Review article

The SR protein family of splicing factors: master regulators of gene expression

Jennifer C. Long, Javier F. Caceres


The SR protein family comprises a number of phylogenetically conserved and structurally related proteins with a characteristic domain rich in arginine and serine residues, known as the RS domain. They play significant roles in constitutive pre-mRNA splicing and are also important regulators of alternative splicing. In addition they participate in post-splicing activities, such as mRNA nuclear export, nonsense-mediated mRNA decay and mRNA translation. These wide-ranging roles of SR proteins highlight their importance as pivotal regulators of mRNA metabolism, and if these functions are disrupted, developmental defects or disease may result. Furthermore, animal models have shown a highly specific, non-redundant role for individual SR proteins in the regulation of developmental processes. Here, we will review the current literature to demonstrate how SR proteins are emerging as one of the master regulators of gene expression.

  • alternative splicing
  • pre-mRNA splicing
  • RS domain
  • SR proteins
  • SR-related proteins
  • translation regulation


Pre-mRNA splicing was discovered in the late 1970s when it was demonstrated that eukaryotic genes contained intervening sequences, or introns, that were not present in the mature mRNA [1,2]. Subsequent studies showed that introns are removed by a macromolecular complex, termed the spliceosome, which consists of five snRNPs [small nuclear RNPs (ribonucleoprotein particles)], U1, U2, U4, U5 and U6, and a large number of protein components (reviewed in [3,4]). Spliceosomal assembly is initiated by the recognition of the 5′ and 3′ ss (splice sites) by the U1 snRNP and the heterodimeric U2AF (U2 snRNP auxiliary factor) respectively, forming the E complex. Recruitment of the U2 snRNP to the BP (branch-point), in an ATP-dependent manner, results in the formation of the A complex. Subsequent recruitment of the U4/U6·U5 tri-snRNP forms the B complex, which is followed by a series of structural rearrangements leading to the formation of the catalytically active spliceosomal C complex (reviewed in [5]). The spliceosome is a dynamic structure and more than 300 proteins have been identified in active splicing complexes [68] (reviewed in [9]).

The aim of this article is to review the contribution of SR protein family members to pre-mRNA splicing, as well as reviewing more recent studies expanding their role in post-splicing activities. Finally, we will discuss how misregulation of SR protein functions can lead to human disease.


The SR proteins were first discovered as splicing factors in the early 1990s (reviewed in [1012]). A protein domain rich in arginine and serine dipeptides, termed the RS domain, was originally observed in three Drosophila splicing regulators, SWAP (suppressor-of-white-apricot) [13], Tra (transformer) [14] and Tra-2 (transformer-2) [15]. Subsequent identification of SF2/ASF (splicing factor 2/alternative splicing factor) [16,17] and SC35 (spliceosomal component 35) [18] revealed that these proteins contained an RS domain, which is also present in the U1 snRNP-associated protein, U1-70K [19,20].

SF2/ASF was the first SR protein to be identified as an activity required to complement an otherwise splicing-deficient HeLa (human cervical carcinoma cell) S100 extract [21] and was also purified from HEK (human embryonic kidney)-293 cells as a factor which could alter 5′ ss selection of an SV40 (simian virus 40) early pre-mRNA [22]. The term ‘SR protein’ was coined following identification of additional RS domain-containing proteins that were recognized by a monoclonal antibody, mAb 104, which binds to active sites of RNA polymerase II transcription [23]. These novel proteins, which were active in splicing complementation, included the SR proteins SRp20, SRp40, SRp55 and SRp75, named after their apparent molecular mass on an SDS/PAGE gel, and are conserved across vertebrates and invertebrates [24]. They have a modular structure containing one or two copies of an RRM (RNA recognition motif) at the N-terminus that provides RNA-binding specificity and a C-terminal RS domain that acts to promote protein–protein interactions that facilitate recruitment of the spliceosome [25,26]. The RS domain can also contact the pre-mRNA directly via the BP and the 5′ ss, suggesting an alternative way to promote spliceosome assembly [27,28] (reviewed in [29]). Furthermore, the RS domain acts as an NLS (nuclear localization signal), affecting the subcellular localization of SR proteins by mediating the interaction with the SR protein nuclear import receptor, transportin-SR [3032].

The prototypical SR protein, SF2/ASF, functions in constitutive splicing and also modulates alternative splicing [22,33]. Further studies demonstrated that other SR protein family members could also affect alternative splicing in vitro [34,35]. Thus, the criteria used to define ‘classical’ SR protein family members are (i) structural similarity, (ii) dual function in constitutive and alternative splicing, (iii) the presence of a phosphoepitope recognised by mAb104; and (iv) their purification using magnesium chloride (Table 1).

