INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Pre-mRNA splicing is a conserved process occurring in a wide variety of eucaryotes with differing exon/intron architectures (reviewed in references 4, 6, 9, 15, 20, and 26). Vertebrates typically have small exons and large introns. Nonmetazoans frequently have the opposite genetic organization, with introns smaller than the minimum permissible for splicing of a vertebrate intron. Drosophila melanogaster possesses a mixture of these two classes of intron sizes (16, 23). In addition, more than half of the small introns in Drosophila are missing a prominent vertebrate splicing signal, the 3' polypyrimidine tract (23). For these reasons, Drosophila provides a model system in which to study potential mechanistic variations operating during recognition of splicing signals.

In the general model of early vertebrate spliceosome complex assembly, U1 snRNP binds to the 5' splice site and U2 snRNP auxiliary factor (U2AF) binds to the 3' polypyrimidine tract, thereby facilitating U2 snRNP interaction with the branch point. Various members of the serine/arginine (SR) family of proteins may participate by promoting or stabilizing these interactions (reviewed in references 13, 22, and 31). This family of proteins may also act as exon or intron bridging factors via their SR-mediated interaction with SR domains on the small subunit of U2AF (U2AF³⁵) and the U1 70K protein (32, 33, 38). SF1, originally discovered as an essential splicing factor in reconstitution assays (19), has also been observed to bind to the branch point (7, 8). In yeast, BBP (branch point bridging protein), the ortholog to SF1, functions as an intron bridging factor via interactions with U1 snRNP-associated proteins and the large subunit of U2AF (U2AF⁶⁵) (1, 2). It is assumed that vertebrate SF1 can play a similar role, although the mammalian equivalents to the yeast U1 snRNP proteins that interact with BBP have not yet been identified. Furthermore, the relationship between bridging by SR proteins and that afforded by SF1 is unclear.

We have previously examined the cis-acting sequences required for efficient splicing of a constitutively spliced small (59-nucleotide [nt]) intron from the D. melanogaster mle gene that lacks a well-defined pyrimidine tract between the branch point and 3' splice site (18, 29). Assembly of initial ATP-dependent spliceosomes (complex A) on the mle intron requires both the 5' and 3' splice sites, suggesting concerted recognition of the entire intron (29). Instead of a classic pyrimidine tract, the mle intron contains two C-rich tracts located between the 5' splice site and branch point that are necessary for efficient splicing of this intron (18). In addition to a requirement for maximal splicing efficiency, the pyrimidine stretches are also necessary for binding of U2AF, interaction of factors with the 5' splice site, and proper assembly of the active spliceosome, suggesting that these sequences affect early assembly events at both ends of this small intron. Interestingly, the upstream C-rich tracts are inhibitory if a classical 3' pyrimidine tract is introduced between the branch point and 3' splice site (18). This observation suggests competing pathways of factor binding to this substrate and also raises the possibility of alternative gene-specific modes of association of constitutive factors with introns.

Here we demonstrate that both U2AF and an SR protein, SRp54, interact with the C-rich tracts in the mle intron. The central location of the pyrimidine tracts, their importance for maximal splicing, and the ability of human SRp54 to interact with U2AF⁶⁵ instead of U2AF³⁵ (37) suggested that the binding of SRp54 to the tracts could replace SF1 in bridging this intron. Immunoprecipitation studies using an antibody specific for SF1 indicated that SF1 did not contact mle precursor RNA unless a pyrimidine tract was introduced downstream of the branch point. Furthermore, antibodies against either SRp54 or U2AF immunoprecipitated both halves of a precleaved mle splicing substrate, suggesting that these factors either directly or indirectly interact with both the 5' and 3' splice sites. We suggest that SRp54 participates in bridging the small mle intron via its ability to bind both the C-rich tracts and the large subunit of U2AF.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Plasmids and oligonucleotides. The following minigenes contained all or part of the first intron from the D. melanogaster mle gene, along with 56 nt of the first exon and 96 nt of the second exon. The wild-type mle substrate contained the entire 59-nt intron. Mle-py1,2, mle+py, and mle-py1,2+py were identical to the wild-type pre-mRNA except for the introduced mutations indicated in Fig. 1, which have been described previously (18). Mutants were constructed by mutagenic PCR. The deletion mutants 5'ss+py and 5'ss-py contained 56 nt of exon 1 plus the first 38 nts of the intron, with wild-type and mutant C-rich tracts, respectively. Mle5' lacked nt +1 to +8 of the intron in the context of the wild-type mle substrate. Mle5'mt contained the 5' splice site nucleotides GGTA (1 to +3) mutated to TTCG. The series of mle intron expansion substrates was constructed by the addition of a NcoI site at either +39 (BP65) or +10 (5'65) of the wild-type intron via PCR. Insertion of 15 or 30 nt of Drosophila troponin T intron sequences (between the TaqI and NlaIII sites in intron 7) into the NcoI site of BP65 or 5'65, respectively, made BP80 and BP95 or 5'80 and 5'95. Readdition of C-rich tracts to the expansion constructs was accomplished by insertion of a fragment from the mle intron (nt +1 to +38) containing both tracts and the natural sequence between them into the NcoI site. Both wild-type (ResWT) and mutant (ResMT) versions of the intron were inserted. The sequences of all plasmids were verified by using a Thermo Sequenase Radiolabeled Terminator Cycle Sequencing kit (Amersham Life Science, Inc.). The A62 substrate was created by deleting +8 to +65 of the MINX splicing substrate (27) to shorten the 120-nt intron to 62 nt.

