Multiple Roles for SR Proteins in trans Splicing

Abstract

The trans-splicing reaction involves the association of 5′; and 3′; splice sites contained on separate transcripts. The mechanism by which these splice sites are juxtaposed during trans-spliceosome assembly and the role of SR proteins at each stage in this process have not been determined. Utilizing a system that allows for the separation of the RNA binding and RS domains of SR proteins, we have found that SR proteins are required for at least two stages of the trans-splicing reaction. They are important both prior to and subsequent to the addition of U2 snRNP to the 3′; acceptor. In addition, we have demonstrated a role for RS domain phosphorylation in both of these activities. Dephosphorylation of the RS domain led to a block in U2 snRNP binding to the substrate. In a separate experiment, RS domain phosphorylation was also determined to be necessary for trans splicing to proceed on a substrate that had U2 snRNP already bound. This newly identified role for phosphorylated SR proteins post-U2-snRNP addition coincides with the recruitment of the 5′; splice site contained on the SL RNP, suggesting a role for SR proteins in splice site communication in trans splicing.

trans splicing involves a productive interaction between 5′; and 3′; splice sites present on separate transcripts. In nematodes, the trans-spliced 5′; splice site is contained on a unique small RNA (spliced leader RNA [SL RNA]) that harbors both the donated exon and a region exhibiting snRNP-like characteristics. The SL RNP contains Sm proteins, which are common to other U snRNPs, as well as specific proteins in addition to a trimethylguanosine cap (for a review, see reference 38). This unusual situation, where the 5′; exon and an RNP component of the reaction are present in a single molecule, highlights the most apparent difference between cis splicing and trans splicing (5, 51, 52). Required RNP components of a trans spliceosome include the SL, U2, and U4/U6.U5 snRNPs but, notably, not U1 snRNP (33). Several lines of evidence have demonstrated that the 5′; end of U1, which normally pairs with the 5′; splice site in cis splicing, is not required for trans splicing (22). Recent work has demonstrated, however, that U1 snRNP can exert a large effect on the efficiency of trans splicing via exon definition-like interactions (2). In addition, there are interactions between several of the snRNPs involved in trans splicing that are not present in a cis spliceosome. These include an interaction between the SL RNP and U6 (57) as well as contact between the exonic portion of the SL RNP and U5, even in the absence of a trans-acceptor substrate (33). These interactions, coupled with the fact that the splice sites are contained on separate RNAs, have led to speculation as to how the splice sites might be initially juxtaposed in a forming trans spliceosome, prior to any extensive RNA-RNA rearrangements that would precede catalysis (3).

The role(s) of protein factors in authentic trans splicing has been examined much less thoroughly. However, the serine/arginine-rich non-snRNP splicing factors, the SR proteins, have been identified as required components for both trans (45) and cis (14, 29) splicing. Members of this family of proteins contain one or two N-terminal RRM-type RNA binding domains and a carboxyl-terminal domain rich in Ser (S) and Arg (R) residues. The RNA binding domain is involved in the interaction of SR proteins with elements within the pre-mRNA substrate termed splicing enhancers. Splicing enhancers serve as a platform for SR proteins to recruit essential factors to the 3′; splice site (17, 40, 62). In addition, splicing enhancers can counteract the negative regulation of splicing imposed by splicing silencers (26). The C-terminal RS domain is involved in mediating protein-protein interactions with other SR proteins as well as with different components of the spliceosome (28, 54). However, recent studies have also demonstrated that the RS domains of SR proteins are dispensable for SR protein function in certain contexts (11, 49, 60, 61). In these cases, it has been suggested that the RNA binding domain of an SR protein can interact with its target sequence and effectively displace negative regulatory factors, namely hnRNP proteins. The separable nature of SR protein domains suggests distinct roles for SR proteins in exon-dependent and exon-independent processes (19, 24). Taken together, it appears that SR proteins are involved at multiple stages of the splicing reaction.

