Green-mol

Molecular Cell, Vol. 13, 367–376, February 13, 2004, Copyright 2004 by Cell Press

Arginine-Serine-Rich Domains Bound at Splicing Enhancers Contact the Branchpoint to Promote Prespliceosome Assembly Haihong Shen, Julie L.C. Kan, and Michael R. Green* Howard Hughes Medical Institute Programs in Gene Function and Expression and Molecular Medicine University of Massachusetts Medical School Worcester, Massachusetts 01605

Summary Exonic splicing enhancers (ESEs) are required for splicing of certain pre-mRNAs and function by providing binding sites for serine-arginine (SR) proteins, which contain an arginine-serine-rich (RS) domain. How an RS domain bound at the ESE promotes splicing is poorly understood. We have developed an RNAprotein crosslinking procedure to identify the target of the ESE-bound RS domain. Using this approach, we show that the ESE-bound RS domain specifically contacts the pre-mRNA branchpoint. The interaction between the ESE-bound RS domain and the branchpoint occurs in the prespliceosome and is dependent upon the same splicing signals, biochemical factors, and reaction conditions required to support prespliceosome assembly. Analysis of RS domain mutants demonstrates that the ability to interact with the branchpoint, to promote prespliceosome assembly, and to support splicing are related activities. We conclude that the ESE-bound RS domain functions by contacting the branchpoint to promote prespliceosome assembly. Introduction Serine-arginine (SR) proteins are a highly conserved family of splicing factors that are present throughout metazoans and have diverse roles in constitutive and regulated splicing (Graveley, 2000; Hastings and Krainer, 2001; Black, 2003). SR proteins are essential for pre-mRNA splicing in vitro, are believed to act early in the splicing pathway (Graveley, 2000; Black, 2003), and can influence alternative splice site choice. All SR proteins have a similar bipartite structure composed of two functional domains: an N-terminal RNA binding domain, comprising multiple RNA-recognition motifs (RRMs), and a C-terminal arginine-serine-rich (RS) domain. Whereas the RRMs are sufficient for sequencespecific RNA binding, the RS domain is required for enhancement of splicing activities (reviewed in Valcarcel and Green, 1996; Tacke and Manley, 1999; Blencowe, 2000; Graveley, 2000). The RS domains of SR proteins are extensively phosphorylated in vivo, which is believed to influence activity (Xiao and Manley, 1997) and subcellular localization (Gui et al., 1994). In addition to SR proteins, several other mammalian splicing factors contain RS domains, including both subunits of the essential splicing factor U2AF (U2AF65 and U2AF35) and the U1 *Correspondence: [email protected]

snRNP 70 kDa (U1 70K) protein (Hoffman and Grabowski, 1992; Zamore et al., 1992; Zhang et al., 1992). Splicing is initiated through recognition of several intron-defining splice signals: the 5⬘ splice site, which is recognized by U1 snRNP and includes a conserved GU dinucleotide at the 5⬘ end of the intron; the 3⬘ splice site, which consists of a polypyrimidine (Py) tract that is recognized by U2AF65 and an AG dinucleotide that is recognized by U2AF35; and the branchpoint, recognized by U2 snRNP, which is located ⵑ17–40 nucleotides (nt) upstream of the 3⬘ splice site. RNA splicing of certain introns requires an additional positive cis-acting element, which is often purine rich, designated a splicing enhancer or exonic splicing enhancer (ESE) (reviewed in Valcarcel and Green, 1996; Tacke and Manley, 1999; Blencowe, 2000; Graveley, 2000). A variety of RNA binding and in vitro genetic selection experiments have shown that ESEs function by providing binding sites for SR proteins. Several models have been proposed for the function of ESEs and SR proteins. One model, based primarily on protein-protein interaction data, postulates that multiple SR proteins engage in a highly specific network of protein-protein interactions through their RS domains that ultimately bridges the 5⬘ and 3⬘ splice sites via interactions with U1 70K and U2AF35, respectively (Wu and Maniatis, 1993). This network of protein-protein interactions has been proposed to increase binding of U2AF65 to the Py tract (Hoffman and Grabowski, 1992; Wu and Maniatis, 1993; Zuo and Maniatis, 1996). It has also been proposed that SR proteins stabilize binding of U1 snRNP to the 5⬘ splice site through an interaction between an SR protein (e.g., ASF/SF2) and U1 70K (Kohtz et al., 1994). Finally, an ESE can promote splicing by counteracting the activity of a splicing inhibitor that is present on certain pre-mRNAs (Kan and Green, 1999). The splicing factor U2AF is not a member of the SR protein family but has an RS domain that is required for function. U2AF binds to the Py-tract/3⬘ splice site and initiates spliceosome assembly by promoting the interaction between U2 snRNP and the branchpoint. We have previously shown that for the large U2AF subunit, U2AF65, the sole determinant for RS domain function is positive charge, which can be supplied by either an arginine or a lysine. On the basis of these data and the results of RNA-protein and RNA-RNA crosslinking experiments, we proposed that binding of U2AF65 to the Py tract directs the RS domain to contact the branchpoint and promote base-pairing with U2 snRNA in the absence of other splicing factors (Valcarcel et al., 1996). Several observations have suggested that the RS domain of U2AF65 may differ from those of RS domains of SR proteins. First, properties associated with the RS domain of SR proteins, such as insolubility at high magnesium concentrations (Zahler et al., 1992) and reactivity to specific antibodies (Roth et al., 1991), are not shared by U2AF65. Second, studies reporting protein-protein interactions among the RS domains of SR proteins failed to detect interactions with the U2AF65 RS region (Wu and Maniatis, 1993). A possible exception is the SR

