Coupling of alternative splicing and alternative polyadenylation Authors: Xueying Zhang, Feiyan Liu, Yu Zhou Published: 2024 ABSTRACT AbstractRNA splicing and 3′-cleavage and polyadenylation (CPA) are essential processes for the maturation of RNA. There have been extensive independent studies of these regulated processing events, including alternative splicing (AS) and alternative polyadenylation (APA). However, growing evidence suggests potential crosstalk between splicing and 3′-end processing in regulating AS or APA. Here, we first provide a brief overview of the molecular machines involved in splicing and 3′-end processing events, and then review recent studies on the functions and mechanisms of the crosstalk between the two processes. On the one hand, 3′-end processing can affect splicing, as 3′-end processing factors and CPA-generated polyA tail promote the splicing of the last intron. Beyond that, 3′-end processing factors can also influence the splicing of internal and terminal exons. Those 3′-end processing factors can also interact with different RNA-binding proteins (RBPs) to exert their effects on AS. The length of 3′ untranslated region (3′ UTR) can affect the splicing of upstream exons. On the other hand, splicing and CPA may compete within introns in generating different products. Furthermore, splicing within the 3′ UTR is a significant factor contributing to 3′ UTR diversity. Splicing also influences 3′-end processing through the actions of certain splicing factors. Interestingly, some classical RBPs play dual roles in both splicing and 3′-end processing. Finally, we discuss how long-read sequencing technologies aid in understanding the coordination of AS-APA events and envision that these findings may potentially promote the development of new strategies for disease diagnosis and treatment.Keywords: alternative polyadenylation, alternative splicing, crosstalk, RNA processing, RNA regulation Introduction In eukaryotes, mRNA synthesis occurs within the cell nucleus, where nascent RNA undergoes a series of processing events to generate mature transcripts, including 5′ capping, splicing, and 3′-end processing. Each of these steps has alternative regulation, such as alternative splicing (AS) in skipping or including an exon and alternative polyadenylation (APA) in selectively choosing proximal or distal polyadenylation site (pPAS or dPAS), or even intronic polyadenylation site (iPAS). These regulatory mechanisms play crucial roles in shaping the final repertoire of mature transcripts, thereby contributing to the complexity of protein isoforms. Hence, both AS and APA contribute significantly to the mRNA diversification and differences in the proteome in many organisms [ 1, 2]. Alternative splicing leads to variations in transcripts through exon skipping, inclusion, intron retention, and etc., determined by the selection of alternative splicing sites [ 1, 3]. Meanwhile, the 3′-end formation in eukaryotes represents a pivotal step in mRNA maturation, involving the recognition of polyadenylation signals and subsequent cleavage and polyadenylation events [ 2, 4]. These two processing events, the majority of which are co-transcriptional, have been studied independently. Contemporary research has revealed an intricate interplay between splicing and 3′-end processing mechanisms [5]. Notably, the coordination of these events is pertinent to APA and AS. This review starts with a brief introduction of the molecular machineries and processes of RNA splicing and 3′-end processing and aims to provide a comprehensive review of the progress on the coupling mechanisms between APA and AS. Fundamentals of Splicing and 3′-End Processing Splicing is a highly regulated molecular process characterized by intricate interactions involving spliceosomes, cis-acting elements, and trans-acting proteins. The spliceosome, a dynamic ribonucleoprotein (RNP) complex, plays a central role in identifying core regulatory sequences within pre-mRNAs, including 5′ splice sites (5′SS), 3′ splice sites (3′SS), branch point sites (BPS), and polypyrimidine tracts [6]. Comprising five major small nuclear ribonucleoprotein particles (snRNPs)—U1, U2, U4/U6, and U5, the principal U2-type spliceosome—orchestrates the removal of approximately 99% of intron sequences [6]. Additionally, various protein factors interact with the human splicing machinery, with heterogeneous nuclear ribonucleoproteins (hnRNPs) and serine/arginine-rich (SR) proteins among the classical trans-acting proteins regulating splicing via splicing silencers and enhancers, respectively [ 7, 8]. The formation of the spliceosome involves sequential interactions among multiple components. Notably, the U1 snRNP recognizes the 5′SS, while the U2 auxiliary factor (U2AF) complex identifies the 3′SS and polypyrimidine tract, marking an intron. Splicing factor 1 (SF1) binds to the branch point sequence to form complex E. Subsequently, SF1 is displaced by U2 snRNA, which binds to the BPS, leading to the formation of