The discovery of an increasing number of alternative splicing events in the human genome highlighted that ∼94% of genes generate alternatively spliced transcripts that may produce different protein isoforms with diverse functions. It is now well known that several diseases are a direct and indirect consequence of aberrant splicing events in humans. In addition to the conventional mode of alternative splicing regulation by ‘cis’ RNA-binding sites and ‘trans’ RNA-binding proteins, recent literature provides enormous evidence for epigenetic regulation of alternative splicing. The epigenetic modifications may regulate alternative splicing by either influencing the transcription elongation rate of RNA polymerase II or by recruiting a specific splicing regulator via different chromatin adaptors. The epigenetic alterations and aberrant alternative splicing are known to be associated with various diseases individually, but this review discusses/highlights the latest literature on the role of epigenetic alterations in the regulation of alternative splicing and thereby cancer progression. This review also points out the need for further studies to understand the interplay between epigenetic modifications and aberrant alternative splicing in cancer progression.
Alternative splicing is one of the major processes that drive transcriptome and proteome diversity, and it is estimated that up to 94% of genes are alternatively spliced in humans [1–3]. The pre-mRNA splicing is an essential step of eukaryotic gene expression in which intronic sequences are removed and exonic sequences are ligated together . Pre-mRNA splicing is catalyzed by the spliceosome, a ribonucleoprotein complex comprising of five small nuclear ribonucleoproteins (snRNPs) and numerous proteins. The pre-mRNA has specific interaction via base-pairing with U1, U2, U4, U5, and U6 snRNPs . During the spliceosome assembly, U1 forms a specific base-pairing with the 5′-splice site and U2 forms a base-pairing with the branch point. Later, U4, U5, and U6 tri-snRNP complex associates with the forming spliceosome, displacing U4 from the complex. This allows U6 to replace U1 at the 5′-splice site and leads to a U6–U2 association that brings the 5′-splice site and the branch point in close vicinity, permitting the first transesterification process. Then, U5 brings the two exons into close proximity, allowing the second transesterification reaction that results in the removal of an intron in the form of lariat and joining of exons . The alternative splicing is the combinatorial selection of splice sites resulting in the generation of more than one alternatively spliced isoforms from a single pre-mRNA. The majority of alternative splicing events undergo cell/tissue-specific regulation in which splicing pathways are modulated according to cell and tissue type, developmental stage, gender, or in response to external factors [6–8]. Deregulation of splicing machinery may generate alternatively spliced protein isoforms that may have altered functional domains that may contribute to different aspects of various disease progression and resistance to treatment [9–12]. In the past decade, a plethora of mRNA isoforms specific to stages of cellular development and disease, including oncogenesis, have been revealed [13,14]. The molecular mechanisms that determine the choice of particular splicing events associated with different stages of normal development and diseases are now beginning to be elucidated [15–17].
Recent studies provided the evidence that deregulated alternative splicing events often reveal abnormalities in splicing regulation machinery. The pre-mRNA splicing regulation is mediated by cis-acting splicing elements and trans-acting splicing factors. Alterations in expression and activity of regulatory splicing factors, mutations in the core components of the splicing machinery and cis-acting splicing sequences may result in deregulated alternative splicing in many diseases, including cancer [18,19]. In addition to the genetic factors, evidence suggests that epigenetic modifications also contribute to alternative pre-mRNA splicing patterns [4,12,20,21].
Increasing evidence in the past several years has revealed the fundamental role of epigenetics in the development and progression of various diseases [22–24]. A growing body of work relates the epigenetic events with alternative splicing in several model systems including lymphocyte differentiation, neuronal differentiation, and epithelial-to-mesenchymal transition [25–27]. Although epigenetics and alternative splicing have been individually very well studied in various diseases [10,16,28–30], there is less information regarding the role of altered epigenetic marks in the generation of disease-specific alternative spliced isoforms. Despite the fact that altered epigenetic modifications are involved in disease progression, they are potentially reversible using therapeutic interventions. In the past few years, researchers have made attempts to delineate the epigenetic mechanisms with the goal of finding novel targets for therapy that can modify the misregulated splicing pattern and relieve the transcriptional repression of key genes and improve clinical outcomes [31–33]. Here, we discuss the role of epigenetic modifications, such as DNA methylation, chromatin structure, and histone modifications, in alternative splicing regulation in cancer.