SR proteins exist in the cell as phosphoproteins, with reversible phosphorylation exerting a large effect on their activities (for a review, see reference 16). Prior to examining the effect of phosphorylation on SR protein function, gross nonspecific effects of phosphorylation were examined in terms of the pre-mRNA splicing reaction. In these earlier experiments, the use of protein phosphatases and phosphatase inhibitors and the addition of ATPγS determined that phosphatase activity was required for pre-mRNA splicing (35, 50). In fact, the essential splicing factor, SCF1, is a protein phosphatase 2Cγ (37). Moreover, phosphatase treatment prior to assembly of the spliceosome resulted in inhibition of complex formation (34, 36) and affected 5′; splice site selection (8). With respect to SR proteins, some degree of phosphorylation is necessary for spliceosome assembly as well as for specificity in RNA binding (7, 41, 48, 55). In fact, in experiments examining a single SR protein (ASF/SF2), dephosphorylation was required for constitutive splicing but not for splicing activator function (56). The overall balance between phosphatase and kinase activity has become readily apparent, as an intermediate level of SR protein phosphorylation seems to be optimal for their activity in pre-mRNA splicing (39, 43, 45).

We have investigated the roles of SR proteins in trans splicing by employing cell extracts derived from the parasitic nematode Ascaris lumbricoides. In this organism, pre-mRNA splicing activity (both cis and trans) in early developing embryos is globally regulated through gross alterations in the level of SR protein phosphorylation. Splicing extracts prepared prior to the 4- to 8-cell stage are inactive and contain hyperphosphorylated SR proteins, whereas extracts prepared from embryos that have developed beyond the 4- to 8-cell stage possess splicing activity as well as SR proteins exhibiting an intermediate level of phosphorylation (43). In this paper, we make use of the MS2-RS fusion protein system (19) in conjunction with site-specific label incorporation and nuclease protection experiments to show that SR proteins are required for trans splicing subsequent to the addition of U2 snRNP to the branch point. As trans splicing is, by definition, a two-part reaction, this result suggests that SR proteins might play a key role in the incorporation of the SL RNP and the U4/U6.U5 tri-snRNP or all four snRNPs in the context of a preformed tetra-snRNP (33) into the assembling trans spliceosome. In addition, we have examined the role of RS domain phosphorylation in both early (the association of U2 snRNP with the branch point) and later (subsequent to U2 snRNP addition) steps in the trans-splicing reaction. Dephosphorylation of the MS2-RS fusion protein inhibits the recruitment of U2 snRNP to the branch point. Moreover, we have determined that RS domain phosphorylation is also critical at a stage in the assembly of the trans spliceosome after U2 snRNP is already bound to the branch point sequence. These experiments demonstrate that SR proteins are required at multiple stages in trans-spliceosome formation, both prior to and after U2 is bound to the branch point. In addition, we provide direct, mechanistic evidence for the importance of RS domain phosphorylation in both of these activities.

MATERIALS AND METHODS

Substrates.

The starting construct (0× MS2) used to make MS2 binding site-containing substrates has been previously described (2). The MS2 coat protein binding sequence (CGUACACCAUCAGGGUACG) was incorporated both singly and in tandem into the starting construct (0× MS2) by using overlap PCR (25). For 1× MS2, the MS2 sequence started 50 nucleotides (nt) downstream of the 3′; splice site (ss). For 2× MS2, MS2 sites began at 50 and 74 nt downstream of the 3′; ss. For 3× MS2, MS2 sites started at 30, 54, and 78 nt downstream of the 3′; ss. A 5-nt linker (AUCGA) separated each of the MS2 sites in 2× MS2 and 3× MS2. Each of the MS2 binding site-containing constructs was linearized with HindIII (NEB) and transcribed with T3 RNA polymerase, generating either a 268-nt (0× MS2), 287-nt (1× MS2), 311-nt (2× MS2), or 335-nt (3× MS2) pre-mRNA.

Preparation of extracts and in vitro trans-splicing assays.

A. lumbricoides whole-cell extracts were prepared as previously described (21). Body-labeled MS2 binding site-containing and control trans-splicing acceptor RNA substrates were generated as previously described (21). In vitro trans-splicing assays were performed essentially as previously described (21, 43). trans-splicing assays employing MS2 binding site-containing substrates (6.3 fmol) were supplemented with 75 ng of MS2-RS9G8 fusion protein (10× saturating for trans-splicing activity) or 41 ng of wild-type MS2 protein. Reaction mixtures used with 2-cell stage extracts were supplemented with 0.5 to 1.5 μg of SR proteins purified from the 32- to 64-cell stage. Amounts of total SR proteins or recombinant 9G8 used in the titration experiments included 0.5, 1.0, and 1.5 μg versus 75 ng of MS2-RS fusion protein or 41 ng of wild-type MS2 protein.

Micrococcal nuclease protection of site-specifically labeled substrates.

