Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

Human Splicing Factor ASF/SF2 Encodes for a Repressor Domain Required for Its Inhibitory Activity on Pre-mRNA Splicing

2002; Elsevier BV; Volume: 277; Issue: 15 Linguagem: Inglês

10.1074/jbc.m107867200

ISSN

1083-351X

Autores

Vita Dauksaite, Göran Akusjärvi,

Tópico(s)

RNA modifications and cancer

Resumo

The essential splicing factor ASF/SF2 activates or represses splicing depending on where on the pre-mRNA it binds. We have shown previously that ASF/SF2 inhibits adenovirus IIIa pre-mRNA splicing by binding to an intronic repressor element. Here we used MS2-ASF/SF2 fusion proteins to show that the second RNA binding domain (RBD2) is both necessary and sufficient for the splicing repressor function of ASF/SF2. Furthermore, we show that the completely conserved SWQDLKD motif in ASF/SF2-RBD2 is essential for splicing repression. Importantly, this heptapeptide motif is unlikely to be directly involved in RNA binding given its position within the predicted structure of RBD2. The activity of the ASF/SF2-RBD2 domain in splicing was position-dependent. Thus, tethering RBD2 to the IIIa intron resulted in splicing repression, whereas RBD2 binding at the second exon had no effect on IIIa splicing. The splicing repressor activity of RBD2 was not unique to the IIIa pre-mRNA, as binding of RBD2 at an intronic position in the rabbit β-globin pre-mRNA also resulted in splicing inhibition. Taken together, our results suggest that ASF/SF2 encode distinct domains responsible for its function as a splicing enhancer or splicing repressor protein. The essential splicing factor ASF/SF2 activates or represses splicing depending on where on the pre-mRNA it binds. We have shown previously that ASF/SF2 inhibits adenovirus IIIa pre-mRNA splicing by binding to an intronic repressor element. Here we used MS2-ASF/SF2 fusion proteins to show that the second RNA binding domain (RBD2) is both necessary and sufficient for the splicing repressor function of ASF/SF2. Furthermore, we show that the completely conserved SWQDLKD motif in ASF/SF2-RBD2 is essential for splicing repression. Importantly, this heptapeptide motif is unlikely to be directly involved in RNA binding given its position within the predicted structure of RBD2. The activity of the ASF/SF2-RBD2 domain in splicing was position-dependent. Thus, tethering RBD2 to the IIIa intron resulted in splicing repression, whereas RBD2 binding at the second exon had no effect on IIIa splicing. The splicing repressor activity of RBD2 was not unique to the IIIa pre-mRNA, as binding of RBD2 at an intronic position in the rabbit β-globin pre-mRNA also resulted in splicing inhibition. Taken together, our results suggest that ASF/SF2 encode distinct domains responsible for its function as a splicing enhancer or splicing repressor protein. small nuclear ribonucleoprotein RNA binding domains amino acids IIIa repressor element HeLa nuclear extracts The human ASF/SF2 protein is a member of the evolutionary conserved SR family of splicing factors (1.Zahler A.M. Lane W.S. Stolk J.A. Roth M.B. Genes Dev. 1992; 6: 837-847Crossref PubMed Scopus (632) Google Scholar). SR proteins are essential splicing factors required for constitutive splicing and, furthermore, play a regulatory role in alternative RNA splicing (reviewed in Refs.2.Blencowe B.J. Trends Biochem. Sci. 2000; 25: 106-110Abstract Full Text Full Text PDF PubMed Scopus (541) Google Scholar, 3.Graveley B.R. RNA (New York). 2000; 6: 1197-1211Crossref PubMed Scopus (886) Google Scholar, 4.Smith C.W. Valcarcel J. Trends Biochem. Sci. 2000; 25: 381-388Abstract Full Text Full Text PDF PubMed Scopus (764) Google Scholar). SR proteins partake at multiple steps during spliceosome assembly. For example, they help to commit a pre-mRNA for splicing by stabilizing U1 snRNP1 and U2AF binding to the 5′ and 3′ splice sites, respectively (5.Jamison S.F. Pasman Z. Wang J. Will C. Luhrmann R. Manley J.L. Garcia-Blanco M.A. Nucleic Acids Res. 1995; 23: 3260-3267Crossref PubMed Scopus (100) Google Scholar, 6.Kohtz J.D. Jamison S.F. Will C.L. Zuo P. Luhrmann R. Garcia-Blanco M.A. Manley J.L. Nature. 