Molecular Cell 368

Figure 1. U2AF65 and U2AF35 RS Domains and Synthetic RS Dipeptide Repeats Can Support Enhancer-Dependent Splicing (A) Schematic diagrams of the pre-mRNA substrates used in this study, in which the ESE was replaced by an MS2 recognition site. Splicing activity was analyzed using the enhancer-dependent pre-mRNA substrates dsxMS2 and IgM-MS2, which contain a splicing inhibitor (In). Splicing complex assembly was analyzed using the 3⬘ half dsx-MS2 substrate. (B) Chimeric proteins in which the MS2 RNA binding domain was fused to the RS domain of ASF/SF2, U2AF65, or U2AF35, or to synthetic RS regions containing seven (MS2-RS7) or fourteen (MS2-RS14) RS dipeptides were analyzed for their ability to support splicing of the dsx-MS2 (left) and IgM-MS2 (right) premRNA substrates. In vitro splicing reactions were performed in either HeLa nuclear extract alone (⫺) or HeLa extract containing the MS2 protein or an MS2-RS fusion-protein as indicated. Identities of the spliced products are shown on the right.

protein, p54, which has been reported to interact with U2AF65, although an involvement of the RS domain has not been demonstrated (Zhang and Wu, 1996). Third, the U2AF65 RS domain cannot fully substitute for the RS domain of ASF/SF2 in vivo (Wang et al., 1998). These observations raise the possibility that there may be distinct classes of RS domains that function differently. Here we elucidate the determinants and role of RS domains in enhancer-dependent splicing. Our experiments reveal that, like the U2AF65 RS domain, a direct target of the ESE-bound RS domain is the pre-mRNA. Results U2AF65 and U2AF35 RS Domains and Synthetic RS Dipeptide Repeats Can Support Enhancer-Dependent Splicing To delineate the features of the RS domain required for enhancer-dependent splicing, we adopted a previously described protein-fusion strategy (Graveley and Maniatis, 1998). We constructed a series of chimeric proteins in which the MS2 RNA binding domain was fused to various natural or synthetic RS domains. The MS2RS fusion-proteins were then tested for their ability to support splicing of pre-mRNA substrates in which the ESE was replaced by the MS2 recognition site (Figure 1A). Splicing activity was analyzed using two enhancerdependent pre-mRNA substrates: dsx-MS2 and IgMMS2. The dsx pre-mRNA is an example of a “simple” enhancer-dependent pre-mRNA substrate, whereas the IgM pre-mRNA contains a splicing inhibitor that must

be counteracted by the enhancer for splicing to occur (Kan and Green, 1999). We have previously shown that the IgM pre-mRNA substrate forms an inhibitor complex (Kan and Green, 1999), which complicates the analysis of splicing complex assembly. Therefore, in the splicing complex assembly experiments presented below, only the dsx-MS2 substrate was analyzed. It has been previously shown that an RS domain from an authentic SR protein can support enhancer-dependent splicing when tethered to the ESE as an MS2 fusion-protein (Graveley and Maniatis, 1998). To determine whether heterologous RS domains unrelated to those of SR proteins could also support enhancerdependent splicing, we constructed MS2-(U2AF65)RS and MS2-(U2AF35)RS domain fusion-proteins and, as a control, an MS2-(ASF)RS domain fusion-protein. Figure 1B shows, as expected, that there was no detectable splicing of the dsx-MS2 or IgM-MS2 pre-mRNA substrates in either HeLa nuclear extract alone or nuclear extract plus the control MS2 protein. Consistent with previous studies (Graveley and Maniatis, 1998), nuclear extract containing the MS2-(ASF)RS fusion-protein supported splicing of both the dsx-MS2 and IgM-MS2 pre-mRNAs. Significantly, both MS2-(U2AF65)RS and MS2-(U2AF35)RS also supported splicing. Thus, the RS domains of U2AF65 and U2AF35, which are not SR proteins and therefore not normally bound to ESEs, can support splicing of both simple and inhibitor-containing enhancerdependent pre-mRNAs when tethered to the ESE. We previously found that an artificial RS domain containing seven RS dipeptide repeats can functionally sub-

Enhancer-Bound RS Domain Contacts the Branchpoint 369

Figure 2. Functional Determinants and Role of the ESE-Bound RS Domain in Splicing Complex Assembly (A) Mutant derivatives of the MS2-RS7 fusion-protein were analyzed for their ability to support splicing of the dsx-MS2 (left) and IgM-MS2 (right) pre-mRNA substrates. Splicing reactions were performed in either HeLa nuclear extract alone (⫺) or HeLa extract containing the MS2 protein or MS2-RS7 mutant derivatives. Identities of the spliced products are shown on the right. (B) MS2-(U2AF65)RS, MS2-(ASF)RS, MS2-RS7, and MS2-RS7 mutant derivatives were analyzed for their ability to support assembly of the prespliceosome (complex A) in HeLa extract.

