Phosphorylation of HP1/Swi6 relieves competition with Suv39/Clr4 on nucleosomes and enables H3K9 trimethyl spreading

Abstract

Heterochromatin formation in Schizosaccharomyces pombe requires the spreading of histone 3 (H3) Lysine 9 (K9) methylation (me) from nucleation centers by the H3K9 methylase, Suv39/Clr4, and the reader protein, HP1/Swi6. To accomplish this, Suv39/Clr4 and HP1/Swi6 have to associate with nucleosomes both nonspecifically, binding DNA and octamer surfaces and specifically, via recognition of methylated H3K9 by their respective chromodomains. However, how both proteins avoid competition for the same nucleosomes in this process is unclear. Here, we show that phosphorylation tunes oligomerization and the nucleosome affinity of HP1/Swi6 such that it preferentially partitions onto Suv39/Clr4's trimethyl product rather than its unmethylated substrates. Preferential partitioning enables efficient conversion from di-to trimethylation on nucleosomes in vitroand H3K9me3 spreading in vivo. Together, our data suggests that phosphorylation of HP1/Swi6 creates a regime that increases oligomerization and relieves competition with the "read-write" mechanism of Suv39/Clr4, together promoting for productive heterochromatin spreading.

Keywords: H3K9 trimethylation; HP1; heterochromatin spreading; phosphorylation; “reader” “writer” competition.

Figure 1:. S18 and S24 in Swi6 are required for spreading, but not nucleation of heterochromatin silencing.

A. Overview of the Swi6 protein domain architecture and previously identified (Shimada et al.) in vivo phosphorylation sites (green residue numbers). NTE: N-terminal extension; CD: chromodomain (H3K9me binding); HINGE: unstructured hinge region; CSD: chromo-shadow domain (dimerization and effector recruitment). B. Strategy for production of swi6 S-A mutants in the MAT locus ΔREIII Heterochromatin Spreading Sensor (HSS, lower diagram) reporter background. C. Swi6 levels are not affected by serine (S) to alanine (A) S-A mutations. Total extracts of swi6 wildtype or indicated mutants were probed with an anti-Swi6 polyclonal antibody, or anti-α-tubulin monoclonal antibody as loading control. In vitro purified Swi6 that was either phosphorylated (pSwi6) or not (unpSwi6) are run as size controls. Note, not all mutant Swi6 proteins display a band shift even if they retain phosphosites. D.-I. 2-D Density hexbin plots examining silencing at nucleation ‘green’ and spreading ‘orange’ reporter in Δswi6, wildtype, and indicated S-A mutants. The dashed orange and blue lines indicate the threshold for full expression of the ‘orange’ and ‘green’ reporters, respectively. Indicated percentages and SD represent the fraction of cells above the line.

Figure 2:. Conversion from H3K9me2 to H3K9me3 is compromised outside nucleation centers in S18 and S24 Swi6 mutants.

A. Overview of the ChIP-seq experiments. B-D. ChIP-seq signal visualization plots. The solid ChIP/input line for each genotype represents the mean of three repeats, while the shading represents the 95% confidence interval. B. Plots of H3K9me2 (TOP) and H3K9me3 (BOTTOM) ChIP signal over input at the MAT ΔREIII HSS mating type locus for wildtype (black), swi6^S18/24A (blue), and Δswi6 (gold). Signal over “green” and “orange” reporters are greyed out. Reads from these reporters map to multiple locations within the reference sequence, as all reporters contain control elements derived from the ura4 and ade6 genes. C. H3K9me2 (TOP) and H3K9me3 (BOTTOM) plots as in A. for subtelomere IIR for wildtype and swi6^S18/24A. The red bar on the H3K9me2 plot indicates the distance from tlh2 to where H3K9me2 levels drop in swi6^S18/24A relative to wildtype. Insets: a zoomed-in view proximal to tlh2 is shown for H3K9me2 and me3. The red arrows in the insets indicate the point of separation of the 95% confidence intervals, which is significantly further telomere-proximal for H3K9me3. D. H3K9me2 (TOP) and H3K9me3 (BOTTOM) plots as in A. for centromere II for wildtype and swi6^S18/24A. Insets: the left side of the pericentromere.