complex A. Subsequent recruitment of the U4/U6·U5 tri-snRNP complex forms complex B, leading to a series of conformational changes culminating in catalytically active complex C formation. Finally, two-step catalytic reactions complete the splicing process [ 9– 13] . Cleavage-and-polyadenylation adds a poly(A) tail to mRNA, and this modification enhances mRNA translation, augmenting its functional capacity, and shields mRNA from degradation. The 3′-end processing machinery comprises four major subunits, including cleavage and polyadenylation specificity factor (CPSF), cleavage stimulation factor (CstF), cleavage factor I and II (CFIm and CFIIm), and other factors such as poly(A) polymerase (PAP), nuclear poly(A) binding protein (PABPN), and symplekin. The CFIm, CPSF, and CstF complexes bind to pre-mRNA via specific recognition motifs—upstream UGUA element, core AAUAAA motif, and downstream GU-rich sequences, respectively—acting synergistically to catalyze cleavage and polyadenylation reactions [ 14– 18]. Interestingly, a recent study has elucidated a novel APA mode named sequential polyadenylation (SPA) for many genes, where the distal polyadenylation site is processed first, anchoring the transcript in the chromatin/nuclear matrix. Subsequently, this transcript serves as a substrate for further 3′-end processing at the proximal polyadenylation site [19]. Recently, SPA has been reported to regulate RNA post-transcriptional m 6A modification in coupled with RNA nuclear retention [20]. Alternative splicing significantly contributes to the complexity of the transcriptome in multicellular eukaryotes, including seven major types classified by splicing pattern: exon skipping or cassette exon, alternative 5′ splice site (A5SS), alternative 3′ splice site (A3SS), mutually exclusive exon (MXE), intron retention (IR), alternative promoter or alternative first exon, and alternative last exon [21]. The diversity of the 3′ UTR can arise from various mechanisms, including genetic variations, RNA modifications, alternative cleavage and polyadenylation, and alternative splicing [22]. APA, a prevalent phenomenon in metazoan transcripts encoding proteins [23], primarily occurs within the 3′ UTR (UTR-APA), resulting in transcripts encoding the same protein but with varying UTR lengths [24]. UTR-APA impacts mRNA stability, translation, nuclear export, and localization [1]. When alternative polyadenylation sites reside within internal introns or exons, this process is termed coding region APA (CR-APA) [25]. CR-APA leads to the expression of distinct protein isoforms, which regulate gene expression through various mechanisms. Thus, APA can be categorized into four types: tandem 3′ UTR APA, intronic APA, alternative terminal exon APA, and internal exon APA [26]. The interactions between splicing and polyadenylation are inherently interconnected, as potential splicing sites and poly(A) sites mutually intertwine along the pre-mRNAs ( Figure 1). Next, we summarize recent progress on the coupling of the two processes in three sections: (1) Regulation and mechanisms of 3′-end processing on alternative splicing; (2) How splicing regulates 3′-end processing vice versa? (3) The dual roles of RNA binding proteins in regulating both splicing and 3′-end processing ( Figure 1). Finally, we will discuss future directions, including new techniques to quantify the coupling of the two events, especially the long-read sequencing technique, and the potential avenues of detecting and targeting these processes in disease diagnostics and therapies. Figure 1 Schematic diagram illustrating the coupling between splicing and 3′-end processing The sequence signals of splicing (5′ or 3′ splicing site, ss) and 3′-end processing (PASs) are interleaving in many pre-mRNAs. The two processes are mutually regulated through multiple mechanisms, including dual-role RNA binding proteins (RBPs). Regulation and Mechanisms of 3′-End Processing on Alternative Splicing Initial studies have revealed that the cleavage and polyadenylation process is not an isolated event but rather a tightly coordinated mechanism that occurs concurrently with splicing. Splicing of the terminal intron can promote CPA; mutations within the 3′ splice site of the final intron impede both splicing and polyadenylation [27]. Conversely, mutations weakening the “AAUAAA” motif have been shown to notably reduce the splicing efficacy of introns, particularly the terminal one [28] . CPA factors and poly(A) tail affect the splicing of the last intron The cleavage and polyadenylation process, a crucial step in eukaryotic mRNA maturation, is intricately regulated by various factors that influence splicing dynamics. Among these factors, CFIm25, a CFIm complex, plays a pivotal role in modulating polyadenylation site selection and, consequently, alternative splicing outcomes. CFIm25 facilitates the utilization of distal polyadenylation sites by binding to upstream of the distal polyadenylation site [ 29, 30], where its interaction with pre-mRNA