View this table:
Table 1 ‘Classical’ SR proteins

A genome-wide survey in metazoans identified a large number of RS domain-containing proteins with a role not only in splicing but also in other cellular processes such as chromatin remodelling, transcription and cell cycle progression [36]. These related proteins contain an RS domain but may lack a defined RRM, however a subset can bind RNA through other domains such as the PWI motif found in the splicing activator SRm160 [37,38] (Tables 2–4). These factors are collectively known as SR-like or SR-related proteins and include both subunits of the U2AF heterodimer, U1-70K and the splicing coactivators SRm 160/300, among others [39]. It was recently proposed that SR proteins should be redefined based on their common structural features and their role in pre-mRNA splicing [40]. Based on this, a ‘bona fide’ SR protein has to contain at least one RRM and an RS domain (irrespective of their positions within the protein) and to function in constitutive or alternative splicing, as assayed by either complementation of splicing-deficient S100 HeLa cytoplasmic extracts or in an alternative splicing assay respectively. The human homologues of the Drosophila splicing regulators Tra2α [41] and Tra2β [42] contain an RRM flanked by two RS domains, which is not the domain structure found in classical SR proteins (Figure 1). Since both proteins function as sequence-specific splicing activators [43], they could be classified as SR proteins (Table 2). Other proteins may contain an RS domain but also have other domains required for their enzymatic activities, as is the case for the RNA helicases HRH1 and hPRP16, that contain a DEAH box domain [44,45] (Table 4).

Figure 1 Schematic diagram of SR and SR-related proteins

The domain structures are depicted. DEAH Box, motif characteristic of RNA helicases; RS: arginine/serine-rich domain; PWI: an alternative RNA binding motif; Zn, zinc finger motif. With the exception of SRm160 and hPRP5, all proteins are drawn to scale.

View this table:
Table 2 Additional SR proteins
View this table:
Table 3 RNA-binding SR-related factors
View this table:
Table 4 Other RS-domain containing proteins


SR proteins are concentrated in nuclear speckles and are recruited from these sites to nascent sites of RNAP II (RNA polymerase II) transcription [46]. It is well documented that RNA splicing occurs co-transcriptionally [47,48]. Interactions between SR-related proteins and the CTD (C-terminal domain) of RNAP II have been reported [49], and members of the SR protein family were identified among the hundreds of proteins present in the RNAP II complex [50]. It was recently reported that SC35 promotes RNAP II elongation in a subset of genes, confirming the existence of coupling between transcription and splicing, and perhaps surprisingly, showing that this coupling can be bidirectional [51] (reviewed in [52]). In this study [51] it was demonstrated that SC35 interacts not only with the CTD but also with CDK9 (cyclin-dependent kinase 9), which is the kinase component of the transcriptional elongation factor P-TEFb (positive transcription elongation factor b), resulting in phosphorylation of Ser2 in the CTD and leading to transcriptional elongation. This activity of SR proteins in transcriptional elongation may be functionally related to their reported effect in the maintenance of genome stability. It has been shown that depletion of SF2/ASF, SC35 and the SR-related protein RNSP1 results in the formation of R-loops (RNA:DNA hybrid structures) leading to a hypermutation phenotype [5355].

The co-transcriptional nature of pre-mRNA splicing underlies a role for the transcriptional machinery in alternative splicing regulation [56]. A kinetic coupling model proposed that changes in the rate of transcriptional elongation affect the timing in which splice sites are presented to the splicing machinery, leading to differential splice site selection [57]. Furthermore, differential recruitment of splicing factors to the CTD of RNAP II may also influence this process [58] (reviewed in [59]).