In vitro spliceosome assembly. Splicing reaction mixtures consisting of 6.5 fmol of radiolabeled substrate, 50% D. melanogaster Schneider 2 (S2) nuclear extract, 1.6 mM MgCl₂, 20 mM phospho-L-arginine, 1.2 mM dithiothreitol, 1.2% polyethylene glycol, and 2 mM ATP were incubated at 22°C as described previously (18). Aliquots were taken at the indicated time points. Spliceosome complexes were analyzed by the addition of heparin to a final concentration of 0.2 mg/ml and electrophoresis on native RNP gels (29).

UV cross-linking and competition. In vitro assembly reaction mixtures were incubated at 22°C for 7 min before the addition of heparin to a final concentration of 2 mg/ml. Mixtures were transferred to ice and UV irradiated immediately for 10 min. Reactions were subsequently digested with RNase A at 37°C for 30 min. Labeled proteins were displayed on sodium dodecyl sulfate-10% polyacrylamide gels and visualized by autoradiography. Competition of cross-linking used unlabeled competitor RNA oligonucleotides, corresponding to either wild-type or mutant C-rich tracts, described above, added at time zero. The amount of competitor added was 0, 50, 100, or 200 pmol.

Immunoprecipitations and immunoblotting. Immunoprecipitations of cross-linked proteins, assembled spliceosomes, and RNA oligonucleotides were performed with antibodies against the 50-kDa subunit of Drosophila U2AF, provided by D. Rio (17), antipeptide antibodies A and B against the human SRp54, provided by N. Chaudhary (12), or anti-SF1 antibodies. The SF1 antibodies were two polyclonal rabbit sera raised against a His-tagged recombinant fragment of human SF1 that includes amino acids 2 to 320, a region of SF1 in which the human and Drosophila proteins are highly homologous (22a). Immunoprecipitation of precleaved radiolabeled pre-mRNA substrates was conducted by incubating the assembly reactions at 22°C for 7 min before the addition of RNase H and a DNA oligonucleotide complementary to nt +7 to +17 of the mle intron. This mixture was then incubated at 30°C for 3 min before the addition of anti-SRp54 or anti-U2AF antibody. All immunoprecipitations were conducted by using protein A-Sepharose (Pharmacia) in a buffer consisting of 20 mM NaCl, 50 mM Tris-HCl (pH 7.5), and 0.05% Nonidet P-40. Immunoprecipitates were washed in this same buffer before elution in protein or RNA sample buffer. For Western blotting, transfer and detection procedures were performed as described in the PolyScreen instruction manual (Dupont).

5' splice site protection. To detect protein interactions at the mle 5' splice site, an oligonucleotide complementary to this region was added in excess along with RNase H after splicing complex assembly had been allowed to proceed for the indicated amount of time. After incubation at 30°C for 20 min to direct cleavage of the substrate, RNA was prepared for electrophoresis on a 5% denaturing polyacrylamide gel.

Cloning of Drosophila SRp54. Drosophila SRp54 cDNA was cloned by obtaining an EST (epitope sequence tag) plasmid (LD 13325) from the Berkley Drosophila Genome Project/HHMI EST Project. This clone was then fully sequenced by using a Thermo Sequenase Radiolabeled Terminator Cycle Sequencing kit (Amersham Life Science), and its amino acid sequence was directly compared to both the human SRp54 amino acid sequence and the entire database, using the Blast+Beauty, Align, or ClustalW computer program.

