Methyl-CpG-binding domain 9 (MBD9) is required for H2A.Z incorporation into chromatin at a subset of H2A.Z-enriched regions in the Arabidopsis genome

Abstract

The SWR1 chromatin remodeling complex, which deposits the histone variant H2A.Z into nucleosomes, has been well characterized in yeast and animals, but its composition in plants has remained uncertain. We used the conserved SWR1 subunit ACTIN RELATED PROTEIN 6 (ARP6) as bait in tandem affinity purification experiments to isolate associated proteins from Arabidopsis thaliana. We identified all 11 subunits found in yeast SWR1 and the homologous mammalian SRCAP complexes, demonstrating that this complex is conserved in plants. We also identified several additional proteins not previously associated with SWR1, including Methyl-CpG-BINDING DOMAIN 9 (MBD9) and three members of the Alfin1-like protein family, all of which have been shown to bind modified histone tails. Since mbd9 mutant plants were phenotypically similar to arp6 mutants, we explored a potential role for MBD9 in H2A.Z deposition. We found that MBD9 is required for proper H2A.Z incorporation at thousands of discrete sites, which represent a subset of the genomic regions normally enriched with H2A.Z. We also discovered that MBD9 preferentially interacts with acetylated histone H4 peptides, as well as those carrying mono- or dimethylated H3 lysine 4, or dimethylated H3 arginine 2 or 8. Considering that MBD9-dependent H2A.Z sites show a distinct histone modification profile, we propose that MBD9 recognizes particular nucleosome modifications via its PHD- and Bromo-domains and thereby guides SWR1 to these sites for H2A.Z deposition. Our data establish the SWR1 complex as being conserved across eukaryotes and suggest that MBD9 may be involved in targeting the complex to specific genomic sites through nucleosomal interactions. The finding that MBD9 does not appear to be a core subunit of the Arabidopsis SWR1 complex, along with the synergistic phenotype of arp6;mbd9 double mutants, suggests that MBD9 also has important roles beyond H2A.Z deposition.

Fig 1. N-TAP-ARP6 and C-TAP-ARP6 transgenes rescue the arp6-1 phenotype.

(A- upper blot) T₂ plants homozygous for N-TAP- or C-TAP-ARP6 transgenes express a fusion protein with the expected size of 67.5 kDa. The fusion protein is specifically detected only in transgenic plants and not in arp6-1 or WT plants using a peroxidase anti-peroxidase (PAP) soluble complex, which binds the protein A moiety of the TAP tag. (A- lower blot) The same protein extracts as in the upper blot were probed with a monoclonal ARP6 antibody. ARP6 presence is specifically detected in all transgenic plants as a 67.5 kDa fusion protein band compared to the 44 kDa ARP6 band in WT plants, and is absent in arp6-1 mutant plants. The ARP6 antibody reacts less strongly with N-TAP-ARP6, most likely because the antibody recognizes the N-terminal region, which is adjoined to the TAP tag in this fusion. (B) Transgenic plants look more similar to WT plants than arp6-1 plants, with more compacted, non-serrated rosette leaves. (C) Early flowering phenotype is rescued in transgenic plants when compared to the arp6-1. (D) The average number of rosette leaves of N-TAP-ARP6 and C-TAP-ARP6 transgenic plants at flowering is significantly higher than in arp6-1 (n = 6 for WT and arp6-1, and n = 12 for N-TAP 11–4 and C-TAP 10–2). Asterisks indicate significant differences from arp6-1 plants with p<0.001, calculated using unpaired t-tests. (E) The loss of apical dominance defects of arp6-1 plants are rescued in N-TAP-ARP6 and C-TAP-ARP6 transgenic plants. (F) The fertility defects of arp6-1 plants are rescued in N-TAP 11–4 and C-TAP 10–2 transgenic plants.

Fig 2. The effects of MBD9 on H2A.Z incorporation and H4 acetylation.

All ChIP-seq profiles of WT (shown in blue), mbd9-1 (shown in green), and arp6-1 (shown in red) were produced using SeqPlots. Standard error is represented as shading around the solid line of the mean signal. (A) Average ChIP-seq H2A.Z profiles plotted over gene body coordinates for all protein-coding Arabidopsis genes, from the transcript start site (TSS) to the transcript end site (TES). (B) H2A.Z signal for each genotype over reproducible H2A.Z-enriched regions from WT plants. (C) Average ChIP-seq H4Ac profiles plotted across gene bodies. (D) H4Ac signals over reproducible H4Ac-enriched regions from WT plants.

Fig 3. Identification of 1391 H2A.Z-enriched sites that require MBD9 for H2A.Z incorporation into chromatin.