promotes the assembly of other processing factors and impedes the usage of proximal polyadenylation signals [30]. Notably, in HeLa cells, loss of CFIm25’s activity induces the selection of the proximal polyadenylation site for the TIMP-2 gene, and interestingly, it also causes retention of the last intron [30] ( Figure 2A), although the mechanism of which remains elusive. These findings implicate the intricate interplay between polyadenylation site utilization and alternative splicing, particularly in the context of the 3′ UTR, highlighting the regulatory role of CPA factors in post-transcriptional mRNA processing [31] . Figure 2 Regulation and mechanisms of 3′-end processing on alternative splicing (A) CFIm knockdown results in the formation of a short 3′ UTR transcript for Timp2, accompanied by retention of the last intron. (B) PABPN1 facilitates the splicing of a last intron with a weak 3′ splice site by recruiting RBM26/RBM27. (C) CPSF binding to the non-functional AAUAAA motif in the intron inhibits splicing of the cryptic exon. (D) Splicing factors and 3′-end processing factors interact on the terminal exon. (E) 3′-End processing factors, together with RBFOX2, NOVA2, HNRNPA1, and other RBPs, mediate exon inclusion and exclusion. (F) Elav promotes the usage of the long 3′ UTR of Dscam1, in coupled with the skipping of upstream exon 19. Furthermore, both PABPN1 and poly(A) tail exert significant influence on the splicing of the terminal intron. PABPN1, a prominent player in RNA metabolism, serves as a major poly(A) binding protein within the nucleus, facilitating poly(A) tail synthesis by stimulating the activity of poly(A) polymerase [32]. Additionally, PABPN1 modulates alternative polyadenylation by inhibiting the recognition of proximal polyadenylation sites, as evidenced by studies demonstrating increased utilization of proximal polyadenylation signals upon PABPN1 deletion [33]. Notably, Li et al . [34] discovered that PABPN1 acts as a splicing enhancer for weaker terminal introns, particularly at the 3′ splice site. Within the nucleus, many pre-mRNAs are often polyadenylated prior to complete splicing of their introns. As the poly(A) tail lengthens, it can recruit PABPN1, which subsequently recruits RNA-binding motif protein 26 (RBM26) and RBM27, facilitating the recognition of the 3′ splice site by the U2 snRNP, thereby enhancing post-transcriptional splicing of the weaker intron ( Figure 2B). Moreover, the length of the poly(A) tail also influences terminal intron splicing. As reported [34], poly(A) tail length enhances the splicing of the terminal intron, with longer poly(A) tails exhibiting increased effectiveness within a certain range. This phenomenon may be attributed to the recruitment of more PABPN1 by longer poly(A) tails, consequently enhancing terminal intron splicing. However, the causal relationship between poly(A) tail length and splicing efficacy requires further investigation. CPA factors affect the splicing of internal and terminal exon Further investigations have revealed that the impact of tailing factors on splicing extends beyond the terminal intron, reaching to regulate the splicing of internal exons. One such example involves the critical splicing factor CPSF1, whose depletion results in enhanced inclusion of exon 6 in the interleukin 7 receptor (IL7R) gene [35]. Intriguingly, IL7R intron 6 has a weak 5′SS and harbors a consensus polyadenylation signal (AAUAAA) close to the 5′SS. The mutation of this signal similarly enhances exon 6 inclusion without concomitant cleavage and polyadenylation. These results suggested that CPSF1’s interaction with intronic IL7R pre-mRNA impedes spliceosome binding to the splice site of exon 6; however, whether this is a widespread mechanism merits further studies. This interference may have implications for individual T cell ontogeny and the pathogenesis of multiple sclerosis ( Figure 2C). Additionally, altered expression of CPSF1 facilitates the selection of the intronic cryptic exon 3 (CE) polyadenylation signal in the androgen receptor (AR) gene, leading to the recruitment of the CPSF complex. This event drives the synthesis of multiple AR variant mRNAs and impedes androgen-independent growth of castration-resistant prostate cancer (CRPC) cells [36]. In models of head and neck squamous cell carcinoma, aberrant CPSF1 expression disrupts alternative splicing events of cancer-related genes, including UBE2C and TGFBI, thereby increasing the likelihood of tumorigenesis [37]. Furthermore, in hepatocellular carcinoma (HCC), CPSF1 expression is markedly elevated and correlates with unfavorable survival outcomes. Knockdown of CPSF1 leads to increased alternative polyadenylation and alternative splicing events, shedding light on the mechanisms of post-transcriptional regulation in HCC. This suggests CPSF1 as a potential prognostic biomarker and therapeutic target for HCC [38]. The coupling of these two processing events requires protein-protein interactions between splicing factors and 