TALE-directed local modulation of H3K9 methylation shapes exon recognition

Abstract

In search for the function of local chromatin environment on pre-mRNA processing we established a new tool, which allows for the modification of chromatin using a targeted approach. Using Transcription Activator-Like Effector domains fused to histone modifying enzymes (TALE-HME), we show locally restricted alteration of histone methylation modulates the splicing of target exons. We provide evidence that a local increase in H3K9 di- and trimethylation promotes inclusion of the target alternative exon, while demethylation by JMJD2D leads to exon skipping. We further demonstrate that H3K9me3 is localized on internal exons genome-wide suggesting a general role in splicing. Consistently, targeting of the H3K9 demethylase to a weak constitutive exon reduced co-transcriptional splicing. Together our data show H3K9 methylation within the gene body is a factor influencing recognition of both constitutive and alternative exons.

Figure 1. Local manipulation of histone modifications by TALE-HME affects alternative splicing of the target exon.

(A) Schematic representation of the approach. A TALE domain was assembled to bind the target exon and fused to HMEs. Upon binding of TALE-HME to the target exon, the local chromatin environment is modified. (B–E) Upon transient transfection with TALE-HME or TALE-GFP as a control, the effect on chromatin was monitored by ChIP (left panel) and EDB inclusion levels were assessed by RT-qPCR (right panel). ChIP signals are calculated as immunoprecipitated DNA over input signal and normalized to total H3 and an intergenic region on chromosome 10. Amplicon positions for FN1 are indicated in the gene diagram at the bottom. EDB inclusion rates are calculated as the ratio of the EDB exon to the upstream constitutive exon 24 and normalized to TALE-GFP. Mean ± SEM are shown, n = 3–4. Statistical significance of all results was analyzed by t-test and significant changes (p < 0.05) were indicated by * .

Figure 2. H3K9me3 signal at internal exons.

Relative H3K9me3 abundance profiles in HeLa cells 1kb up and downstream of transcriptional start sites (TSS, upper left panel), polyadenylation sites (p(A) site, upper right panel), 3′ splice sites (lower left panel) and 5′ splice sites (lower right panel) of internal exons were plotted versus relative genomic positions 1 kb up and downstream of gene architecture hallmark of interest. Genome wide abundance profiles were derived from chromatin immunoprecipitation sequencing (ChIP-seq) experiments performed in the ENCODE project. Schematic gene architecture hallmarks are displayed above. See also Supplementary Fig. S2 for K562 and MCF-7 analysis.

Figure 3. H3K9me3 promotes co-transcriptional splicing of FOSL1.

(A) Stable U2OS-TO cell lines for doxycycline inducible overexpression of JMJD2D and Suv39H1 were established to test the effect of H3K9me3 on co-transcriptional splicing of exons 3 to 4 of FOSL1. Schematic representation of the Tet-On constructs and the endogenous FOSL1 gene are shown. To select nascent transcripts a primer downstream of the poly(A) cleavage site was used for reverse transcription. qPCR primers used for detection of pre-mRNA or mRNA are indicated below the gene diagram; above are the primers for ChIP. (B) H3K9me3 ChIP on FOSL1 exon 3 showing reduced H3K9me3 levels after JMJD2D overexpression, while Suv39H1 had no additional effect. ChIP signals are calculated as immunoprecipitated DNA over input and normalized to the uninduced control cells. Mean ± SEM are shown, n = 3. Statistical significance of all results was analyzed by t-test and significant changes (p < 0.05) were indicated by*. (C) Unspliced pre-mRNA accumulates after JMJD2D overexpression and depletion of H3K9me3 levels, while Suv39H1 had no effect on co-transcriptional splicing of FOSL1 exons 3–4 as determined by RT-qPCR. Ratios of unspliced pre-mRNA to spliced mRNA are normalized to uninduced control cells. Mean ± SEM are shown, n = 3. Statistical significance of all results was analyzed by t-test and significant changes (p < 0.05) were indicated by*. (D) Co-transcriptional splicing of FOSL1 exons 3–4 was assessed after targeting H3K9 methyltransferases and demethylases by TALE-HMEs binding exon 3. A primer downstream of the poly(A) cleavage site was used for reverse transcription to select nascent transcripts. Tethering of JMJD2D, but not the catalytically inactive JMJD2D-N202M, resulted in accumulation of unspliced transcripts. In contrast, tethering of methyltransferases G9a or Suv39H1 did not further improve co-transcriptional splicing. Ratios of unspliced pre-mRNA to spliced mRNA are normalized to TALE-GFP. Mean ± SEM are shown, n = 3. Statistical significance of all results was analyzed by t-test and significant changes (p < 0.05) were indicated by*.