(A-B) Volcano plots of–log10 of the adjusted P value (y-axis) versus log2 fold change of H2A.Z ChIP-seq reads (x-axis) between wild type plants and arp6-1 mutant plants (A) or mbd9-1 mutant plants (B). Each point corresponds to a called H2A.Z peak that was analyzed by DESeq2. Peaks that had a log2 fold change equal to or greater than 0.6 and an adjusted p value of 0.05 or less are colored red. Out of 7,039 peaks analyzed, there are 6,266 peaks that are significantly depleted of H2A.Z in arp6-1 plants (A, red dots), and 1,391 peaks that are significantly depleted of H2A.Z in mbd9-1 plants (MBD9-dependent peaks, panel B, red dots). Peaks that had an absolute log2 fold change from 0 to 0.25 in the WT to mbd9-1 comparison are colored light blue (MBD9-independent peaks, panel B, light blue dots). (C) Heatmaps (left) and average plots (right) of the 1391 MBD9-dependent (red bar, top of heatmaps, top average plot) peaks that are significantly depleted of H2A.Z in mbd9-1 plants and 1,505 MBD9-independent (blue bar, bottom of heatmaps, bottom average plot) peaks that had an absolute log2 fold change from 0 to 0.25 in the WT to mbd9-1 plants. Plots are centered on each peak and show a 2 kb window around the peak centers.

Fig 4. H2A.Z sites that require MBD9 have distinct chromatin properties.

Average ChIP-seq profiles of H3K9Ac (A), H3K18Ac (B), H3K27me3 (C), H3K9me2 (D), H3K4me3 (E), and H3K36me3 (F) at MBD9-dependent H2A.Z sites (shown in red) and MBD9-independent H2A.Z sites (shown in blue). Plots show a 2 kb window around the center of the MBD9-dependent and MBD9-independent sites. Average read density for either set of sites is shown as a single line depicting the mean with standard error denoted as dark shading, and 95% confidence interval denoted as light shading.

Fig 5. MBD9 has distinct binding preferences for modified histone tails.

The intensities of each spot on the histone peptide array after interaction with a full-length MBD9 protein were quantified using software provided by the array manufacturer and were further compared to the spot intensities of unmodified H4 peptide (A), or unmodified H3 peptide (B). Y-axis represents average spot intensities with values ranging from 0 (no signal) to 1 (maximum signal strength). Each bar with standard deviations represents an average of 4 array spot intensities (two experimental replicates, each with two technical replicates) for a given modified histone peptide. Only the normalized average spot intensities with values higher than 0.49 are presented here except for the spot intensities of control peptides (unmodified H4 in panel A, and unmodified H3 peptide in panel B), and of two additional peptides in panel B (H3K4me3 and H3K27me3), which are included for comparison. Asterisks indicate level of significant differences in spot intensities between given histone peptides and control peptides (* = p<0.05, ** = p<0.01, *** = p<0.001), calculated using unpaired, one-tail t-tests. (A) MBD9 shows significant interaction preference for acetylated H4 residues. (B) MBD9 also shows higher preference for mono- and di-methylated H3K4 over tri-methylated H3K4, strongly interacts with methylated arginines in combination with other modifications of histone H3, including K9Ac, and it does not recognize H3K27me3 above background.

Fig 6. MBD9 is not a core subunit of the Arabidopsis SWR1 complex.

(A) Schematic representation of the Arabidopsis SWR1 protein complex based on TAP-tag experiments, including MBD9 as a potential SWR1 subunit. (B) Estimated size of the Arabidopsis SWR1 complex with and without MBD9 as a subunit. The estimated size was calculated based on the known stoichiometry of the yeast SWR1 complex [59] and predicted molecular weights of the Arabidopsis SWR1 subunits listed in Table 1. (C) Protein gel blots of even-numbered SEC fractions from WT (top blot) and mbd9-1 (bottom blot) plants. The blots were incubated with the ARP6 monoclonal antibody [40]. Asterisks indicate the ARP6 peak fractions. The average molecular weights of ARP6-containing protein complexes in WT and mbd9-1 plants were calculated from two biological replicates and presented on the right side of the corresponding protein blots.

Fig 7. arp6-1;mbd9-1 double mutant plants have a more severe phenotype than either single mutant.

WT, arp6-1, mbd9-1, and arp6-1;mbd9-1 plants grown under long-day conditions were individually photographed at 7, 14, 21, and 35 days after stratification. arp6-1;mbd9-1 double mutant plants have severely delayed development and are dwarfed compared to single arp6-1 and mbd9-1 mutant plants.