1994; 368: 119-124Crossref PubMed Scopus (543) Google Scholar, 7.Staknis D. Reed R. Mol. Cell. Biol. 1994; 14: 7670-7682Crossref PubMed Scopus (249) Google Scholar, 8.Wu J.Y. Maniatis T. Cell. 1993; 75: 1061-1070Abstract Full Text PDF PubMed Scopus (640) Google Scholar). Also, they facilitate U4/U6-U5 snRNP recruitment to the spliceosome and are believed to bridge between splicing factors binding to the ends of the intron during spliceosome formation (9.Chew S.L. Liu H.X. Mayeda A. Krainer A.R. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 10655-10660Crossref PubMed Scopus (46) Google Scholar, 10.Roscigno R.F. Garcia-Blanco M.A. RNA (New York). 1995; 1: 692-706PubMed Google Scholar, 11.Tarn W.Y. Steitz J.A. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 2504-2508Crossref PubMed Scopus (76) Google Scholar). SR proteins contain one or two N-terminal RNP-type RNA binding domains (RBDs) and a highly charged variable-length C-terminal domain rich in arginine-serine dipeptide repeats (the RS domain), hence the name SR proteins (reviewed in Ref. 3.Graveley B.R. RNA (New York). 2000; 6: 1197-1211Crossref PubMed Scopus (886) Google Scholar). ASF/SF2 contain two RBDs, and both are required for high affinity RNA binding (12.Caceres J.F. Krainer A.R. EMBO J. 1993; 12: 4715-4726Crossref PubMed Scopus (222) Google Scholar, 13.Zuo P. Manley J.L. EMBO J. 1993; 12: 4727-4737Crossref PubMed Scopus (161) Google Scholar). SR proteins are modular in structure with the RBDs making sequence-specific contact with the RNA (14.Tacke R. Manley J.L. EMBO J. 1995; 14: 3540-3551Crossref PubMed Scopus (305) Google Scholar, 15.Shi H. Hoffman B.E. Lis J.T. Mol. Cell. Biol. 1997; 17: 2649-2657Crossref PubMed Scopus (78) Google Scholar) and the RS domain mediating protein-protein interactions with multiple general splicing factors (3.Graveley B.R. RNA (New York). 2000; 6: 1197-1211Crossref PubMed Scopus (886) Google Scholar, 6.Kohtz J.D. Jamison S.F. Will C.L. Zuo P. Luhrmann R. Garcia-Blanco M.A. Manley J.L. Nature. 1994; 368: 119-124Crossref PubMed Scopus (543) Google Scholar, 8.Wu J.Y. Maniatis T. Cell. 1993; 75: 1061-1070Abstract Full Text PDF PubMed Scopus (640) Google Scholar, 16.Xiao S.H. Manley J.L. Genes Dev. 1997; 11: 334-344Crossref PubMed Scopus (314) Google Scholar). Previous studies (3.Graveley B.R. RNA (New York). 2000; 6: 1197-1211Crossref PubMed Scopus (886) Google Scholar) have shown that SR proteins function as enhancer proteins, stimulating both constitutive and alternative splicing. SR proteins bind to splicing enhancer elements, which typically are located downstream of the affected intron (reviewed in Ref. 2.Blencowe B.J. Trends Biochem. Sci. 2000; 25: 106-110Abstract Full Text Full Text PDF PubMed Scopus (541) Google Scholar). SR protein binding to enhancer elements helps to recruit U2AF65 to a nearby weak 3′ splice site (8.Wu J.Y. Maniatis T. Cell. 1993; 75: 1061-1070Abstract Full Text PDF PubMed Scopus (640) Google Scholar, 17.Wang Z. Hoffmann H.M. Grabowski P.J. RNA (New York). 1995; 1: 21-35PubMed Google Scholar, 18.Zuo P. Maniatis T. Genes Dev. 1996; 10: 1356-1368Crossref PubMed Scopus (251) Google Scholar). Thus, the RS domain of ASF/SF2 appears to be required for U2AF recruitment to weak polypyrimidine tracts, whereas the RS domain is dispensable for ASF/SF2 activation of strong 3′ splice sites that bind U2AF efficiently (19.Zhu J. Krainer A.R. Genes Dev. 2000; 14: 3166-3178Crossref PubMed Scopus (96) Google Scholar). Interestingly, this study also suggested that the ASF/SF2-RBDs provide a specific splicing enhancer function and are not simply required for RNA binding. Collectively, available data suggest that the RBDs and the RS domains in SR proteins partake at distinct steps in splicing activation; the RS domain is sufficient for the second exon splicing enhancer-dependent function (reviewed in Ref. 3.Graveley B.R. RNA (New York). 