RT-PCR. RT-PCRs were performed on total RNA isolated from either S2 cells, Drosophila third-instar larvae (male and female), or HeLa cells. The 5' primer used in the reactions containing S2 or larval RNA corresponded to nt 131 to 148 of the Drosophila cDNA (dSRp54 RRM), and the 3' primer was complementary to either nt 683 to 700 (dSRp54 spacer region) or 860 to 877 (dSRp54 antigenic peptide) of the cDNA. Primers used to amplify the HeLa RNA were complementary to nucleotides 183 to 200 (hSRp54 RRM) and nucleotides 708 to 728 (antigenic peptide) of the human SRp54 cDNA. Reverse transcription was performed with 1 µg of both the appropriate RNA and 3' primer, 10 nmol of deoxynucleoside triphosphates and avian myeloblastosis virus RT (Promega) at 50°C for 30 min under standard buffer conditions. To begin the PCR, 1 µg of the 5' primer as well as recombinant Taq polymerase (Perkin-Elmer) was added. The reaction cycled 30 times between 94°C for 40 s, 55°C for 40 s, and 72°C for 1 min, and the reaction product was then subjected to electrophoresis on an ethidium bromide-stained 1% agarose gel.

Nucleotide sequence accession number. The Drosophila SRp54 sequence has been assigned GenBank accession no. AF055719.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

The SR protein SRp54 cross-links to the mle intron, and its binding is dependent on wild-type C-rich tracts located upstream of the branch point. We have previously observed that two short pyrimidine tracts located upstream of the branch point within the small mle intron are necessary for maximal spliceosome assembly and association of U2AF (18). The sequences of these tracts (designated py1 and py2) and available mutations (py1,2, +py, and py1,2+py) are indicated in Fig. 1. Despite the requirement of the tracts for maximal U2AF binding, the tract sequences (CUCCCCC and UUUCCC) do not resemble optimal binding sites for U2AF (10, 28), suggesting the participation of another factor during early pre-mRNA recognition.

View larger version (43K): [in this window] [in a new window]   FIG. 1.   Sequence of the mle intron. (A) Top, diagram of the mle intron depicting the locations of various splicing signals. The two upstream C-rich pyrimidine tracts located between the 5' splice site and branch point are designated py1 and py2. The poor pyrimidine tract located downstream of the branch point is labeled py3. Bottom, sequences of the wild-type and mutant pyrimidine tracts. The introduced point mutations are underlined. Mutation of the upstream tracts is denoted py1,2; improvement of the downstream pyrimidine tract is denoted +py. (B) Other substrates used in immunoprecipitation and/or UV cross-linking experiments. The C-rich tracts are represented by the letter p, while mutations are indicated by an X. (C) Sequences of both the wild-type and mutant RNA oligonucleotides used in competition and immunoprecipitation experiments.

View larger version (81K): [in this window] [in a new window]   FIG. 2.   A 75-kDa protein interacts with the upstream C-rich tracts. (A) UV cross-linking of proteins to the wild-type (WT) and mutant mle introns. The position of a 75-kDa protein (p75) whose cross-linking is dependent on the presence of wild-type upstream pyrimidine tracts is marked by an arrow. The precursor RNAs used are diagrammed in Fig. 1. A second band appearing below p75 in lane 7 and designated p70 in the text is discussed with reference to Fig. 5. Positions of markers here and in all other relevant figures are indicated in kilodaltons. (B) Competition of cross-linking of p75 by an oligonucleotide containing wild-type but not mutant C-rich pyrimidine tracts. Cross-linking of p75 to mle wild-type precursor RNA was monitored in the presence of increasing amounts of RNA oligonucleotides corresponding to either wild-type or mutant C-rich tracts. The amount of competitor used is indicated above the corresponding lane, and the competitor sequences are shown in Fig. 1C. The bands corresponding to p75 and U2AF are indicated, and positions of markers are shown on the right.

