The RNA polymerase II C-terminal domain

Abstract

Work over the last two decades has provided a wealth of data indicating that the RNA polymerase II transcriptional machinery can play an important role in facilitating the splicing of its transcripts. In particular, the C-terminal domain of the RNA polymerase II large subunit (CTD) is central in the coupling of transcription and splicing. While this has long been assumed to involve physical interactions between splicing factors and the CTD, few functional connections between the CTD and such factors have been established. We recently used a biochemical approach to identify a splicing factor that interacts directly with the CTD to activate splicing and, in doing so, may play a role in the process of spliceosome assembly.

Introduction

Splicing can occur faithfully in vitro in the absence of transcription. In contrast, in vivo, transcription by RNA polymerase II (RNAP II) is often required for efficient intron removal.^1,2 This observation implies that there is a functional connection between the transcription and splicing machineries, a phenomenon known as coupling. Why might coupling be necessary? Introns in mammalian cells can be extremely long, sometimes exceeding 100,000 nucleotides (nt). However, there is no correlation between the length of an intron and the time required for its excision.³ Introns spanning 1,500 to well over 100,000 nt were all found to be spliced within 5–10 min of the transcription of the downstream exon. In contrast, in vitro splicing rarely occurs efficiently on substrates longer than 500 nt, with a lag time of 20–30 min. Splicing/transcription coupling may ensure the accurate and swift pairing of splice sites of lengthy mammalian introns.

An appealing possibility encompassed by a coupling model is that exons are recognized co-transcriptionally and, through multiple interactions between RNA-bound splicing factors and the RNAP II elongation complex, the exon becomes “tethered” to the elongating polymerase, which then scans the nascent RNA for an appropriate acceptor site. Evidence for the tethering hypothesis was provided by Proudfoot and colleagues, who showed that the presence of a self-cleaving hammerhead ribozyme placed within an intron was insufficient to disrupt splicing in a transfected β-globin derived minigene.⁴ However, these conclusions have more recently been called into question by Bentley and colleagues, who showed that use of the faster-cleaving hepatitis delta ribozyme in the same position was in fact sufficient to disrupt splicing.⁵ To resolve the controversy, it will be important to test the hypothesis using ribozymes situated in longer introns in constructs stably integrated in the genome.

Additional observations also seem to support functional links between transcription and splicing. For example, considerable evidence has been provided showing that alternate promoter usage can result in changes in the pattern of alternative splicing.⁶ This phenomenon may in some cases be explained in ways that do not require physical interactions between the splicing and transcription machineries. For example, alterations in the elongation rate of RNAP II, which can be affected by promoter choice, can affect the inclusion of alternatively spliced exons.⁷ While some of the promoter effects on splicing may be explained through alterations in RNAP II kinetics, promoter-bound transcription factors can also affect splicing by directly altering the composition of the transcription elongation complex (TEC). For example, a strong transcriptional activator, GAL4-VP16, was shown to recruit splicing factors, such as PSF, to the promoter.⁸ These splicing factors then join the TEC, through interactions that will be discussed below, and enhance the efficiency of splicing of the transcribed pre-mRNA. In addition, some of the effects of promoters on pre-mRNA splicing are mediated by specialized proteins that serve as dual transcription/splicing factors. For example, the CAPERα/β proteins, both of which are homologous to the essential splicing factor U2AF65, appear to be capable of both trans-activating transcription and influencing splicing of the resulting transcript.⁹ Bolstering the idea that splicing factors can be an integral part of the TEC, in vitro evidence has been provided showing that factors involved in splicing associate with RNAP II and with the nascent pre-mRNA.¹⁰ Specific protein-protein interactions that contribute to the interaction of the splicing machinery with the transcriptional apparatus will be discussed below.

Spliceosome Assembly

Splicing is certainly the most complex of the major pre-mRNA processing reactions. The splicing reaction is carried out by a large complex known as the spliceosome, which consists of at least 150 protein components and five snRNAs.^11–13 The spliceosome is a highly dynamic entity that undergoes sequential compositional and conformational changes between the initial recognition of pre-mRNA splice sites and the eventual excision of the intron and ligation of the exons.