3′-end processing factors on the terminal exon. Experimental evidence suggests that point mutations in polyadenylation signals can impede splicing of adjacent upstream introns [28]. Moreover, inhibition or depletion of key CPA factors such as CPSF has been shown to hinder splicing of upstream introns, possibly by disrupting the function of the U2 snRNP cofactor large subunit U2AF65 [ 39 – 41]. The interplay between the CPA apparatus and splicing extends further, with CPA factors enhancing splicing before polyadenylation site cleavage occurs [ 39, 42], which involves a series of steps. Initially, the splicing factor U2AF65 interacts with the CPA factor CFIm. Subsequently, in the remodeling step, CFIm is displaced from U2AF65 by poly(A) polymerase. Finally, U2 snRNP binds to CPSF, facilitating the cooperative interaction between splicing and CPA [43] ( Figure 2D). In vivo, the Pol II CTD also stimulates coupling between splicing and 3′-end processing [44], probably by loading or recruiting both splicing and 3′-end processing factors. 3′-End factors affect splicing through RNA binding proteins 3′-End processing factors exert unexpected influences on alternative splicing via different RBPs. One such RBP is RBFOX2, known for its role in promoting exon inclusion or exclusion. Interestingly, two mRNA 3′-end processing factors, CPSF and SYMPK, act as cofactors of RBFOX2. RBFOX2-regulated exons can be categorized into two groups: those where CPSF/SYMPK is crucial for selective splicing, and those where CPSF/SYMPK is neither required nor present in the vicinity. For CPSF/SYMPK-regulated exons, RBFOX2 facilitates exon inclusion by interacting with and recruiting CPSF/SYMPK to pre-mRNAs. Subsequently, CPSF/SYMPK promotes the recruitment of U1 70K and U1 snRNP to the 5′ splice site, thereby facilitating exon inclusion. Conversely, on exons where RBFOX2 promotes exclusion, the binding of RBFOX2 and CPSF/SYMPK interferes with the binding of U2AF at the upstream 3′ SS, leading to exclusion of the exon. Moreover, CPSF/SYMPK serves as a cofactor for other splicing regulators such as NOVA2 and HNRNPA1, which also modulate exon inclusion and exclusion similar to RBFOX2 [45] ( Figure 2E). This revelation sheds light on the intricate molecular mechanisms underlying the crosstalk between mRNA 3′-end formation factors and the spliceosome, offering valuable insights into AS regulation. 3′ UTR-mediated exon skipping The preceding examples elucidate the impact of 3′-end processing factors on splicing, offering insights into potential interaction mechanisms and coupling between alternative polyadenylation and alternative splicing. However, while these examples indicate that individual factors can influence AS processes, they provide only partial evidence of such interactions. Two hypothetical scenarios have been proposed to elucidate the relationship between APA events and upstream AS events. In the first scenario, APA and AS events are considered to be mechanistically uncoupled, operating relatively independently of each other. Conversely, the second scenario posits that APA and AS events are mechanistically coupled, suggesting that mRNA isoforms exhibiting a specific AS event may exhibit a selective preference for a particular APA type [31]. To test this hypothesis, the following examples may be utilized. Zhang et al. [46] utilized Pull-a-Long-Seq (PL-seq), a long-read sequencing method based on cDNA capture, to explore the intricate relationship between alternative splicing and alternative polyadenylation. Their study focused on Drosophila late embryos and heads, revealing cryptic exon splicing events in 23 genes associated with the 3′ UTR. Notably, the Dscam1 gene was identified to undergo APA, generating a short 3′ UTR isoform (Dscam1-S) and a long 3′ UTR isoform (Dscam1-L). Elav protein was found to bind to the proximal PolyA site of Dscam1 pre-mRNA, promoting the biogenesis of the long 3′ UTR isoform (Dscam1-L). Long-read sequencing data showed that Dscam1-L exhibits a preference for skipping upstream exon 19 while containing the long 3′ UTR, whereas Dscam1-S retains exon 19 and has a short 3′ UTR [47] ( Figure 2F). Intriguingly, the skipping of exon 19 in Dscam1-L leads to alterations in the Dscam1 amino acid sequence, resulting in an abnormal protein. Deletion of the Dscam1-L protein was found to impair the growth of Drosophila axons, indicating the critical role of ELAV-mediated exon skipping and variable polyadenylation of the Dscam1 gene in axon growth. Further investigations using minigene reporter experiments revealed that exon 19 and its flanking introns alone are insufficient to promote exon skipping. Elav-induced exon 19 skipping was observed only in the presence of the long 3′ UTR. However, the precise mechanism underlying specific 3′ UTR selection and the coupling of exon skipping remains unclear. It is possible that these 3′ UTRs confer distinct RNA stability through variations in the length of polyA tails or via microRNA and RBP targets within the 3′ UTR [48]. Moreover, differences in the expression of these RNA isoforms across vairous cell types and subcellular localizations may also contribute to their functional impact. Additionally, the long 3′ UTRs of the Dys and Khc-73 genes were identified to modulate cassette exon splicing in the short 3′ UTR [46]. These findings underscore the complex interplay between APA, AS, and mRNA processing factors in regulating gene expression and cellular functions. Elucidating these mechanisms further promises to enhance our comprehension of RNA biology and its impact on biological processes. Regulation and Mechanisms of Splicing on 3′-End Processing The regulation of splicing and 3′-end processing encompasses many intricate mechanisms, involving the competition of the two processes, splicing within 3′ UTR, and diverse splicing factors. This will be discussed in detail below. Splicing competes with intronic cleavage and polyadenylation For many genes, especially with long introns, where there are often intronic PASs, the two processes (splicing and polyadenylation) would compete with each other, and intronic PASs tend to be preferentially recognized when large introns have weak 5′ splice sites [49]. Maria Vlasenok et al. [50] defined a phenomenon termed spliced polyadenylated introns (SPIs). Through polyA-read analysis and 3′ end sequencing, SPI is observed to undergo both splicing and cleavage-and-polyadenylation. SPI represents an intermediate formed when splicing and CPA machinery operate simultaneously with transcription elongation. If CPA proceeds faster than splicing, IPA results in truncated transcripts. Conversely, if splicing outpaces CPA, introns containing PAS are excised, with PAS components degraded as part of the lariat. If PAS-mediated cleavage within the intron occurs after spliceosome assembly and is committed to splicing, the second catalytic step of splicing removes the lariat and all CPA products, forming SPI. The enrichment of PAS within introns and the presence of SPI collectively indicate competitive and cooperative interactions between splicing and CPA processes ( Figure 3A). Co-transcriptional pre-mRNA splicing may function to salvage eukaryotic transcripts from premature transcription termination. Figure 3 Regulation and mechanisms of splicing on 3′-end processing (A) Splicing competes with intronic cleavage and polyadenylation to form IPA, SPI or lariat. (B) Schematic depiction of how increased 3′ UTR splicing may promote carcinogenesis. (C) U1 snRNP serves as a crucial factor in safeguarding pre-mRNA from premature cleavage and polyadenylation, thereby ensuring the production of full-length mRNA transcripts. (D) SRSF7 and SRSF3 exert opposing effects on APA. SRSF7 promotes the utilization of the proximal polyadenylation site, leading to the formation of a short 3′ UTR. Conversely, SRSF3 facilitates the use of the distal polyadenylation site, promoting the generation of a long 3′ UTR. 3′ UTR splicing 3′ UTR diversity also encompasses internal 3′ UTR splicing. While splicing research traditionally emphasizes changes in protein-coding regions, the presence of introns within the 3′ UTR can lead to alternative splicing that alters 3′ UTR length. For transcripts of the heterogeneous nuclear ribonucleoprotein D (hnRNPD/AUF1) gene, splicing of exon 9 and intron 9 can generate four alternatively spliced AUF1 3′ UTR variants. This intronic splicing within the 3′ UTR regulates AUF1 expression, suggesting a significant role in gene expression regulation [51]. Another example is the Arc gene, where splicing of its 3′ UTR introns induces its own nonsense-mediated mRNA decay (NMD) and also facilitates mRNA translation efficiency, revealing a novel mechanism in neuronal post-transcriptional gene regulation that confines Arc expression to be transient and bursting [52]. Another noteworthy factor, Cstf64, has the potential to be an AS regulator. Knockdown of Cstf64 impacts splicing events within the 3′ UTR of hnRNP A2/B1, culminating in the reduced protein level of hnRNP A2/B1. Notably, the 3′ UTR of hnRNP A2/B1 harbors three potential alternative polyadenylation sites. Using the most distal APA site makes the splice sites in the last exon available and generates multiple unique spliced isoforms. Among these isoforms, hnRNP B1 isoform is susceptible to nonsense-mediated mRNA decay [53]. Consequently, knockdown of Cstf64 leads to an increase in the abundance of the unstable hnRNP B1 isoform, thereby contributing to a reduction in protein levels [31]. Recent pan-cancer analyses indicate that widespread upregulation of isoforms with splicing within 3′ UTR drives tumorigenesis. Notably, CTNNB1 3′ UTR splicing emerges as one of the most consistently dysregulated events across cancers. Experimental results demonstrate that CTNNB1 3′ splicing transcripts (3′SPs) are major translational isoforms, predominantly localized in the cytoplasm, whereas 3′ full-length transcripts (3′FLs) are primarily located in the nucleus. These findings indicate that an