View larger version (30K): [in this window] [in a new window]   FIG. 3.   p75 is Drosophila SRp54. (A) Immunoprecipitation of cross-linked proteins with an antipeptide antibody specific for SRp54. Products of UV cross-linking reactions using wild-type (WT) or mutant mle precursor RNAs were immunoprecipitated from S2 nuclear extract with anti-SRp54 antibodies specific for amino acids 209 to 225 of the human protein (antibody A [12]). The positions of Drosophila SRp54 and of the markers are indicated. Here and in all other relevant figures, total X-link (lane 1) indicates all of the proteins that UV cross-link to mle wild-type precursor mRNA. (B) Western blot of Drosophila and HeLa nuclear extracts using the antipeptide anti-SRp54 antibody. Lane 1, Drosophila S2 nuclear extract proteins; lanes 2 and 3, HeLa nuclear extract proteins partially purified on poly(U)-Sepharose; lane 4, HeLa nuclear proteins. The multiple bands presumably reflect differential phosphorylation of the SR domain. Similar results for both blotting and immunoprecipitation were observed with a different anti-SRp54 antipeptide antibody (antibody B [12]). Markers are indicated.

Antibodies against SRp54 and U2AF immunoprecipitate RNA oligonucleotides corresponding to the C-rich tracts. To more fully determine which factors interact with the upstream pyrimidine stretches, further immunoprecipitation experiments were carried out. In these studies, RNA oligonucleotides containing mle intron sequence from nt +19 to +38, either wild type or mutated for the upstream pyrimidine tracts (py1,2 mutant), were incubated in S2 extract and anti-SRp54 or anti-U2AF antibodies were used to immunoprecipitate the radiolabeled oligonucleotides. As can be seen in Fig. 4, the wild-type RNA oligonucleotide was immunoprecipitated with either antibody (lanes 3 and 5). This ability was lost when the C-rich tracts were mutated (lanes 4 and 6). The results from this experiment support the hypothesis that both SRp54 and U2AF interact with the C-rich tracts located between the 5' splice site and branch point in the mle intron.

View larger version (47K): [in this window] [in a new window]   FIG. 4.   Both SRp54 and U2AF interact with the upstream C-rich tracts. RNA oligonucleotides corresponding to either a wild-type (WT; lanes 2, 3, and 5) or mutant (MT; lanes 4 and 6) C-rich tract (diagrammed at the bottom) were incubated in S2 extract for 7 min and then immunoprecipitated with either anti-SRp54 (lanes 3 and 4) or anti-U2AF (lanes 5 and 6) antibody. Immunoprecipitated RNA was displayed on a 10% denaturing polyacrylamide gel. The position of the immunoprecipitated oligonucleotide is indicated. Markers (nucleotide lengths indicated) are in lanes 1 and 7.

The Drosophila homolog of SRp54 is similar to human SRp54 within the RNA binding domain. Only the human version of SRp54 has been reported to date (12). A computer search of the Berkley Drosophila Genome Project revealed an EST clone that showed homology with the human SRp54 RRM region. The clone (LD 13325) was provided, and sequencing revealed that it contained a likely cDNA for Drosophila SRp54. Figure 5A compares the sequences of the human and Drosophila proteins. The RRMs located at the N termini of the two proteins exhibited 60% identity and 83.8% similarity, indicating that the Drosophila and human proteins could exhibit very similar RNA binding specificity. Such strong conservation of the RRM between the human and Drosophila proteins is striking, given the observation that the RRM of human SRp54 shares only 17 to 21% identity with other human SR proteins (37). The SR regions of the two proteins were also similar.

View larger version (32K): [in this window] [in a new window]   FIG. 5.   Drosophila SRp54. (A) The sequence of Drosophila SRp54 compared with the sequence of human SRp54. The N-terminal RRM is underlined; the SR region is in bold, and the peptide epitope for the anti-SRp54 antibody used in this study is boxed in both proteins. (B) Sequence comparison of the central putative RRM in Drosophila SRp54 (dSRp54) compared to a central putative RRM in a candidate SRp54 gene from C. elegans (cSRp54; GenBank accession no. 1813908) and the N-terminal RRMs of the closest RNA binding domains in the sequence database from a U1 snRNP 70K-like protein (3), Drosophila U1 snRNP 70K (d70 kDa), and human U1 snRNP 70K (h70 kDa). Black shading indicates amino acid identity; gray shading depicts conservation.

View larger version (49K): [in this window] [in a new window]   FIG. 6.   RT-PCR analysis of SRp54 RNA in D. melanogaster and human cells reveals only a single SRp54 mRNA species. Whole-cell RNA from S2 cells, Drosophila third-instar larvae (male and female), or HeLa cells was subjected to RT-PCR amplification using primers that flank the region between the N-terminal RRMs and the epitopes of both proteins to detect potential variants within the spacer region. The 5' primers were complementary to a region within the N-terminal RRM domain of each gene. Two sets of 3' primers were used. The first was complementary to nucleotides which encode the peptide against which the anti-SRp54 antibody was raised (labeled P). This region is present in both Drosophila and human SRp54 (dSRp54 and hSRp54) and was used to amplify RNAs from all sources. The expected products are 746 nt (using Drosophila RNA) or 546 nt (using HeLa RNA). The second 3' primer was specific for the Drosophila RNA and was complementary to the central putative RRM domain (marked S). This amplification product is 569 nt. Marker sizes are given in nucleotides, positions of the primers are diagrammed below, and the sizes of the RT-PCR products are indicated at the right.