stitute for the natural U2AF65 RS domain (Valcarcel et al., 1996). To determine whether comparable artificial sequences could support enhancer-dependent splicing when tethered to the ESE, we constructed MS2 fusionproteins containing seven (MS2-RS7) or fourteen (MS2RS14) RS dipeptides, and analyzed their ability to support enhancer-dependent splicing. Figure 1B shows that both the MS2-RS7 and MS2-RS14 fusion-proteins supported splicing of the dsx-MS2 and IgM-MS2 pre-mRNA substrates in HeLa nuclear extract. Because the RS regions used in Figure 1B differ in size, distribution of RS dipeptides, and content of other amino acids, we conclude that the primary sequence of the ESE-bound RS domain is not essential for activity. Functional Determinants and Role of the ESE-Bound RS Domain in Splicing Complex Assembly The activity provided by the U2AF65 RS domain depends solely on its positive charge (Valcarcel et al., 1996). To determine whether this is also the case for an ESEbound RS domain or whether there are additional determinants for enhancer-dependent splicing, we constructed and analyzed a series of MS2-RS7 derivatives in which the arginine or serine residues were mutated to other amino acids. Figure 2A shows that substitution of arginine by lysine did not significantly affect the ability of the MS2 fusion-protein to support enhancer-dependent splicing of dsx-MS2 or IgM-MS2 substrates, whereas splicing activity was abolished by alanine substitution. Substitution of serine by alanine, aspartic acid, or glycine destroyed the ability of the MS2 fusion-protein to support splicing. A ␥-32P-ATP labeling assay revealed that in nuclear extract MS2-RS7, MS2-KS7, and MS2AS7 were phosphorylated, whereas MS2-RA7, MS2RD7, and MS2-RG7 were not (data not shown). Thus, the ESE-bound RS domain requires both a positive charge, provided by either arginine or lysine, and a serine, which presumably functions as a phosphorylation site. We next sought to determine the relationship between

the ability of MS2-RS fusion-proteins to support splicing and splicing complex assembly. For analysis of prespliceosome assembly, we constructed a 3⬘ half dsxMS2 pre-mRNA substrate. Figure 2B shows that MS2(U2AF65)RS, MS2-(ASF)RS, MS2-RS7, and MS2-KS7, which supported splicing, also supported assembly of the prespliceosome (complex A). By contrast, MS2, MS2-AS7, MS2-RA7, MS2-RG7, and MS2-RD7, which did not support splicing, failed to form a prespliceosome. Therefore, there is a perfect correlation between the ability of ESE-bound RS domain derivatives to support splicing and prespliceosome assembly. The ESE-Bound RS Domain Contacts the Pre-mRNA The ability of the U2AF65 RS domain to support enhancer-dependent splicing and the relatively nonstringent sequence requirements for ESE-bound RS domain function raised the possibility that, like U2AF65, the ESEbound RS domain might function through contact with the pre-mRNA rather than a protein. To test this idea, we developed an ultraviolet light (UV) crosslinking assay that could detect RS domain-pre-mRNA interactions (Figure 3, bottom). An MS2-(ASF)RS derivative was constructed in which a TEV protease cleavage site and FLAG epitope were inserted between the MS2 RNA binding and RS domains. The fusion-protein was added to HeLa nuclear extract containing a uniformly 32 P-labeled pre-mRNA substrate, and the reaction mixture was irradiated with UV light to induce RNA-protein crosslinks. Following RNase treatment, TEV protease cleavage, and immunoprecipitation with an anti-FLAG antibody, 32P-tagged polypeptides were fractionated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and detected by phosphorimager analysis or autoradiography. Figure 3A shows that in the absence of TEV cleavage a 26 kDa 32P-tagged polypeptide, the expected size of the intact MS2-ASF(RS) fusion-protein, was detected using RNA substrates containing an MS2 RNA binding

Molecular Cell 370

Figure 3. The ESE-Bound RS Domain Contacts the Pre-mRNA (A) The MS2-TEV-FLAG-(ASF)RS fusion-protein was analyzed for its ability to bind a uniformly 32P-labeled pre-mRNA substrate using a UV crosslinking assay, shown schematically in the bottom panel. Reactions were performed using RNA substrates containing (MS2, dsx-MS2, and IgM-MS2) or lacking (dsx-ASLV and IgM1-2) an MS2 RNA binding site. Assays were performed in the presence of nuclear extract (NE) and ATP except where indicated. Molecular weight markers are shown on the left in kDa, and the positions of the 26 kDa 32P-tagged MS2RS fusion-protein and 8 kDa 32P-tagged RS TEV-cleaved product are indicated on the right. (B) Time course analysis of UV crosslinking assays before and after TEV cleavage using the dsx-MS2 (left) or IgM-MS2 (right) substrates. (C) The MS2-TEV-FLAG-(ASF)RS fusion-protein was analyzed for its ability to bind the dsx-MS2 pre-mRNA substrate in S100 extract and in an EDTA-treated nuclear extract. (D) The MS2-TEV-FLAG-(ASF)RS fusion-protein was analyzed for its ability to bind the dsx-MS2 pre-mRNA substrate in a micrococcal nuclease (MNase)-treated nuclear extract and in MNase-treated extract in the presence of EGTA. (E) The MS2-TEV-FLAG-(ASF)RS fusion-protein was analyzed for its ability to bind the dsx-MS2 pre-mRNA substrate following RNase H-directed cleavage of U1 or U2 snRNA. Assays were performed in nuclear extracts in which RNase H cleavage was directed by a nonspecific control DNA oligonucleotide (C), or DNA oligonucleotides complementary to the 5⬘ end of U1 snRNA (U1) or to the U2 snRNA branchpoint basepairing region (U2).