increase in nuclear CTNNB1 3′ UTR splicing events generates more 3′SP transcripts. Subsequent cytoplasmic translation of these transcripts leads to the overexpression of CTNNB1, potentially conferring oncogenic advantages to cancer cells [54] ( Figure 3B). Thus, 3′ UTR diversity not only enhances overall transcriptomic and proteomic diversity but also fuels cancer development and progression. Splicing factors affect APA The aforementioned findings underscore the bidirectional relationship between cleavage-and-polyadenylation and splicing processes, wherein CPA can influence splicing, and vice versa. Notably, certain splicing factors also exert regulatory effects on alternative polyadenylation. For instance, U1 snRNP, a pivotal component of the spliceosome machinery, plays a crucial role in safeguarding pre-mRNAs from premature cleavage and polyadenylation. Upon binding to nascent transcripts through complementary pairing with homologous sequences, U1 snRNP effectively inhibits cleavage and polyadenylation at cryptic polyadenylation sites, thereby facilitating the production of full-length mRNAs [55]. Conversely, a decline in the availability of splicing factors or disruption in splice site recognition can lead to enhanced intronic polyadenylation. Diminished levels of U1 snRNP result in the preferential utilization of proximal PASs over normal PASs, giving rise to truncated mRNA transcripts [56] ( Figure 3C). In addition to the mechanisms mentioned earlier, SNRNP70, a component of the spliceosomal U1 snRNP, can drive phase transitions of PABPN1 to reduce its inhibitory activity at proximal polyA sites. It can also recruit CPSF6 to promote the usage of proximal polyA sites [57]. The above phenomenon raises another question: why frequently used PASs are not inhibited by U1 snRNP? Shi et al. [58] found that actionable PASs are often located near regions that lack optimal U1 snRNP docking sites. In addition, the RNA-RNA base-pairings between short 5′-SS-like motifs (U(A/G)(A/G)G) and the 5′-end sequences of U1 snRNA are not strong enough to stably assemble U1 snRNP, which prevents its effect on these PASs. Deng et al. [59] discovered that U1 snRNP binding at the 5′ SS may promote the binding of U1A (U1 small nuclear ribonucleoprotein A) near the intronic PAS region, thereby interfering with the recruitment of the 3′-end processing factor CstF64 and inhibiting the cleavage of intronic PASs in pre-mRNA. In HeLa cells, inhibition of U4 snRNP leads to premature CPA events similar to those resulted from U1 snRNP inhibition [60]. Several other splicing factors have been identified to influence polyadenylation processes [61]. Among these, serine/arginine-rich splicing factors (SRSFs) are crucial in regulating selective splicing. Notably, SRSF3 has been implicated in promoting cellular senescence through alternative polyadenylation mechanisms. In a study by Shen et al. [62], transcriptome-wide APA analysis was performed using polyadenylation sequencing (PA-seq) and strand-specific sequencing methods. The analysis revealed a global shortening of the 3′ UTR in human and mouse cells following SRSF3 knockdown. Importantly, among the genes exhibiting 3′ UTR shortening, SRSF3 displayed a higher binding density at the proximal polyadenylation site than at the distal site. Furthermore, the effects of 3′ UTR shortening were recapitulated by the overexpression of three candidate genes ( PTEN, PIAS1, and DNMT3A), all of which have been linked to senescence-associated phenotypes. These findings provide compelling evidence supporting the notion that deletion of SRSF3 triggers cellular senescence through APA mechanisms. SRSF7 shares similarity with SRSF3 in its binding affinity upstream of the proximal polyadenylation signal. However, these two splicing factors exert opposing effects on 3′ UTR length regulation. The distinct protein structural domain of SRSF7 facilitates the recruitment of the cleavage factor FIP1, thereby promoting the utilization of the proximal polyA site and leading to the formation of a short 3′ UTR. Conversely, SRSF3 counteracts the action of SRSF7 by directly antagonizing its function. Specifically, SRSF3 favors the utilization of the distal polyA site, thereby facilitating the formation of a long 3′ UTR. Moreover, SRSF3 indirectly modulates the levels of active CFIm through selective splicing, orchestrating the assembly of non-productive cleavage complexes. This complex interplay ultimately enhances the utilization of the distal PAS while inhibiting the use of the proximal PAS, thereby promoting the generation of a long 3′ UTR [63] ( Figure 3D). RBPs Play Dual Roles in AS and APA There are RBPs that exhibit dual functionalities in both alternative splicing and alternative polyadenylation. For instance, the ELAV protein is known for its multifaceted roles in mRNA processing. In Drosophila, ELAV, specifically expressed in neurons, plays a pivotal role in inducing the expression of long 3′ UTR and facilitating the skipping of upstream exons. Within