Splice site consensus sequences are generally not sufficient to direct assembly of a functional spliceosome, and auxiliary elements known as ESEs and ISEs (exonic and intronic splicing enhancers respectively) and ESSs and ISSs (exonic and intronic silencers respectively) are involved in both constitutive and, to perhaps a greater extent, alternative splicing. Binding of SR proteins to ESEs acts as a barrier that prevents exon skipping, thus ensuring the correct 5′ to 3′ linear order of exons in spliced mRNA [60]. Two main models have been proposed to explain the mechanism by which SR proteins regulate exon inclusion. The ‘recruitment model’ focuses on the ability of ESE-bound SR proteins to recruit and stabilize interactions between the U1 snRNP at the 5′ ss and U2AF65 at the 3′ ss [61], in a process known as exon definition [62] (Figure 2A). The hnRNP (heterogeneous nuclear RNP) family comprises a structurally diverse group of RNA-binding proteins with roles in many aspects of RNA biogenesis, including pre-mRNA splicing (reviewed in [63]). In the ‘inhibitor model’, ESE-bound SR proteins may act by antagonizing the negative activity of hnRNP proteins recognizing ESSs [64] (Figure 2B). The SR proteins may also form a network of protein–protein interactions across introns to juxtapose the 5′ and 3′ ss early in spliceosomal assembly, as shown by the reported interactions of SF2/ASF and SC35 with U1-70K at the 5′ ss and with U2AF35 at the 3′ ss in an RS domain-dependent manner [25] (Figure 2C). Additionally, the enhancer-bound RS domain of the SR protein SF2/ASF has been shown to interact directly with RNA at the BP to promote pre-spliceosomal assembly [27,28]. SR proteins may also facilitate the recruitment of the U4/U6·U5 tri-snRNP to the pre-spliceosome [65] via RS domain-mediated interactions with the SR-related proteins SRrp65 and SRrp110 [66]. The function of SF2/ASF in pre-mRNA splicing depends on the context of the pre-mRNA sequence to which it binds, as shown by the fact that SF2/ASF inhibits adenovirus IIIa pre-mRNA splicing when bound to an intronic repressor element [67]. The second RRM of SF2/ASF, and in particular a phylogenetically conserved heptapeptide, SWQDLKD, which is located in the first α-helix of this domain [68], is essential for the splice site selection activity of SF2/ASF [69,70]. The structure of this domain revealed an atypical RRM fold that binds to RNA in a novel manner [71].

Figure 2 Roles of SR proteins in splice site selection

(A) SR proteins bound to ESE elements recruit U2AF35 to an upstream 3′ ss and U1-70K to the downstream 5′ ss. (B) ESSs recruit hnRNP proteins which block 3′ ss selection by U2AF. SR proteins bound to ESEs can antagonize the action of these splicing repressors, thereby promoting splice site selection. (C) SR proteins can facilitate intron bridging interactions by binding, via the RS domain, to U1-70K and U2AF 35, thereby juxtaposing the 5′ and 3′ ss.

Use of FRAP (fluorescence recovery after photobleaching) approaches revealed a high mobility for SF2/ASF within the nucleus, with kinetics compatible with a diffusion mechanism [72,73]. Advances in imaging have allowed analysis of splicing factors both in speckles and at other sites in the nucleoplasm by FRET (fluorescence resonance energy transfer) [74,75]. A recent study provided a map of SR protein splicing complexes in the nucleus, and showed that they act in exon and intron definition in vivo [76].

The U12-type class of pre-mRNA introns, also known as AT-AC introns, are spliced by the less abundant U12-dependent (minor) spliceosome. The 5′ ss and BP sequences are highly conserved in AT-AC introns, unlike the degenerate sequences found in GT-AG introns (reviewed in [77]). The SR proteins have been shown to participate in AT-AC intron splicing where they promote binding of the U11 and U12 snRNPs to the 5′ ss and BP respectively [78]. There is also evidence that SR proteins contact the pre-mRNA of U12-type introns directly via their RS domain, again in an analogous fashion to that seen in conventional splicing [79].

A delicate interplay of cis-acting sequences and trans-acting factors modulate the splicing of regulated exons in a combinatorial fashion [80]. SR family proteins antagonize the activity of hnRNP A/B proteins in splice site selection, with an excess of hnRNP A1 favouring distal 5′ ss, whereas SF2/ASF promotes the use of proximal 5′ ss [8185]. Thus, the ratio of hnRNP A1 to SR proteins in the nucleus is of great importance in alternative splicing regulation and may have a crucial role in the tissue-specific and developmental control of regulated splicing. Accordingly, the protein levels of SF2/ASF and hnRNP A1 have been found to vary naturally over a very wide range in rat tissues and also in immortal and transformed cell lines [35,86]. SF2/ASF and hnRNP A1 have also been found to have an antagonistic role in the regulation of the neuronal-specific N1 exon of the c-src gene [87]. Antagonism between hnRNP proteins and SR proteins has also been shown to regulate a highly complex pattern of mutually exclusive exons in the Dscam (Down's syndrome cell adhesion molecule) gene in Drosophila [88]. A subset of SR proteins has been shown to activate alternative splicing of the cTNT (cardiac troponin T) exon 5 by directly interacting with a purine-rich ESE. Thus regulation of the levels of individual SR proteins may contribute to the developmental regulation of alternative splicing in cTNT [89]. Interestingly, individual SR proteins can sometimes have antagonistic effects on splice site selection, as is the case with SRp20 and SF2/ASF in the regulation of SRp20 pre-mRNA alternative splicing [90] and of SF2/ASF and SC35 in the regulation of β-tropomyosin [91] and human growth hormone pre-mRNA alternative splicing [92]. Other, non-classical, SR proteins, including p54, SRp38 and SRp86, function solely as negative regulators of alternative splicing, antagonizing classical SR proteins and promoting exon skipping [9395].