Enhancer-Bound RS Domain Contacts the Branchpoint 371

Figure 4. The ESE-Bound RS Domain-Pre-mRNA Interaction Occurs in the Prespliceosome (A) UV crosslinking analysis of the RS domain-pre-mRNA interaction in the prespliceosome (complex A), mature spliceosome (complexes B and C), or nonspecific H complex using the dsx-MS2 (left) or 3⬘ half dsx-MS2 (right) substrates. (B) Mutant derivatives of the MS2-RS7 fusion-protein were analyzed for their ability to contact the dsx-MS2 (left) and IgM-MS2 (right) premRNA substrates using the UV crosslinking assay.

site (MS2, dsx-MS2, and IgM-MS2) but not control RNA substrates lacking an MS2 RNA binding site (dsx-ASLV and IgM1-2). Upon TEV cleavage we detected an 8 kDa 32 P-tagged polypeptide, the expected size of the RS domain, using RNA substrates containing both an MS2 RNA binding site and an intron (dsx-MS2 and IgM-MS2) but not RNA substrates lacking either an MS2 binding site (dsx-ASLV and IgM1-2) or an intron (MS2). Most importantly, using the dsx-MS2 and IgM-MS2 RNA substrates the 8 kDa 32P-tagged polypeptide was observed only in nuclear extract and was ATP dependent (Figure 3A) and time dependent (Figure 3B). Figure 3C shows that the ESE-bound RS domain-premRNA interaction also occurred in S100 extract, which lacks SR proteins, and in EDTA-treated nuclear extract (Abmayr et al., 1988), both of which support assembly of the prespliceosome but not the complete spliceosome. The interaction between the ESE-bound RS domain and the pre-mRNA did not occur in a micrococcal nuclease-treated nuclear extract (Figure 3D), indicating the involvement of an essential RNA component. To identify this RNA component, we analyzed binding in extracts following RNase H-directed cleavage of U1 or U2 snRNAs. The results of Figure 3E show that the ESEbound RS domain-pre-mRNA interaction occurred in a control RNase H-treated extract, an extract lacking U1 snRNA, but not in an extract lacking U2 snRNA. Significantly, U2 snRNA is required for mammalian prespliceosome assembly whereas U1 snRNA is dispensable (Reed, 2000). On the basis of these results, we conclude

that, when bound to the ESE, an RS domain contacts the pre-mRNA under conditions that support prespliceosome assembly.

The ESE-Bound RS Domain-Pre-mRNA Interaction Occurs in the Prespliceosome To understand the role of the RS domain-pre-mRNA interaction in greater detail, we analyzed splicing complex assembly. Splicing complexes were fractionated on a nondenaturing gel and analyzed for the RS domainpre-mRNA interaction using the UV crosslinking assay. Figure 4A shows that the RS domain-pre-mRNA interaction occurred in the prespliceosome (complex A) but not in the mature spliceosome (complexes B and C) or the nonspecific H complex. As expected, the intact MS2RS fusion-protein (no TEV cleavage) contacted the premRNA in all complexes (H, A, B, and C). The interaction between the ESE-bound RS domain and the pre-mRNA also was evident on a dsx-MS2 3⬘ half RNA substrate, consistent with its occurrence in the prespliceosome and ruling out a requirement for the 5⬘ splice site. To identify the functional determinants required for interaction between the RS domain and the pre-mRNA, we analyzed the series of MS2-RS7 derivatives using the UV crosslinking assay. Figure 4B shows that MS2RS7 and MS2-KS7 contacted the pre-mRNA whereas MS2-RA7, MS2-AS7, MS2-RD7, and MS2-RG7 did not. Thus, the RS domain determinants required to support splicing (Figure 2A), to support prespliceosome assem-

Molecular Cell 372

Figure 5. The ESE-Bound RS Domain Contacts the Pre-mRNA Branchpoint UV crosslinking analysis of the RS domain-pre-mRNA interaction using the dsx-MS2 (top) or IgM-MS2 (bottom) pre-mRNA substrate labeled with a single 32P-G at the branchpoint, Py tract, 3⬘ splice site, or exon 2, as indicated by asterisks on the sequence on the right. Exon and intron sequences are indicated in upper- and lowercase letters, respectively. In the IgM-MS2 substrate, the sequence of the inhibitor is underlined.

bly (Figure 2B), and to contact the pre-mRNA (Figure 4B) are identical. The ESE-Bound RS Domain Contacts the Pre-mRNA Branchpoint The results described above reveal that in the prespliceosome the ESE-bound RS domain contacts the pre-mRNA, but they do not indicate whether this occurs at a specific site. To address this issue, we prepared a series of pre-mRNAs containing a single radioactive phosphate either at the branchpoint, Py tract, 3⬘ splice site, or exon 2. Figure 5 shows that for both the dsxMS2 and IgM-MS2 pre-mRNA substrates, the 8 kDa 32Ptagged RS domain polypeptide was detected only with a pre-mRNA substrate labeled at the branchpoint. We conclude that the ESE-bound RS domain contacts the pre-mRNA specifically at the branchpoint. To delineate the interaction site of the ESE-bound RS domain in greater detail, a series of site-specifically labeled IgM-MS2 pre-mRNAs were prepared and analyzed in the UV crosslinking assay. The results of Figure 6A indicate that crosslinking of the RS domain was detected with IgM-MS2 pre-mRNA substrates labeled 7 or 2 nt upstream and 1 or 3 nt downstream of the branchpoint adenosine, but not with substrates labeled 11 or 56 nt upstream and 5 or 11 nt downstream of the branchpoint adenosine. Thus, the ESE-bound RS domain interaction is highly specific, encompassing an 11–15 nt region surrounding the branchpoint adenosine. A Functional Branchpoint Sequence Is Required for the ESE-Bound RS Domain-Branchpoint Interaction We next asked whether a functional branchpoint sequence was required for interaction with the ESE-bound RS domain. We derived a mutant IgM-MS2 pre-mRNA substrate in which the branchpoint adenosine and surrounding adenosines were changed to guanosines. Figure 6B (left panel) shows that this IgM-MS2 branchpoint mutant substrate was unable to undergo splicing. The