In vitro systems have been useful in elucidating the highly complex sequence of events that define the process of mammalian spliceosome assembly.^12,13 During in vitro splicing, a series of stable intermediate complexes form on the pre-mRNA substrate. These complexes are amenable to biochemical analysis, and thus provide “snapshots” of the assembly pathway. The spliceosomal complexes, termed E, A, B and C (in order of formation) can be separated by native gel electrophoresis, a technique that allowed the initial characterization of this process.¹⁴ More recently, affinity purification approaches coupled with proteomic analysis have provided significant insight into the compositional remodeling that occurs along the spliceosome assembly pathway.^15,16

Spliceosome assembly begins with the formation of the ATP-independent E complex, in which the 5′ splice site is recognized by U1 snRNP, while the 3′ splice site is recognized by the dimeric splicing factor U2AF.¹² The binding of U2AF and U1 snRNP often requires the assistance of SR proteins, a class of essential splicing factor characterized by an N-terminal RNA binding domain and a C-terminal arginine-serine rich domain.¹⁷ An important feature of the earliest stages in splice site recognition is crosstalk between splice sites flanking an individual exon. This phenomenon, known as exon definition, involves communication between factors that recognize the 5′ and 3′ splice sites, resulting in the cooperative recognition of the exon.¹⁸

Formation of the next intermediate in spliceosome assembly, the A complex, requires ATP hydrolysis and results in the stable recruitment of additional factors to the pre-mRNA, most notably the U2 snRNP. Proteomic analysis has shown that the Prp19 complex (PRP19C), which plays an important role in spliceosome activation,¹⁹ is also associated with the spliceosome in A complex.¹⁶

Following A complex formation, the U4/U6 and U5 snRNPs become associated with the spliceosome as part of a pre-assembled unit known as the tri-snRNP, forming the B complex. Additional rearrangements, notably ejection of two snRNA components, U1 and U4, precede the formation of the activated B complex (B*). This is followed by the execution of the first catalytic step, which generates C complex. The transition from the B to C complex involves the destabilization of the SF3a/b proteins, while PRP19C becomes more stably integrated in the catalytic core of the spliceosome.²⁰ Further rearrangements are necessary for the second catalytic step and, then, for spliceosome disassembly.¹²

The RNAP II CTD: A Critical Mediator of Transcription/Splicing Coupling

Our understanding of transcription/splicing coupling reached an important milestone when it was shown that truncation of the C-terminal domain of the RNAP II largest subunit (CTD) severely impaired splicing of a transcript produced from a transiently transfected minigene.²¹ Evidence was soon provided that this was a direct effect, when Hirose et al.²⁶ (1999) showed that exogenously added RNAP II was capable of increasing the kinetics of in vitro splicing, but only when the CTD was present and phosphorylated. This observation led to speculation that the phosphorylated CTD directly enhances spliceosome assembly by acting as a scaffold that binds multiple splicing factors (see below).

CTD Structure and Modification

The CTD in mammals is composed of 52 heptad repeats of the consensus sequence YSPTSPS. Heptads in the N-terminal half of the CTD mostly conform to the consensus sequence, while heptads in the C-terminal half are more degenerate. The CTD is subject to extensive phosphorylation, which has major impacts on its function. RNAP II recruited to promoters is hypophosphorylated. Upon transcriptional initiation, the CTD is phosphorylated by the cdk7 subunit of the GTF TFIIH at serine 5 of the heptad repeat (S5). S5 phosphorylation levels peak near the 5′ ends of genes, then decline as RNAP II moves towards the 3′ end.^22,23 The other major phosphorylation event occurs on serine 2 (S2), which in contrast to S5, increases towards the 3′ ends of genes.²² Activities of multiple S5 or S2 specific kinases and phosphatases determine the pattern of CTD phosphorylation along each gene.^24,25

CTD phosphorylation appears to play an important role in coupling splicing and transcription. As mentioned above, CTD phosphorylation has been shown to be essential for the stimulatory effect of the CTD on splicing in vitro.^26,27 And, more recently, the coupling of transcription and splicing through CTD phosphorylation has emerged as regulatory point in the control of gene expression. Medzhitov and colleagues described a set of inducible genes, which, when in the “off” state, are actively transcribed by RNAP II phosphorylated on S5 but not S2.²⁸ The transcripts that resulted from transcription by the S5-phosphorylated RNAP II, while full length, remained unspliced. When the gene is induced, P-TEFb, an S2-specific CTD kinase, was recruited to the gene, resulting in proper splicing of the transcribed RNA. In addition to its repercussions for gene regulation, this important finding strongly implicates S2 phosphorylation as essential to the integration of splicing and transcription.