neuronal contexts, ELAV/Hu RBP binds to the downstream UUGUUUU motif of the polyadenylation signal, thereby exerting local repression on the cleavage and polyadenylation process. Subsequently, CPA cleaves transcripts at the downstream PAS, leading to the generation of neural-extended 3′ UTR isoforms [ 64, 65]. This intricate regulatory mechanism underscores the dynamic interplay between RNA-binding proteins and mRNA processing machinery, contributing to generating transcriptomic diversity essential for neuronal function and plasticity. Another illustrative example involves the Drosophila Srrm234 gene, which exhibits a distinctive neural-specific alternative last exon isoform harboring the eMIC domain. Notably, the RNA-binding protein ELAV plays a pivotal role in orchestrating the selection of this last exon with ELAV binding sites (UUUNUUU motifs) at the 3′ end. By repressing the proximal polyadenylation signal site, ELAV facilitates the expression of the eMIC-containing isoform. Subsequently, this isoform plays a crucial role in directing the widespread inclusion of neural microexons [66] ( Figure 4A). This intricate interplay between alternative polyadenylation and splicing underscores their collaborative roles in executing essential functions within the cellular transcriptomic landscape. Such coordination between APA and splicing mechanisms highlights the dynamic regulatory networks that govern mRNA processing, ultimately contributing to cellular diversity and functionality. Figure 4 Dual-role RBPs in AS and APA (A) ELAV facilitates the expression of the eMIC-containing Drosophila Srrm234 isoform, which mediates microexon inclusion. (B) The MBNL protein exhibits a dual regulatory role in AS and APA. When the binding site of MBNL protein overlaps with the polyadenylation site, it impedes the recruitment of 3′-end processing factors, thereby influencing APA. Conversely, when MBNL protein binds to upstream of the polyadenylation site, it activates 3′-end processing at the downstream site, thereby modulating APA. (C) The splicing factors HNRNPC, PTBP1, and PCBP1 also modulate cleavage and polyadenylation. The loss of MBNL protein function is the underlying cause of myotonic dystrophy (DM), a genetic disorder characterized by muscle weakness and other symptoms. Throughout development, the MBNL protein family exerts significant influence over alternative splicing. Specifically, MBNL1 and MBNL2 modulate AS in muscle and brain development by binding to nascent transcripts. MBNL1 preferentially recognizes the YGCY motifs and promotes exon inclusion when binding to motifs downstream of the 5′SS, whereas it facilitates exon skipping when binding to motifs upstream of the 3′SS [67]. Additionally, an important binding site for the MBNL protein family resides within the 3′ UTR of target RNA. Research conducted by Batra et al. [68] shed light on the widespread dysregulation of alternative polyadenylation resulted from loss of MBNL function in cellular and animal models. This dysregulation manifests as a shift in APA patterns in adult tissues towards those typically observed in the fetal stage. Through techniques such as HITS-CLIP and minigene reporter analysis, it was demonstrated that MBNL proteins directly regulate APA by binding to specific sites. When the MBNL binding site overlaps with the polyadenylation site, it inhibits the recruitment of 3′-end processing factors. Conversely, when the MBNL protein predominantly binds with the upstream site, it activates 3′-end processing at the downstream site. Furthermore, aberrant activation of normally expressed polyadenylation sites during embryonic development and silencing of adult polyadenylation sites are observed upon inhibition of MBNL protein activity by CUG exp RNA. Hence, it is evident that MBNL proteins exert dual regulatory control over both AS and 3′-end processing of target gene transcripts ( Figure 4B). This intricate interplay underscores the multifaceted roles of MBNL proteins in post-transcriptional regulation, highlighting their significance in cellular homeostasis and disease pathology. The polypyrimidine (Py) tract-binding proteins (PTBPs) combines with Py-rich motif at the 3′ splice site to inhibit the splicing of target exons [ 69, 70]. In glioblastoma tumors, PTBP1 has been observed to bind to approximately 75-nucleotides downstream of the poly(A) site, where it antagonizes CSTF2 and inhibits the usage of the PAS [71]. hnRNPC can compete with U2AF65 to suppress the exonization of Alu elements to protect transcriptome integrity [ 72, 73]. When HnRNPC binds to U-rich elements near the poly(A) site, it inhibits the usage of the poly(A) site [74]. The poly(C)-binding proteins (PCBPs) facilitate both splicing [75] and polyadenylation [76] ( Figure 4C). It is anticipated that more dual-role RBPs will be revealed in future studies. Perspectives It is known that traditional RNA-seq is not well-suited for the precise identification of poly(A) sites. For that purpose, many library preparation