Coupling between transcription and splicing
Researchers working in the field of RNA splicing in the dawn of the 21st century focused solely on the role of RNA-binding proteins that bind to enhancer or silencer sequences within the pre-mRNA to regulate alternative splicing events [19,34]. Recent research has highlighted the importance of transcriptional dynamics in regulating interactions between the spliceosome and individual exons with the pre-mRNA. Reports suggest that spliceosome assembly occurs co-transcriptionally and is co-ordinated with the transcription elongation rate and chromatin structure [35,36]. Splicing is a co-transcriptional process, and RNA polymerase II (RNA Pol II) is a key player, co-ordinating transcription with splicing [35,37,38]. The RNA Pol II is distinguished from the other eukaryotic RNA polymerases by the presence of highly conserved heptapeptide repeats in the C-terminal domain (CTD), the modification pattern of which constitutes a code known as the CTD code. Post-translational modifications of these heptapeptides repeat in the CTD domain of Pol II, particularly phosphorylation of serine and threonine residues along with other key residues (such as proline isomerization) plays a key role in the regulation of its activity and helps in coupling the transcription and numerous RNA processing events [39,40]. During the initiation of transcription, TFIIH places bivalent marks in the form of phosphorylation on ser5 and ser7, and during elongation ser2 is phosphorylated by the P-TEFb elongation factor [41,42]. The phosphorylation shift from ser5 to ser2 of the RNA Pol II CTD during elongation plays a key role in the co-transcriptional splicing process. Further, RNA Pol II CTD directly interacts with RNA splicing factors, recruits them to the nascent RNA transcript, and regulates the pre-mRNA splicing [43,44]. Additionally, the serine–arginine (SR) families of proteins have been shown to functionally interact with the RNA Pol II CTD to affect pre-mRNA splicing . Another important regulator of splicing is nucleosome positioning. Approximately, 145–147 base pairs of DNA is wrapped around the histone octamer to form a nucleosome core. Recent high-throughput genomic profiling studies have identified the nonrandom distribution of nucleosomes at exonic regions when compared with intronic regions [46,47]. The correlative evidence suggests that nucleosomes may cause a physical barrier to RNA Pol II transcription and can affect splicing by modulating RNA Pol II dynamics [48,49]. The RNA Pol II pause at nucleosomes enriched on the exons allows the spliceosome to recognize the splice sites and helps in regulating the pre-mRNA splicing. Various studies have suggested nucleosome positioning as a key determinant of exon recognition and alternative pre-mRNA splicing [50,51]. Similarly, the SWI/SNF protein complex primarily studied for chromatin remodeling at promoter regions has also been found to be enriched at the intragenic regions. These chromatin remodelers thus may function as a regulator of alternative splicing [52,53]. The SWI/SNF complex protein, Brm, associates with several components of the spliceosome and favors the alternative exon inclusion in CD44. Brm induces the accumulation of RNA Pol II with a modified CTD on the variant exons of the CD44 gene. In addition, Brm also regulates the alternative pre-mRNA splicing of the Cyclin D1, E-cadherin, and BIM genes. Thus, Brm facilitates the cross-talk between transcription and RNA processing by decreasing RNA Pol II elongation kinetics and aids in the recruitment of the splicing machinery to the exons with weak splice sites . In brief, the regulation of co-transcriptional splicing can be explained by two well-defined models: (1) kinetic and (2) adaptor models. As suggested by the kinetic model of alternative splicing, the RNA Pol II elongation rate plays a key role in regulating the pre-mRNA splicing by providing sufficient time to the spliceosome machinery to recognize the alternative exons and thereby leads to exon inclusion or exclusion in a context-dependent manner [54,55], whereas, in the adaptor model, different chromatin modifications recruit chromatin adaptor proteins and splicing factors, leading to differential splice site selection that ultimately regulates the alternative pre-mRNA splicing process [20,56]. Although the kinetic and adaptor models have individually been well defined to play a role in the regulation of alternative splicing, these models may not be mutually exclusive and an outcome of the alternative splicing process might depend on the integration of both, as suggested in some of the previous reports [57,58].