Physical Links between the CTD and Splicing

A number of physical links between the phosphorylated CTD and the splicing apparatus have been established. The S. cerevisiae protein Prp40, which is associated with U1 snRNP, binds to phosphorylated CTD repeats through its WW domain.²⁹ Phospho-CTD binding appears to be a frequent function of WW domains, and also of the related FF domain. These domains are found on multiple splicing factors and, in addition to interactions with the CTD, also mediate interactions with other splicing factors. For example, the spliceosome-associated protein TCERG1 (formerly known as CA150) interacts with the phosphorylated CTD through FF domains,³⁰ while interacting with other splicing factors, such as SF1, through its WW domains.³¹ Based on these observations, an appealing hypothesis is that proteins such as TCERG1 provide a link between the CTD and factors that play a central role in spliceosome assembly, although direct evidence to support this idea is still lacking. In addition, the RNA binding proteins PSF and p54^nrb, also implicated in splicing, were shown to bind directly to the CTD regardless of its phosphorylation status.³² Binding of PSF to the CTD had in fact provided the only demonstration until recently of a functional significance for an interaction between a splicing factor and the CTD. Rosonina et al.⁸ (2005) showed that PSF, which can be recruited to promoters by the transcriptional activator GAL4-VP16, promotes enhanced splicing of transcripts from promoters containing GAL4-VP16 binding sites. This effect was abolished when transcription was carried out by RNAP II bearing a truncated CTD, suggesting that PSF is recruited to promoters and then travels along the gene with RNAP II through interactions with the CTD. Indeed, Kaneko et al.³³ (2007) provided evidence that p54^nrb/PSF is present along the length of a transcribed gene and functions in transcription termination.³³

Functional links between the CTD and alternative splicing have also been provided.³⁴ For example, CTD deletion results in increased inclusion of the fibronectin EDI exon, an effect that appears to stem from failure to efficiently recruit the SR protein SRSF3, which inhibits EDI inclusion, to nascent transcripts.³⁵ These results not only provide a functional connection between a splicing factor and the CTD, but also highlight the possibility that the CTD can participate in the physiological modulation of AS (see below).

A U2AF65-CTD Interaction Links the Splicing/Transcription Machineries

While several splicing-related proteins have been shown to bind to the CTD, demonstrated functional connections between the CTD and splicing factors remain few. In an effort to identify functional links between the CTD and splicing, we developed a biochemical complementation assay that allowed identification of an activity that stimulated in vitro splicing in a CTD-dependent manner.²⁷ The assay utilized a chimeric protein containing the SR protein SRSF1 fused to the CTD. Addition of this protein to splicing reactions containing an IgM-based model pre-mRNA substrate containing SRSF1 binding sites resulted in activation of splicing, but only in the presence of a fraction from nuclear extract. In contrast, SRSF1 itself was incapable of activating splicing under the same conditions, indicating that the splicing activity in the nuclear fraction was CTD-dependent. We recently reported purification of the CTD-dependent splicing activity, which unexpectedly revealed it to be a complex containing two well-known splicing factors, U2AF65 and Prp19C.³⁶ Further experiments showed that U2AF65 directly binds the CTD, an interaction that requires CTD phosphorylation. The RNA-bound CTD stimulated U2AF65-RNA binding to the suboptimal IgM polypyrimidine tract, likely accounting for the CTD-dependent splicing activation observed in the presence of the SRSF1-CTD fusion.

The fact that U2AF65 binds to the phosphorylated CTD, together with earlier observations that U1 snRNP associates tightly with RNAP II in vitro and in vivo,^10,37 indicates that the major factors that recognize the 5′ and 3′ splice sites at the earliest stage of spliceosome assembly can associate with the RNAP II transcriptional elongation machinery (Fig. 1). Given evidence that the earliest splice site recognition occurs at the level of exon definition (discussed above), a possible outcome of the U2AF/U1 snRNP association with RNAP II is enhanced efficiency of cross-exon interactions, bridged by the transcriptional apparatus. In fact, Berget and colleagues showed that GST-CTD added to in vitro splicing reactions was capable of activating splicing, but only in the presence of a substrate in which the downstream exon contains an intact 5′ splice site.³⁸ This result supports the idea that cross-exon interactions bridged by the CTD may play a role in the splicing stimulatory properties of the CTD. The U2AF65-CTD interaction may contribute to the cooperative recognition of exons in this way.