SF1 does not bind the wild-type mle intron. SF1 has been postulated to be a branch point binding protein that bridges introns via binding to U2AF and U1 snRNP-associated proteins (2, 7, 8). Observation of SRp54 binding to the centrally located C-rich pyrimidine tracts in the mle intron raised the question of whether SRp54 obviated the need for participation of SF1 as a splice site bridging factor for this small intron. To investigate whether SF1 is used during mle splicing, we immunoprecipitated UV cross-linked proteins and in vitro-assembled spliceosomes, using antibodies specific for SF1 that were raised against an N-terminal fragment (amino acids 2 to 320) of human SF1 (5) that is highly conserved between humans and Drosophila (22a). As a positive control, an adenovirus-based substrate known to require SF1 (A62) was used. SF1 did interact with the A62 substrate (Fig. 7A, lane 6), as revealed by an immunoprecipitated, cross-linked band of 70 kDa. In contrast, when mle wild-type pre-mRNA was used in the reaction, no detectable cross-linked and immunoprecipitated band was observed (Fig. 7A, lane 3). Therefore, SF1 does not appear to UV cross-link to the mle intron. When the 3' splice site of the mle intron was strengthened by the addition of a more classical 3' pyrimidine tract downstream of the branch point, SF1 cross-linking was detected (Fig. 7A, lane 5).

View larger version (54K): [in this window] [in a new window]   FIG. 7.   SF1 does not UV bind to the wild-type (WT) mle splicing substrate. (A) UV cross-linking of SF1. The radiolabeled precursor RNAs indicated above the gel were subjected to UV cross-linking as described for Fig. 2. mle precursor RNAs are diagrammed in Fig. 1; the A62 construct is an adenovirus-based substrate that has had its single intron internally deleted to 62 nt (see Materials and Methods). Polyclonal antibodies against human SF1 protein were then added to immunoprecipitate cross-linked proteins. Immunoprecipitated proteins were displayed on a sodium dodecyl sulfate-10% polyacrylamide gel. The band corresponding to SF1 is indicated. Markers are as shown. Pre-imm, preimmune serum. (B) Immunoprecipitation of spliceosomes with anti-SF1 antibodies. The indicated precursor RNAs were incubated under standard splicing conditions in S2 extract for 7 min to permit assembly of complex A. Reaction mixtures were immunoprecipitated with the two anti-SF1 peptide antibodies indicated. Radiolabeled RNAs within the immunoprecipitates were displayed on denaturing urea gels. The amount of immunoprecipitated RNA was quantified in the PhosphorImager; relative density units are indicated below the lanes.

SRp54 and U2AF contact both the 5' and 3' ends of the mle intron. The association of SRp54, but not SF1, with the mle pre-mRNA raised the possibility that either SRp54 or U2AF⁶⁵ bound to the centrally located C-rich elements functions as a bridging factor(s) for this intron. To address this question, immunoprecipitation of precleaved substrates was used. In these experiments, radiolabeled substrates were incubated in S2 extract for a short period of time sufficient for the assembly of complex A but not subsequent assemblies and subjected to RNase H-mediated cleavage using a DNA oligonucleotide complementary to a region between the 5' splice site and C-rich pyrimidine tracts. Following cleavage, antibodies against SRp54 or U2AF were used to immunoprecipitate RNA associated with the factors. If the protein that the antibody recognizes interacts with both sides of the intron, directly or indirectly, then both halves of the cleaved substrate should be immunoprecipitated. If, on the other hand, the factor makes contact with one of the ends, then only that part of the substrate should be present in the immunoprecipitate. Figure 8 (lanes 3 and 6) shows that anti-SRp54 and anti-U2AF antibodies immunoprecipitated both the 3' and the 5' halves of the mle wild-type intron. This ability was lost and the signal dropped to background levels when the upstream elements were mutated (Fig. 8, lanes 4 and 7). These results support the idea of SRp54 and/or U2AF performing a bridging function in the mle intron, such that one or both proteins contact sequences and/or factors bound to the 5' and 3' splice sites. When a substrate that contained only intron sequence from nt +1 to +38 (down to and including the C-rich tracts) was used, both anti-SRp54 and anti-U2AF antibodies were still able to immunoprecipitate the 5' half of the construct. This observation indicates that the association of SRp54 and U2AF with the 5' end of the mle intron is independent of any contribution from splicing signals present in the 3' half of the intron (branch point, 3' pyrimidine tract, 3' splice site, or others). As neither a deletion of the 5' splice site nor a mutation of this signal affected SRp54 cross-linking (Fig. 2, lanes 4 and 5), it is possible that SRp54 and U2AF interact with the upstream C-rich tracts first and then make contact with other factors binding to the 5' splice site.