2000; 6: 1197-1211Crossref PubMed Scopus (886) Google Scholar), whereas the RBDs provide a distinct enhancer function besides U2AF recruitment (19.Zhu J. Krainer A.R. Genes Dev. 2000; 14: 3166-3178Crossref PubMed Scopus (96) Google Scholar). In addition to functioning as splicing enhancer proteins, SR proteins also have the capacity to repress splicing (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar, 21.Cook C.R. McNally M.T. Virology. 1998; 242: 211-220Crossref PubMed Scopus (24) Google Scholar). We have shown previously that SR proteins inhibit splicing of the regulated adenovirus L1 IIIa pre-mRNA (reviewed in Ref. 22.Akusjärvi G. Recent Res. Dev. Virol. 1999; 1: 621-630Google Scholar) by binding to the so-called IIIa repressor element (3RE), located immediately upstream of the IIIa branch site (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar). Mechanistic studies showed that SR proteins inhibit IIIa splicing by preventing U2 snRNP recruitment to the spliceosome (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar, 23.Petersen-Mahrt S.K. Estmer C. Öhrmalm C. Matthews D.A. Russell W.C. Akusjärvi G. EMBO J. 1999; 18: 1014-1024Crossref PubMed Scopus (141) Google Scholar). The inhibitory effect of the 3RE on IIIa splicing is simply due to its high affinity for SR proteins. Thus, replacing the 3RE with consensus binding sites for ASF/SF2 resulted in splicing repression (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar). Also, moving the 3RE to the second exon in the IIIa pre-mRNA converted the 3RE from a repressor element to a classical splicing enhancer element (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar). Collectively, our results suggested that they either activate or repress splicing, dependent on where in the pre-mRNA SR proteins bind. In our previous study (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar) we noted that a truncated ASF/SF2 protein, lacking the RS domain, was capable of inhibiting IIIa splicing. We proposed that binding of SR proteins to the 3RE blocked IIIa splicing by sterically interfering with U2 snRNP recruitment to the IIIa branch site. Here we tested this model. The results strongly argue against this simplistic model. Thus, an MS2-LacZ fusion protein, with a size exceeding that of MS2-ASF/SF2, did not interfere with IIIa splicing. We further show that ASF/SF2 encodes for a specific splicing repressor domain required for its inhibitory effect on IIIa pre-mRNA splicing. Thus, we show that the second RNA binding domain (RBD2) of ASF/SF2 is both necessary and sufficient for ASF/SF2-mediated repression of IIIa splicing. Furthermore, we show that the SWQDLKD motif conserved in all SR proteins containing an RBD2 is required for ASF/SF2-RBD2 repression of splicing. Our results also confirm that the ASF/SF2-RS domain functions as a splicing enhancer domain (24.Graveley B.R. Maniatis T. Mol. Cell. 1998; 1: 765-771Abstract Full Text Full Text PDF PubMed Scopus (169) Google Scholar), although its activity was very weak under our experimental conditions. However, our results further suggest that RBD2 may augment the splicing enhancer function of the RS domain. Interestingly, the functions of RBD2 and the RS domains in splicing regulation are opposite and position-dependent. Thus, RBD2 inhibits IIIa splicing when tethered to the IIIa intron but has no effect when bound to the second exon. In contrast, the ASF/SF2-RS domain activates IIIa splicing weakly when tethered to the IIIa second exon but has no effect on splicing when bound at an intronic position. Importantly, the repressor activity of ASF/SF2-RBD2 was not unique to the IIIa pre-mRNA. Thus, tethering ASF/SF2-RBD2 to the intron of the rabbit β-globin pre-mRNA similarly results in splicing inhibition, suggesting that the repressor activity of RBD2 may be a general feature of ASF/SF2. The gene encoding the MS2 coat protein was cloned by reverse transcription of the genomic MS2 RNA followed by PCR amplification using primers 5′-GTGACATATGGCTTCTAACTTTAC-3′ and 5′-CGGGATCCGTAGTAGCCGGAGTTTGCTGC-3′. The PCR product was cloned as an NdeI-BamHI fragment into the pET15b vector (Novagen), generating plasmid pET15b-H6MS2 (N-terminal 6-His tag). The ASF/SF2 gene was excised from plasmid pAdG5Trip(His)-ASF/SF2 (25.Molin M. Akusjärvi G. J. Virol. 