Involvement of PARP1 in the regulation of alternative splicing

Abstract

Specialized chromatin structures such as nucleosomes with specific histone modifications decorate exons in eukaryotic genomes, suggesting a functional connection between chromatin organization and the regulation of pre-mRNA splicing. Through profiling the functional location of Poly (ADP) ribose polymerase, we observed that it is associated with the nucleosomes at exon/intron boundaries of specific genes, suggestive of a role for this enzyme in alternative splicing. Poly (ADP) ribose polymerase has previously been implicated in the PARylation of splicing factors as well as regulation of the histone modification H3K4me3, a mark critical for co-transcriptional splicing. In light of these studies, we hypothesized that interaction of the chromatin-modifying factor, Poly (ADP) ribose polymerase with nucleosomal structures at exon-intron boundaries, might regulate pre-mRNA splicing. Using genome-wide approaches validated by gene-specific assays, we show that depletion of PARP1 or inhibition of its PARylation activity results in changes in alternative splicing of a specific subset of genes. Furthermore, we observed that PARP1 bound to RNA, splicing factors and chromatin, suggesting that Poly (ADP) ribose polymerase serves as a gene regulatory hub to facilitate co-transcriptional splicing. These studies add another function to the multi-functional protein, Poly (ADP) ribose polymerase, and provide a platform for further investigation of this protein's function in organizing chromatin during gene regulatory processes.

Keywords: Epigenetics; PARP1; chromatin; cotranscriptional splicing; gene regulation.

Figure 1

PARP1 is enriched around the promoters of active genes and depleted at the ends of genes. Dyad density plot of PARP1-bound nucleosomes and total nucleosomes in S2 cells around (a) transcription start sites (TSSs) and (b) transcription termination ends (TTEs). To control for the difference in the total number of tags, dyad density scores are normalized by the average density over the genome.

Figure 2

PARP1 demarcates exons and is enriched at intron/exon and exon/intron boundaries irrespective of the transcriptional state of the genes. For this calculation, intron-containing genes were used and curves were normalized by genome-wide average. Dyad density of PARP1 nucleosomes at (a) start of exons (intron/exon) and (b) end of exons (exon/intron) of active genes, respectively. Dyad density of PARP1 nucleosomes at (c) start of exons (intron/exon) and (d) end of exons (exon/intron) of inactive or silent genes, respectively. To control for the effect of nucleosomes, dyad density scores are normalized by the average density over the genome (see Supplementary Figure S1).

Figure 3

Characteristic features of PARP1-bound nucleosomes. (a) PARP1-bound nucleosomes (red curve) are not enriched at first exons boundaries but are (b) highly enriched at internal exons (tag densities were measured at ±1000 bp), even if the effect of total nucleosome (green curve) is subtracted. For this calculation only intron-containing genes were used. (c) PARP1 associates with GC-rich nucleosomes. Plot shows PARP1-bound nucleosome density relative to total nucleosome density as a function of GC content. The black curve is the fitted smoothing line by loess method (local regression) using locally weighted polynomial regression analyses. (d) Heatmap showing that PARP1 binding overlaps with chromatin regions occupied by specific histone modifications. Pearson correlations of PARP1 and the following epigenetic marks (all with P-values<10¹⁰).

Figure 4

Global gene regulatory events mediated by PARP1 and its PARylation activity. (a) Four-way Venn diagram showing the different types of ASEs that are detected by either MATS or MISO detected by both methods. (b) Sashimi plots showing example of ASEs mediated by PARP1 and PARylation. RNA-seq read densities supporting ASEs and the estimated confidence levels are shown in the figure. (c) Visualization of the Gene Ontology Biological Process (BP) categories of PARP1 and PARylation-mediated ASEs. Circles are shaded based on types of ASEs as indicated on the legend. (d) Four-way Venn diagram summarizes the number of shared proteins in each combination of the four groups. Yellow and green: PARP1-mediated DEGs and ASEs, respectively; purple and red: PARylation-mediated DEGs and ASEs, respectively (P<0.01, by Student’s t-test). Numbers depicted in the intersections between circles represent the numbers of genes that are commonly regulated in two, three, or four conditions.

