Structural Basis of H2B Ubiquitination-Dependent H3K4 Methylation by COMPASS

Abstract

The COMPASS (complex of proteins associated with Set1) complex represents the prototype of the SET1/MLL family of methyltransferases that controls gene transcription by H3K4 methylation (H3K4me). Although H2B monoubiquitination (H2Bub) is well known as a prerequisite histone mark for COMPASS activity, how H2Bub activates COMPASS remains unclear. Here, we report the cryoelectron microscopy (cryo-EM) structures of an extended COMPASS catalytic module (CM) bound to the H2Bub and free nucleosome. The COMPASS CM clamps onto the nucleosome disk-face via an extensive interface to capture the flexible H3 N-terminal tail. The interface also sandwiches a critical Set1 arginine-rich motif (ARM) that autoinhibits COMPASS. Unexpectedly, without enhancing COMPASS-nucleosome interaction, H2Bub activates the enzymatic assembly by packing against Swd1 and alleviating the inhibitory effect of the Set1 ARM upon fastening it to the acidic patch. By delineating the spatial configuration of the COMPASS-H2Bub-nucleosome assembly, our studies establish the structural framework for understanding the long-studied H2Bub-H3K4me histone modification crosstalk.

Keywords: COMPASS; H3K4; MLL; chromatin; epigenetics; histone crosstalk; methylation; nucleosome; ubiquitin.

Figure 1.. Biochemical Analyses of COMPASS Nucleosomal Binding and Activity

(A) Domain organization and construct design of COMPASS subunits utilized in biochemical analysis and structure determination. Set1 is not drawn to scale. The dotted box between the nSET and RRM domains indicates a long linker region. The CM boxed in dashed lines contains the Set1 SET domain and the four WRAD subunits. The eCM extends the CM by including the nSET domain and its binding partner, Spp1. (B) H3K4 methyltransferase activity of purified full-length COMPASS on NCP and uNCP substrates. COMPASS exhibits little to no activity on unmodified nucleosomes while displaying robust methylation on ubiquitinated nucleosomes like its S. cerevisiae ortholog. (C) Native Tris-borate-EDTA (TBE) gel shift assay of unmodified nucleosomes with increasing concentrations of COMPASS, stained by SybrGold. (D) H3K4 methyltransferase activity of both the COMPASS catalytic module (CM) and extended CM (eCM) on NCP and uNCP substrates. eCM activity is largely suppressed on unmodified nucleosomes and greatly stimulated by H2Bub nucleosomes. (E) Native TBE gel shift assay of unmodified nucleosomes with increasing concentrations of COMPASS-eCM, stained by SybrGold. (F) Native TBE gel shift assay of H2Bub nucleosomes with increasing concentrations of COMPASS-eCM, stained by SybrGold.

Figure 2.. Overall Structure of the COMPASS eCM-uNCP Complex

(A) Model of COMPASS-eCM bound to the H2Bub modified nucleosome. Set1 (blue), Swd1 (magenta), Swd3 (green), Bre2 (orange), Sdc1-A and B (turquoise and yellow), Spp1 (pink), and Ub (red) are shown in cartoon form. Nucleosomal DNA (dark salmon) is shown in cartoon form. The histone octamer is shown in cylinders. The SAM cofactor is shown in space-filling form. Residues forming the H2A-H2B acidic patch are shown in red spheres. The positions of the Swd1 WDRP and Set1 ARM helix are indicated by arrows. (B) View of (A) rotated 90° and colored in the same scheme. The positions of the Set1 WIN motif (tube form), Swd3 SMART motif (tube), and Swd1 WDRP (tube) are indicated by arrows. A dashed line indicates the flexible histone H3 N-terminal tail spanning residue 9–38, which connects the first stretch of H3 seen in the NCP body (yellow tube) and the H3 (A1-R8) peptide (sticks) found in the Set1 SET domain.