crosslinking assay (right panel) using wild-type and mutant substrates site-specifically labeled within the branchpoint showed that the ESE-bound RS domainbranchpoint interaction occurred on the wild-type but not mutant pre-mRNA substrate. Thus, a functional branchpoint sequence is required for interaction with the ESE-bound RS domain. The Natural SR Protein, ASF, Contacts the Pre-mRNA Branchpoint during Enhancer-Dependent Splicing The experiments described above were all performed with MS2 fusion-proteins. We next sought to verify that a natural ESE-bound SR protein also contacted the branchpoint. We constructed an IgM1-2 pre-mRNA derivative in which the normal IgM1-2 enhancer was replaced by the dsx purine-rich enhancer (PRE), which is bound by the SR protein ASF (Lynch and Maniatis, 1995). This IgM-PRE substrate and a control IgM⌬E substrate were site-specifically labeled within the branchpoint and added to S100 extract supplemented with a recombinant His-tagged ASF protein. Following UV crosslinking, RNase treatment, and immunoprecipitation with an antiHis antibody, 32P-tagged ASF was detected by SDSPAGE and autoradiography or phosphorimager analysis. The results of Figure 7A show that ASF crosslinked to the branchpoint of IgM-PRE but not IgM⌬E. To characterize the ASF-branchpoint interaction in greater detail, we performed RNA-protein crosslinking experiments using a series of IgM-PRE RNAs containing a site-specific label 2, 7, 11, or 56 nt upstream or 1, 3, 5, or 11 nt downstream of the branchpoint adenosine. Significantly, the pre-mRNA region contacted by ESEbound ASF (Figure 7B) and the ESE-bound MS2-RS domain (Figure 6A) were identical. Discussion In this report, we have performed a series of structurefunction and RNA-protein crosslinking experiments to

Enhancer-Bound RS Domain Contacts the Branchpoint 373

Figure 6. Delineating the Boundaries of the ESE-Bound RS Domain-Branchpoint Interaction and Requirement for a Functional Branchpoint Sequence (A) UV crosslinking analysis of the RS domain-branchpoint interaction using a series of IgM-MS2 pre-mRNA substrates site-specifically labeled at the positions indicated by asterisks. The branchpoint adenosine is denoted by the gray box. The minimal (black) and maximal (gray) boundaries of the RS domain interaction site are indicated. (B) A mutant IgM-MS2 pre-mRNA substrate in which the branchpoint adenosine and surrounding adenosines were changed to guanosines, as indicated, was site-specifically labeled 7 nt upstream of the branchpoint adenosine and analyzed for its ability to undergo splicing (left) and to bind the RS domain (right).

determine the role of an ESE-bound RS domain. We found that heterologous RS domains unrelated to SR proteins as well as synthetic RS dipeptides could support splicing when tethered to the ESE. A similar conclusion was reached in a recently published study (Philipps et al., 2003). These structure-function results are reminiscent of those obtained with U2AF65 (Valcarcel et al., 1996). However, for U2AF65, the RS domain required only positively charged amino acids, whereas the ESE-bound RS domain also required a serine, almost certainly reflecting an essential role for phosphorylation. Cartegni and Krainer (2003) analyzed the ability of

synthetic RS domains, tethered to the RNA through oligonucleotide hybridization, to restore splicing to a premRNA containing a mutated ESE in a nuclear extract containing all SR proteins. They reported that a negatively charged amino acid could functionally substitute for serine, whereas we found that a serine-to-aspartic acid substitution abolished activity. The conflicting results may be due to one or more experimental variables, as the two studies differed substantially in the premRNA substrates used, the enhancers analyzed, and the presentation of the RS domains. In particular, Cartegni and Krainer (2003) analyzed an ESE weakened

Molecular Cell 374

Figure 7. An ESE-Bound SR Protein, ASF, Contacts the Pre-mRNA Branchpoint (A) UV crosslinking analysis of the ASF-branchpoint interaction using IgM1-2 pre-mRNA derivatives in which the normal IgM1-2 enhancer was replaced by the dsx purine-rich enhancer (IgM-PRE) or was deleted (IgM⌬E). Both substrates were site-specifically labeled within the branchpoint and added to S100 extract supplemented with a recombinant His-tagged ASF protein. The position of the 32P-tagged ASF protein is indicated on the right. (B) UV crosslinking analysis of the ASF-branchpoint interaction using a series of IgM-PRE substrates site-specifically labeled at the positions indicated by asterisks. The branchpoint adenosine is denoted by the gray box. The minimal (black) and maximal (gray) boundaries of the ASF interaction site are indicated.

