|Home | About | Journals | Submit | Contact Us | Français|
Epstein-Barr virus (EBV) SM protein is an essential nuclear protein produced during the lytic cycle of EBV replication. SM is an RNA-binding protein with multiple mechanisms of action. SM enhances the expression of EBV genes by stabilizing mRNA and facilitating nuclear export. SM also influences splicing of both EBV and cellular pre-mRNAs. SM modulates splice site selection of the host cell STAT1 pre-mRNA, directing utilization of a novel 5′ splice site that is used only in the presence of SM. SM activates splicing in the manner of SR proteins but does not contain the canonical RS domains typical of cellular splicing factors. Affinity purification and mass spectrometry of SM complexes from SM-transfected cells led to the identification of the cellular SR splicing factor SRp20 as an SM-interacting protein. The regions of SM and SRp20 required for interaction were mapped by in vitro and in vivo assays. The SRp20 interaction was shown to be important for the effects of SM on alternative splicing by the use of STAT1 splicing assays. Overexpression of SRp20 enhanced SM-mediated alternative splicing and knockdown of SRp20 inhibited the SM effect on splicing. These data suggest a model whereby SM, a viral protein, recruits and co-opts the function of cellular SRp20 in alternative splicing.
SM protein (EB2, Mta, and BMLF1) is a nuclear phosphoprotein synthesized by Epstein-Barr virus (EBV) during the early stage of lytic replication (for a review, see reference 38). SM has multiple functions in enhancing EBV gene expression posttranscriptionally, binds target gene mRNA, enhances nuclear mRNA export and stability, and modulates cellular and EBV RNA splicing (2, 9, 17, 18, 23, 31, 35, 39, 40). SM is essential for EBV replication and EBV recombinants with insertional deletion of the SM gene are defective for virus production (12). SM is required for the efficient accumulation of ca. 60% of EBV lytic transcripts (13). SM is required for efficient expression of both EBV DNA primase (BSLF1) and EBV DNA polymerase (BALF5) mRNAs, leading to severely impaired lytic EBV DNA replication in the absence of SM (13). SM also directly enhances accumulation of specific late gene mRNAs in addition to enabling DNA replication (13). This combination of effects on DNA replication and late gene mRNAs leads to a global deficiency of late gene expression in the absence of SM.
We recently demonstrated that SM acts as an alternative splicing factor and modulates cellular splicing (40). The effects of SM on host cellular gene expression during lytic EBV replication remain to be fully characterized. When inducibly expressed in EBV-negative cells, SM has a broadly inhibitory effect on cellular mRNA accumulation (30). Nevertheless, SM causes several cellular transcripts to accumulate at higher levels (30). These transcripts include STAT1 and several interferon-stimulated genes. The STAT1 protein is an integral mediator of both type I (alpha/beta interferon [IFN-α/β]) and type II (IFN-γ) IFN signal transduction pathways (for a review, see reference 7). STAT1 is expressed as two isoforms, STAT1α and STAT1β. STAT1β mRNA is generated by cleavage and polyadenylation at an alternative site in the last intron of the STAT1 pre-mRNA, leading to production of a protein which lacks the transactivating domain encoded in the last exon of the STAT1 gene (see Fig. 5A). STAT1β homodimers are not capable of activating GAS sequences, and STAT1β may therefore act as a dominant-negative repressor of STAT1α (3, 27, 41). Consistent with a role for STAT1β as an antagonist of STAT1α, the ratio of STAT1α and -β isoforms has been shown to affect cellular apoptosis and resistance to viral infection (1, 26). Interestingly, SM disproportionately increases the relative amounts of STAT1β mRNA.
Further investigation of previous findings that SM changed the ratio of two functionally distinct STAT1 isoforms generated by alternative processing (30) led to the finding that SM directed splicing of STAT1 to an alternative 5′ splice site with high efficiency and specificity (40). This activity was based on preferential binding of SM to specific regions of the pre-mRNA, indicating that SM may function in a manner similar to cellular splicing factors. Although SM does bind to RNA directly (14, 29), it does not possess arginine-serine (RS) repeats typically found in cellular SR proteins that act as alternative splicing factors (11). We report here the interaction of SM with SRp20, a cellular SR protein, and its role in modulation of splicing by SM.