Several approaches have been taken to identify physiological RNA targets of SR proteins. One such approach, termed SELEX (selected evolution of ligands through exponential enrichment) involves the selection of high-affinity binding sites from randomized pools of RNA sequences [96]. This has resulted in the identification of high-affinity binding sites for SF2/ASF and SC35 [97], SRp40 [98], 9G8 and SRp20 [99] (Table 5). These binding sites consist of purine-rich sequences that resemble 5′ ss or exonic sequences, known to function as splicing enhancers. SELEX can be used in conjunction with large-scale bioinformatic screens to identify further potential binding sites. An alternative to the SELEX approach was the development of a functional SELEX strategy, which involves selection for a sequence that will promote splicing, rather than binding alone [100]. This can be modified further by performing the splicing reactions in a S100 extract with the addition of a single SR protein [101,102]. The motifs identified for the SR proteins SF2/ASF, SC35, SRp40 and SRp55 using functional SELEX were more redundant than those found by conventional SELEX, suggesting that the specificity of binding in vivo depends on factors other than just sequence recognition (Table 5). These motifs have been integrated into a web-based program, known as ESE finder, where input sequences can be scanned for potential ESEs responsive to the above proteins [103].

View this table:
Table 5 RNA sequences identified as SR protein binding sites

N: any nucleotide; R: purine; Y: pyrimidine; D: A, G or U; K: U or G; M: A or C; S: G or C; W: A or U.

A number of computational approaches have also been used to define splicing sequence motifs that regulate exon inclusion. RESCUE (relative enhancer and silencer classification by unanimous enrichment)-ESE predicts sequences which could function as ESEs by statistical analysis of exon–intron and splice site composition [104]. This is based on the observation that ESEs function in a highly position-dependent fashion and are present in constitutively spliced exons and absent in introns. This approach identified 238 candidate ESEs that occurred more frequently in exons with weak splice sites than in exons with strong splice sites. By sequence similarity, these were condensed to ten RESCUE-ESE motif clusters and were shown to have enhancer activity in vivo. Another computational study found 2000 RNA octamers, identified as PESEs (putative ESEs), that were found more frequently in exons than in pseudo-exons or intronless genes [105]. Validation of a subset of these PESEs resulted in 82% exhibiting decreased splicing efficiency when the PESE was mutated [106].

ChIP (chromatin immunoprecipitation) can be used to study nascent RNA–protein interactions, but a variation of this technique named RIP (RNP immuno-precipitation) provides more information on protein–RNA interactions in vivo [107]. RIP involves cross-linking the protein–RNA interactions using formaldehyde, followed by immuno-precipitation of the protein–RNA complexes. After reversing the cross-links, the RNA can be amplified by RT–PCR (reverse transcription–PCR). This technique relies on random hexamer primers to identify unknown RNAs or can be used in conjunction with microarray technology. Reassociation of RNA-binding proteins after cell lysis can complicate the analysis of these results, since observed protein–RNA interactions may not necessarily reflect true in vivo interactions [108]. An adaptation of the SELEX method described above, known as genomic SELEX, uses real genomic sequences rather than random pools, which allows for identification of authentic protein-binding RNA sequences [109,110].

Previously, a novel technique named CLIP (cross-linking and immunoprecipitation) was developed in order to identify in vivo RNA targets [111]. CLIP involves an in vivo photo cross-linking step to capture the protein–RNA interactions, followed by partial RNase digestion to generate RNA tags of approximately 60 nucleotides followed by specific immunoprecipitation of the protein of interest. An advantage of this method is that by using an in vivo photo cross-linking step, which induces a covalent protein–nucleic acid bond, these interactions are preserved in an intact cell. CLIP was used to characterize the in vivo RNA binding targets of the neuronal-specific splicing factor Nova and has allowed the generation of an RNA map to predict splicing regulation dependent on this protein [112,113]. It has also been used for other RNA binding proteins, including SF2/ASF [114] (Table 5). The identification of SR protein targets and the study of how tissue-specific patterns of splicing change, depending on the complement of SR proteins present, can also be analysed by alternative splicing microarrays [115117].

The additional importance of structural elements in splice site selection should also be taken into account. For example, RNA structure elements associated with alternative splice-site selection have been recently identified in the human genome [118]. In addition, RNA folding has been shown to affect the recruitment of SR proteins to mouse and human ESE elements in the fibronectin EDA exon [119,120]. Finally, competing intronic RNA secondary structures help to define a complex pattern of mutually exclusive exons in the Dscam gene [121].


