Interaction of HP1 and Brg1/Brm with the globular domain of histone H3 is required for HP1-mediated repression

Abstract

The heterochromatin-enriched HP1 proteins play a critical role in regulation of transcription. These proteins contain two related domains known as the chromo- and the chromoshadow-domain. The chromo-domain binds histone H3 tails methylated on lysine 9. However, in vivo and in vitro experiments have shown that the affinity of HP1 proteins to native methylated chromatin is relatively poor and that the opening of chromatin occurring during DNA replication facilitates their binding to nucleosomes. These observations prompted us to investigate whether HP1 proteins have additional histone binding activities, envisioning also affinity for regions potentially occluded by the nucleosome structure. We find that the chromoshadow-domain interacts with histone H3 in a region located partially inside the nucleosomal barrel at the entry/exit point of the nucleosome. Interestingly, this region is also contacted by the catalytic subunits of the human SWI/SNF complex. In vitro, efficient SWI/SNF remodeling requires this contact and is inhibited in the presence of HP1 proteins. The antagonism between SWI/SNF and HP1 proteins is also observed in vivo on a series of interferon-regulated genes. Finally, we show that SWI/SNF activity favors loading of HP1 proteins to chromatin both in vivo and in vitro. Altogether, our data suggest that HP1 chromoshadow-domains can benefit from the opening of nucleosomal structures to bind chromatin and that HP1 proteins use this property to detect and arrest unwanted chromatin remodeling.

Figure 1. The CSD of HP1 proteins is a histone-binding domain.

(A) Indicated histones either epitope-tagged recombinant (rH3) or purified bovine (H3 and H4) were resolved by SDS-PAGE, blotted to nitrocellulose membrane and probed with either HA-HP1α or HA-HP1γ. Bound HP1 proteins were detected with anti-HA antibodies and chemiluminescence. (B) Purified bovine core histones were incubated with indicated HP1α-derived GST-fusion proteins bound to agarose beads. After washing, retained proteins were eluted, resolved by 4–12% gradient SDS-PAGE, and detected by western blotting with anti-histone H3 antibodies. (C) Isoleucine 25 and 126 (I25 and I126) positioned on the structure of the chromo-domain (1Q3L) and the monomeric chromoshadow-domain (1DZ1) respectively, visualized with CN3D. (D) Schematic representation of the MMTV(Gal4)-Luc reporter construct. Black boxes symbolize glucocorticoid receptor (GR) binding sites (GRE). (E) MCF7 cells were transfected with 1mg of the Gal4-MMTV-Luc reporter construct in the absence or in the presence of dexamethasone (Dex – induces activation of the MMTV promoter by the glucocorticoid receptor) and the indicated amounts (in µg) of Gal4-HP1α or Gal4-HP1αI126F expression constructs. (F) Schematic representation of the B10/6xHIS-tagged recombinant histone H3 constructs expressed in E. coli. Black boxes represent the B10/6xHIS tag. (G,H) Indicated histone H3 mutants were incubated with indicated HP1-derived GST-fusion proteins and probed for interaction as in (B). Western blotting was performed with anti-B10 monoclonal antibodies.

Figure 2. Brg1 binds histone H3.

(A) Positioning on the nucleosome of the Shadock (V35 to P66, yellow) of histone H3 interacting with HP1 proteins. H3 histones are lilac and gray. (B) Schematic representation of the truncated Brg1 construct. HP1α: HP1α interaction domain . Helicase: catalytic domain. Bromo: bromodomain. (C) Purified core histones were incubated in the absence (lane 1) or in the presence (lane 2) of recombinant flag-tagged full length Brg1. Immunoprecipitation was carried out with anti-flag antibodies. Immunoprecipitate was resolved by SDS-PAGE and analyzed by western blot using anti-Brg1 or anti-H3 antibodies. (D) Purified core histones were resolved by SDS-PAGE, blotted to nitrocellulose membrane and probed with indicated GST fusion-proteins. Bound proteins were detected with anti-GST antibodies and chemiluminescence. (E) Purified core histones or recombinant histone H3 (rH3) were incubated with GST or GST-ΔBrg1-1 bound to agarose beads. After washing, retained proteins were eluted, resolved by 4–12% gradient SDS-PAGE, and detected by western blotting with anti-histone H3 antibodies. (F) As in (E), with agarose beads bound to the indicated Brg1 truncation mutants. (G) As in (E), with the indicated H3 truncation mutants. (H) REA assays: 5S polynucleosome template at 1 nM was digested by HhaI in the presence or absence of hSWI/SNF pre-incubated with the indicated B10-tagged histone H3 polypeptides. Digestion products were separated on 1% agarose gels. NC: Not cut. (I) Quantification of three independent REA assays described in (H).