Site-specifically labeled substrates were generated as previously described (32). For reactions in which the U2 snRNP was blocked, the reaction mixtures included 200 ng of a 2′;-OMe oligonucleotide complementary to nt 29 to 45 (branch point interaction region). Nuclease protection assays were carried out essentially as previously described (32, 40), except that 420 U of micrococcal nuclease (Worthington) was used and the digestion occurred on ice for 30 min.

Protein purification.

Sf21 cells were infected with MS2-RS9G8 recombinant virus (19) and were allowed to grow for 4 days at 27°C. Cells were then spun at ∼450 × g in a clinical centrifuge. Cells were resuspended in binding buffer (50 mM sodium phosphate buffer [pH 8.0], 1.5 M NaCl, 5 mM imidazole, 0.5% NP-40, 4 mM 2-mercaptoethanol). Cells were then sonicated. Insoluble material was removed by spinning the cell sonicate at 17,000 × g in a Beckman JA-20 rotor for 20 min at 4°C. Cleared lysate was added to Ni²⁺-nitrilotriacetic acid resin (Qiagen), which had been previously equilibrated with binding buffer. Binding occurred in batch at 4°C for 1 h. Following binding, the resin was washed two times with 25 ml of wash buffer (50 mM NaP buffer [pH 8.0], 1.5 M NaCl, 20 mM imidazole). Resin was then transferred to Eppendorf tubes, where protein was eluted sequentially five times with 0.5× bed volumes of elution buffer (50 mM NaP buffer [pH 8.0], 1.5 M NaCl, 250 to 400 mM imidazole). Following sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis, protein was dialyzed stepwise into BC100 (20 mM Tris [pH 7.5], 100 mM KCl, 0.2 mM EDTA, 20% glycerol). Protein concentration was determined by Bradford analysis. The bacterial expression construct for His-tagged MS2 coat protein was obtained from B. Graveley (University of Connecticut). MS2 coat protein was purified essentially as previously described (19). A. lumbricoides SR proteins from the 32- to 64-cell stage were purified by using the ammonium sulfate-magnesium precipitation procedure previously described (59).

In vitro protein dephosphorylation.

Dephosphorylation reaction mixtures included 0.1 to 1.0 U of alkaline phosphatase (Roche Molecular Biochemicals) and were incubated at 37°C for 1 h. Following incubation, modified proteins were placed on ice and used promptly. Aliquots of modified MS2-RS fusion proteins (75 ng) were used in micrococcal nuclease protection assays. Aliquots of modified intact SR protein (9G8, 0.5 μg) were added to splicing reaction mixtures along with 75 ng of active MS2-RS fusion proteins. A portion of the modified MS2-RS fusion protein was resolved by SDS-12% PAGE and visualized by Coomassie blue staining.

RESULTS

MS2-RS fusion protein-dependent trans splicing.

While initial studies demonstrated that SR proteins were required for the trans-splicing reaction (45), the nature of their participation was not investigated. SR proteins were subsequently shown to be necessary for U2 snRNP addition to the branch point sequence of a natural trans-splicing substrate (40). The question remained, however, whether SR proteins were required for a subsequent stage(s) of trans-spliceosome formation and, ultimately, the generation of trans-spliced product. The nature of the events in trans-spliceosome assembly that occur after the incorporation of U2 snRNP is not clear beyond the requirement for the SL and U4/U6.U5 snRNPs. In order to address the question of SR protein participation in later stages of the trans-splicing reaction, we sought to generate a trans-splicing substrate that was fully dependent upon the presence of an MS2-RS domain fusion protein (19). In this manner, we reasoned that a tethered RS domain would promote an early complex on the 3′; acceptor molecule that included U2 snRNP. If the reaction required SR proteins beyond this stage, the complex would stall if it were allowed to assemble in 2-cell stage extracts, which contain inactive SR proteins.