293 is a cell line derived from human embryonic kidney cells (10). 293T and HeLa cells were maintained in Dulbecco modified Eagle medium containing 10% fetal calf serum supplemented with Glutamax (Invitrogen). HeLa cell transfections were performed with Lipofectamine Plus (Invitrogen) in six-well plates using 1 μg of DNA per transfection, according to the manufacturer's protocols. 293T cells transfections were performed with TransIT293 reagent according to the manufacturer's protocols (Mirus). Cells were harvested 48 h after transfection. P3HR1/ZHT cell line was derived from P3HR-1 (28) by transfection with a plasmid expressing a tamoxifen-inducible EBV Zta activator of lytic gene expression (pCDNA3-ZHT), followed by selection in G418 (20). P3HR1/ZHT cells were cultured in RPMI supplemented with 10% estrogen-free fetal bovine serum and 0.8 mg of G418 (AG Scientific)/ml. EBV lytic replication was induced by adding 100 nM 4-hydroxytamoxifen (Sigma) to the culture medium as previously described (30).
To generate the STAT1 minigene construct, pDV1, 1.088 kb of the STAT1 gene (AY865620, nucleotides 40293 to 41381), which contains exon 23, exon 24, and the intron between exons 23 and 24, was amplified by high-fidelity PCR and cloned in mammalian expression vector pCDNA3 (Invitrogen) at the EcoRV site (40). SRp20 expression vector (amino acids [aa] 1 to 164) and various truncated mutants of SRp20 (aa 1 to 104,1 to 83, 1 to 50, and 1 to 25) were generated by PCR using AccuPrime Pfx DNA polymerase (Invitrogen) and cloned into pcDNA3 expression vector. All SRp20 constructs were tagged with a hemagglutinin (HA) epitope at N terminus by cloning into pCDNA3 with an HA epitope inserted between the HindIII and BamHI sites. DV30 contains the SM response element cloned in the intron of the splicing reporter plasmid pI-12 (40). The PI-12 plasmid provided by Mariano Garcia-Blanco has been previously described (5). The SRp20 expression plasmid was kindly provided by H. Lou (24). The SM expression plasmid was constructed by PCR amplification from B95-8 EBV DNA as previously described (31). Glutathione S-transferase (GST)-SM fusion plasmids were generated by cloning various PCR-generated fragments of SM into pGEX plasmid and have been reported previously (29). A FLAG-calmodulin-binding peptide (CBP)-tagged SM expression plasmid was constructed by sequential cloning of PCR-amplified fragments into pCDNA3. The FLAG, CBP, and tobacco etch virus (TEV) protease sites were amplified from pMZ1 (42) and inserted with the three FLAG epitopes, followed by a TEV protease site and the CBP epitope located upstream of the amino terminus of SM. All plasmids constructed in our laboratory were verified by DNA sequencing.
SM complexes were purified from cell lysates 48 h after transfection. 293T cells were transfected with FLAG-tagged SM expression vector or empty vector. Cells were detached from plates with phosphate-buffered saline (PBS) containing EDTA, washed, and lysed in lysis buffer (50 mM Tris-HCl [pH 7.4], 1 mM EDTA, 1% Triton X-100) with mammalian protease inhibitor cocktail (Sigma), followed by incubation at 4°C with occasional mixing for 30 min, followed by sonication at 4°C. Lysates were clarified by centrifugation for 10 min at 12,000 × g and incubated for 2 h with anti-FLAG M2 antibody-conjugated affinity gel. The agarose matrix was washed extensively and SM complexes were eluted with 150 ng of FLAG peptide/μl in Tris-buffered saline with shaking for 30 min at room temperature. Vector-transfected cells and SM-transfected cells were processed in parallel, and eluates were analyzed by SDS-PAGE and silver staining to assess recovery and purity. Tandem mass spectrometry (MS/MS) analysis of protein complexes in solution was performed at the University of Florida ICBR Proteomics facility, and the data were analyzed by using Scaffold software.