Figure 3:. Swi6 phosphorylation increases oligomerization working through, or in parallel to, the ARK loop and NTE acidic stretch.

A. Production of phosphorylated Swi6 (pSwi6) in E. coli. Casein Kinase II (CKII) is co-expressed with Swi6. After lysis and purification, the 6XHis tag is removed from the pSwi6 or unpSwi6 protein by TEV cleavage. B. Mass Spectrometry on pSwi6. Shown is a domain diagram of Swi6. Phosphorylation sites identified in pSwi6 by 2D-ETD-MS are indicated and grouped by detection prevalence in the sample. For prevalence of pS18 and/or p24 peptides, see SFigure 4A C. Size Exclusion Chromatography followed by Multi-Angle Light Scattering (SEC-MALS) on EDC/NHS cross-linked unpSwi6 (black) and pSwi6 (green). Relative refractive index signals (solid lines, left y-axis) and derived molar masses (lines over particular species, right y-axis) are shown as a function of the elution volume. The Swi6 concentration was 100μM. D. Overview of NTE and CD residues targeted for mutation in subsequent panels. S18/S24 phosphoserines, the CD ARK loop (R93AK94A; loopX), and the CD-preceding acidic stretch (E74–80A; acidicX). E.-J. 2-D Density hexbin plots examining silencing at nucleation ‘green’ and spreading ‘orange’ reporter, wildtype, S18/24A, loopX, and acidicX mutants and combinations thereof. Note this experiment was a fully separate run from Figure 1/SFigure 1.

Figure 4:. Swi6 phosphorylation decreases nucleosome affinity without affecting specificity.

A. Overview of fluorescence polarization (FP) experiments with fluorescein (star)- labeled H3 tail peptides (1–20) and nucleosomes to assess pSwi6 and unpSwi6 substrate affinity and specificity. B. FP of H3K9me0 (open circles) and H3K9me3 (filled circles) tail peptides with pSwi6 (green) or unpSwi6 (black). The binding affinity was too low to be fit for unpSwi6 and H3K9me0 peptides. C. FP with H3K9me0 (open circles) or H3K_c9me3 (MLA, filled circles) mononucleosomes. Fluorescein (green star) is attached by a flexible linker at one end of the 147 bp DNA template. For B.&C., the average of three independent fluorescent polarization experiments for each substrate is shown. Error bars represent standard deviation. D. Summary table of affinities and specificities for B. and C. E. Representative maximum projection live microscopy images of indicated Swi6-GFP/ Sad1-mKO2 strains. F. Analysis of signal intensity in Swi6-GFP foci in indicated strains. Wt Swi6, n=242; Swi6^S18/24A, n=251; Swi6^{S18/24/117−220A} (6S/A), n=145; Swi6^{S46/52/117−220A}, n=192. n, number of foci analyzed.

Figure 5:. Swi6 phosphorylation focuses Swi6 onto heterochromatin nucleation sites and away from euchromatin.

A. Schematic of Swi6 ChIP-seq experiments. The anti-pS18-pS24 experiment was carried out with wildtype swi6 (and swi6^S18/24A as a negative control, see SFigure 8). For the anti-FLAG experiments, the endogenous swi6 locus was 3XFLAG tagged in the context of wildtype swi6 (black), swi6^S18/24A (blue), or swi6^{S18/24/117−220A} (6S/A, red). For quantitative normalization, ChIP reactions were supplemented with Drosophila chromatin spike-in. B. Heatmaps of spike-in normalized FLAG ChIP-seq signal (in Counts Per Million, CPM) for swi6, swi6^S18/24A, or swi6^6S/A at features previously classified as nucleators(Greenstein et al. 2022). C. As in B. but for features classified as regions of H3K9me2 spreading(Greenstein et al. 2022). D. FLAG ChIP-seq signal visualization plot for the MAT ΔREIII HSS mating type locus. The solid ChIP/input line for each genotype represents the mean of three repeats, while the shading represents the 95% confidence interval. E. As in D. but for cen II with zoom-in of the left and right pericentromeric region. The brown dashed boxes indicate siRNAi-generating centers as mapped in [(Djupedal et al. 2009)]. F. Heatmap as in B. and C., but for H3K9me2 negative regions. G. ChIP-seq signal visualization plots for H3K9me2 negative regions. The solid ChIP/input line for each genotype represents the mean of two repeats, while the shading represents the 95% confidence interval. A high signal replicate was removed for both genotypes. ChIP-seq signal visualization for all three genotypes with all three repeats in SFigure 7C.