We began with a poor trans-splicing substrate that requires splicing enhancer elements in the exon for activity (2). Into this substrate we cloned 1, 2, or 3 copies of the MS2 RNA binding site, since it had previously been shown that increasing the number of natural enhancers (23) or heterologous RNA binding sites (18) led to an additive increase in splicing activity. Along with these substrates, we overexpressed both wild-type MS2 protein and MS2-RS fusion proteins in bacteria and insect cells, respectively (19). The substrates as well as the trans-acting protein factors are shown schematically in Fig. 1A. We next tested the trans-splicing activity of substrates containing 0, 1, 2, or 3 copies of the MS2 protein binding site in splicing-competent extracts prepared from ∼32- to 64-cell stage embryos. Each of these substrates was completely inactive in this extract (Fig. 1B, lanes 4 and 5, 8 and 9, 13 and 14, and 17 and 18). For comparison, a standard trans-splicing substrate was spliced in the same extract (Fig. 1B, lanes 1 and 2). As a control, each of these substrates was incubated under splicing conditions in the presence of the wild-type MS2 protein lacking the RS domain. There was no increase in trans-splicing activity with any of these transcripts (Fig. 1B, lanes 6, 10, 15, and 19). Finally, we performed the reactions in the presence of an MS2-RS fusion protein (RS domain from the mammalian SR protein, 9G8). In this case, trans-splicing activity increased with the inclusion of increasing copies of the MS2 binding site (Fig. 1B, lanes 7, 11, 16, and 20). We titrated the amount of MS2-RS fusion protein added to the reaction in order to ensure that we had saturated trans-splicing activity (data not shown). Thus, the trans-splicing activity of these substrates is absolutely dependent upon the binding of an MS2-RS fusion protein to the exon downstream of the 3′; splice site.

FIG. 1.

Development of an MS2-RS fusion protein-dependent trans-splicing system. (A) Schematic diagram representing MS2 binding site-containing substrates (see Materials and Methods for the sequence) and MS2-containing proteins. MS2 coat protein binding sites were incorporated singly and in tandem (1 to 3× MS2) downstream of the 3′; splice site of a poor trans-splicing substrate (0× MS2). Schematic representations of both the wild-type MS2 coat protein and a fusion of the MS2 coat protein to the RS domain of an SR protein are also depicted. (B) Body-labeled (0 to 3×) MS2 substrates were incubated in 32- to 64-cell stage extract under splicing conditions for the indicated times. Wild-type MS2 or MS2-RS fusion proteins were included as indicated. Open boxes represent trans-spliced exons, a line indicates intronic regions, and a striped box corresponds to the trans-spliced leader exon. 3′;Acc, a standard trans-splicing substrate, was used as a positive control. Lanes are derived from multiple gels analyzing reactions from the same experiment. M, markers (pBR322 cut with MspI).

As mentioned above, SR proteins are required for the addition of U2 snRNP onto the branch site of a naturally trans-spliced pre-mRNA (40). We therefore carried out experiments to determine whether the MS2-RS fusion protein is capable of recruiting U2 snRNP to substrates containing MS2 binding sites (Fig. 2A). We performed this assembly reaction in 2-cell stage extracts, where the SR proteins are hyperphosphorylated and inactive (43). If an MS2-RS fusion protein allowed for U2 snRNP recruitment, we could also assess whether SR proteins were required for a later stage in trans-spliceosome assembly (Fig. 2B). While the RNA components required for the complete assembly of a functional trans spliceosome are known, the nature of their association with the complex formed on the 3′; acceptor is not.

FIG. 2.

Hypothetical model for MS2-RS fusion protein-dependent trans-spliceosome assembly. (A) Schematic representation of the initial complex formed on an MS2 binding site-containing substrate in the presence of an MS2-RS fusion protein; (B) possible mechanism by which a nontethered SR protein may function to move from a U2-stalled complex generated by an MS2-RS fusion protein to a catalytically active trans spliceosome.

SR proteins are required for trans splicing subsequent to U2 snRNP addition to the branch point sequence.

In order to determine the effect of the MS2-RS fusion protein on U2 snRNP interaction with the branch point sequence, we employed a combination of site-specific label incorporation and nuclease protection (32). Modified trans-splicing substrates containing MS2 binding sites were labeled at the 3′; splice site A∗G dinucleotide. Incubation of substrates containing 0, 1, 2, or 3 MS2 binding sites with the MS2-RS fusion protein in 2-cell stage extracts resulted in increased protection of a region adjacent to the labeled 3′; splice site (Fig. 3A, lanes 2, 4, 6, and 8, respectively). In order to identify the protected fragments, parallel reactions were performed after the extract was preincubated with a 2′;-OMe oligonucleotide that blocks the branch point interaction region of the U2 snRNP. In each of these reactions, the protected bands were abolished (Fig. 3A, lanes 3, 5, 7 and 9), demonstrating that the protections resulted from U2 snRNP interaction with the substrate. The addition of a 2′;-OMe oligonucleotide complementary to the 5′; end of U1 snRNP did not affect the protection pattern (data not shown). Control reactions in extract alone (Fig. 3A, lanes 12 and 13) or in the presence of wild-type MS2 protein (Fig. 3A, lanes 10 and 11) led to low levels of U2 snRNP protection, similar to the levels obtained with the substrate containing no MS2 binding sites (0× MS2; Fig. 3A, lanes 2 and 3). Thus, the MS2-RS fusion protein, bound to a trans-splicing substrate containing one or several MS2 binding sites, functions to recruit U2 snRNP to the substrate. Furthermore, increasing levels of U2 binding were observed as the number of MS2 binding sites present in the substrate increased.