by a point mutation, whereas we replaced the entire ESE with an MS2 binding site. Thus, in our assay, the requirements for RS domain function are likely to be more stringent. The relatively nonstringent sequence requirements for RS domain function indicate that the interaction between the ESE-bound RS domain and its target is not highly specific. Thus, if the target of an ESE-bound RS domain is a protein, it would have to be capable of binding a wide variety of RS domains. To date, a protein able to bind RS domains as diverse as those shown to function in this study (i.e., both U2AF RS domains, the ASF/SF2 RS domain, and RS and KS dipeptide repeats) has not been identified. However, our results do not rule out the possibility that RS domains could also participate in protein-protein interactions. The Pre-mRNA Branchpoint Is a Target of the ESE-Bound RS Domain Our RNA-protein crosslinking experiments directly demonstrate that, when tethered to the ESE, the RS domain contacts the branchpoint. The possibility that an arginine-rich region could function through RNA contact is precedented by, first, the family of sequence-specific RNA binding proteins that contain an arginine-rich RNA binding domain (e.g., the HIV-1 Tat and Rev proteins [Weiss and Narayana, 1998]) and, second, the U2AF65 RS domain (Valcarcel et al., 1996). It is intriguing that the ESE-bound RS domain is re-

quired for prespliceosome assembly and, conversely, prespliceosome assembly is required to detect the interaction between the ESE-bound RS domain and the branchpoint. One possible explanation for this mutual dependence is that prespliceosome assembly is a highly cooperative process involving multiple interactions each of which on its own is relatively weak and unstable. In this regard, several other splicing factors are known to contact the branchpoint, including U2 snRNA (Parker et al., 1987; Wu and Manley, 1989; Zhuang and Weiner, 1989), U2AF65 (Valcarcel et al., 1996), and SF1 (Berglund et al., 1997), and one or more of these interactions may occur cooperatively with that between the ESE-bound RS domain and the branchpoint. By analogy, eukaryotic transcription complex assembly is also a highly cooperative process involving multiple, weak interactions (Ptashne and Gann, 2001). In progression of the prespliceosome to the spliceosome, the RS domainbranchpoint interaction is disrupted (Figure 4A), presumably reflecting the major conformational change known to occur during this transition (Staley and Guthrie, 1998; Brow, 2002). How is the ESE-bound RS domain directed to contact the branchpoint, and why does this interaction promote prespliceosome assembly? The relatively nonstringent determinants for RS domain function are inconsistent with a sequence-specific branchpoint recognition. RS domains have a general affinity for RNA (Zamore et al., 1992; Tacke et al., 1997), and thus when tethered to the

Enhancer-Bound RS Domain Contacts the Branchpoint 375

ESE, the RS domain could randomly contact the premRNA at multiple positions. When contact occurs at the branchpoint, prespliceosome assembly is potentiated and the RS domain-branchpoint interaction is stabilized; by contrast, interactions at other, random pre-mRNA regions are transient, unstable, and thus not detected in the UV crosslinking assay. By contacting the branchpoint the RS domain could promote prespliceosome assembly through several mechanisms, including modulating or stabilizing RNA base-pairing, acting like an RNA chaperone to promote RNA folding or RNA-protein interactions, or preventing association of proteins that could inhibit RNA splicing (e.g., hnRNP proteins). The results presented here raise the possibility that direct contact with the pre-mRNA, which also occurs for the U2AF65 RS domain (Valcarcel et al., 1996), is a general mechanism by which RS domains promote pre-mRNA splicing. Experimental Procedures Pre-mRNA Substrates Plasmids encoding the dsx-MS2 and IgM-MS2 pre-mRNA substrates were constructed by inserting the MS2 binding site into the plasmids dsx⌬E (Tanaka et al., 1994) and IgM⌬E (Watakabe et al., 1991), which lack the ESE. To construct the 3⬘ half dsx-MS2 plasmid, PCR products containing the 3⬘ half intron and 3⬘ exon from plasmid dsx-MS2 were inserted into the plasmid pCR II-TOPO (Invitrogen). The control MS2 plasmid used in the crosslinking experiment of Figure 3A was constructed by first inserting the MS2 binding site into the vector pSP72 (Promega), and then inserting the sequence corresponding to the multiple cloning site of pSP72 upstream of the MS2 binding site. All three of the MS2 RNAs in Figure 3A (MS2, dsxMS2, and IgM-MS2) contain approximately the same length of RNA upstream of the MS2 binding site (ⵑ250–300 nt). Plasmids encoding the IgM-MS2 branchpoint mutants were constructed using an overlapping PCR method and appropriate oligonucleotides. The IgMPRE plasmid was constructed by inserting annealed oligonucleotides corresponding in sequence to the dsx purine-rich enhancer into the plasmid IgM⌬E (Watakabe et al., 1991). Protein Expression and Purification Plasmids expressing MS2-RS fusion-proteins were constructed in pFastBac (Invitrogen) for baculovirus production. The MS2 protein was tagged with a His6 epitope at the N terminus and a flexible linker (GGGGGS) at the C terminus and fused in-frame to the RS domain of ASF/SF2, U2AF65, U2AF35, or the synthetic dipeptides RS7, RS14, KS7, RD7, AS7, RG7, or RA7. All constructions were verified by sequencing. SF9 cells were infected with recombinant baculovirus and harvested 3–4 days postinfection. The MS2-TEV-FLAG-RS(ASF) fusion-protein expression plasmid was constructed using the starting plasmid pET15b-MS2-RS(ASF) (Dauksaite and Akusjarvi, 2002) by inserting a linker encoding TEVFLAG in between the MS2 RNA binding and RS domains. MS2-TEVFLAG-RS7 and mutant derivatives were constructed using pET15bMS2 (Dauksaite and Akusjarvi, 2002) by inserting a TEV-FLAG-RS7 sequence or appropriate mutant derivative after the MS2 RNA binding domain. All fusion-proteins were coexpressed in E. coli BL21 cells together with SR protein kinase as previously described (Yue et al., 2000). All MS2-fusion proteins were purified under native conditions on Ni-NTA agarose (Qiagen) and dialyzed against buffer D (20 mM HEPES [pH 7.9], 10 mM KCl, 20% glycerol, 0.2 mM EDTA, 0.5 mM DTT, and 0.5 mM PMSF). All MS2 fusion-proteins bound to the MS2 recognition site with comparable affinities (data not shown). The His6-tagged ASF/SF2 protein (pET19b-ASF) was expressed and purified as described (Caceres and Krainer, 1993). Immunoblot analysis confirmed expression of all recombinant proteins. In Vitro Splicing Assays and Spliceosome Assembly Reactions Splicing reactions and spliceosome assembly reactions were performed essentially as described previously (Kan and Green, 1999),