View larger version (57K): [in this window] [in a new window]   FIG. 8.   SRp54 and U2AF contact both ends of the mle intron. Radiolabeled RNA substrates were incubated in S2 extract for 7 min. Following the addition of RNase H and a DNA oligonucleotide (GCAACATAACC) complementary to the region between the 5' splice site and upstream pyrimidine stretches (nt +7 to +17 of the intron; diagrammed at the bottom), the reaction mixtures were incubated at 30°C for 3 min. They were next subjected to immunoprecipitation using anti-SRp54 or anti-U2AF antibody. Immunoprecipitated RNAs were prepared for electrophoresis on a 10% denaturing polyacrylamide gel. The bands corresponding to precursor RNA or RNAs matching the 5' or 3' halves of the substrate following RNase H-mediated cleavage are indicated. Predicted products from each precursor used are depicted below the gel. Marker sizes are given in nucleotides. WT, wild type.

The C-rich tracts need to be located near both the branch point and the 5' splice site in order for spliceosome formation to proceed. Another possible outcome of a bridging model using centrally located intron signals is that the distance between the interacting cis elements might be critical. To test this possibility, internal expansions were made in the 59-nt mle intron. An NcoI site was introduced either between the C-rich tracts and the branch point (BP65) or between the tracts and the 5' splice site (5'65). The Drosophila troponin T intron sequence was inserted at this site in BP65 or 5'65 to create BP80 and BP95 or 5'80 and 5'95, respectively. These introns have total lengths of 80 or 95 nt, compared to the 59 nt of the wild-type mle intron. The ability of these various splicing substrates to be recognized was then tested in spliceosome assembly assays. Figure 9A depicts the results obtained with the substrates containing an expansion between the upstream pyrimidine-rich elements and the branch point in an assembly assay. Even a moderate expansion of 6 nt began to inhibit complex A formation, while larger insertions of 21 or 36 nt completely abolished all assembly. Figure 9B displays the phenotypes of the 5' splice site-pyrimidine tract expansions. These substrates also lose the ability to form complex A when expanded. Thus, there is a requirement for the C-rich tracts to be in close proximity to both the 5' splice site and branch point for maximal assembly of complex A. 

View larger version (59K): [in this window] [in a new window]   FIG. 9.   Expanding the distance between the C-rich sequences and either the 5' splice site or branch point inhibits spliceosome complex assembly. (A and B) Spliceosome assembly of expanded mle introns. The 59-nt wild-type mle intron [WT (59)] was expanded to a total length of 65, 80, or 95 nt by additions between the upstream pyrimidine tracts and the branch point (A, the BP series) or between the 5' splice site and the pyrimidine tracts (B, the 5' series). All constructs were incubated in S2 extract, aliquots were taken at the indicated time points, and heparin was added to a final concentration of 0.2 mg/ml. The reactions were then displayed on native polyacrylamide gels in order to visualize spliceosome complex assembly. Complexes A and H are indicated. dTroponin T seq, Drosophila troponin T sequence. (C) Spliceosome assembly of the mle intron mutated for the upstream pyrimidine tracts. Assembly of the mle-py1,2 mutant (right) was compared to that of the wild type (left) in a native assembly assay under the conditions described for panel A. (D) Spliceosome assembly of expanded mle introns in which two copies of the wild-type (ResWT) or mutant (ResMT) C-rich tracts were inserted into the expansion cassette. This construct effectively created an expanded intron containing four copies of the C-rich tracts between the 5' splice site and branch point by duplicating the region containing the C-rich tracts.

