A new link between transcriptional initiation and pre-mRNA splicing: The RNA binding histone variant H2A.B

Abstract

The replacement of histone H2A with its variant forms is critical for regulating all aspects of genome organisation and function. The histone variant H2A.B appeared late in evolution and is most highly expressed in the testis followed by the brain in mammals. This raises the question of what new function(s) H2A.B might impart to chromatin in these important tissues. We have immunoprecipitated the mouse orthologue of H2A.B, H2A.B.3 (H2A.Lap1), from testis chromatin and found this variant to be associated with RNA processing factors and RNA Polymerase (Pol) II. Most interestingly, many of these interactions with H2A.B.3 (Sf3b155, Spt6, DDX39A and RNA Pol II) were inhibited by the presence of endogenous RNA. This histone variant can bind to RNA directly in vitro and in vivo, and associates with mRNA at intron-exon boundaries. This suggests that the ability of H2A.B to bind to RNA negatively regulates its capacity to bind to these factors (Sf3b155, Spt6, DDX39A and RNA Pol II). Unexpectedly, H2A.B.3 forms highly decompacted nuclear subdomains of active chromatin that co-localizes with splicing speckles in male germ cells. H2A.B.3 ChIP-Seq experiments revealed a unique chromatin organization at active genes being not only enriched at the transcription start site (TSS), but also at the beginning of the gene body (but being excluded from the +1 nucleosome) compared to the end of the gene. We also uncover a general histone variant replacement process whereby H2A.B.3 replaces H2A.Z at intron-exon boundaries in the testis and the brain, which positively correlates with expression and exon inclusion. Taken together, we propose that a special mechanism of splicing may occur in the testis and brain whereby H2A.B.3 recruits RNA processing factors from splicing speckles to active genes following its replacement of H2A.Z.

Fig 1. H2A.B.3 is located at the intron—exon boundary of active genes in the testis.

Input nucleosomes, nucleosomes immunoprecipitated with H2A.B.3 or H3K36me3 affinity purified antibodies, and poly (A)-transcripts obtained from 28–30 day old mice testes were sequenced yielding 100 base pair paired-end reads (see Material and methods). (a) The individual lines represent the normalised H2A.B.3 and total input reads aligned between -1 and +10 kb from the TSS in the testis. (b) The individual line represents the normalized input nucleosome reads (mean reads per base pair per million reads mapped (RPM)) aligned between -1 and +1 kb from the intron—exon boundary for all exons in the testis ranked according their expression level (repressed, low, medium and high). The colour-map panel shows the relationship between colour and the gene expression rank. (c) Normalized testis H2A.B.3 ChIP-Seq reads ranked according to their expression level aligned with the intron—exon boundary. (d) At each base position relative to the intron exon-boundary, a linear model was fitted to the mean H2A.B.3/input ratio versus gene log expression across all intron—exon boundaries, and the slope of the fitted model plotted for the testis. (e) Normalized testis H3K36me3 ChIP-Seq reads ranked according to their expression level aligned with the intron—exon boundary. (f) Pearson correlation of the log coverage, across 50 base pair windows, was calculated between testes H2A.B.3 ChIP-Seq reads and H3K36me3 ChIP-Seq reads for each base pair relative to the intron—exon boundary. (g) Normalized testis H3K36me3 ChIP-Seq reads ranked according to the incorporation of H2A.B.3 (very low, low, medium and high) aligned with the intron—exon boundary. 95% confidence bands are shown in grey.

Fig 2. H2A.B.3 replaces H2A.Z on the coding region of active genes.

(a) Normalized testis H2A.Z ChIP-Seq reads ranked according to their expression level (repressed, low, medium and high) aligned with the intron—exon boundary. (b) Normalized hippocampus H2A.Z ChIP-Seq reads ranked according to their expression level aligned with the intron—exon boundary. (c) Normalised testis H2A.Z ChIP-Seq reads ranked according to the incorporation of histone H3 trimethyl K36 (very low, low, medium and high) aligned with the intron—exon boundary. (d) Quantitative H2A.B.3 and H2A.Z ChIP assays were performed in the testis at three exons for each of three individual genes (Akap4, Il2rg and Akap14) located on the X chromosome that are repressed at day 18 (the pachytene stage) but activated at day 30 (the late round spermatid stage). Standard deviation of three replicates is shown. (e) Quantitative H2A.B.3 and H2A.Z ChIP assays were performed in the hippocampus at an exon and its neighbouring intronic sequences for genes that are expressed more highly in the brain (Ctnnd2, Mpped1 and Ctnn1) versus the testis (a brain to testis expression ratio of 27.1, 17.4 and 75.4, respectively). Standard deviation of three replicates is shown. (f) Quantitative H2A.B.3 and H2A.Z ChIP assays were performed in the testis at an exon and its neighbouring intronic sequences for genes that are expressed more highly in the testis (Pkib, Tbata and Slain2) versus the brain (a testis to brain expression ratio of 7.3, 23.4 and 22.2, respectively). Standard deviation of three replicates is shown.

Fig 3. The incorporation of H2A.B.3 at alternatively spliced exons is positively correlated with the level of inclusion.