The interplay between epigenetics and splicing
Epigenetic marks, such as DNA methylation, histone post-translational modifications, and nucleosome occupancy, are involved in determining the chromatin structure, while some transcription factors can bind to specific regulatory regions to recruit the epigenetic modifiers to direct the specific changes in chromatin structure and thereby regulate gene expression . Earlier it was believed that these epigenetic factors were found to be enriched only at promoter regions . However, it has become increasingly clear that they are also enriched in intragenic regions. Moreover, genome-wide mapping analysis of methylation, histone modifications, and nucleosomes have revealed their nonrandom distribution around exons, with exonic regions having high levels of nucleosome occupancy and modified histones compared with intronic regions, which suggests the potential link between epigenetics and splicing (Table 1) [47,61–63]. Several recent studies in honey bees and Arabidopsis using high-definition profiling of DNA methylation by single-molecule-resolution bisulfite sequencing found enriched methylation at exons compared with the flanking introns [63–65]. These studies suggest the possible role of DNA methylation in exon definition and thereby in the regulation of alternative pre-mRNA splicing. In support, DNA methylation is primarily confined to exons and correlates with inclusion or exclusion of exons during transcription in invertebrates . Similarly, in the honey bee, genome-wide studies showed a positive association between alternative exon inclusion and DNA methylation . It is evident from a genome-wide study that >20% of all alternative exons are affected by DNA methylation . The DNA methylation either acts as an enhancer or silencer of exon recognition in a position-dependent manner in pre-mRNA splicing . It was reported that DNA methylation-dependent alternative exons are longer in length comparable with constitutive exons and also possess stronger 3′-splice sites [67–70]. Recent studies have shown the mechanistic link between DNA methylation and the regulation of alternative splicing. In 2011, Shukla et al. analyzed the DNA methylation-mediated splicing of CD45 pre-mRNA. CTCF is a zinc finger DNA-binding protein, which binds to the exon 5 of CD45 gene and acts as a roadblock for the RNA Pol II elongation rate, resulting in the inclusion of exon 5. However, DNA methylation at the exon 5 region inhibits CTCF binding, which enables a fast elongation rate of RNA Pol II and thereby leads to skipping of exon 5 . On the other hand, DNA hydroxymethylation at the CD45 exon 5 region leads to inclusion of exon 5 . The above-described mechanism of alternative splicing modulated by DNA methylation supports a kinetic model which states that reduced RNA Pol II elongation may favor recognition of weak alternative exons by the spliceosomal complex and thereby inclusion of alternative exons, whereas faster RNA Pol II elongation does not allow efficient recognition of weak alternative exons, which leads to the exclusion of alternative exons . Alteration of DNA methylation at exonic regions affects the binding of methyl-sensitive DNA-binding proteins, such as CTCF and MeCP2, and alters the inclusion and exclusion level of exons. DNA methylation may affect the alternative splicing events not only by inhibiting the binding of DNA-binding proteins but also by altering nucleosome positioning [63,73,74]. DNA methylation and nucleosome positioning alter each other in a complementary and bidirectional manner . DNA methylation can determine nucleosome positioning and affects the transcriptional elongation rate . It is known that nucleosome positioning may act as a determinant of exon recognition and affects alternative splicing. The positioned nucleosome in the exonic region reduces the elongation rate of the RNA Pol II, which favors the interaction of RNA Pol II-associated splicing factors with the splice sites, thus mediating the alternative splicing events [44,50].