2000; 74: 9002-9009Crossref PubMed Scopus (32) Google Scholar) as a BamHI fragment and cloned into the BamHI site in pET15b-H6MS2, generating plasmid pET15b-H6MS2-ASF/SF2. Fragments encoding different domains of ASF/SF2 (Fig. 1A) were generated by cleavage with suitable restriction enzymes; 5′-ends were converted to BamHI sites by linker addition and ligated into the BamHI site in pET15b-H6MS2. Substitution mutants of MS2-RBD2+RS (Fig. 6A) were constructed by standard PCR mutagenesis using oligonucleotide primers with the desired amino acid changes (details available upon request). The C-terminal part of the lacZ gene, aa 711–1023, was taken from plasmid pCH110 (Amersham Biosciences AB) and cloned as a BamHI fragment into pET15b-H6MS2. The gene encoding SRPK1 (26.Gui J.F. Lane W.S. Fu X.D. Nature. 1994; 369: 678-682Crossref PubMed Scopus (463) Google Scholar) was cloned into pET24a (Novagen) to create pET24a-T7-SRPK1 (N-terminal T7 tag). Sequences of all hybrid proteins or junction points were confirmed by DNA sequencing.Figure 6The conserved SWQDLKD motif contained in the putative α-helix 1 of ASF/SF2-RBD2 is essential for splicing repression.A, schematic picture showing the structure of MS2-RBD2+RS and MS2-RBD2+RS mutant proteins. The amino acid sequence of RBD2 is shown expanded, and the residues changed in each mutant protein are indicated. The completely conserved SWQDLKD motif is shown boxed in the sequence.B, splicing reactions were carried out in HeLa-NE using IIIa-MS2I pre-mRNA (depicted at the top) minus/plus 4 pmol of MS2-RBD2+RS (wild type, (wt)) or mutant proteins. Products were resolved by gel electrophoresis and visualized by autoradiography. Positions of pre-mRNA, splicing intermediates, and splicing products are indicated on the left.View Large Image Figure ViewerDownload Hi-res image Download (PPT) To generate the IIIa-MS2I and IIIa-MS2I(inv) constructs, two self-complementary oligonucleotides (5′-GATCTCCGGTTGAGGATCACCCAACCGGT-3′ and 5′-GATCACCGGTTGGGTGATCCTCAACCGGA-3′), encoding one MS2 operator sequence (27.SenGupta D.J. Zhang B. Kraemer B. Pochart P. Fields S. Wickens M. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 8496-8501Crossref PubMed Scopus (439) Google Scholar), were annealed, the ends filled in with the Klenow polymerase and ligated in the direct or inverted orientations into the blunt-ended EagI site in plasmid pGDIIIaβ(Eag-Sca) (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar). In IIIa-MS2(−15) the same oligonucleotide was ligated into the BglII site positioned just upstream of the IIIa branch site in pGDIIIaβ(Eag-Sca). To generate the IIIa-MS2E and IIIa-MS2E(inv) constructs, the same double-stranded oligonucleotide was ligated into the blunt-ended MluI site in pGDIIIaβ(Eag-Sca). Plasmid glob-MS2I was generated by PCR amplification, using as an upstream primer 5′-GAAGATCTCCGGTTGAGGATCACCCAACCGGTGATCAAGAAACTG-3′ (which positions an MS2 operator site 53 nucleotides upstream of the β-globin branch site) and downstream primer 5′-CAAGGGTCCCCAAACTCA-3′. The PCR product was cloned as a BglII/BamHI fragment into plasmid pSPβ(IIIaHindIII/Sca)β (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar). IIIa-3RE(−53) was constructed by introducing a double-stranded oligonucleotide encoding the 49-nucleotide-long 3RE (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar) into at the Eagsite in plasmid pGDIIIaβ(Eag-Sca). The relevant sequences in all constructs were confirmed by DNA sequencing. Capped and32P-labeled pre-mRNAs were made by run-off transcription from purified PCR products with T7 RNA polymerase, as described previously (28.Mühlemann O. Akusjärvi G. Wold W.S.M. Methods in Molecular Medicine. 