RNA from transfected HeLa cells were harvested 48 h after transfection. Cells were washed with PBS and lysed in 750 μl of QIAzol reagent (Qiagen), and RNA was isolated by using miRNeasy mini columns according to the manufacturer's protocol (Qiagen). Reverse transcription-PCR (RT-PCR) was performed with a One-Step Access Quick RT-PCR kit (Promega) according to the manufacturer's instructions. The primers used for amplification of the STAT1 splice products were previously reported (40). RT-PCR products were analyzed by electrophoresis in 1.5% agarose gels, followed by ethidium bromide staining. The sequence of the spliced products was verified by DNA sequencing as necessary. The relative intensity of the bands in ethidium-stained agarose gels was quantified by using both ImageQuant and Photoshop software.
293T cells were harvested 48 h after transfection and lysed in ice-cold lysis buffer (Tris-buffered saline [pH 7.4], 1% Triton X-100, protease inhibitor cocktail [Sigma-Aldrich], phosphatase inhibitor cocktail [Roche]). The cells were lysed in 200 μl of lysis buffer for 10 min on ice with frequent pipetting and were also sonicated for maximum lysis. The lysed cell suspension was centrifuged for 30 min at 10,000 × g at 4°C. The cleared supernatant was precleared with 1.0 μg of rabbit immunoglobulin G (Bethyl), followed by incubation with protein A-conjugated agarose beads. The precleared samples were used for immunoprecipitation. Mouse monoclonal HA antibody (Covance), control IgG (Bethyl), or rabbit polyclonal anti-SM antibody (31) was added to the lysate, followed by incubation for 1 h at 4°C. Immune complexes were incubated with either protein A-conjugated beads or protein G-conjugated beads for 2 h at 4°C. RNase treatment was performed with 100 μg of RNase A/ml for 30 min at 37°C. The beads were washed four times in wash buffer (Tris-buffered saline [pH 7.4], 500 mM NaCl, 1% Triton X-100) and once in low-salt wash buffer (Tris-buffered saline [pH 7.4], 150 mM NaCl, 1% Triton X-100) and then boiled for 5 min in 1× SDS-PAGE loading buffer. After electrophoresis, proteins were transferred to nitrocellulose membranes (Bio-Rad) and probed with specific antibodies. The following conditions were used for immunoblotting: for SM, a 1:500 dilution of rabbit polyclonal antibody; for HA, a 1:600 dilution of mouse monoclonal antibody (16B12; Covance); for SRp20 antibody, a neat hybridoma supernatant was used (CRL-2384; American Type Culture Collection). After washing, secondary antibodies were then applied to primary antibodies using horseradish peroxidase-conjugated sheep anti-mouse or goat anti-rabbit antibodies (Amersham) diluted to 1:3,000 for monoclonal antibodies and 1:7,500 for polyclonal antibodies. Antigen and antibody complexes were then detected with enhanced chemiluminescence detection reagents (Pierce). Quantitation of the band intensity was performed by using ImageQuant software and Photoshop.
Cell-free coupled transcription-translations were performed according to the manufacturer's protocols (TnT-Quick kit; Promega). Full-length SRp20 mRNA was transcribed from 1 μg of linearized plasmid DNA and translated in the presence of [35S]methionine (Perkin-Elmer). Fusion proteins consisting of GST fused to SM were synthesized in bacteria (E. coli DH5α). Portions (25 μg) of GST or GST fusion proteins were incubated with glutathione-Sepharose beads in binding buffer (1× PBS, 0.1% NP-40, 0.5 mM dithiothreitol, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride) in the presence of bacterial protease inhibitors (P8465; Sigma). Then, 25 μl of [35S]methionine-labeled SRp20 protein was incubated with each fusion protein bound to Sepharose beads for 2 h at 4°C. The beads were washed five times with binding buffer, and bound proteins were treated for 5 min in 1× SDS-protein loading buffer at 95°C and analyzed by SDS-PAGE and autoradiography.