FIG. 3.

Tethered RS domains recruit U2 snRNP to the branch point but are not sufficient for assembling a catalytically active trans spliceosome. (A) Each (0 to 3×) MS2 binding site-containing substrate was site-specifically labeled between the A and G at the 3′; splice site. Indicated substrates were incubated under splicing conditions in 2-cell stage extract in the presence or absence of a 2′;-OMe oligonucleotide that blocked the branch point interaction region of U2 snRNP. Wild-type MS2 or MS2-RS fusion protein was included in the reactions as indicated. RNA fragments protected from micrococcal nuclease digestion that are dependent upon interaction with U2 snRNP are indicated by the “U2” arrow. (B) trans-splicing substrates containing (0 to 3×) MS2 binding sites were incubated in 2-cell stage extract under splicing conditions for 90 min. Reactions were supplemented with MS2-RS fusion protein, active SR proteins, or both as indicated. The standard trans-splicing substrate, 3′;Acc, was included as a positive control for the rescue of splicing activity in 2-cell stage extract upon the addition of active SR proteins. trans-splicing substrates and products are indicated as described in the legend to Fig. 1. Lanes are derived from multiple gels analyzing reactions from the same experiment. M, markers (pBR322 digested with MspI).

With the U2 snRNP bound to the substrate, the next question was whether the trans-splicing reaction would be able to proceed or whether the reaction would depend upon a separate source of active SR proteins for subsequent catalysis. We performed trans-splicing reactions in the same early developmental, inactive extract that was used in the nuclease protection assays. As previously shown (43), 2-cell stage extracts fail to catalyze trans splicing of a control substrate unless they are supplemented with active SR proteins (Fig. 3B, lanes 2 and 3). Each of the substrates, containing 0 to 3 MS2 binding sites, was subjected to the same reaction conditions. For example, see Fig. 3B, lanes 18 to 21, for the 3× MS2 substrate. Each reaction was incubated for 90 min subsequent to the addition of buffer (lane 18), MS2-RS fusion protein (lane 19), active SR proteins (lane 20), or MS2-RS fusion protein and active SR proteins together (lane 21). Neither the addition of the MS2-RS fusion protein nor that of active SR proteins alone led to an increase in trans-splicing activity beyond the level observed with the addition of buffer alone (compare lanes 19 and 20 to lane 18). Only upon addition of both MS2-RS fusion protein and active SR proteins was trans-splicing activity rescued. Thus, although the MS2-RS fusion protein alone led to U2 snRNP occupancy of the branch site in early (2-cell stage) extracts containing inactive SR proteins (Fig. 3A), this was not sufficient to allow for trans-splicing activity. Therefore, subsequent events in trans-spliceosome assembly, including the addition of the SL and U4/U6.U5 snRNPs, require active SR proteins.

In order to determine whether the trans-splicing activity observed upon the inclusion of an MS2-RS fusion protein and a separate source of SR proteins was indeed indicative of the requirement for these two components, we titrated SR proteins into the reactions. When SR proteins were titrated into trans-splicing reactions performed on the 3× MS2 substrate against buffer alone (Fig. 4, lanes 5 to 9) or against a constant amount of wild-type MS2 protein (Fig. 4, lanes 12 to 15), very little activity was observed. Each of these reaction conditions led to a maximal level of 6% trans-splicing activity over the course of the incubation. In contrast, when SR proteins were titrated against a constant amount of MS2-RS fusion protein, trans-splicing efficiency was seen to increase dramatically (Fig. 4, lanes 17 to 20). In this case, the reaction reached 32% turnover with respect to the levels of product formed relative to the amount of SR proteins added.

FIG. 4.

trans-splicing efficiency increases with increasing amounts of SR proteins, indicating an absolute requirement for SR protein function post-U2 snRNP addition. The 3× MS2 substrate was incubated in 2-cell stage extract under splicing conditions for the indicated times. Reactions were performed with increasing amounts of SR proteins in the presence or absence of a fixed concentration of either wild-type MS2 or MS2-RS fusion protein. The standard trans-splicing substrate, 3′;Acc, was included as a control. trans-splicing substrates and products are indicated as described in the legend to Fig. 1. Lanes are derived from multiple gels analyzing reactions from the same experiment. M, markers (pBR322 digested with MspI).