Initially, the ability of different individual SR proteins to complement splicing-deficient extracts suggested that SR proteins may have redundant functions. However, the sequence-specific RNA binding ability of individual SR proteins and differences in their ability to regulate alternative splicing suggested otherwise [35,122,123].

A growing body of evidence showed that individual SR proteins were not functionally equivalent in Drosophila, Caenorhabditis elegans and mouse models. For instance, SF2/ASF was shown to be an essential factor for cell viability in a chicken cell line and its depletion could not be rescued by expression of SC35 or SRp40, indicating a non-redundant function of SF2/ASF [124]. Other studies have shown that the SR protein B52/SRp55 is essential for Drosophila development [125,126]. B52/SRp55 was shown not to be essential for the splicing of a number of substrates [127], but specific substrates that were mis-spliced in B52-deficient flies were identified [128,129]. Furthermore, B52/SRp55 regulates the inclusion of alternative exon 2 in eyeless, a master regulator of eye development in Drosophila, resulting in the production of a protein isoform that gives rise to a small-eye phenotype. Conversely, the canonical eyeless isoform induces eye overgrowth [130].

Use of RNAi (RNA interference) to inhibit SR protein function during C. elegans development revealed that depletion of the orthologue of the mammalian SF2/ASF (CeSF2/ASF) resulted in embryonic lethality, which indicates an essential, non-redundant, role for this gene during nematode development. By contrast, RNAi-mediated depletion of other SR genes resulted in no obvious phenotype, which is indicative of functional redundancy [131133]. The function of SR proteins has also been studied in mouse model systems (recently reviewed in [134]). All SR-null mice for SRp20 [135], SC35 [136,137] and SF2/ASF [138] show an early embryonic phenotype indicating that SR proteins are not redundant. However, these essential functions appear to be tissue- or developmental stage-specific, as cultured cells from the knockout mice are viable. The generation of conditional knockouts has allowed further characterization of SR protein function in different tissue types or at various developmental time points. Deletion of SC35 in mice results in decreased thymus size and a major defect in T-cell maturation [136], whereas tissue-specific ablation of SC35 in the heart has been shown to cause dilated cardiomyopathy [137]. SF2/ASF has also been shown to have a role in cardiac function; however its main function is in the developmental process of postnatal heart remodelling [138]. Mice knockouts of other splicing factors, including Nova, U2AF26, hnRNP U and hnRNP C, also result in embryonic lethality or developmental defects, which highlights the importance of splicing for the correct regulation of biological processes such as embryogenesis and tissue maintenance [134].


SR proteins also function in mRNA processing reactions that occur after splicing, including mRNA nuclear export, NMD (nonsense-mediated decay) and translation (reviewed in [139]). SR proteins display a nuclear localization pattern and are found to accumulate in splicing speckles [140]. However, a subset of SR proteins, which includes SF2/ASF, SRp20 and 9G8, shuttle continuously between the nucleus and the cytoplasm [141], reminiscent of what was found for a subset of hnRNP proteins [142]. This suggested that the shuttling SR proteins may function in cytoplasmic processes, or be involved in the transport of spliced mRNA. Indeed, SRp20, 9G8 and SF2/ASF function in the nucleocytoplasmic export of mRNA by interacting with the mRNA nuclear export receptor TAP/NFX1 [143,144], exhibiting a higher affinity when hypophosphorylated [145].

SR proteins have also been implicated in regulating the NMD pathway, whereby mRNAs containing premature termination codons are targeted for degradation. Increased expression of a subset of SR proteins, including SF2/ASF, SC35, SRp40 and SRp55, strongly enhanced NMD [146]. Interestingly, this effect does not appear to be dependent on their nucleocytoplasmic shuttling, suggesting a role for SR proteins in enhancing nuclear steps of NMD. A recent study showed that SF2/ASF has the potential to affect the cellular site of NMD, shifting this process to the nuclear compartment before mRNA release from nuclei [147].