In order to gain insight into the mechanism of SM function in enhancing gene expression and regulating splicing, we performed a series of purifications to isolate cellular protein complexes containing SM. To identify SM-interacting proteins, we isolated protein complexes from cells transfected with an epitope-tagged SM expression vector. We constructed a version of SM tagged with three FLAG epitopes and a CBP epitope separated by a TEV protease site (Fig. (Fig.1A).1A). Epitope tagging was performed by the fusion of PCR-generated epitopes to the amino terminus of SM. We confirmed that the epitope-tagged SM retained function in enhancing reporter gene expression by performing cotransfection assays (data not shown). Although tandem affinity purification has been used successfully to isolate complexes from a variety of cell types, we found that the yield of SM was severely reduced after binding to calmodulin-Sepharose and TEV protease cleavage. We therefore utilized a single-step purification with FLAG antibody-conjugated matrix alone. Optimization of washing conditions and elution with a triple FLAG peptide resulted in adequate reduction of background to permit analysis of SM complexes by MS. Verification of the presence of SM and its associated proteins in complexes purified from SM-transfected cell lysates was performed by silver staining of polyacrylamide gels (Fig. (Fig.1B).1B). Negligible background was observed in eluates from vector-transfected cells (Fig. (Fig.1B).1B). The identity of the prominent band on the silver-stained gel was confirmed to be SM by immunoblotting with anti-FLAG antibody (Fig. (Fig.1C).1C). Liquid eluate samples were digested with trypsin and subjected to a solid-phase extraction and Nanoflow liquid chromatography-MS/MS on an ABI QSTAR XL.
One of the proteins identified in this manner was SRp20 (SFRS3), a nucleocytoplasmic shuttling cellular SR protein involved in both splicing and RNA export (6, 19). SRp20 has a structure typical of highly conserved SR proteins that play multiple roles in constitutive and regulated splicing and contain one or two RNA-binding domains (RBDs) and a carboxy-terminal RS domain with various numbers of RS repeats (Fig. (Fig.2A).2A). To confirm the interaction of SM and SRp20, we performed cotransfections of HA-tagged SRp20 and SM in 293T cells and immunoprecipitated the cell lysates with either anti-SM antibody or preimmune serum. Immunoprecipitates were analyzed by SDS-PAGE, followed by immunoblotting with anti-HA monoclonal antibody to detect SRp20 (Fig. (Fig.2B).2B). SRp20 was coimmunoprecipitated by anti-SM antibody but not by preimmune serum. We next sought to determine whether SM could be shown to interact with endogenous cellular SRp20 in EBV-infected cells. 293T cells were transfected with either SM or empty vector, and SM was immunoprecipitated with anti-SM antibody. The immunoprecipitates were analyzed by immunoblotting with anti-SRp20 monoclonal antibody. As shown in Fig. Fig.2C,2C, SRp20 was specifically immunoprecipitated with SM. In order to determine whether SM interacts with SRp20 during the course of EBV lytic replication, we performed the analogous experiment in an EBV-infected Burkitt lymphoma cell line. P3HR1/ZHT is a derivative of the human Burkitt lymphoma P3HR1 cell line engineered to allow efficient induction of EBV replication by tamoxifen. P3HR1/ZHT was generated by stable transfection of a fusion between Zta and the hormone-binding domain of estrogen receptor, which allows nuclear translocation of Zta and triggering of EBV lytic replication when tamoxifen is added to the culture medium (20). SM was immunoprecipitated from tamoxifen-treated P3HR1/ZHT cells, and the coimmunoprecipitated SRp20 was detected by immunoblotting with anti-SRp20 monoclonal antibody. As shown in Fig. Fig.2C,2C, endogenous SRp20 and SM were coimmunoprecipitated from P3HR1/ZHT lysates (lower panel).
RNA-binding proteins may coimmunoprecipitate because they both bind to the same RNA molecules, rather than due to direct protein-protein interactions. Since SRp20 and SM are both RNA-binding proteins, it was possible that the SM-SRp20 interaction was stabilized by RNA binding. To determine whether such RNA bridging was required for SM-SRp20 interaction, we transfected cells with SM and either empty vector or HA-SRp20, as in Fig. Fig.2B2B above, and performed immunoprecipitations with anti-SM antibody. Prior to SDS-PAGE and immunoblotting, the immunoprecipitates were either treated or mock treated with RNase A. As shown in Fig. Fig.2D,2D, coimmunoprecipitation of SRp20 and SM was resistant to RNase treatment, indicating that the SM-SRp20 interaction is not RNA dependent.