Figure 3.. DNA Binding Stabilizes COMPASS on Nucleosomes

(A) Zoom-out view of the complex structure superimposed on the cryo-EM map. Major DNA-binding sites are boxed. Swd1 is shown in magenta, Swd3 in green, Set1 in blue, Bre2 in orange, Spp1 in pink, Ub in red, histones in dark gray, and nucleosomal DNA in wheat. (B) Close-up views of the Spp1 DNA-binding domain density (pink) on nucleosomal DNA (wheat) in two different orientations. (C) Structure-based sequence alignment of yeast Spp1 orthologs from K. lactis (Kl), S. cerevisiae (Sc), Z. rouxii (Zr), K. pastoris (Kp), A. gossypii (Ag), L. thermotolerans (Lt), W. ciferri (Wc), C. glabrata (Cg), and C. albicans (Ca). Cylinders denote α helices, while the dashed line indicates an unmodeled region, predicted to lack secondary structure. A basic region with potential DNA-binding function is boxed. (D) Left: close-up view of Bre2 (orange in surface representation) looking down the nucleosomal DNA entry/exit point (cartoon representation, dark salmon). Right: 90° rotation of the left panel, highlighting the position of auxiliary β sheet (gray) on Bre2 relative to DNA. Two highly conserved lysine residues on the surface of this structural element are colored in blue. (E) H3K4 methyltransferase activity of the COMPASS CM assembled with Bre2 mutants against nucleosomal substrates. The IK→AA double mutant of two conserved surface residues, Ile251 and Lys254, drastically reduces activity of the CM against H3K4.

Figure 4.. Swd1 Is a Histone-Binding Protein

(A) Side view of Swd1 (magenta, cartoon) positioned over the histone surface (dark gray, surface). The “top” face of the WD40 is pointed toward the nucleosome, while the “bottom” is facing COMPASS, interacting with Swd3 (green, surface). The top surface loops interacting with the histone surface are shown in tube representation. The 5D-6A loop (tube form) and strand D from blade 6 are labeled. The two loops interacting with H3 and H4 are indicated by triangles. (B) Top-down view, looking down the C-terminal H2B helix (cyan). Swd1 (magenta) wraps around this helix using the 5D-6A loop (tube form) and strand 6D. Set1 is shown in blue surface representation. H2A and H4 are shown in green and deep red, respectively, in cartoon form. The C-terminal tail of Ub (red) is fused to H2B via a disulfide linkage. (C) Local structure-based sequence alignment of the 5D-6A loop across K. lactis (Kl), S. cerevisiae (Sc), Z. rouxii (Zr), C. glabrata (Cg), A. gossypii (Ag), H. sapiens (Hs), M. muscuius (Mm), G. gallus (Gg), D. rerio (Dr), and D. meianogaster (Dm) orthologs. Straight line denotes an element with no secondary structure. The four-residue segment that inserts itself into the H2A-H2B three-helix cleft is boxed. (D) Close-up view centered on the two conserved hydrophobic residues in the 5D-6A loop (magenta) at the three-helix juncture (H2B in cyan, H2A in green). (E) Close-up view centered on the two polar residues in the 5D-6A loop (magenta). Asn273 forms a hydrogen bond with H2B Gln96 (cyan), while Arg274 interacts with the phosphate backbone of nucleosomal DNA (red, sticks). Polar and salt bridge interactions are shown as dotted yellow lines. (F) H3K4 methyltransferase activity of the COMPASS CM assembled with Swd1 5D-6A loop mutants against nucleosomal substrates. A double mutant of two hydrophobic residues (II→AA) reduces methyltransferase activity to monomethylation. A quadruple mutant (IINR→A4) nearly renders the CM inactive.

Figure 5.. The SET N Terminus Unravels to Bind H2A

(A) Zoom-out view of Set1-SET (blue, cartoon) being positioned over histone H2A (green) and sandwiched by Bre2 (orange, surface) and Swd1 (magenta, surface). Swd3 (green) and the nSET domain (blue) are also shown in surface representation. The cofactor SAM and the H3 N terminus are shown in space-filling and stick form, respectively. Boxed is the region of interaction between Set1 and H2A. (B) View looking down the N-terminal helix of the Set1 SET domain. The SET domain found in the NCP-bound structure is shown in blue, while the free form of the SET domain is shown in gray. Residues involved in interacting with H2A upon helical unraveling are shown in sticks. (C) Close-up view of the rearranged Set1 N terminus (blue) presenting both Lys855 and Arg856 into a polar tip found on H2A α2 helix (green). Hydrogen bonds and salt bridges are shown as dashed yellow lines. (D) Close-up view of the Set1 hydrophobic claw (blue) interaction with H2A (green). H2B is shown in cyan. H3 and H4 are omitted for clarity. Set1, H2A, and H2B amino acids are labeled in blue, white, and cyan, respectively.