RS domain phosphorylation is critical for U2 snRNP interaction with the branch point sequence.

We took advantage of the ability of an MS2-RS fusion protein to promote the assembly of a U2 snRNP-containing complex in order to examine the role of RS domain phosphorylation on this activity. We had demonstrated that an MS2-RS fusion protein overexpressed and purified from insect cells was able to function in loading U2 snRNP onto the substrate. Thus, the endogenous phosphatases and kinases in insect cells are capable of activating an overexpressed MS2-RS fusion protein. However, we sought to modify the phosphorylation state of the fusion protein and assess the potential effect on U2 loading. We employed alkaline phosphatase as well as the SR protein kinases SRPK1 (20) and Clk/Sty (10) in order to alter the phosphorylation state of the MS2-RS fusion protein. We had demonstrated previously that the early developmental extracts did not exhibit appreciable kinase or phosphatase activity over the course of incubation under splicing conditions (45). Upon dephosphorylation, the MS2-RS fusion protein exhibited enhanced mobility on SDS-PAGE when compared to the nonmodified control (Fig. 5A, compare lanes 3 and 2, respectively). This change in mobility is consistent with a decrease in the state of phosphorylation.

FIG. 5.

The state of RS domain phosphorylation is critical for U2 snRNP addition to the branch site. (A) In vitro modified (A.P.) and native MS2-RS fusion protein was run on SDS-PAGE and analyzed for gel mobility shift via Coomassie blue staining. M, markers (NEB broad-range prestained protein marker). (B) The trans-splicing substrate 3× MS2, site-specifically labeled between the A and G at the 3′; splice site, was incubated under splicing conditions in 2-cell stage, inactive extract in the presence or absence of a 2′;-OMe oligonucleotide that blocked the branch point interaction region of U2 snRNP. In vitro modified (A.P.), mock-treated, or native MS2-RS fusion protein was included in the reactions as indicated. RNA fragments protected from micrococcal nuclease digestion due to U2 snRNP interaction with the branch point are indicated by the arrow. M, markers (pBR322 digested with MspI).

Aliquots of these same modified MS2-RS fusion proteins were also used in U2 snRNP loading assays employing site-specifically labeled 3× MS2 as the substrate. The level of U2 addition achieved with the active fusion protein in its native state of phosphorylation (as purified from insect cells) is shown in Fig. 5B, lanes 1 and 2, representing reactions carried out in the absence or presence of the 2′;-OMe oligonucleotide that blocks the branch point interaction region of U2 snRNP. In contrast, when the MS2-RS fusion protein was treated with alkaline phosphatase prior to its addition to the reaction mixture, we observed a dramatic reduction in the recruitment of U2 snRNP to the substrate (Fig. 5B, lanes 3 and 4) similar to levels seen following incubation of the substrate in extract alone (Fig. 5B, lanes 8 and 9). Mock-treated reactions in which alkaline phosphatase was diluted into the reaction mixture as opposed to being allowed to act on the MS2-RS fusion protein demonstrated that the effects observed on U2 loading were the result of an alteration of the state of phosphorylation of the fusion protein and not an indirect effect on some other component of the extract (Fig. 5B, lanes 6 and 7). Upon excess phosphorylation with SRPK1, Clk/Sty, or both, the levels of U2 loading were unaffected (data not shown). In addition, we performed RNA binding assays with the modified MS2-RS fusion proteins and a transcript that contained the 1× MS2 exon downstream of the 3′; splice site. By utilizing the His tag on the fusion protein, we examined RNA binding via pullout assays with nickel resin. All of the modified MS2-RS fusion proteins interacted with the 1× MS2-containing RNA indistinguishably (data not shown). Thus, in the context of the MS2-RS fusion protein, alteration of the level of phosphorylation of the RS domain did not affect RNA binding activity. These results directly demonstrate that there is a requirement for RS domain phosphorylation in the initial stage of trans-spliceosome formation, namely the addition of U2 snRNP onto the branch site.

Phosphorylation is required for SR protein activity subsequent to U2 snRNP addition to the branch point.