In order to map the region of SRp20 required for SM binding, we generated an influenza virus HA epitope-tagged series of SRp20 deletions by PCR and cloning in an expression vector (Fig. (Fig.3A).3A). These SRp20 expression vectors were cotransfected with SM, and immunoprecipitation with anti-HA monoclonal antibody was performed, followed by immunoblotting with an ti-SM antibody. As shown in Fig. Fig.3B,3B, deletion of aa 105 to 164 of SRp20 did not reduce SM binding. The first 104 aa of SRp20 contained the SM interacting motif, and the deletion of additional amino acids eliminated SM association, locating the SM interaction motif between aa 84 and 104. To confirm these findings, we performed the reciprocal experiment, immunoprecipitating with anti-SM antibody and immunoblotting with anti-HA antibody to detect SRp20 peptides. As shown in Fig. Fig.3C,3C, aa 84 to 104 of SRp20 were required for binding to SM.
In order to map the region of SM that binds SRp20, we used bacterially derived GST-SM fusion proteins encompassing defined portions of the SM protein (Fig. (Fig.4A)4A) (29). We then performed “pulldown” assays with 35S-labeled SRp20 that was synthesized in vitro. It should be noted that we used SM fusion proteins with an internal deletion of aa 149 to 185 that consisted of RXP tripeptide repeats. Deletion of the RXP motif allows more efficient production of intact SM protein in bacteria (29). The expression of all GST-SM fusion proteins was verified by Coomassie blue staining of the bacterial lysates used in the SRp20 pulldown assays (Fig. (Fig.4B).4B). The full-length GST-SM fusion protein with the internal RXP deletion interacted with SRp20, demonstrating that aa 149 to 185 are not required for SRp20 binding (Fig. (Fig.4C).4C). The carboxy-terminal amino acids of SM from aa 281 to 479 were also dispensable for SRp20 interaction. Although SM aa 1 to 281 bound SRp20, the deletion of aa 133 to 281 eliminated binding. Thus, the major SRp20 interacting region lies within aa 185 to 281. This portion of SM also contains the putative Ref interaction domain and RNA-binding regions (17, 18).
SM increases total cellular STAT1 expression, a central mediator of IFN signal transduction, but disproportionately increases the abundance of the STAT1β isoform, which can act as a dominant-negative suppressor of STAT1 (1). Constitutive splicing of STAT1 joins exons 23 and 24, the latter exon encoding the STAT1α transactivation domain. STAT1β is generated by alternative processing of the pre-mRNA, with cleavage and polyadenylation occurring in the intron between exons 23 and 24 (Fig. (Fig.5A).5A). We have shown that SM induces bypassing of the constitutive (α) 5′ donor splice site of exon 23 and instead induces utilization of a cryptic alternative 5′ splice site (α′). Utilization of α′ results in a STAT1 mRNA incapable of producing STAT1α protein due to inclusion of an in-frame stop codon in between exons 23 and 24 (Fig. (Fig.5A)5A) (40). SM-induced alternative splicing is dependent on the presence of an RNA sequence in the STAT1 pre-mRNA to which SM binds directly and which can confer SM-dependent splicing on heterologous RNA (40). Based on a series of mutational analyses, we also showed that the effect of SM was not due to suppression of the constitutive splice site and unmasking of the cryptic splice site. SM therefore acts like an SR protein, promoting utilization of an alternative splice site by binding to exonic splicing enhancers (6). However, SM does not possess any of the motifs typically associated with such functions, such as RS repeats. The association of SM with SRp20 suggested the possibility that SM was recruiting SRp20 to pre-mRNA, allowing SRp20 to perform its function as a splicing enhancer. We therefore tested the effect of overexpressing SRp20 on SM-directed alternative splicing of STAT1. The effect of SM on alternative splicing of STAT1 can be assessed by means of a splicing assay in which splicing substrates are transfected along with putative splicing factors and the products of splicing are analyzed by RT-PCR and sequencing. Using our previously published alternative splicing assay (40), we transfected cells with pDV1, a STAT1 pre-mRNA construct, and an SM plasmid, with or without SRp20. As shown in Fig. Fig.5B,5B, SRp20 enhanced the SM effect on STAT1 splicing by promoting α′ 5′ splice-site utilization and inclusion of the alternative, longer exon 23. Importantly, overexpression of SRp20 alone did not induce alternative splicing at the SM-directed site, indicating that SRp20 itself does not induce alternative STAT1 splicing, and exerts its effect via SM.