It has been widely accepted that chromatin state plays essential roles in regulating alternative splicing process. Certain histone modifications, including Histone H3 lysine 4 trimethylation (H3K36me3), Histone H3 lysine 27 dimethylation (H3K27me2) and Histone H3 lysine 27 trimethylation (H3K27me3), were specifically enriched at exons . H3K4me3 is often stated as a mark of transcriptionally active promoters and is closely located after the first exon at the 5′-splice site and overlaps with a CpG island in mammalian cells [78,79]. The trimethylated H3K4 mediates the association between chromatin and U2 snRNP [80,81]. The U2 spliceosomal RNA is a small nuclear RNA component of the spliceosome, and the association of U2 snRNP with the pre-mRNA branch site plays a fundamental role in splicing assembly since the branch site directly participates in chemical catalysis . The U2 snRNP binds to chromatin via the chromatin remodeling protein CHD1 that binds to H3K4me3. Inhibition of CHD1 and H3K4me3 levels reduced the U2 snRNP association and, thus, influences the pre-mRNA splicing efficiency. It signifies the role of cross-talk between histone marks and splicing regulators/effectors in the regulation of alternative splicing and supports the adaptor model wherein specific chromatin marks may be recognized by different adapter proteins that can interact with splicing regulators and thus regulate alternative splicing . In addition to histone methylation [61,83,84], histone acetyltransferase GCN5 also regulates the recruitment of U2 snRNP to the site of transcription and modulates the alternative splicing process . MRG15 is a chromatin-binding protein that contains a chromodomain that can recognize methylated H3K36  and can modulate the splicing process by the recruitment of splicing regulator PTB  and regulate FGFR2 splicing, which controls tumor growth and invasiveness in lung cancer . However, MRG15 knockdown in lung cancer cells influences alternative splicing of >180 genes independently of PTB . The above statement suggests that there exists an unknown mechanism that does not depend on PTB. Eaf3, an ortholog of MRG15 in yeast, interacts with H3K36me3 and plays an important role in restoring repressive chromatin structure after the passage of elongating RNA Pol II . The MRG15 participates in the regulation of chromatin acetylation  and transcription that was shown to modulate splicing outcome . These studies suggest that epigenetic regulation not only determines which parts of the genome are expressed but also decides the fate of splicing events in cells.
The role of epigenetic modulators on alternative splicing in cancer
Approximately 94% of human genes express more than one mRNA by alternative splicing, a process by which functionally diverse protein isoforms can be expressed according to different regulatory programs. It is not surprising that disruption of normal splicing patterns can cause human disease . It is now established that several human diseases are a direct consequence of aberrant splicing events . Diseases such as spinal muscular atrophy [92,93], retinitis pigmentosa [94,95], and Hutchinson–Gilford progeria syndrome  are a direct consequence of misregulated splicing events. The precise contribution of alternatively spliced transcripts to various disease progressions is likely to be cell type-specific [19,97]. In addition to the diseases mentioned above, the involvement of varied and complex accumulation of alternatively spliced transcripts to other diseases will need further study.
The transition from normal cell growth to neoplasia and then to malignancy is a multistep selection for the most aggressive cells. The changes in alternative splicing patterns of oncogenes are associated with neoplasia and metastasis. The alternative splicing events are implicated in almost all types of cancers including breast , lung , gastric , colorectal , head and neck , liver , lymphoma , and melanoma . Increasing evidence also highlighted that alternative splicing events are considered as one of the hallmarks for cancers . The alternative splicing might be used to identify and subclassify cancer to predict outcomes and evaluate treatment responsiveness [107,108]. So, it is necessary to identify molecular regulations of the alternative splicing event in cancer and other diseases for better diagnosis and therapy.