Up to this point, we had rescued trans-splicing activity upon the addition of total SR proteins purified from an active point in development. We wanted to determine whether a single, overexpressed SR protein would be able to fulfill this same role in allowing the U2-containing complex to proceed through the trans-splicing reaction. In this way, we would be able to assay the activity of a single SR protein at a later stage of trans-spliceosome formation as a function of its state of phosphorylation.

We began by testing whether the full-length SR protein 9G8 was able to allow trans splicing to proceed in conjunction with the MS2-RS fusion protein on the 3× MS2 substrate. Previously, we showed that the MS2-RS fusion protein allowed for U2 snRNP addition to the branch point yet did not allow for trans splicing in extracts where the SR proteins were inactive (Fig. 3). Upon addition of either the MS2-RS fusion protein (Fig. 6A, lane 3) or increasing amounts of purified 9G8 alone (Fig. 6A, lanes 4 to 6), only low levels of trans-spliced product were formed. However, when the same amounts of 9G8 were added to reaction mixtures containing a constant amount of MS2-RS fusion protein, trans-splicing activity increased markedly (Fig. 6A, lanes 7 to 9). Thus, the single SR protein 9G8 is able to fulfill the requirement for SR proteins in trans splicing after U2 snRNP addition to the 3′; acceptor molecule. We then examined the ability of 9G8 to promote this later step in trans splicing as a function of phosphorylation. Purified 9G8 was modified in vitro in a manner similar to that done with the MS2-RS fusion protein. In this case, while the trans-splicing reaction was able to proceed in an MS2-RS fusion protein-dependent manner in the presence of native 9G8, no activity was observed when dephosphorylated 9G8 was added to the reaction mixture (Fig. 6B, compare lanes 5 and 6). A mock control in which the phosphatase was diluted into the reaction as opposed to being preincubated with 9G8 prior to addition demonstrated that modification of the phosphorylation state of 9G8 was responsible for the loss of activity (Fig. 6B, lane 7). Phosphorylation with SRPK1 did not affect the ability of 9G8 to perform this later function (data not shown). Therefore, phosphorylated SR proteins are required for trans splicing after U2 snRNP is already bound to the substrate.

FIG. 6.

SR protein phosphorylation is required for later functions in trans splicing. (A) The trans-splicing substrate 3× MS2 was incubated under splicing conditions in 2-cell stage, inactive extract. Reactions were supplemented with increasing amounts of the SR protein 9G8 in either the absence or the presence of a fixed concentration of the MS2-RS domain fusion protein. trans-splicing substrates and products are indicated as described in the legend to Fig. 1. (B) trans-splicing reactions were performed as described for panel A, with either native or dephosphorylated 9G8. Mock reaction mixtures included native 9G8 and alkaline phosphatase diluted directly into the reaction. M, markers (pBR322 digested with MspI).

DISCUSSION

We have demonstrated that SR proteins are required at a later stage(s) in the formation of the trans spliceosome, beyond the addition of U2 snRNP. It is at this point that other required snRNPs, U4/U6.U5, and the SL RNP harboring the 5′; exon are incorporated. Thus, SR proteins appear critical for the assembly of a complete, catalytically active trans spliceosome. In addition, we investigated the requirement for RS domain phosphorylation both in the addition of U2 snRNP to the initial complex formed on the acceptor RNA and later in the reaction. We determined that RS domain phosphorylation is necessary for both of these activities.

Possible roles for SR proteins throughout the trans-splicing reaction.

Previous work demonstrated that SR proteins are required for the trans-splicing reaction (45) and that they can function to recruit U2 to the branch point of a natural trans-splicing substrate (40). Beyond these studies, there was no information regarding other potential points in the reaction where SR proteins might play a critical role. While the cis-splicing reaction has been more thoroughly examined, questions as to the function(s) of SR proteins and the nature of the factors with which they interact at specific steps remain unanswered. Several of the activities attributed to SR proteins in the cis-splicing reaction are directly relevant to trans splicing. The first of these is the role of SR proteins in bridging splice sites, or splice site communication. This can occur either across an intron or across an exon. Interaction spanning an exon, or exon definition (for a review, see reference 1), occurs when a 5′; splice site downstream of a 3′; splice site influences the ability of complexes to form on that 3′; splice site, and as a result, the splicing activity of the upstream intron is increased. In trans splicing, particularly in nematodes, pre-mRNAs that are trans spliced normally contain internal introns and thus 5′; splice sites downstream of the 3′; splice site used in trans splicing. Recently, U1 snRNP bound to a 5′; splice site-like sequence downstream of the 3′; splice site of a trans acceptor was shown to dramatically enhance the efficiency of trans splicing (2). In fact, this activity depends upon the presence of SR proteins to mediate interactions between the U1 snRNP bound downstream and the interaction of U2 snRNP with the upstream branch site (L. Boukis and J. P. Bruzik, unpublished data).