except that 30% HeLa nuclear extract or 40% S100 extract was used. Spliced products were resolved on 12% denaturing polyacrylamide gels (19:1) in 8 M urea in Tris-Borate-EDTA buffer. Spliceosomal complexes were resolved on native 4% acrylamide:bisacrylamide (80:1)-0.5% agarose gels in 50 mM Tris base-50 mM glycine buffer. UV Crosslinking Assay UV crosslinking was performed as described previously (Wu and Green, 1997). TEV (Invitrogen) cleavage and immunoprecipitation with M2 agarose (Sigma) were performed according to the manufacturer’s instructions. Micrococcal nuclease, EGTA, and EDTA treatments were carried out as described previously (Wu and Green, 1997). Site-specific labeling of the pre-mRNA substrates was performed as described previously (Wu and Green, 1997). For the IgMMS2 and IgM-PRE substrates, the nucleotide at position ⫺2, ⫹1, or ⫹3 was changed to guanosine prior to the labeling reaction. RNase H-Mediated Inactivation of U1 and U2 snRNA Inactivation of U1 and U2 snRNA in HeLa nuclear extracts by RNase H-directed cleavage was performed as described previously (Merendino et al., 1999) using DNA oligonucleotides complementary to the 5⬘ end of the U1 snRNA or to the branchpoint base pairing region in the U2 snRNA. Acknowledgments We thank Adrian Krainer, Goran Akusjarvi, and Jan-Peter Kreivi for providing plasmids, and Sara Evans for editorial assistance. This work was supported in part by an NIH grant to M.R.G. M.R.G. is an investigator of the Howard Hughes Medical Institute. Received: June 3, 2003 Revised: December 4, 2003 Accepted: December 4, 2003 Published: February 12, 2004 References Abmayr, S.M., Reed, R., and Maniatis, T. (1988). Identification of a functional mammalian spliceosome containing unspliced premRNA. Proc. Natl. Acad. Sci. USA 85, 7216–7220. Berglund, J.A., Chua, K., Abovich, N., Reed, R., and Rosbash, M. (1997). The splicing factor BBP interacts specifically with the premRNA branchpoint sequence UACUAAC. Cell 89, 781–787. Black, D.L. (2003). Mechanisms of alternative pre-messenger RNA splicing. Annu. Rev. Biochem. 72, 291–336. Blencowe, B.J. (2000). Exonic splicing enhancers: mechanism of action, diversity and role in human genetic diseases. Trends Biochem. Sci. 25, 106–110. Brow, D.A. (2002). Allosteric cascade of spliceosome activation. Annu. Rev. Genet. 36, 333–360. Caceres, J.F., and Krainer, A.R. (1993). Functional analysis of premRNA splicing factor SF2/ASF structural domains. EMBO J. 12, 4715–4726. Cartegni, L., and Krainer, A.R. (2003). Correction of disease-associated exon skipping by synthetic exon-specific activators. Nat. Struct. Biol. 10, 120–125. Dauksaite, V., and Akusjarvi, G. (2002). Human splicing factor ASF/ SF2 encodes for a repressor domain required for its inhibitory activity on pre-mRNA splicing. J. Biol. Chem. 277, 12579–12586. Graveley, B.R. (2000). Sorting out the complexity of SR protein functions. RNA 6, 1197–1211. Graveley, B.R., and Maniatis, T. (1998). Arginine/serine-rich domains of SR proteins can function as activators of pre-mRNA splicing. Mol. Cell 1, 765–771. Gui, J.F., Lane, W.S., and Fu, X.D. (1994). A serine kinase regulates intracellular localization of splicing factors in the cell cycle. Nature 369, 678–682. Hastings, M.L., and Krainer, A.R. (2001). Pre-mRNA splicing in the new millennium. Curr. Opin. Cell Biol. 13, 302–309.