21. Humana Press Inc., Totowa, NJ1998: 203-216Google Scholar). MS2 hybrid proteins were coexpressed in Escherichia coli BL21(DE3) (Novagen) together with SRPK1, as described previously (29.Yue B.G. Ajuh P. Akusjärvi G. Lamond A.I. Kreivi J.P. Nucleic Acids Res. 2000; 28: e14Crossref PubMed Scopus (31) Google Scholar). His-tagged proteins were purified under native conditions by standard nickel column chromatography, as described by the manufacturer (Novagen). The eluted proteins were dialyzed against 20 mm HEPES (pH 7.9), 100 mm KCl, 20% glycerol, 0.2 mm EDTA, and stored at −70 °C. The final protein concentration was determined by the Bradford method (30.Ausubel F.M. Brent R. Kingston R.E. Moore D.D. Seidman J.G. Smith J.A. Struhl K. Current Protocols in Molecular Biology. Wiley & Sons, Inc., New York1995: 10.1.4-10.1.5JohnGoogle Scholar), with bovine serum albumin as a standard. Native SR proteins were purified from HeLa cells as described previously (1.Zahler A.M. Lane W.S. Stolk J.A. Roth M.B. Genes Dev. 1992; 6: 837-847Crossref PubMed Scopus (632) Google Scholar,31.Kanopka A. Mühlemann O. Petersen-Mahrt S. Estmer C. Öhrmalm C. Akusjärvi G. Nature. 1998; 393: 185-187Crossref PubMed Scopus (169) Google Scholar). HeLa nuclear extracts (HeLa-NE) were prepared and in vitro splicing reactions performed as described previously (28.Mühlemann O. Akusjärvi G. Wold W.S.M. Methods in Molecular Medicine. 21. Humana Press Inc., Totowa, NJ1998: 203-216Google Scholar). Briefly, splicing reactions contained 15–20% of HeLa-NE, 2.6% polyvinyl alcohol, 20 mmcreatine phosphate, 2 mm ATP, 12% glycerol, 12 mm HEPES (pH 7.9), 60 mm KCl, and 15 fmol of32P-labeled pre-mRNA in a total volume 25 μl. The final MgCl2 concentration varied between 2.5 and 3.5 mm, depending on the substrate (3.2 mm for IIIa-MS2I and IIIa-MS2I(inv) transcripts; 3.5 mm for glob-MS2I transcript; 2.5 mm for IIIa-MS2E and IIIa-MS2E(inv) transcripts). The indicated amounts (see figure legends) of MS2 hybrid proteins were preincubated for 10 min at 30 °C with the pre-mRNA in a reaction mixture containing all components except the nuclear extract. Splicing was initiated by addition of HeLa-NE and reaction mixtures incubated at 30 °C for 1–3 h. Reaction products were resolved on denaturing 8% polyacrylamide denaturing gels and visualized by autoradiography. The identity of spliced products and intermediates were assigned based on size. Dried gels were subjected to PhosphorImager quantification as described previously (23.Petersen-Mahrt S.K. Estmer C. Öhrmalm C. Matthews D.A. Russell W.C. Akusjärvi G. EMBO J. 1999; 18: 1014-1024Crossref PubMed Scopus (141) Google Scholar). Quantitative data given in the text were based on three, or more, independent experiments. Short 32P-labeled RNA transcripts containing the MS2 operator site in a direct or inverted orientation were generated by T7 transcription (28.Mühlemann O. Akusjärvi G. Wold W.S.M. Methods in Molecular Medicine. 21. Humana Press Inc., Totowa, NJ1998: 203-216Google Scholar). Binding conditions and gel analysis was essentially as described previously (24.Graveley B.R. Maniatis T. Mol. Cell. 1998; 1: 765-771Abstract Full Text Full Text PDF PubMed Scopus (169) Google Scholar). Briefly, 15 fmol of 32P-labeled RNA, 0.8 μg of bovine serum albumin, 0.25 μg of tRNA, 3.2 mm MgCl2, 1.3% polyvinyl alcohol, 1 mm dithiothreitol, 6 μl of buffer D or MS2-ASF hybrid protein in buffer D (1 pmol) was incubated for 10 min at 30 °C in a 10-μl reaction mixture, and heparin was added (final concentration 0.5 μg/μl) and resolved on a 3.5–0.4% agarose native gel (60:1 acrylamide/bisacrylamide). The gel was run in a cold room at 350 V for 3 h in a 75 mm Tris glycine buffer. We showed previously (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar) that the ASF/SF2ΔRS protein efficiently represses IIIa splicing. The aim of this study was to determine whether a specific sequence element within