methods that enrich fragments carrying poly(A) tails have been developed, which can be categorized into two types based on their strategies [77]. The first category is oligo(dT) priming-based, where mRNA molecules are reverse transcribed into cDNA using oligo(dT) primers, and encompasses techniques such as PA-seq [78], 3′-Seq [79], PAS-seq [80], 3PC [81], EXPRSS [82], MAPS [83], SAPAS [84], and WTTS-seq [85]. The second category is RNA manipulation-based, where adapters are added to the 3′-ends of RNA fragments, followed by reverse transcription using primers annealed to the adapters, including 3P-seq [86], 3′ READS [87], and PAT-seq [88]. It is a general strategy to integrate those high-resolution 3′-end sequencing technologies with RNA-seq to investigate the regulation of RNA processing. Additionally, third-generation sequencing technologies offer more powerful approaches for studying various RNA processing steps, especially the coupled regulation of different events. The advancement of third-generation sequencing technologies has propelled long-read RNA sequencing (LRS) into a potent tool for investigating RNA biology. The leading long-read sequencing platforms include Pacific Biosciences (PacBio) REVIO system, Oxford Nanopore devices, and BGI CycloneSEQ released recently. Unlike short-read high-throughput sequencing methods, which typically generate reads shorter than 150 or 300 nucleotides, long-read sequencing can capture transcripts spanning tens of kilobytes without fragmenting RNA molecules, thus preserving transcript integrity [89]. Consequently, long-read RNA sequencing facilitates a comprehensive exploration of different mRNA processing events simultaneously at the single-transcript level. It enables the investigation of the interplay between various RNA processing phenomena, thereby enhancing our understanding of underlying biological processes [90]. Zhang et al. [46] introduced a targeted Nanopore long-read sequencing strategy named Pull-a-Long-Seq. Through PL-Seq, they identified 23 genes exhibiting AS linked to 3′ UTR variations in neuron-enriched tissues and quantified the regulatory role of ELAV in coordinating AS and APA. PL-seq (Pull-a-Long Seq) is a probe-based cDNA pulldown strategy, in which SMARTer cDNA synthesis is first performed using oligo(dT) primers, followed by cDNA PCR. Synthesized and biotinylated probes are then used to capture the target genes, followed by PCR amplification and Nanopore long-read sequencing. The targeted long-read single-molecule sequencing enables studying the coordination between alternative splicing and alternative polyadenylation in single transcripts with sufficient depth, different from the traditional NGS-based RNA-seq good at detecting local AS or APA events. PL-Seq offers a cost-effective approach and can be broadly applied to quantify linked AS-APA events across diverse organisms, tissues, and cell types. Furthermore, PL-seq holds promise for single-cell analysis, complementing single-cell RNA-seq methodologies in deciphering the interplay of co-transcriptional events within complex tissues. In addition, the removal of introns occurs predominantly in a defined sequence order during splicing. However, current global analyses of intron removal order are limited to adjacent pairs of introns. Multi-intron splicing order exceeding two introns is rare and has been achieved only in a few individual genes through RT-PCR or targeted next-generation sequencing. Therefore, long-read transcriptome sequencing represents a promising approach to studying the order of RNA processing events within single RNA molecules [91]. Dysregulation of splicing and 3′-end processing is implicated in the mechanisms of numerous diseases. For instance, alternative splicing of the 3′ UTR may be a widely exploited mechanism by cancer cells to produce transcripts resistant to nonsense-mediated decay and splicing-mediated decay, thereby promoting oncogene expression and tumorigenesis [54]. Therefore, new mechanisms uncovered through long-read transcriptome sequencing may elucidate the etiology of various diseases and play significant roles in both physiological and pathological contexts. Furthermore, discovering mechanisms coupling splicing and 3′-end processing could provide novel insights for developing new drug targets. It is possible that diseases caused by splicing dysregulation might be addressed through targeting 3′-end processing, or conversely. Thus, detecting and targeting these processes might represent new avenues for more precise diagnostics and therapies. Acknowledgments We thank the members of the Zhou laboratory for their helpful comments. COMPETING INTERESTS The authors declare that they have no conflict of interest. Funding Statement This work was supported by the grants from the National Natural Science Foundation of China (No. 82341023), the National Key R&D Program of China (No. 2023YFC2307802), and the Fundamental Research Funds for the Central Universities (No. 2042022dx0003) to Y.Z. References