Figure 1.

A model for splicing-transcription coupling. (A) Promoter-bound RNAPII is hypo-phosphorylated on its CTD. (B) Upon transcriptional initiation, the CTD is phosphorylated by kinases contained in complexes such as TFIIH and P-TEFb, resulting in elongating polymerase hyperphosphorylated on the CTD. Through unknown interactions, this promotes the association of U1 snRNP and SR proteins with the CTD, as well as the direct binding of U2AF65. (C) Upon transcription of a 3′ splice site containing U2AF65 binding sequences, U2AF65 shifts from protein-protein interactions to protein-RNA interactions, promoting spliceosome assembly by interacting with multiple factors, including PRP19C.

While the above studies indicated a likely role for the CTD in recruiting U2AF65 to pre-mRNA, some in vivo evidence suggests that CTD phosphorylation may be insufficient to result in efficient U2AF65 recruitment to a transcription unit. Neugebauer and colleagues showed by chromatin immunoprecipitation (ChIP) that U2AF65 is recruited co-transcriptionally to intron-containing genes, but not to intron-lacking histone genes.³⁹ While this might be interpreted to mean that U2AF65 recruitment is intron-dependent, and thus necessitates RNA binding, another possibility is that CTD phosphorylation patterns differ between histone and other protein-coding genes. A more recent study examined more directly the dependence of splicing factor recruitment on the presence of a functional intron. Using immunofluorescence, Spiluttini et al.³⁷ showed that U2AF65 recruitment to a transcription unit was reduced if the introns contained mutations that abrogated splicing, in contrast to U1 snRNP, which was still robustly recruited. While this result suggests that the CTD is not sufficient for stable U2AF65 recruitment to transcribed genes, it does not address the possibility that it plays a necessary role in the process.

The finding that CTD phosphorylation is important for U2AF65 binding also raises the possibility that differential CTD phosphorylation might contribute to the regulation of alternative splicing (AS). The CTD is already well known to contribute to AS regulation.³³ As discussed above, changes in the CTD phosphorylation pattern can profoundly influence the splicing fate of the transcribed pre-mRNA.²⁸ It has also recently been shown that the extent of CTD phosphorylation on elongating RNAP II can be modulated in response to environmental factors, such as UV light, resulting in physiologically important changes in AS.⁴⁰ While the effect on AS described by Munoz et al. was shown to be a result of changes in RNAP II elongation rates, it is possible that increased recruitment of splicing factors, such as U2AF65, to the more highly phosphorylated CTD can influence AS decisions. In the case of U2AF65, this additional recruitment might favor the inclusion of exons with weak polypyrimidine tracts. Additional work will be necessary to uncover the importance of this potential mechanism in regulation of AS.

A U2AF65-PRP19C Interaction Connects the CTD with Later Stages in Spliceosome Assembly

The finding that U2AF65 interacts with PRP19C adds another degree of previously unrecognized interconnectedness between components of the transcription/splicing machineries. PRP19C becomes tightly associated with the spliceosome only at a later stage of assembly, where it plays a central role in mediating the conformational changes necessary for splicing catalysis.¹² The weaker interactions that result in the initial recruitment of PRP19C to the spliceosome have been previously unknown. The data of David et al.³⁶ indicates that one of the many functions of U2AF65 in spliceosome assembly is to facilitate recruitment of PRP19C, thus bridging RNAP II and splicing activation.³⁶

U2AF65 has already been shown to interact dynamically with many factors early in the splicing cycle, facilitating their association with the spliceosome. Interacting factors include SF1, U2 snRNP and the RNA helicase UAP56.¹² Our observation that U2AF65 interacts with a late-acting factor, PRP19C, raises many questions. Do the interactions between PRP19C and other factors occur simultaneously? The RS domain of U2AF65, implicated in the PRP19 interaction,³⁶ is also known to facilitate binding of U2 snRNP to the pre-mRNA.⁴¹ If the interactions are mutually exclusive, which comes first? The current paradigm for ordered spliceosome assembly would predict that the U2 snRNP interaction would be first, but the fact that a pre-formed U2AF65-PRP19C complex can be purified from nuclear extracts indicates that this is not necessarily the case. Future work will be necessary to understand whether the U2AF65-PRP19C interaction is required for PRP19C association with the spliceosome, or whether additional pathways for PRP19C recruitment exist.