the RBDs was necessary for the splicing repressor phenotype of ASF/SF2. However, mutating the ASF/SF2-RBDs would be expected to impair the RNA binding capacity of the ASF/SF2 mutant proteins. We therefore reconstructed the ASF/SF2 deletion mutants as fusion proteins with the MS2 coat protein (Fig. 1A), which binds to a well defined RNA stem-loop structure, the MS2 operator (32.Uhlenbeck O.C. Carey J. Romaniuk P.J. Lowary P.T. Beckett D. J. Biomol. Struct. Dyn. 1983; 1: 539-552Crossref PubMed Scopus (54) Google Scholar). However, replacing the 3RE in the IIIa pre-mRNA with an MS2 operator site positioned 15 nucleotides upstream of the IIIa branch site (IIIa-MS2(−15), Fig. 2A) resulted in inhibition of constitutive IIIa splicing (Fig. 2B, lane 1), most likely because the MS2 site sequesters the so called anchoring sequence as a double-stranded RNA (see "Discussion"). In agreement with this hypothesis inverting the MS2 operator site resulted in a restoration of constitutive IIIa splicing (Fig. 2B,lane 2). To overcome this negative effect of the MS operator a modified IIIa reporter construct was generated by moving the MS2 operator site to a position located 53 nucleotides upstream of the IIIa branch site (IIIa-MS2I, Fig. 2A). At this position the MS2 operator does not interfere with basal IIIa splicing and functions as an MS2-ASF/SF2-responsive splicing repressor element (see below). However, since we shifted the position of the repressor element in the IIIa-MS2I pre-mRNA, it became important to show that the 3RE also functions as splicing repressor element when moved further away from the IIIa branch site. We therefore constructed a control transcript in which the 3RE was moved from its natural position at −6 relative to the IIIa branch site to position −53 (IIIa-3RE(−53), Fig. 2A). As shown in Fig. 2C, addition of purified HeLa SR proteins efficiently repressed IIIa-3RE(−53) splicing (lanes 1–4). This inhibition was specific since SR proteins did not inhibit splicing of the IIIa-MS2I transcript (lanes 5–8), here used as a control. We conclude that the 3RE, also at a distance from the branch site, functions as a splicing repressor element (see also "Discussion"). MS2-ASF/SF2 hybrid proteins were expressed and purified from E. coli as His6-tagged proteins together with SR protein kinase 1 (SRPK1), as described previously (29.Yue B.G. Ajuh P. Akusjärvi G. Lamond A.I. Kreivi J.P. Nucleic Acids Res. 2000; 28: e14Crossref PubMed Scopus (31) Google Scholar). As shown in Fig. 3A, coexpression with SRPK1 was necessary for the MS2-ASF/SF2 protein to function as a splicing repressor protein. Thus, addition of increasing amounts of an unphosphorylated MS2-ASF/SF2 protein did not inhibit IIIa-MS2I splicing (Fig. 3A, lanes 1–5). In contrast, addition of the same molar amount of an MS2-ASF/SF2 protein purified from SRPK1 coexpressing cells resulted in a more than 5-fold repression of IIIa splicing (Fig. 3A, lanes 6–10). Unexpectedly, SRPK1 coexpression resulted in an increase in the RNA binding capacity of the MS2 fusion proteins (data not shown). Thus, the lack of activity of the unphosphorylated MS2-ASF/SF2 fusion protein appears to result from a failure to bind efficiently to the MS2 operator. Although SRPK1 is not expected to phosphorylate all MS2-ASF/SF2 hybrid proteins used in this study (Fig. 1A), we chose to purify all hybrid proteins from SRPK1-coexpressing cells. Fig. 1B shows a Coomassie-stained gel of typical batches of MS2-ASF/SF2 hybrid proteins used in the experiments presented here. As shown in Fig. 4, the inhibitory effect of MS2-ASF/SF2 requires an MS2 operator site. Thus, using a IIIa-MS2I transcript with an inverted MS2 operator site IIIa-MS2I(inv) completely annulled the inhibitory effect of MS2-ASF/SF2 on IIIa splicing (Fig. 4,lanes 9–12). This loss of splicing inhibition results from the fact that the MS2-ASF/SF2 protein does not bind to the inverted MS2 operator site (Fig. 3B, lane 9). Fig. 3B further shows that all tested MS2-ASF/SF2 hybrid proteins bind efficiently (lanes 1–8) and specifically (lanes 9–15) to the MS2 operator site. The appearance of multiple complexes most likely reflects the capacity of the MS2 protein to multimerize (see Ref. 33.Valegard K. Murray J.B. Stockley P.G. Stonehouse N.J. Liljas L. Nature. 