Figure 6.. H2B~Ub Packs against Swd1 and Set1

(A) Zoom-out view of H2B~Ub (red, cartoon) against COMPASS. Swd1 (magenta), Set1 (blue), and Bre2 (orange) are shown in surface form. The disulfide linkage between H2B (cyan, cylinder) and Ub is shown in stick form. The critical WDRP necessary for Set1 activity is labeled. (B) Close-up view of the interface between the Ub Ile44 hydrophobic patch and Swd1. Ub packs extensively against a series of conserved residues on Swd1 spread across its N terminus, WD40, and C-terminal tail. Key side chains are shown in sticks and labeled. Hydrogen bonds and salt bridges are shown as dashed yellow lines. (C) Close-up view of the interface between the Set1 SET N terminus (blue) and Ub Ile36 underside site (red). The N terminus of the SET domain makes several backbone interactions with Ub between Thr7 and Thr12 (shown in sticks, side chains omitted for clarity). A series of Set1 hydrophobic residues pack against the base of the Ub helix. The Ub Leu71/Leu73 pair identified in Holt et al. (2015) critical for H2Bub-H3K4me crosstalk is boxed. Hydrogen bonds and salt bridges are shown as dashed yellow lines.

Figure 7.. The Set1 ARM Regulates COMPASS H2Bub Sensitivity

(A) Sequence alignment of the nSET ARM across K. lactis (Kl), S. cerevisiae (Sc), Z. rouxii (Zr), C. glabrata (Cg), A. gossypii (Ag), H. sapiens (Hs), M. musculus (Mm), G. gallus (Gg), D. rerio (Dr), and D. melanogaster (Dm) orthologs. The dotted cylinder represents the unraveled SET N-terminal helix. Cylinders represent helices, and lines indicate coiled structure. The strictly conserved arginine residues are highlighted by a yellow background. Letters above the alignment indicate the interactions the residue is involved in. (H) indicates a histone interaction, (S) indicates a SET domain interaction, (U) indicates a Ub interface, and (A) indicates a Sett ARM interaction. (B) View of the tunnel formed by the junction between Set1 (blue, surface), Swd1 (magenta, surface), Ub (red, surface), and the histone octamer (gray, surface). Swd3 sits at the periphery (green, surface). The Set1 ARM helix (blue, cartoon) sits in the center of this tunnel. The loop connecting the SET domain and the ARM helix is shown in tube form. (C) Close-up view of the Set1 ARM helix (blue) against the acidic surface of H2A (green). Side chains are shown in sticks. Two of the three strictly conserved arginine residues that impair H2Bub-H3K4me crosstalk (Kim et al., 2013) are underlined. H2A residues identified in (Nakanishi et al., 2008) critical for H2Bub-dependent H3K4 methylation are marked by an asterisk. Hydrogen bonds and salt bridges are shown as dashed yellow lines. (D) Close-up view of the Set1 ARM (blue) against the H2B αC helix (cyan). Side chains are shown in sticks. The third strictly conserved arginine residue involved in the H2Bub-H3K4me crosstalk (Kim et al., 2013) is underlined. H2B residues identified by Nakanishi et al. (2008) to be important for H2Bub-dependent H3K4 methylation are marked by an asterisk. Hydrogen bonds and salt bridges are shown as dashed yellow lines. (E) Close-up view of the three-way interface between the Set1 (blue) ARM, SET N terminus, and the Ub (red) Ile36 site. Side chains are shown in sticks. The ARM helix packs against the SET N terminus using an interlocking series of hydrophobic amino acids. Ub contributes to this interface by helping form the connecting loop between the ARM and SET regions through hydrophobic packing. Ub Leu71/Leu73, identified by Holt et al. (2015) to be important for H2Bub-stimulated H3K4me activity, are boxed. (F) Comparison of the Set1 ARM density (blue mesh) between the unmodified NCP-bound (left) and uNCP-bound (right) structure of the COMPASS eCM. The neighboring H2B density (cyan, surface) was used to normalize the ARM densities. The key side-chain feature on H2B used for controlling the map contour level is indicated by a triangle. The uNCP density has the model of the ARM superposed in the mesh. The positions of where Ub and the SET domain dock on the ARM helix are indicated by a red and blue circle, respectively. (G) H3K4 methyltransferase activity of the COMPASS CM and CM plus (CM+) on NCP and uNCP substrates. CM+ activity is suppressed relaive to the CM on NCPs and regains robust methyltransferase activity on uNCPs. (H) Comparison of the H3 N-terminal tail density (gray mesh) at the Set1 catalytic cleft between the unmodified NCP-bound (left) and uNCP-bound (right) structure of the COMPASS eCM. The density of a nearby Set1 sequence was used to normalize the H3 N-terminal tail densities. The key features of the Set1 density for controlling the map contour level is indicated by triangles. The models of the H3 N-terminal tail centered at the H3K4 methylation site are shown in sticks.