Figure 5

PARP1 regulates alternative splicing. (a–g) Measurements of ASEs at alternative exons. (h) Measured gene expression of constitutive exons. Representative gel images of the effect of PARP1 and PARylation on ASEs. Total RNA from wt and Parp ^C03256 flies were tested for changes in ASEs (lanes 1 and 2). S2 cells treated with PJ34 (lane 3), S2 cells treated with (i) LacZ non-targeting siRNA (lane 4), (ii) PARP1 siRNA1, (iii) PARP1 siRNA2 (KD1 and KD2, lanes 5 and 6, respectively). Bar charts represent the alternative arbitrary units from qRT-PCR measured as a rate of alternative exon included over the sum of all the alternative exons calculated from the mean intensities (n⩾3 biological replicates, ±S.D.; P<0.05 with Student’s t-test; see Supplementary Figures S2 and 3). Black and white boxes represent constitutive exons 5′ and 3′ to the alternative exons (gray boxes), respectively. Red arrows depict locations of primers.

Figure 6

Co-occupancy of PARP1 and H3K4me3 at PARP1-target exons. S2 cells were transfected with siRNA targeting lacZ (NT) and PARP1 (PARP1 KD). ChIP-qPCR analysis of PARP1, nucleosome density (as measured by H3) and H3K4me3 occupancies at PARP1-target exons in (a) Stau, (b) Fl(2)d and (c) Capt1 genes were analyzed in NT, PARP1 KD and PJ34-treated S2 cells. As proof of the specificity of the observed effect, these same factors were measured in Histone H1 Knockdown cells (H1 KD). Representative inverted agarose gel images of qPCR products stained with Gelstar are shown (far left). Bar graphs show ChIP-qPCR (quantitative real-time PCR) analyses normalized to IgG control of the measured occupancies. Antibodies used are indicated above each graph; results are represented as mean plus s.e.m. (n⩾3; Student's t-test, P<0.05). Primers that target the alternative exons as in Supplementary Figure S5B were used in ChIP PARP1, H3K4me4 co-occupancy assays.

Figure 7

Identification of PARP1 interactome in vivo. PARP1 is covalently cross-linked to nascent RNAs using 365 nm iUV light and 4-thiouridine. PARP1-bound nascent RNAs are immunoprecipitated using PARP1 antibody and then purified under stringent conditions. (a) PAR-CLIP procedure. (b) Radiolabeled PARP1-bound RNAs are blotted onto nitrocellulose, released by proteinase K and analyzed on a phosphoimager. The same blot was probed with anti-PARP1 antibody confirming PARP1-RNA binding. Knockdown of PARP1 or stringent RNase treatment of immunoprecipitated samples eliminated the PARP1-RNA band. Furthermore, protein samples resulting from PAR-CLIP experiments were split into three aliquots (i) no further treatment; (ii) stringent DNase1 treatment; (iii) stringent RNase A. These samples were subjected to mass spectrometry (Supplementary Table S5). (c) Proteins released from stringent RNase A and DNase1 digests were analyzed on a coomassie-stained gel and also probed with anti-H3 antibody to show depletion of histones.

Figure 8

PARP1 regulation of co-transcriptional splicing. (a) PARP1 and SF3B1 bind to the same nucleosomes. SF3B1 (U2 snRNP) ChIP shows that most SF3B1-bound nucleosomes (SF3B-IP) also bind PARP1. However, PARP1 binds other nucleosomes as indicated by H3 ChIP. Knockdown of PARP1 impaired the association of SF3B1 to nucleosomes, whereas inhibition of PARylation had no significant effect. (b) ChIP experiments showing co-occupancy of PARP1 and SF3B1 in HeLa cells. Knockdown of PARP1 resulted in reduction of SF3B1-nucleosome association, whereas PARylation inhibition (PJ34 treatment had no such effect). The results (bar graph) are represented as mean plus s.e.m. (n⩾3; Student's t-test, *P<0.05). (c) Model of PARP1 in mediating co-transcriptional splicing. PARP1 binds to specific nucleosomes at exons (specified by specific histone PTMs, for example, H3K4me3) and also binds to the nascent pre-mRNA and recruits SF3B1 a U2 component. U2 binds to the branch-point recognized by the splicing machinery, allowing PARP1 to influence exon recognition. RNA polymerase II generating the nascent pre-mRNA is shown on the right.