A dynamic structural unit of phase-separated heterochromatin protein 1α as revealed by integrative structural analyses

Abstract

The heterochromatin protein HP1α consists of an N-terminal disordered tail (N-tail), chromodomain (CD), hinge region (HR), and C-terminal chromo shadow domain (CSD). While CD binds to the lysine9-trimethylated histone H3 (H3K9me3) tail in nucleosomes, CSD forms a dimer bridging two nucleosomes with H3K9me3. Phosphorylation of serine residues in the N-tail enhances both H3K9me3 binding and liquid-liquid phase separation (LLPS) by HP1α. We have used integrative structural methods, including nuclear magnetic resonance, small-angle X-ray scattering (SAXS), and multi-angle-light scattering combined with size-exclusion chromatography, and coarse-grained molecular dynamics simulation with SAXS, to probe the HP1α dimer and its CSD deletion monomer. We show that dynamic intra- and intermolecular interactions between the N-tails and basic segments in CD and HR depend on N-tail phosphorylation. While the phosphorylated HP1α dimer undergoes LLPS via the formation of aggregated multimers, the N-tail phosphorylated mutant without CSD still undergoes LLPS, but its structural unit is a dynamic intermolecular dimer formed via the phosphorylated N-tail and a basic segment at the CD end. Furthermore, we reveal that mutation of this basic segment in HP1α affects the size of heterochromatin foci in cultured mammalian cells, suggesting that this interaction plays an important role in heterochromatin formation in vivo.

Graphical Abstract

Figure 1.

Comparison of phosphorylated and unphosphorylated HP1α by NMR. (A) Sequence of HP1α and numbering of basic and acidic segments. Phosphorylation sites are shown as phosS. (B) Schematic representations of the HP1α mutants used for NMR, SAXS, and MD experiments. Mutants phosphorylated at the serine residues marked as SSSS are defined as pHP1α, pN-tail-CD, or pΔCSD. The mutated amino acid residue is shown as S97A. (C) Chemical shift differences (Δδ) between 500 and 50 mM NaCl.

Figure 2.

Comparison of phosphorylated and unphosphorylated HP1α by SAXS. (A) R_g values (see Supplementary Fig. S4A and B) plotted against molar concentration of dimer for HP1α and pHP1α at 50 mM NaCl. (B) SAXS profiles obtained by SEC-SAXS for HP1α and pHP1α at 50 mM NaCl. The lines represent the fits for both profiles obtained from EOM calculations (see Supplementary Fig. S4F). (C and D) Heatmaps of residue–residue interaction probabilities of HP1α (C) and pHP1α (D), calculated from the reweighted ensembles from CGMD simulations. Left and right panels show, respectively, intra- and inter-subunit residue–residue interactions, respectively.

Figure 3.

Differences in intramolecular interactions between phosphorylated and unphosphorylated HP1α. (A and B) Representative structures of the top six clusters from the reweighted ensembles of HP1α (A) and pHP1α (B). NT, pNT, and CT represent the N-tail, phosphorylated N-tail, and C-tail respectively.

Figure 4.

Comparison of phosphorylated and unphosphorylated ΔCSD by NMR and SAXS. (A) Differences in chemical shift between 500 and 50 mM NaCl. (B) R_g values (see Supplementary Fig. S9C and D) plotted against molar concentration of monomer for ΔCSD and pΔCSD. (C) SAXS profiles obtained by SEC-SAXS for ΔCSD and pΔCSD at 50 mM NaCl. The lines represent the fits for both profiles obtained from EOM calculations (see Supplementary Fig. S9F). (D) Heatmaps of residue–residue interaction probabilities of pΔCSD calculated from the reweighted ensemble from the CGMD simulation of the two-molecule system. Left and right panels show, respectively, intra- and intermolecular residue–residue interactions. (E) Representative structures of the top four clusters from the reweighted ensemble of pΔCSD. pNT represents the phosphorylated N-tail.

Figure 5.

LLPS of the phosphorylated ΔCSD. (A) Difference in appearance of the condensed solution between ΔCSD and pΔCSD. (B) Chemical shift differences (Δδ) of pΔCSD between mid (120 μM) and condensed (400 μM) solutions at 50 mM NaCl. (C) Difference in appearance of the condensed solution between pΔCSD and the pΔCSD_b4 mutant. (D) SAXS profile for the pΔCSD_b4 mutant at 50 mM NaCl (black). The gray fitted line is derived from the EOM calculation (see Supplementary Fig. S15E).

Figure 6.

Model of LLPS of the phosphorylated ΔCSD. LLPS is mediated by dynamic intermolecular dimers formed via the phosphorylated N-tail (pS) and an essential basic segment located at the end of CD (b4).

Figure 7.

Effect of b4 mutation on heterochromatic localization of HP1α. (A) Schematic diagram of EGFP-fused HP1α showing the amino acid sequence of the N-terminal region containing the phosphorylation sites (S11–S14) and the b4 segment (K68–K72). Serine residues that can be phosphorylated are underlined; mutated amino acid residues are indicated in bold; two boxes in HP1α represent the conserved CD and CSD, respectively. (B) Example images of NIH3T3 cells transfected with EGFP-fused WT or mutant (SA, b4, or SAb4) HP1α. Scale bar: 10 μm. (C) Number of HP1α foci detected in the nucleus of transfected cells. Statistical significance of differences relative to WT was determined by Mann–Whitney’s U test. ns, not satisfied; **P < 0.01. (D) Size of HP1α foci detected in the nucleus of transfected cells calculated and shown by beeswarm plot. (E) Percentages of HP1α foci classified in accordance with size (<1 μm²; ≥1 and < 2 μm²; ≥2 and < 3 μm²; ≥3 μm²). (F) Schematic diagram of Schizosaccharomyces pombe (Sp) Swi6, human (hs) HP1α, and a chimeric protein (Chimera-WT) containing HP1α NCD and Swi6 CSD showing the amino acid sequences of the N-terminal region containing the phosphorylation sites (S11–S14) and the b4 segment (K68–K72). (G) Immunoblotting analysis of WT Swi6 and chimeric HP1α/Swi6 proteins. Whole-cell extracts prepared from WT cells (control) or cells expressing WT or mutant chimeric proteins were subjected to immunoblotting using anti-Swi6 antibody raised against full-length Swi6 protein. Due to the lower reactivity of chimeric proteins, a longer exposure image is shown to confirm expression. Anti-tubulin antibody was used as a control. (H) Spotting assays for Kint2::ura4⁺ silencing. A serially diluted culture of the indicated strains was spotted onto nonselective medium (N/S) or medium containing 5FOA (FOA). (I) Expression of the ura4⁺ silencing reporter evaluated by quantitative RT-PCR analysis. Results are means ± s.d. of at least three independent experiments. Statistical significance relative to WT was determined by Mann–Whitney’s U test; **P < 0.01.