Expanding the distance between the upstream pyrimidine tracts and either the branch point or 5' splice site depresses factor association with the 5' splice site. The assembly results discussed above indicate that the upstream pyrimidine tracts are necessary for the association of U2 snRNPs with the spliceosome so as to form complex A. To assess the requirement for the upstream tracts on events at the 5' splice site, we analyzed the ability of splicing factors (e.g., U1 snRNP) to bind to and protect the 5' splice site. Expansion of the mle intron on either side of the upstream pyrimidine tracts lowered the ability of factors to protect the 5' splice site against digestion with RNase H and a complementary oligonucleotide (Fig. 10). Expansion upstream of the C-rich tracts severely inhibited binding of factors to the 5' splice site (22% protection compared to the wild type), whereas expansion downstream of the tracts afforded 50% of the protection given by the wild-type substrate. We have previously reported that mutation of the upstream tracts severely inhibited protection of the 5' splice site and that the protection was U1 dependent (18). Therefore, the upstream tracts are necessary for the binding of both U2 and U1 snRNPs to the ends of the intron.

View larger version (56K): [in this window] [in a new window]   FIG. 10.   Association of factors with the 5' splice site is depressed by expanding the distance between the upstream pyrimidine tracts and either the 5' splice site or branch point. To monitor 5' splice site protection, RNase H along with an oligonucleotide complementary to the 5' splice site (nt 4 to +13) was added during an assembly reaction using the depicted expansion constructs. Time points indicate time (minutes) of addition of the oligonucleotide. RNA products were displayed on a 5% denaturing polyacrylamide gel. Protection is indicated by the appearance of a band of the length of precursor RNA and a reduction in the amount of cleavage product. Precursor RNAs are as in Fig. 7. Anticipated cleavage and protection products are diagrammed below the gel. Marker sizes are given in nucleotides. WT, wild type.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Examination of the factors involved in pre-mRNA splicing has revealed an amazing conservation from yeasts to humans. Both snRNPs and other factors are conserved, suggesting that constitutive recognition of splicing signals may be very similar for all introns. U2AF is one such conserved factor participating in one of the earliest steps of recognition by binding to sequences at the 3' end of the intron and facilitating binding of U2 snRNPs to the branch point (17, 30, 35, 36). Another conserved protein is SF1, which binds to the branch point region (5, 7). U2AF and SF1 cooperatively interact to recognize the branch point and pyrimidine tract (8). In the yeast Saccharomyces cerevisiae, genetic interactions have established that SF1 acts to bridge the intron by making contacts with both U2AF and proteins bound to U1 snRNPs (2). Because the U1 snRNP proteins implicated in SF1-mediated bridging have not yet been observed in vertebrates, it is unclear if SF1 plays this role in assembly of the mammalian spliceosome. The conservation between the yeast and mammalian proteins, however, argues that SF1-mediated bridging should be universal.

Intron size and sequence diverges greatly across the eucaryotic kingdom. Typical vertebrate introns are large and contain noticeable pyrimidine tracts in the region between the branch point and 3' splice site (16). These tracts are the binding site for U2AF (35). In contrast, many nonvertebrates have small introns, many of which are so small that they cannot function in a vertebrate system (14, 29). In addition, they frequently lack recognizable pyrimidine tracts downstream of the branch point (23). Here we suggest that one of these introns from the Drosophila mle gene may utilize a mode of early assembly different from that defined for other introns, in either yeasts or vertebrates.

The mle intron has unusual C-rich pyrimidine tracts located in the short interval between the 5' splice site and branch point that are required for maximal spliceosome assembly and the binding of U2AF and the SR protein SRp54. Unlike most SR proteins, SRp54 contacts U2AF via its large subunit (U2AF⁶⁵) rather than its small subunit (U2AF³⁵) (37). Our data suggest that this mode of interaction could potentially serve to create a bridge for small introns, obviating the need for the involvement of SF1. Bridging was revealed by the requirement for the sequence and location of the C-rich tracts for the association of U1 and U2 snRNPs, as well as the ability of SRp54 and U2AF to contact both the 5' and 3' splice sites, either directly or indirectly. This bridging appeared to operate in the absence of SF1 because, in contrast to spliceosomes formed with precursor RNAs with normal 3' pyrimidine tracts, mle spliceosomes could not be immunoprecipitated with anti-SF1 antibodies.