The ability of SR proteins to mediate communication between 5′; and 3′; splice sites (4, 13, 46, 47, 54), which may be related to the demonstration that they have an exon-independent function (24), could clearly be important in trans splicing where the splice sites originate on separate transcripts. In fact, this activity is an obvious candidate for the role of SR proteins that we have observed, beyond the addition of U2 snRNP to the 3′; acceptor. It is at this point that the half of the trans spliceosome that includes the SL and U4/U6.U5 snRNPs, potentially in the form of a novel tetra-snRNP (33), joins the 3′; splice site complex. In fact, SR proteins have been shown to promote the entry of the U4/U6.U5 tri-snRNP into the cis spliceosome (41), and this multi-snRNP complex may well participate in earlier stages of spliceosome assembly than previously thought (31). Three RS domain-containing proteins of 27, 65, and 110 kDa have been identified that are associated with the U4/U62.U5 tri-snRNP, and at least two of these (65 and 110 kDa) are required for splicing in vitro (12, 30). These factors may participate in the association of the U4/U62.U5 particle with the spliceosome through RS domain interactions. In the trans-splicing reaction, the potential recruitment of a tetra-snRNP or separate complexes containing the SL RNP and the U4/U6.U5 tri-snRNP suggests that SR proteins might be a key component involved in the assembly of a complete trans spliceosome. In addition, if the tetra-snRNP is indeed a preformed complex, then the 5′; splice site in trans splicing (within the SL RNP) may well be defined very early in the reaction with the participation of the U4/U6.U5 tri-snRNP. Thus, two separate complexes containing either the 5′; or the 3′; splice site may be preformed or form very early in the reaction, and SR proteins may be critical for their interaction, especially since they are not tethered by an intron.

The effect of RS domain phosphorylation on SR protein activity.

Many studies have examined the importance of both phosphorylation and dephosphorylation on the activity of SR proteins. In a general sense, both hyperphosphorylation (39, 43) and hypophosphorylation (7, 27, 45, 55) can lead to an inhibition of pre-mRNA splicing. More specifically, phosphorylation of the SR protein ASF/SF2 was shown to be required for activity in the splicing reaction (55). Subsequently, more detailed analyses have demonstrated that while SR protein phosphorylation is required for spliceosome formation, dephosphorylation is necessary for the first transesterification reaction to proceed (7). In fact, dephosphorylation appears to be required for SR protein function in constitutive splicing but not for splicing activation (56). In the present study, we have determined that SR protein phosphorylation is required for the recruitment of U2 snRNP to the 3′; acceptor molecule. In addition, SR protein phosphorylation is also required for their later function(s) in trans splicing.

While an intermediate level of SR protein phosphorylation seems to promote their activity in pre-mRNA splicing in vitro, the situation in vivo may be more complex. There are multiple kinases in the cell that target SR proteins. Along with SRPK1 (20) and Clk/Sty (10), at least two other kinases, including SRPK2 (53) and DNA topoisomerase I (42), have been described. Each of these enzymes can be regulated and also exhibits specificity for distinct phosphorylation sites within RS domains (for an example, see reference 9). Recently, it was demonstrated that the phosphorylation of a single serine in one RS dipeptide (out of a total of eight) within the yeast protein Npl3 had dramatic effects on subcellular localization and RNA binding (15, 58). Finally, phosphorylation might serve as a signal for other posttranslational modifications or interacting factors that functionally block SR protein activity or influence their subcellular localization (6, 44). In fact, although we have previously demonstrated that SR proteins are hyperphosphorylated and inactive early in development, prior to zygotic gene activation, phosphorylation-dependent mechanisms similar to those listed above may be critical in the regulation of SR protein activity.

The results presented here demonstrate that SR proteins are required in at least two capacities in order for the trans-splicing reaction to proceed. The early function, involving the addition of U2 snRNP to the branch point, may be more prevalent in trans splicing than in cis splicing, since there is no opportunity for interactions across an intron. The later function occurs at a point where the bimolecular nature of the reaction is the key feature for the assembly of a complete, splicing-competent trans spliceosome. Both of these activities require RS domain phosphorylation. These results underscore the importance of the RS domain as well as its phosphorylation in trans splicing.