RS domain-truncated SRp20 (ΔRS-SRp20) acts as a dominant-negative mutant when expressed in the presence of wild-type SRp20, because the RS domain is required for interaction with other splicing factors and with specific regions of the RNA undergoing splicing (24). If SM were recruiting SRp20 to promote alternative splicing, we expected that ΔRS-SRp20, which still binds SM strongly, would repress alternative splicing induced by SM, similar to the manner in which it represses the splicing factor activity of full-length SRp20 on SRp20 substrates. The effect of ΔRS-SRp20 on SM-induced alternative splicing was therefore tested by substituting the truncated SRp20 for full-length SRp20 in the splicing assay. Cotransfection of ΔRS-SRp20 with did indeed inhibit SM-directed alternative splicing of the STAT1 minigene, indicating that the effect of SRp20 on SM splicing requires the RS domain (Fig. (Fig.5B5B).
We have also shown that the effect of SM on splicing of STAT1 is mediated via nucleotides in the distal STAT1 RNA alternative exon, i.e., between exon-intron boundary of exon 23 and the alternative 5′ donor splice site (40). This SM response element (SMRE), when placed between two adenovirus-derived exons in a minigene directs alternative splicing at the STAT1 alternative 5′ splice site, but only in the presence of SM (Fig. (Fig.5C)5C) (40). This additional assay thus serves as a confirmation of the specificity of SM activity as an alternative splicing factor. In order to confirm that the synergy of SRp20 with SM did not require additional STAT1 sequences besides the SMRE, we transfected HeLa cells with the minigene containing the SM response element and either empty vector or SM plasmid, with or without SRp20. Parallel transfections were also performed with ΔRS-SRp20. As shown in Fig. Fig.5D,5D, SM leads to usage of the alternative 5′ SS, and SRp20 overexpression synergizes with SM, increasing the proportion of RNA spliced at the alternative 5′ SS. As was seen when STAT1 was used as the target pre-mRNA, ΔRS-SRp20 antagonizes SM activity in directing alternative splicing.
In order to confirm the involvement of SRp20 in SM function, we performed siRNA knockdown of SRp20 prior to performing STAT1 splicing assays. First, efficacy of SRp20 knockdown was assessed by immunoblotting cell lysates from small interfering RNA (siRNA)-transfected cells. HeLa cells were transfected with either a non-target control siRNA or SRp20-specific siRNA. Efficiency of knockdown was assessed by preparing cell lysates and immunoblotting for SRp20. SRp20 siRNA reduced SRp20 expression by ca. 60% as quantified by densitometry of the immunoblot (Fig. (Fig.6A).6A). We then examined the effect of SRp20 knockdown on constitutive STAT1 splicing by using the previously described splicing assay with the STAT1 minigene. As seen in Fig. Fig.6B,6B, SRp20 knockdown did not affect constitutive STAT1 splicing. The effect of SRp20 knockdown on SM-induced alternative STAT1 splicing was similarly assessed by transfecting SM with the STAT1 minigene after SRp20 knockdown. Knockdown of SRp20 significantly reduced SM-directed alternative splicing compared to negative control siRNA (Fig. (Fig.6C).6C). These data thus indicate that endogenous SRp20 plays a role in the ability of SM to direct alternative splicing.
We and others have previously shown that SM plays multiple roles in posttranscriptional processing of EBV mRNA and cellular mRNA (2, 9, 17, 18, 23, 31, 35, 39, 40). Although the target sequences for SM binding have not been biochemically defined, abundant evidence indicates that SM contacts RNA directly. SM can be covalently cross-linked to mRNAs in vitro and exhibits RNA-binding specificity when immunoprecipitated from EBV-infected cells undergoing lytic replication (14, 40). Further, SM enhances the expression of EBV mRNAs preferentially, with some genes being SM dependent for expression, while others are SM independent (13). The posttranscriptional effect of SM in enhancing mRNA accumulation has been attributed to an ability to act as a nuclear export factor. According to this model, SM binds to RNA via an RBD and recruits Ref/Aly, TAP, or CRM1 cellular export proteins (2, 18, 22). SM also enhances the nuclear accumulation of target mRNAs posttranscriptionally, suggesting that SM stabilizes mRNA, similar to its KSHV homolog ORF57 (25, 31, 32).