Figure 6:. Swi6 phosphorylation mitigates inhibition of the Clr4-mediated conversion of H3K9me2 to H3K9me3

A. Most Swi6 molecules in the cell are phosphorylated at S18 and S24. Quantitative western blots against total Swi6 and phosphorylated Swi6 at S18 and/or S24. A standard curve of pSwi6 isolated as in Figure 3 is included in both blots. Total protein lysates from wildtype swi6 and swi6^S18/24A strains were probed with a polyclonal anti-Swi6 antibody (α-Swi6) or an antibody raised against a phosphorylated S18/S24 peptide (α-pS18-pS24). α-tubulin was used as a loading control. One of two independent experiments is shown. L; ladder. Total fraction of Swi6 phosphorylated in vivo at S18 and/or S24 is adjusted by the prevalence of phosphorylation in the in vitro produced standard (~0.76, see SFigure 4A) B. Experimental scheme to probe the impact of Swi6 on H3K9 trimethylation. C. Quantitative western blots on the time-dependent formation of H3K9me3 from H3K9me2 mononucleosomes in the presence of pSwi6 or unpSwi6 under single turnover conditions. The same blots were probed with α-H3K9me3 and α-H4 antibodies as a loading and normalization control. D. Single exponential fits of production of H3K9me3 tails over time for indicated concentrations of unpSwi6 or pSwi6. E. plot of the observed single turnover rate constant measured from exponential fits (k_obs) against the Swi6 concentration in μM.

Figure 7:. S18 and/or S24 phosphorylation contributes to pSwi6’s biochemical behaviors.

A. Schematic of phosphorylated Swi6 molecules used in this Figure. B. pSwi6^S18/24A is defective in oligomerization. pSwi6 or pSwi6^S18/24A was crosslinked or not (−) at indicated concentrations, separated on SDS-PAGE, and probed with a polyclonal anti-Swi6 antibody. M, monomer; D, dimer; T, tetramer; O, octamer; O⁺, higher molecular weight species. Below: Quantification of oligomer signal divided by dimer signal for crosslinked species. C. FP with H3K9me0 (open circles) or H3K_c9me3 (MLA, filled circles) mononucleosomes as in Figure 4C, with pSwi6 (green) and pSwi6^S18/24A (magenta). Error bars represent the standard deviation of three repeats. Relative dissociation constant (K_d) values in SFigure 10D. D. Quantitative western blots on the time-dependent formation of H3K9me3 from H3K9me2 mononucleosomes in the presence of pSwi6 or pSwi6^S18/24A, as in Figure 6C. Quantification of the signal below. Note that the reactions were not fast enough in this experiment to derive a single exponential observed rate. E. Model of the impact of pSwi6 on Clr4 activity. Left: pSwi6 does not engage with H3K9me0 nucleosomes, clearing the substrate for Clr4, and has reduced interactions with the nucleosome core. The pS18/pS24 NTE releases the ARK loop (light blue), allowing CD-CD contacts. Right: unpSwi6 binds H3K9me3 and me0 nucleosomes, occluding Clr4 access. The S18/S24 unphosphorylated NTE blocks the ARK from engaging CDs and additionally may contribute to increased nucleosome affinity by contacting DNA or octamer.