Molecular Cell 376

Hoffman, B.E., and Grabowski, P.J. (1992). U1 snRNP targets an essential splicing factor, U2AF65, to the 3⬘ splice site by a network of interactions spanning the exon. Genes Dev. 6, 2554–2568. Kan, J.L., and Green, M.R. (1999). Pre-mRNA splicing of IgM exons M1 and M2 is directed by a juxtaposed splicing enhancer and inhibitor. Genes Dev. 13, 462–471. Kohtz, J.D., Jamison, S.F., Will, C.L., Zuo, P., Luhrmann, R., GarciaBlanco, M.A., and Manley, J.L. (1994). Protein-protein interactions and 5⬘-splice-site recognition in mammalian mRNA precursors. Nature 368, 119–124. Lynch, K.W., and Maniatis, T. (1995). Synergistic interactions between two distinct elements of a regulated splicing enhancer. Genes Dev. 9, 284–293. Merendino, L., Guth, S., Bilbao, D., Martinez, C., and Valcarcel, J. (1999). Inhibition of msl-2 splicing by sex-lethal reveals interaction between U2AF35 and the 3⬘ splice site AG. Nature 402, 838–841. Parker, R., Siliciano, P.G., and Guthrie, C. (1987). Recognition of the TACTAAC box during mRNA splicing in yeast involves base pairing to the U2-like snRNA. Cell 49, 229–239. Philipps, D., Celotto, A.M., Wang, Q.Q., Tarng, R.S., and Graveley, B.R. (2003). Arginine/serine repeats are sufficient to constitute a splicing activation domain. Nucleic Acids Res. 31, 6502–6508. Ptashne, M., and Gann, A. (2001). Genes and Signals (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory). Reed, R. (2000). Mechanisms of fidelity in pre-mRNA splicing. Curr. Opin. Cell Biol. 12, 340–345. Roth, M.B., Zahler, A.M., and Stolk, J.A. (1991). A conserved family of nuclear phosphoproteins localized to sites of polymerase II transcription. J. Cell Biol. 115, 587–596. Staley, J.P., and Guthrie, C. (1998). Mechanical devices of the spliceosome: motors, clocks, springs, and things. Cell 92, 315–326. Tacke, R., and Manley, J.L. (1999). Determinants of SR protein specificity. Curr. Opin. Cell Biol. 11, 358–362. Tacke, R., Chen, Y., and Manley, J.L. (1997). Sequence-specific RNA binding by an SR protein requires RS domain phosphorylation: creation of an SRp40-specific splicing enhancer. Proc. Natl. Acad. Sci. USA 94, 1148–1153. Tanaka, K., Watakabe, A., and Shimura, Y. (1994). Polypurine sequences within a downstream exon function as a splicing enhancer. Mol. Cell. Biol. 14, 1347–1354. Valcarcel, J., and Green, M.R. (1996). The SR protein family: pleiotropic functions in pre-mRNA splicing. Trends Biochem. Sci. 21, 296–301. Valcarcel, J., Gaur, R.K., Singh, R., and Green, M.R. (1996). Interaction of U2AF65 RS region with pre-mRNA branch point and promotion of base pairing with U2 snRNA. Science 273, 1706–1709. Wang, J., Xiao, S.H., and Manley, J.L. (1998). Genetic analysis of the SR protein ASF/SF2: interchangeability of RS domains and negative control of splicing. Genes Dev. 12, 2222–2233. Watakabe, A., Sakamoto, H., and Shimura, Y. (1991). Repositioning of an alternative exon sequence of mouse IgM pre-mRNA activates splicing of the preceding intron. Gene Expr. 1, 175–184. Weiss, M.A., and Narayana, N. (1998). RNA recognition by argininerich peptide motifs. Biopolymers 48, 167–180. Wu, S., and Green, M.R. (1997). Identification of a human protein that recognizes the 3⬘ splice site during the second step of premRNA splicing. EMBO J. 16, 4421–4432. Wu, J.Y., and Maniatis, T. (1993). Specific interactions between proteins implicated in splice site selection and regulated alternative splicing. Cell 75, 1061–1070. Wu, J., and Manley, J.L. (1989). Mammalian pre-mRNA branch site selection by U2 snRNP involves base pairing. Genes Dev. 3, 1553– 1561. Xiao, S.H., and Manley, J.L. (1997). Phosphorylation of the ASF/SF2 RS domain affects both protein-protein and protein-RNA interactions and is necessary for splicing. Genes Dev. 11, 334–344. Yue, B.G., Ajuh, P., Akusjarvi, G., Lamond, A.I., and Kreivi, J.P.

(2000). Functional coexpression of serine protein kinase SRPK1 and its substrate ASF/SF2 in Escherichia coli. Nucleic Acids Res. 28, E14. Zahler, A.M., Lane, W.S., Stolk, J.A., and Roth, M.B. (1992). SR proteins: a conserved family of pre-mRNA splicing factors. Genes Dev. 6, 837–847. Zamore, P.D., Patton, J.G., and Green, M.R. (1992). Cloning and domain structure of the mammalian splicing factor U2AF. Nature 355, 609–614. Zhang, M., Zamore, P.D., Carmo-Fonseca, M., Lamond, A.I., and Green, M.R. (1992). Cloning and intracellular localization of the U2 small nuclear ribonucleoprotein auxiliary factor small subunit. Proc. Natl. Acad. Sci. USA 89, 8769–8773. Zhang, W.J., and Wu, J.Y. (1996). Functional properties of p54, a novel SR protein active in constitutive and alternative splicing. Mol. Cell. Biol. 16, 5400–5408. Zhuang, Y., and Weiner, A.M. (1989). A compensatory base change in human U2 snRNA can suppress a branch site mutation. Genes Dev. 3, 1545–1552. Zuo, P., and Maniatis, T. (1996). The splicing factor U2AF35 mediates critical protein-protein interactions in constitutive and enhancerdependent splicing. Genes Dev. 10, 1356–1368.