(a) Normalised testis H2A.B.3 ChIP-Seq reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons (very low, 20%; low, 40%; medium, 60%; and high, 80%). (b) Normalised hippocampus H2A.B.3 ChIP-Seq reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons. (c) Normalised testis input reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons. (d) Normalised hippocampus input reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons. (e) Normalised testis H2A.Z ChIP-Seq reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons. (f) Normalised hippocampus H2A.Z ChIP-Seq reads aligned with the intron—exon boundary ranked according to the level of inclusion of alternatively spliced exons.

Fig 4. H2A.B.3 co-immunoprecipitates with proteins involved in RNA processing and transcription.

Mononucleosome-enriched chromatin from 28–30 day old mouse testes was immunoprecipitated with anti-H2A.B.3 or anti-H2A.Z antibodies. Co-immunoprecipitated proteins were identified by subsequent western blotting with the indicated antibodies selected to detect proteins involved in different aspects of RNA synthesis, processing and export. The presence of H2A.B.3, H2A.Z, histone H3 and H3K36me3 in the H2A.B.3 or H2A.Z co-immunoprecipitate were also examined by western blotting. Input chromatin (1/20 of the amount used for a ChIP-Seq experiment) was used as a loading control.

Fig 5. H2A.B.3 can bind RNA in vitro and in vivo.

(a) Total cellular lysates were prepared from UV treated mouse testes and treated with RNase I or not. H2A.B.3 was then immunoprecipitated and the co-immunoprecipitated proteins were identified by western blotting with the indicated antibodies selected to detect proteins involved in different aspects of RNA synthesis, processing, and export. (b) Amino acid sequence alignment of the N-terminal region of histone H2A and the variants H2A.Z, H2A.B.3, and H2A.B. Compared to H2A, the N-terminus of H2A.B.3 and H2A.B are 6.3% and 23.5% identical, respectively. The red box demarcates the sequences corresponding to the N-terminal peptides used for the pulldown experiments in panel d, and corresponds to the unstructured region (dashed line) preceding the first alpha helix of H2A (α1; orange box). Arginine residues are highlighted in blue. (c) Histone dimer samples (0.6, 1.1, 2.3, 4.5 μM) were incubated with 20 ng in vitro transcribed RNA (222 nt and 152 nt, top and bottom panels, respectively) and analysed on 5% acrylamide 1X TB gels. The asterisk (*) denotes shifted bands corresponding to H2A.B—H2B-RNA complexes. (d) An RNA pulldown assay using biotinylated histone N-terminal peptides (n-H2A, n-H2A.Z, n-H2A.B and n-H2A.B; 130 pmol). Samples were run on 15% TBE-Urea gels, along with input RNA (5 pmol; 3% of total input) for comparison. (e) CLIP assays demonstrating that H2A.B.3 but not H2A.Z directly interacts with RNA in germ cells. Also show is the western blot analysis of the immunoprecipitated H2A.B.3 and H2A.Z. Following the RNA—IP procedure (see Methods), cells isolated from 28–30 day old testes were UV crosslinked, the chromatin sheared and following the immunopurification of H2A.B.3-containing chromatin fragments, the released RNA was sequenced to yield 100 base pair paired end reads. (f) H2A.B.3 RNA plot ranked according to expression aligned with all intron—exon boundaries. (g) A H2A.B.3 RNA plot ranked according to the level of exon inclusion (20 to 80%) aligned with the intron—exon boundary of alternatively spliced exons.

Fig 6. H2A.B.3 co-localises with proteins involved in RNA processing and transcription.

(a) Hypotonic spreads of germ cells from adult mouse testis showing round (i-v, vii) and elongating spermatids (vi) were indirectly immunostained with antibodies against H2A.B.3 (i-vi) or H2A.Z (vii), the splicing speckle marker, Y12 (i and vii) and representative RNA-binding proteins that co-immunoprecipitated with H2A.B.3 in panel a (ii-vi). White arrows show splicing speckles. Scale bar, 10μm. (b) Round spermatids were isolated by gravity sedimentation and then fractionated into a chromatin fraction (chrom), a cytoplasmic and nucleoplasmic fraction (Cyt+Nucl) and a loosely bound chromatin faction (loosely chrom bound). (c) Hypotonic spreads of round spermatids from adult mouse testis were stained with DAPI and indirectly immunostained with antibodies against the splicing speckle marker, Y12 (i-iv), initiating and elongating states of RNA Pol II S5 (i), and S2 (ii) respectively, H3K36me3 (iii), H4K20me1 (iv). White arrowheads show accumulation of a signal in splicing speckles, empty arrowheads show depletion. Scale bar, 10μm.

Fig 7. A model depicting the role of H2A.B.3 in recruiting RNA processing factors to the gene body during gene activation.

Following the replacement of H2A.Z with H2A.B.3 to assemble active chromatin, H2A.B.3 directly recruits RNA processing factors from splicing speckles to an active gene. Upon transcriptional elongation and the synthesis of mRNA, H2A.B.3 binds and ‘holds’ onto this RNA thus releasing the splicing factors to facilitate the splicing process.