SF2/ASF controls alternative splicing of pre-mRNAs encoding the kinases MNK2 [MAPK (mitogen-activated protein kinase)-interacting kinase 2] and S6K1 (S6 kinase 1) that are involved in translational regulation. Increased expression of SF2/ASF results in the production of an isoform of MNK2, which promotes MAPK-independent eIF4E (eukaryotic initiation factor 4E) phosphorylation, and an unusual oncogenic isoform of S6K1, thereby enhancing cap-dependent translation [148]. SR proteins have also been shown to directly affect translational regulation. SF2/ASF associates with polyribosomes in cytoplasmic extracts and enhances the translation of an ESE-containing luciferase reporter both in vivo and in vitro [149]. This direct effect of SF2/ASF in the regulation of the translation of SF2/ASF-bound mRNA targets is mediated by the recruitment of components of the mTOR (mammalian target of rapamycin) signalling pathway, resulting in phosphorylation and release of 4E-BP, a competitive inhibitor of cap-dependent translation [150]. The role of mTOR in the activation of S6K1, which phosphorylates eIF4B and S6, promoting translation initiation, may also be enhanced by SF2/ASF [151] (Figure 3). Other SR proteins have also been reported to function in translation. SRp20 has been shown to function in IRES (internal ribosome entry site)-mediated translation of a viral RNA [152], whereas 9G8 plays a role in translation of unspliced mRNA containing a CTE (constitutive transport element) [153].

Figure 3 Role of SF2/ASF in translation

SF2/ASF-bound mRNAs recruit the mTOR kinase resulting in the phosphorylation and release of 4E-BP, leading to in enhanced translation initiation. The mTOR kinase phosphorylates S6K1, which promotes translation initiation, and this may also be enhanced by SF2/ASF. The Mnk2b isoform induced by SF2/ASF-dependent alternative splicing also leads to translation activation.

The results described in this section demonstrate that SR protein function is not restricted to nuclear mRNA splicing, and it seems sensible that proteins already bound to spliced mRNA may function in subsequent processing events as they are already in place to facilitate future interactions. However, it also highlights the requirement for exquisite regulation of SR proteins in order to maintain their role in cytoplasmic processing of mRNAs without disrupting nuclear processes, which are highly sensitive to the relative concentration of splicing factors.


A dynamic cycle of phosphorylation and dephosphorylation is required for pre-mRNA splicing [154], this being related, at least in part, to the phosphorylation status of SR proteins. The RS domain of SR proteins is extensively phosphorylated on serine residues and this plays an important role in regulating the subcellular localization and activity of SR proteins (reviewed in [40]). For instance, phosphorylation of the RS domain in SF2/ASF acts to enhance protein–protein interactions with other RS domain-containing splicing factors, such as U1-70K [155], whereas dephosphorylation of SR and SR-related proteins is required for splicing catalysis to proceed [156,157].

Several protein kinase families have been shown to phosphorylate the RS domain of SR proteins, including the SRPK (SR protein kinase) family [158,159], the Clk/Sty family of dual-specificity kinases [160] and topoisomerase I [161]. SRPK1 is a serine-specific kinase that binds to a ‘docking motif’ in SF2/ASF that restricts phosphorylation to the N-terminus of the RS domain [162]. In contrast, Clk/Sty can phosphorylate the whole of the RS domain, resulting in a hyperphosphorylated state [163]. The phosphorylation status of the RS domain of SR proteins is also important in the post-splicing activities of SR proteins. A hypophosphorylated RS domain is required for the interaction of nucleocytoplasmic shuttling SR proteins with the TAP/NFX1 nuclear export receptor [145]. SR protein kinases present in the cytoplasm are required to re-phosphorylate the RS domain before the SR protein can return to the nucleus [164]. RS domain dephosphorylation also plays an important role in sorting SR proteins in the nucleus, where shuttling SR proteins and non-shuttling SR proteins are recycled via different pathways [165]. In the cytoplasm, dephosphorylation of the RS domain enhances mRNA binding of SF2/ASF and contributes to its role in translation [166]. SR protein phosphorylation is also important in developmental regulation, as demonstrated in the nematode Ascaris lumbricoides [167].

Importantly, alternative splicing is extensively regulated by signal transduction pathways, whereby signalling cascades can link the splicing machinery to the exterior environment [168]. For instance, the SR protein SRp38 is dephosphorylated upon heat shock by the phosphatase PP1 and becomes a potent splicing repressor [169,170]. Two other well described examples are the insulin-induced promotion of protein kinase C beta II alternative splicing as a result of SRp40 phosphorylation by Akt [171], and the growth factor induced alternative splicing of the fibronectin EDA exon, via phosphorylation of SF2/ASF and 9G8 by Akt [172]. Interestingly, growth factors not only modify the alternative splicing pattern of the fibronectin gene but also affect its translation in an SR protein-dependent fashion, providing an example where modification of SR protein activity in response to extracellular stimulation leads to a concerted regulation of splicing and translation [173]. Caffeine regulates the alternative splicing of a subset of cancer-associated genes, including the tumor suppressor KLF6. This response is mediated by the SR protein, SC35, which is in turn induced by caffeine, and its overexpression is sufficient to recapitulate this regulated event [174]. Another example of the tight regulation of the SR protein family members is exemplified by the common existence of unproductive splicing of SR genes. This is associated with ultraconserved elements that overlap alternatively spliced exons and target the resulting mRNAs for degradation by NMD [175,176].