Figure 3. HP1α and HP1γ inhibits chromatin remodeling in vitro.

(A) REA assay performed on a 5S polynucleosome template. Template at 1 nM was pre-incubated with indicated concentrations of recombinant F-HP1α (produced in baculovirus) before digestion by HhaI in the absence or presence of hSWI/SNF. At the end of the reaction (1hr), digestion products were separated on a 1% agarose gel. (B) REA assay performed as in (A) on a 202 bp mononucleosome template and the restriction enzyme PstI. Digestion products were separated on a 5% polyacrylamide gel. (C) REA assay performed as in (A) with indicated concentrations of recombinant F-HP1γ (produced in baculovirus). (D) Schematic representation of the HA-tagged HP1α point and deletion mutants produced in E. coli. Note that HA-HP1α(V22M) also carries a V21A mutation. (E) REA assays performed as in (A) with increasing concentrations of indicated HA-HP1α proteins. Values are averaged from 3 independent experiments.

Figure 4. Brm and HP1α/HP1β have opposite effects on common target genes.

(A) Western blots of extracts from HeLa cells transfected with the indicated siRNA and used for the preparation of the RNAs used in (B,C). The blots shown are representative of experimental triplicates. (B) HeLa cells were transfected with Brm siRNAs. The mRNA abundance from the indicated genes was measured by RT–qPCR and normalized to levels of HPRT. Values were averaged from experimental triplicates and normalized to levels of HPRT mRNA. (C) As in (B), with the indicated HP1 siRNAs. (D) As in (B), using HeLa HA-HP1γ cells (see Figure 5) and HP1γ or Brm siRNAs and a treatment with 0.5 nM interferon-α2 for either 0 or 10 hours as indicated.

Figure 5. Brm facilitates recruitment of HP1γ to chromatin.

(A) (Frame) Western blot with anti-HP1γ antibodies on total extract from HeLa expressing an HA-tagged version of HP1γ (HeLa HA-HP1γ). (B) Chromatin immunoprecipitation (ChIP): HeLa HA-HP1γ were stimulated with 0.5 nM interferon-α2 for the indicated times after siRNA-mediated knock down of either GAPDH or Brm. ChIP experiments were performed with the anti-Brm, anti-HP1γ, anti-HA epitope tag antibodies, or with total IgG as indicated. Enrichment in IFIT3 promoter chromatin was quantified by qPCR using primers spanning over the transcription start (TS) region. Values are averaged from 2 independent experiments. IgGs bring down approx. 1% of the input. (C) ChIP-reChIP: HeLa HA-HP1γ were stimulated with 0.5 nM interferon-α2 for 60′ after siRNA–mediated knock down of either GAPDH or Brm. Sequential ChIPs were carried out first with anti-Brm antibodies then with anti-HA antibodies or total IgG, or vice versa. Enrichment in IFIT3 promoter chromatin was quantified as in B. Values are averaged from 2 independent experiments. (D) ChIP: J-Lat A1 cells that carry a single integration of an HIV1–derived reporter construct were treated with phorbol ester PMA for the indicated times. ChIP experiments were performed with antibodies specific for RNAPII, Brg1, and HP1γ. Enrichment in HIV1 LTR chromatin was quantified by qPCR using primers spanning over the transcription start (TS) region. Values are averaged from 2 independent experiments.

Figure 6. Brg1 remodeling facilitates binding of HP1 to nucleosomes.

(A) Schematic representation of the chromatin reconstitution protocol. The DNA used for chromatin reconstitution is a linearized biotinylated fragment containing 12 repeats of the 5S nucleosome positioning sequence. (B) Micrococcal digestion pattern of salt-reconstituted chromatin with unmodified histones. (C) dHP1a was assayed for binding to chromatin in the absence or in the presence of Brg1 and either ATP or γSATP as indicated. (D) Quantification of experiment in (C).

Figure 7. Model suggesting how HP1 proteins may gain access to internal nucleosomal regions and control remodeling by SWI/SNF.

(A) When the chromatin is methylated on H3K9, HP1 is brought in the vicinity, attracted by the methylation mark. The interaction with the nucleosome is very dynamic. (B) Remodeling creates access to hidden HP1-binding sites on the globular domain of histone H3. (C) The exposure of the internal binding sites allows HP1 to detect the remodeling and to regulate it, gaining at the same time a more stable interaction with the nucleosome.