), also showed excellent correlation with SUMO-SIM affinities. The correlation was also recapitulated in three different human cell lines and demonstrated our ability to generate BMCs with predictable properties in living cells. Finally, I demonstrated that the activity of the enzyme pyroglutamyl peptidase 1 that plays an important role in the cellular stress response could be regulated by its recruitment to engineered BMCs of different mobilities. Thus, my mechanistic studies with the small ubiquitin-like modifier protein have led me to identify negative crosstalk between H4K12 sumoylation and H2BK120 monoubiquitylation, to define the specific mode of interaction between SUMO and CoREST1, and to establish an exciting and novel modular framework for condensate formation in vitro and in living cells, that directly links SUMO-SIM affinities to predictable condensate properties and tunable enzyme activities.