Disruption of the many roles of SR family proteins can lead to human disease. Approximately 15% of mutations that cause genetic disease affect pre-mRNA splicing [177], targeting conserved splicing signals including the 5′ ss, 3′ ss and BP, as well as enhancer and silencer sequences. Indeed, analysis of a database of 50 single-base substitutions associated with exon-skipping in human genes revealed that more than 50% of these mutations disrupted at least one of the target motifs for the SR proteins SF2/ASF, SRp40, SRp55 and SC35 [178,179] (reviewed in [180]).


There is emerging evidence that establishes a connection between the mis-expression of SR proteins and the development of cancerous tissues, mainly as a result of change in the alternative splicing patterns of key transcripts. Increased expression of SR proteins usually correlates with cancer progression, as shown by elevated expression of SF2/ASF, SC35 and SRp20 in malignant ovarian tissue [181] and of several classical SR proteins in breast cancer [182]. However, the mRNA levels of SF2/ASF, SRp40, SRp55 and SRp75 are lower in non-familial colon adenocarcinomas than adjacent non-pathological tissue, suggesting the levels of SR proteins in cancerous tissues may be tissue-specific [183].

SF2/ASF was found to be upregulated in several human tumours, including lung, colon, kidney, liver, pancreas and breast tumours [148]. Accordingly, gene amplification of SFRS1, which codes for SF2/ASF, is commonly found in breast cancers [184]. Furthermore, increased expression of SF2/ASF transforms immortal rodent fibroblasts and leads to the formation of sarcomas in nude mice, whereas downregulation of SF2/ASF reverses these phenotypes. Other SR proteins, such as SC35 and SRp55, did not have transforming activity, indicating a highly specific role of SF2/ASF in cancer development. Altogether, these results support the notion that SFRS1 is a proto-oncogene [148]. Another RNA target for SF2/ASF that can explain its transforming activity is the proto-oncogene Ron (macrophage-stimulating 1 receptor). SF2/ASF regulates the alternative splicing of Ron pre-mRNA by binding to an ESE in exon 12 and promoting skipping of exon 11 [185]. This results in production of ΔRon, a constitutively active isoform which confers increased motility on expressing cells, a characteristic required for tumour metastasis. Importantly, abnormal accumulation of ΔRon occurs in breast and colon tumours and the levels of SF2/ASF mirror those of ΔRon [185].


HIV-1 uses a combination of several alternative 5′ and 3′ ss to generate more than 40 different mRNAs from its full-length genomic pre-mRNA [186]. Several SR proteins have been shown to regulate different splicing events affecting the viral transcripts. For instance, SRp75 binds a viral ESE [187], whereas SF2/ASF and SRp40 bind a guanosine-adenosine-rich ESE identified in exon 5 of HIV-1 leading to its inclusion [188]. Furthermore, HIV infection induces changes in the levels of splicing factors, including SR proteins, that regulate viral alternative splicing and therefore virus replication [189]. Current drugs used to treat HIV-infected patients involve the use of combinations of retrovirals that specifically target viral proteins such as reverse transcriptase, protease and gp120 (reviewed by [190]). HIV viral production is tightly linked with alternative splicing of the viral HIV-1 pre-mRNA. Therefore, an alternative and novel approach to circumvent the problem of resistance of HIV-1 to current inhibitors is to target the role of SR proteins in HIV pre-mRNA splicing [191]. A screen for chemical inhibitors of pre-mRNA splicing identified indole derivatives that specifically inhibit ESE-dependent splicing by interacting directly and selectively with individual SR proteins [192]. One such small chemical compound was shown to prevent the production of key viral HIV-1 regulatory proteins whose splicing depends on weak 3′ ss [193].

SMA (spinal muscular atrophy)

SMA is a severe hereditary neurodegenerative disorder that results from the lack of a functional SMN1 (survival of motor neuron 1) gene product, which is a key component of the snRNP biogenesis pathway. An SMN1 paralogue, the centromeric SMN2 gene, differs by a single nucleotide change, a C>T transition in exon 7, that causes substantial skipping of this exon and results in the production of a non-functional protein. This exon-skipping event has been attributed either to the loss of an SF2/ASF-dependent exonic splicing enhancer [194] or to the creation of an hnRNP A/B-dependent exonic splicing silencer [195].