Earlier, it was believed that splicing factors are the sole determinant that can alter the alternative splicing events . However, there is a growing awareness that histone modifications and chromatin organization influence pre-mRNA splicing. The complex regulation of alternative splicing is mediated by transcription rate, histone post-translational modifications, DNA methylation, DNA-binding proteins, and transcription factors [34–35,109]. Epigenetic modifications and deregulated alternative splicing are individually known to play a major role in tumor progression. However, a definite link between the two is yet to be found. Very few studies have connected the dots between epigenetics and alternative splicing in cancer (Table 2). Recent genome-wide studies have identified that DNA methylation is observed more often within gene bodies than at promoters [110–113]. Notably, increased proportions of DNA methylation were observed in exonic regions compared with intronic and other intergenic regions. Increased exonic methylation frequency suggests the speculation for its pivotal role in alternative splicing regulation [63,114,115]. Maunakea et al. reported that intragenic DNA methylation modulates alternative splicing by recruiting methyl CpG-binding protein 2 (MeCP2) to promote exon recognition in colon cancer cells. The MeCP2 binds to an exonic region of methylated DNA to promote alternative splicing in colon cancer cells. The binding of MeCP2 to methylated exonic region facilitates the recruitment of histone deacetylases (HDACs) to maintain a low acetylation level at the alternatively spliced exon region compared with other exon regions for its efficient inclusion in the spliced mRNA transcripts . Histone acetylation is associated with euchromatin state and may promote RNA Pol II elongation , which has been proposed to affect the alternative splicing outcome [26,118]. HDACs recruitment to the exonic region influences the rate of local RNA Pol II elongation and supports the inclusion of the alternatively spliced exon. Also, MeCP2 binding on the methylated exonic region alone can cause RNA Pol II pause and facilitates the inclusion of an alternatively spliced exon. Lower RNA Pol II elongation rates may be necessary for selection of an alternative splice site, whereas a higher RNA Pol II elongation rate can facilitate exon-skipping events [119,120]. In addition, Saint-André et al.  identified that elevated levels of H3K9me3 favor the inclusion of variant exons of the CD44 gene in cervical cancer. The elevated levels of H3K9 methylation at CD44 variant exons provide a binding site for the chromodomain of HP1γ protein. HP1 is a family of nonhistone proteins that contain the methyl-lysine-binding chromodomain that recognizes methylated H3K9 , are responsible for the establishment and maintenance of heterochromatin, and associate with other nonhistone proteins . HP1γ interacts with chromatin via its chromodomain and facilitates inclusion of alternative exons via a mechanism involving a decreased RNA Pol II elongation rate . Although the induced heterochromatin formation close to the alternative exons reduces the processivity of RNA Pol II, which causes the increased inclusion of alternative exons, it is still possible that methylation of H3K9 influences splicing of certain genes directly via tethering of splicing regulators (hnRNP A1, hnRNP L, hnRNP K, hnRNP A2/B1, and hnRNP A/B) to the alternative exon region of pre-mRNAs . Iannone et al. identified that dynamics of nucleosome occupancy and RNA Pol II affect alternative splicing in breast cancer. Exon inclusion upon progesterone stimulation in breast cancers is associated with higher nucleosome occupancy at the alternatively spliced exonic region . On the other hand, exon exclusion was associated with low levels of nucleosome density in the exonic regions. Nucleosome positioning can influence splice site recognition possibly by altering the transcription elongation rates. A stably positioned nucleosome at the exonic region may cause the RNA Pol II pause and, provides the opportunity for the other splicing factors to recognize the splice site, and facilitates the inclusion of the exon . It suggests that changes in nucleosome positioning may affect the transcription elongation rate, which might lead to aberrant alternative splicing in cancer. Alló et al.  reported that siRNA targeting a human fibronectin 1 (FN1) gene region extra domain I located near an alternative exon creates a compact chromatin structure and thus prevents efficient RNA Pol II elongation. Delay in elongation opens a window for the splicing machinery to recognize the alternative exon and, in turn, affects the splicing process . A recent study in gastric cancer shows the correlation between increased H3K36me3, reduced acetylation, and exclusion of exon 8 from CDH1 transcripts. . In another study, Salton et al. revealed the role of HP1γ in alternative splicing events. Alternative splicing of the VEGFA gene involves an adaptor system consisting of the chromatin modulator HP1γ, which binds preferentially to the methylated H3K9 and recruits splicing regulator SRSF1 to