1994; 371: 623-626Crossref PubMed Scopus (324) Google Scholar). It is interesting to note that the complexes formed were not retarded in migration, as one would have predicted from the relative size of the MS2 fusion proteins (Fig. 1B), a result suggesting conformational variations. We have shown previously (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar, 23.Petersen-Mahrt S.K. Estmer C. Öhrmalm C. Matthews D.A. Russell W.C. Akusjärvi G. EMBO J. 1999; 18: 1014-1024Crossref PubMed Scopus (141) Google Scholar) that ASF/SF2 inhibits IIIa splicing by preventing U2 snRNP recruitment to the IIIa branch site. Since the ASF/SF2ΔRS protein also inhibited IIIa splicing, we determined whether ASF/SF2 inhibition of splicing resulted from a steric interference. As shown in Fig. 4, the wild type MS2 coat protein did not interfere with IIIa-MS2I splicing (lanes 1–4), arguing that protein interaction at position −53 in the IIIa intron is not enough to cause splicing repression. Potentially, MS2-ASF/SF2 inhibits IIIa splicing because binding of a bulky protein near the IIIa branch site physically excludes U2 snRNP recruitment to the IIIa 3′ splice site. Such a model predicts that the size of the protein, binding to the intron, is the critical parameter. To test this hypothesis we constructed an MS2-LacZ fusion protein that was engineered to have a size exceeding that of the full-length MS2-ASF/SF2 protein (445 amino acids compared with 382; Fig. 1). Importantly, the MS2-LacZ fusion protein did not inhibit IIIa-MS2I splicing (Fig. 4,lanes 5–8). Collectively, these results suggest that ASF/SF2 repression of IIIa splicing is specific and does not simply result from a steric interference, caused by binding of a large protein near the IIIa branch site. Since the RBDs of ASF/SF2 are sufficient for IIIa splicing inhibition (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar), we next sought to identify the sequence element in the ASF/SF2-RBDs required for splicing repression. As shown in Fig. 5, the MS2-RBD2+RS protein repressed IIIa-MS2I splicing (lanes 13–16). A further dissection of ASF/SF2 demonstrated that the ASF/SF2-RS domain did not inhibit IIIa-MS2I splicing (Fig. 5, lanes 9–12), whereas the MS2 fusion protein encoding for ASF/SF2-RBD2 inhibited IIIa-MS2I splicing ∼5-fold (Fig. 5, lanes 5–8). The inhibitory effect of RBD2 was specific since the MS2-ASF/SF2-RBD1 protein did not block IIIa-MS2I splicing (Fig. 5, lanes 1–4). Taken together these results suggest that the second RNA binding domain of ASF/SF2 encodes for a splicing repressor domain, which is both necessary (see also Fig. 6) and sufficient for repression of IIIa splicing. Systematic analysis of sequence motifs in pre-mRNA splicing factors have shown that many contain one or two RNP-type RNA binding domains (RBDs or RNA recognition motifs (34.Birney E. Kumar S. Krainer A.R. Nucleic Acids Res. 1993; 21: 5803-5816Crossref PubMed Scopus (595) Google Scholar)). The crystal structure of the RBDs in several proteins suggests a common β1, α1, β2, β3, α2, and β4 organization with the RNP2 and RNP1 motifs in the anti-parallel β1 and β3 strands providing the essential residues for RNA binding (reviewed in Ref. 35.Varani G. Nagai K. Annu. Rev. Biophys. Biomol. Struct. 