Several therapeutic approaches, which focus on altering the splicing of SMN2 to induce exon 7 inclusion and would result in functional SMN protein in affected patients, have made use of antisense technology [196]. The first uses bifunctional ASOs (antisense oligonucleotides) which are comprised of oligonucleotides complementary to exon 7, with a non-complementary tail containing exonic-splicing enhancer motifs recognized by SR proteins. This approach has been shown to mediate the binding of SF2/ASF to SMN2 exon 7 and promote exon inclusion [197]. An alternative strategy has recently been developed based on bifunctional U7 snRNAs that contain both an antisense sequence targeting exon 7 and a splicing enhancer sequence to improve recognition of the exon. These RNAs are stably introduced into cells and the U7 snRNAs become incorporated into snRNPs, inducing a prolonged restoration of SMN protein in SMA fibroblasts [198]. Another approach uses ESSENCE (exon-specific splicing enhancement by small chimaeric effectors) molecules, which also contain an antisense moiety complementary to the target exon and a minimal RS domain peptide designed to mimic the effect of SR proteins. The ESSENCE molecules have also been shown to restore SMN2 levels to that of wild-type SMN1 levels by exon inclusion [199]. Interestingly, the antisense moiety alone stimulated exon 7 inclusion, and functional full-length SMN protein was produced in primary fibroblasts from a type I SMA patient [200,201]. An in vivo delivery system has been developed for bifunctional RNAs using a viral vector [202].

Other human diseases

It has been demonstrated that SR proteins are autoantigens in patients with systemic lupus erythematosus [203]. SR family proteins have also been shown to have regulatory roles in the splicing of several pre-mRNAs associated with human disease. For example, SF2/ASF and SRp40 bind to an ISS and promote exclusion of exon 9 of CFTR (cystic fibrosis transmembrane conductance regulator) [204]. Lack of exon 9 correlates with the occurrence of monosymptomatic and full forms of cystic fibrosis disease [205]. SC35 has a role in the aberrant splicing of the E1αPDH (E1α pyruvate dehydrogenase) mRNA, resulting in a defect of mitochondrial energy metabolism. An intronic mutation of the E1αPDH gene that activates a cryptic 5′ ss leads to mis-spliced mRNA and defective protein [206]. SC35 has been shown to significantly activate splicing at this cryptic site. Accordingly, RNAi-mediated depletion of SC35 in primary fibroblasts from the affected patient could restore the normal E1αPDH splicing pattern [207]. It has also been recently reported that the expression of SRp20 is elevated in bipolar patients, which may explain the aberrant splicing of glucocorticoid receptor α in these patients [208].


This article has reviewed the many roles of SR proteins in gene expression. An obvious question is, why are SR proteins involved in so many cellular functions? These are very abundant nuclear proteins and a subset of them shuttle to the cytoplasm where they are involved in NMD and translation regulation. Their function in different biochemical activities may underlie the extensive network of coupling amongst gene expression machines [209]. It should be noted that SR proteins are not the only proteins coupling nuclear and cytoplamic RNA processing events, since the EJC (exon junction complex), a multiprotein complex deposited as a consequence of pre-mRNA splicing, links pre-mRNA splicing with mRNA export, NMD and translation (reviewed in [210]). Individual SR proteins regulate subsets of pre-mRNAs via splicing in the nucleus and post-splicing processes in the cytoplasm. The next few years will see considerable efforts to identify physiological RNA targets of SR proteins and gain a better understanding of the many cellular functions of these master regulators of RNA processing.


This work was supported by the Medical Research Council and by the European Alternative Splicing Network of Excellence, EURASNET [grant number S18238].


We are grateful to Sonia Guil (Barcelona) for critical reading of this manuscript.

Abbreviations: BP, branch-point; CLIP, cross-linking and immunoprecipitation; CTD, C-terminal domain; Dscam, Down's syndrome cell adhesion molecule; E1αPDH, E1α pyruvate dehydrogenase; eIF, eukaryotic initiation factor; ESE, exonic splicing enhancer; ESS, exonic splicing silencer; hnRNP, heterogeneous nuclear RNP; ISS, intronic splicing silencer; mTOR, mammalian target of rapamycin; NMD, non-sense-mediated decay; PESE, putative exonic splicing enhancer; RESCUE, relative enhancer and silencer classification by unanimous enrichment; RNAi, RNA interference; RNAP II, RNA polymerase II; RNP, ribonucleoprotein particle; RRM, RNA recognition motif; S6K1, S6 kinase; SC35, spliceosomal component 35; SELEX, selected evolution of ligands through exponential enrichment; SF2/ASF, splicing factor 2/alternative splicing factor; SMA, spinal muscular atrophy; SMN, survival of motor neuron; snRNP, small nuclear RNP; 3′/5′ ss, 3′/5′ splice site; Tra, transformer; U2AF, U2 snRNP auxiliary factor


View Abstract