The mle intron lacks a classical pyrimidine tract between the branch point and 3' splice site. When such a site was added, the intron functioned; but in this context, the upstream C-rich tracts were inhibitory for assembly and activity (18), suggesting alternative competitive modes of early recognition. Addition of the downstream tracts was accompanied by the ability to detect 10- to 15-fold higher levels of immunoprecipitation of the mle spliceosome with antibodies specific for SF1, suggesting that SF1 contacts the mle intron only in the presence of downstream tracts. Furthermore, these results suggest that simultaneous association of SRp54 and SF1 with the intron is not optimal for assembly or activity.

Taken together, these observations suggest that when introns are very small, U2AF and SRp54 may function to provide early recognition and intron bridging. SR proteins have been observed to provide both cross-exon and cross-intron interactions in a number of vertebrate pre-mRNAs (reviewed in references 9, 13, 22, and 25). Models for these interactions invoke associations with U2AF that involve the SR domain of the small subunit of U2AF. In contrast, SF1 bridging models invoke an interactions with the large subunit of U2AF (2). The relationship between the two modes of bridging has been unclear.

Our results suggest that individual introns may differ in the mechanism of bridging employed and raise the possibility of gene-specific early assembly. In S. cerevisiae, branch points are well conserved but pyrimidine tracts are relatively rare and play only a weak role in intron recognition. Accordingly, MUD2 (the gene encoding the U2AF⁶⁵ homolog) is inessential for viability; instead SF1 (BBP) plays a pivotal role (2). In contrast, in vertebrates, U2AF is required and binds to a prominent splicing signal, the 3' polypyrimidine tract (36) and also contacts the branch site (30). Bridging is afforded by SR proteins and presumably by SF1. Our data suggest a third scenario operating in small introns such as those in Drosophila or C. elegans. In these cases, U2AF and SRp54 function to bind to pyrimidine tracts in the vicinity of the branch point, thereby providing intron bridging. One scenario for this bridging is that SRp54, via its ability to bind the upstream pyrimidine tracts in the mle intron, facilitates U2AF binding upstream of the branch point and positions the small subunit of U2AF to make optimal interactions with the U1 snRNP 70K protein (Fig. 11A). This model presents the mle intron as a recognition unit equivalent to the SR-mediated recognition units postulated in exon definition in vertebrates.

View larger version (49K): [in this window] [in a new window]   FIG. 11.   Suggested models for participation of U2AF and SRp54 in splice site bridging. In model A, the SR-rich small subunit of Drosophila U2AF (dU2AF) contacts the U1 snRNP 70K protein to affect interactions across a small intron. Use of pyrimidine tracts upstream of the branch point is postulated to position U2AF such that participation of SF1 or other normal bridging proteins is unnecessary. In model B, the SR protein SRp54 is used to make contacts with factors bound to the 5' splice site, whereas the small subunit of U2AF is used to bridge the downstream exon. This model suggests how both intron and exon definition could operate in adjacent areas of a single pre-mRNA.

Also possible, however, is an alternative mechanism whereby U2AF participates in simultaneous interactions across a neighboring exon and intron. In recent years, both intron and exon bridging models for recognition of splice sites in multiexon RNAs have been proposed (1, 2, 6, 9, 25, 27, 38). In general, organisms with small introns are thought to operate via intron bridging, whereas in organisms with large introns, exon bridging is proposed. D. melanogaster presents an example of an organism with a split exon/intron organization such that small and large introns are mixed within individual transcription units. The mle gene is one such split unit. The intron studied in this report is quite small; the adjacent intron is considerable larger. Such mixed architecture presents problems for models of splice site pairing that use either consistent exon or intron bridging. The results in this study suggest a mechanism whereby both could occur simultaneously (Fig. 11B). As this model depicts, a small upstream intron using a bridging factor such as SRp54 or SF1 that contacts the large subunit of U2AF would permit simultaneous interaction of the small subunit of U2AF with factors binding the downstream exon.

As part of this study, we sequenced the Drosophila homolog of SRp54. While considerable conservation was found within both the RRM and SR domains, the Drosophila protein contained extra sequence within the spacer region. Analysis of this sequence revealed that it resembled an RRM. In fact, computer searches attempting to find motifs similar to this sequence yielded the RRM regions of both the human and Drosophila U1 snRNP 70-kDa proteins. The proposed C. elegans SRp54 also contains this second RRM domain, suggesting that in these nonvertebrates, the protein may possess some unique functions. The exact role of this second RRM is unknown. It could affect the affinity and specificity of the first RRM, it could function in protein-protein interactions, or it could perform a completely unique activity related to small-intron recognition (11, 21, 24, 34). Experiments comparing and contrasting the Drosophila and human proteins should lend insight into the function of this protein.