|Home | About | Journals | Submit | Contact Us | Français|
The cause and consequence of enriched DNA methylation within gene bodies of actively transcribed genes is a mystery that has garnered significant attention in recent years. In particular, elevation of methylation at exons relative to introns has fueled investigations into whether methylation aids the spliceosome in the process of exon “definition."1 However, efforts to modulate methylation led to both increased and decreased inclusion of variable exons,2,3 arguing against a direct role in spliceosome assembly. Accordingly, we uncovered an indirect link between methylation and splicing that is based on variable intragenic binding of a methyl-sensitive transcription factor. Specifically, we found that intragenic binding of the zinc-finger protein CCCTC-binding factor (CTCF) promotes inclusion of weak upstream exons in spliced mRNA, whereas overlapping 5-methylcytosine (5mC) promotes exon exclusion through CTCF eviction.4 The mechanistic basis for the reciprocal effects on pre-mRNA processing extends from the ability of CTCF to kinetically regulate splicing through transient obstruction of RNA polymerase II (pol II) elongation. In other words, CTCF-associated pol II pausing favors spliceosome assembly at weak splice sites that may otherwise be outcompeted by stronger, downstream splice sites. Accordingly, loss of pol II pausing through 5mC-mediated CTCF eviction has the opposite effect, leading to upstream exon exclusion.4 However, the mechanism supporting regulated exchange of the 5mC/CTCF “epigenetic splicing switch” to achieve cell-appropriate alternative pre-mRNA splicing was unclear. The importance of this putative regulation is heavily underscored by significant biological context: 1) to represent a bona fide mechanism of alternative splicing the 5mC/CTCF switch should be temporally modulated in response to specific cues and 2) given the well-known role of CTCF as the master architect of the genome, the switch should be spatially regulated such that only select CTCF sites are modified.
To explore the mechanisms regulating CTCF/5mC dependent splicing, we examined a developmentally regulated switch in the CD45 gene. Naïve peripheral lymphocytes bind CTCF at CD45 DNA and include an upstream alternative exon in spliced mRNA. In contrast, activating antigenic signals cause the CTCF-binding site to become methylated, resulting in upstream exon exclusion. Focusing on this tractable switch, we investigated the mechanism leading to increased 5mC at CD45 DNA in activated lymphocytes. To our surprise, bisulfite analysis revealed a steady level of overlapping methylation at the CTCF-binding site in both the naïve and activated populations. As bisulfite conversion fails to discriminate between 5mC and the TET protein-catalyzed oxidation product 5-hydroxymethylcytosine (5hmC), we examined for overlapping 5hmC in the CTCF-binding populations. Utilizing antibodies specific to 5mC or 5hmC, we revealed opposing patterns: 5hmC was consistently elevated in CTCF-binding cells whereas 5mC was reciprocally elevated in the absence of CTCF binding. We further showed that TET-catalyzed oxidation of 5mC was required for CD45 exon inclusion in naïve lymphocytes, and determined that declining TET1 and TET2 levels following lymphocyte activation were responsible for emerging 5mC at the CTCF binding site.5 These dynamic methylation changes were relatively restricted to the CTCF dependent exon, as were the observed changes in splicing, thus demonstrating the predicted spatial and temporal regulation of the epigenetic splicing switch.
The activation-induced decline in TET levels further allowed for a generalization of our findings in a physiologically relevant context. The involvement of TET-catalyzed 5mC oxidation in CTCF-dependent splicing was examined at the genome-wide level in naïve and activated primary CD4+ T cells. Determined CTCF binding sites proximal to alternatively spliced exons were examined for overlapping 5mC and 5hmC. The resulting integrated analysis revealed strong adherence to the previously established model for alternative exons with downstream CTCF binding: locations with increased 5hmC and decreased 5mC (CTCF binding) upon T cell activation were associated with upstream exon inclusion, whereas locations with decreased 5hmC and increased 5mC (CTCF eviction) were associated with upstream exon exclusion.5 These genome-wide results establish the TET proteins as global regulators of CTCF-dependent splicing, and provide a basis for dynamic regulation of epigenetic splicing switches through variations in TET activity.
Notably, our study uncovered an additional layer of complexity in CTCF-dependent splicing. Biochemical efforts to examine CTCF association with oxidized 5mC derivatives revealed that CTCF preferentially interacts with 5-carboxylcytosine (5caC) containing DNA. While levels are generally low within genomic DNA, 5caC was readily detected within CTCF binding sites in the CD45 gene and genome-wide in thymine-DNA glycosylase (TDG) knockout embryonic stem cells, which accumulate 5caC.5,6 While the direct significance of these findings to splicing regulation in vivo remains unclear, they raise a number of intriguing questions. First, enhanced interaction of CTCF with 5caC-containing DNA suggests that 5caC adopts unique structural and/or electrophysical properties. If true, does 5caC act to preserve CTCF interaction with DNA in the context of active transcription? More generally, how does the full spectrum of 5mC-oxidated derivatives contribute to the lexicon of the “splicing code”? DNA binding factors that are specific to each of the oxidized derivatives have been identified,7 raising the possibility that dynamic intragenic methylation creates a platform for widespread modulation of transcription factors with the potential to impact pre-mRNA splicing decisions (Fig. 1). An expanded description of the intragenic epigenome and associated factors will aid our understanding of these and related questions in the coming years.
No potential conflicts of interest were disclosed.