promote the alternative splicing process . The recruitment of SRSF1 to the methylated exonic region influences the exclusion of an alternatively spliced exon of the VEGFA gene. In 2014, Khan et al.  demonstrated that the nucleosome density levels on the alternatively spliced exon of the MCL1 gene are highly dynamic in relation to histone acetylation, which greatly affects the alternative splicing process. The nucleosomes marked by H3K4 methylation are preferentially engaged in acetylation levels by recruiting lysine acetyltransferase (KATs) and HDACs to the alternatively spliced exonic region in a position-dependent manner. The SRSF1 preferentially binds to H3K4-methylated exonic regions and further recruits HDACs to the alternatively spliced exonic region. The recruitment of KATs and HDACs to the exonic region alters the acetylation levels dynamically. Alterations of acetylation levels and recruitment of SRSF1 by HDACs to the exon 2 region influence the alternative splicing of MCL1 gene . In addition, acetylation levels at the promoter region of E-cadherin lead to the exon 11 skipping. HDAC inhibitors increased the acetylation of histones H3 and H4 in the E-cadherin promoter region and lead to the inclusion of exon 11 . In 2015, Nakka et al. reported that deacetylation of SAM68 by SMAR1 and HDAC6 inhibits the inclusion of CD44 alternative exons in breast cancer cell line MCF7. The SMAR1 is a scaffold/matrix attachment region-binding protein 1, which in co-operation with HDAC6 mediates the deacetylation status of splicing factor SAM68 and inhibits CD44 variant exon inclusion. Moreover, activation of ERK-MAPK pathway or SMAR1 knockdown enhances SAM68 acetylation, leading to inclusion of CD44 variant exons, and thus promote the invasion and metastasis in breast tumor cells . These independent studies cumulatively reveal the epigenetic control of alternative splicing in tumorigenesis. Presumably, DNA methylation, histone modifications, and nucleosome alterations influence the dynamics of spliceosome assembly and can affect splicing with or without altering the transcriptional elongation, indicating a direct role of epigenetic modifications in the modulation of alternative splicing in cancer (Figure 1). Delineation of the complex interactions between epigenetics and alternative splicing in cancers will open doors for developing better therapeutic strategies involving modulation of reversible epigenetic marks to treat human cancers that are a consequence of aberrant splicing.
Conclusion and future perspectives
Key progress is being made in understanding the complex relationship between epigenetics and alternative splicing events in various model systems. Discovery of the interplay between epigenetics and alternative splicing has greatly expanded our understanding of eukaryotic gene regulations. However, in disease models interplay between these two is less understood. Uncovering the complex relationship between epigenetic modifications and alternative splicing events in each disease type may provide new insights for therapeutic targeting strategies. Delineating the epigenetic fingerprints in each cell or tissue type and correlating those with alternative splicing events during disease progression may open the doors for treatment opportunities. Future research work on epigenetics and alternative splicing should focus on understanding how aberrant epigenetic marks cause widespread dysregulation of alternative splicing in cancer cells and other disease models. Further studies in this direction will open doors to develop better therapeutic strategies involving modulation of reversible epigenetic marks to alter splicing defects in human diseases.
The Authors declare that there are no competing interests associated with this manuscript.
We apologize to our colleagues whom we could not cite due to space constraints. We thank Mr Amit Gupta, Ms Sandhya Yadav, Ms Neha Ahuja, and Mr Dhananjay Kumar for their critical reading of the review. We thank IISER Bhopal for Intramural research funds. We thank the Department of Science and Technology (SERB) for a fellowship to Dr Sathiya Pandi and the Ministry of Human Resource and Development for a fellowship to Ms Smriti Singh. This work is supported by the Science and Engineering Research Board (SERB) [grant no. EMR/2014/000716].
Abbreviations: CTCF, CCCTC-binding factor; CTD, C-terminal domain; ES, embryonic stem; FGFR2, fibroblast growth factor receptor 2; FN1, fibronectin 1; H3K27me2, histone H3 lysine 27 dimethylation; H3K27me3, histone H3 lysine 27 trimethylation; H3K36me3, histone H3 lysine 4 trimethylation; HDACs, histone deacetylases; hMSC, human mesenchymal stem cells; hnRNP, heterogeneous nuclear ribonucleoproteins; HP1, heterochromatin protein 1; MeCP2, methyl CpG-binding protein 2; N2A, neuro 2A; RNA Pol II, RNA polymerase II; SMAR1, scaffold/matrix-associated region-binding protein 1; snRNPs, small nuclear ribonucleoproteins; SR, serine–arginine; SRSF1, serine/arginine-rich splicing factor 1; VEGF-A, vascular endothelial growth factor A.
- © 2017 The Author(s); published by Portland Press Limited on behalf of the Biochemical Society