SM also has distinct effects on spliced mRNA transcripts. SM inhibits the expression of spliced cellular targets in transfection assays (31). Recently, we demonstrated that SM inhibits expression of the spliced immediate-early EBV transactivator Rta (39). Inhibitory effects of SM on spliced gene expression have been attributed to premature export of unspliced pre-mRNAs by SM, analogous to the mechanism of HIV Rev protein (4). However, the inhibitory effect of SM on spliced gene expression is target specific, suggesting that SM interacts specifically with the process of splicing (25, 30, 39). Our recent demonstration that SM can affect splice-site selection lends further support to the hypothesis that SM exerts direct effects on splicing in the nucleus (40). The interaction of SM with SRp20 and the involvement of this association in alternative splice site selection suggest a model whereby SM binds to specific pre-mRNA sites and recruits SRp20 via protein-protein interactions (Fig. (Fig.7).7). The RS domain of SRp20 would then direct the formation of spliceosome assembly by recruiting snRNPs and potentially by contacting RNA at discrete sites in the intron (33, 36, 37). In this manner, SM could utilize SRp20 and co-opt its functions of directing splice-site selection even at sites of mRNA devoid of SRp20 binding sites. Recruitment of SRp20 by SM would thereby enable usage of a set of novel splice sites to generate mRNA isoforms unique to cells in which EBV is replicating lytically.
SRp20 is important for export of intronless histone mRNAs and is thought to act as an export factor by virtue of its ability to bind RNA via its RBD and undergo nucleocytoplasmic shuttling (19, 20). SRp20 also interacts with the export mediator TAP via a region in the central portion of the SRp20 molecule (15). Interestingly, this region overlaps with the SM-binding region of SRp20. Thus, it is possible that binding to SM may interfere with SRp20's ability to enhance export of its own RNA cargoes. Alternatively, SRp20 binding to SM may be compatible with TAP recruitment in a manner analogous to the process whereby Ref/Aly, a cellular RNA-binding protein, is thought to bind mRNA and then hand off the mRNA to TAP (16). The net effect of the interaction of SM with SRp20 on export of cellular histone mRNAs and other RNAs remains to be determined. Herpes simplex virus (HSV) ICP27, a homolog of EBV SM, has been shown to inhibit splicing by altering the function of SRPK1, an SR protein kinase, leading to hypophosphorylation of SRp20 and other SR proteins (34). Recently, SRp20 has been implicated in the export of HSV intronless mRNAs (8). Because hypophosphorylation of SRp20 is linked to its ability to export RNA (19), ICP27 may facilitate HSV RNA export by SRp20. Whether SRp20 plays a role in enhancing intronless EBV mRNA stabilization and export is another important question that is raised by these findings. The net effect of the SRp20-SM interaction on EBV mRNA expression is likely to be complex, since the SRp20 effects may also be RNA target dependent.
In summary, we have shown that SM protein binds SRp20 independently of RNA, via discrete protein-protein interactions, and that the interaction is important for the ability of SM to direct specific alternative splice-site selection. Further, recruitment of SRp20 by SM suggests a model by which SM may utilize the RNA and spliceosome-recruiting properties of SR proteins to redirect these proteins to convert SM-binding sites to splicing enhancers. These findings implicate mRNA splicing as another target pathway by which EBV modulates the cellular milieu during lytic replication. Inasmuch as >70% of cellular genes undergo alternative splicing (21), it is likely that EBV affects a number of genes in addition to those currently described, thereby altering the “splicing code” of the cell to facilitate EBV replication.
This study was supported by grant RO1 CA81133 from the NCI, National Institutes of Health.
Proteomic analyses were carried out by the Proteomics Center at the ICBR, University of Florida.
Published ahead of print on 1 September 2010.