1998; 27: 407-445Crossref PubMed Scopus (257) Google Scholar). To characterize further the sequences in ASF/SF2-RBD2 required for splicing repression, we constructed four mutants of MS2-RBD2+RS (Fig. 6A). We chose MS2-RBD2+RS as the parental protein for mutant construction since this fusion protein was more robust, retaining the splicing regulatory activity better than MS2-RBD2, i.e.after multiple thawings or storage. The ASF/SF2 pre-mRNA is alternatively spliced in vivo, generating three protein isoforms that differ at their C terminus (36.Ge H. Zuo P. Manley J.L. Cell. 1991; 66: 373-382Abstract Full Text PDF PubMed Scopus (287) Google Scholar). Two of the proteins lack the RS domain and ∼12 amino acid residues from the C terminus of RBD2, amino acids corresponding to the β4 strand of RBD2 (Fig. 6A). Removal of these residues inactivates ASF/SF2 as a constitutive splicing factor and regulator of alternative splicing (13.Zuo P. Manley J.L. EMBO J. 1993; 12: 4727-4737Crossref PubMed Scopus (161) Google Scholar). As shown in Fig. 6B, deletion of the β4strand abolished the inhibitory effect of MS2-RBD2+RS on IIIa-MS2I splicing. This result argues that the C terminus of RBD2 is required for the repressor function of ASF/SF2. Interestingly, the SWQDLKD motif contained in α-helix 1 is completely conserved in the RBD2 in all SR proteins containing two RBDs (37.Lutzelberger M. Gross T. Kaufer N.F. Nucleic Acids Res. 1999; 27: 2618-2626Crossref PubMed Scopus (30) Google Scholar). In the predicted structure of ASF/SF2-RBD2, this helix is positioned opposite to the β1and β3 strands making contact with the RNA. As shown in Fig. 6B, introduction of conservative amino acid changes, which are predicted to preserve the α-helical structure of the motif (37.Lutzelberger M. Gross T. Kaufer N.F. Nucleic Acids Res. 1999; 27: 2618-2626Crossref PubMed Scopus (30) Google Scholar), annulled the splicing repressor activity of the MS2-RBD2+RS protein. In contrast, mutating the RNP-1 motif or the highly conserved KLD motif at the C terminus of α-helix 2 (37.Lutzelberger M. Gross T. Kaufer N.F. Nucleic Acids Res. 1999; 27: 2618-2626Crossref PubMed Scopus (30) Google Scholar) did not significantly affect the splicing repressor phenotype of MS2-RBD2+RS. Collectively, these results suggest that the conserved α-helix 1 and the integrity of the β4 strand are critical for ASF/SF2 function as a splicing repressor protein. We showed previously (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar) that inserting the SR protein-responsive 3RE element at an intronic position in the rabbit β-globin pre-mRNA converted this transcript from an SR-protein-activated to an SR-protein-repressed transcript. Collectively, our results suggested that the inhibitory effect of SR protein may be a general feature of SR protein, binding to the intron of a pre-mRNA, and not restricted to the viral IIIa pre-mRNA. To determine whether the ASF/SF2-RBD2 also was sufficient to inhibit β-globin splicing, we inserted an MS2 operator site at position −53 in the rabbit β-globin first intron (transcript glob-MS2I). As shown in Fig. 7, MS2-RBD2 repressed glob-MS2I splicing ∼5-fold, whereas the MS2-RS protein failed to block splicing. In contrast, MS2-RBD2 or MS2-RS did not have any negative effects on wild type β-globin splicing (data not shown). Collectively, these results suggest that the inhibitory effect of the ASF/SF2-RBD2 domain on splicing, when tethered to the intron, may be a general feature and not restricted to the adenoviral IIIa pre-mRNA. Previous studies (24.Graveley B.R. Maniatis T. Mol. Cell. 1998; 1: 765-771Abstract Full Text Full Text PDF PubMed Scopus (169) Google Scholar, 38.Graveley B.R. Hertel K.J. Maniatis T. Curr. Biol. 1999; 9: R6-R7Abstract Full Text Full Text PDF PubMed Google Scholar) have shown that tethering the ASF/SF2-RS domain to the downstream exon is sufficient to reproduce the exonic splicing enhancer activity of ASF/SF2. As we show here, tethering the ASF/SF2-RBD2 to the intron results in repression of splicing (Figs. 5 and 7). Next we asked whether the opposite activity of the ASF/SF2-RS and RBD2 domains on splicing was position-dependent. For this experiment the MS2 operator was inserted at the IIIa second exon, 131 nucleotides downstream of the IIIa 3′ splice site. In agreement with previous results (24.Graveley B.R. Maniatis T. Mol. Cell. 1998; 1: 765-771Abstract Full Text Full Text PDF PubMed Scopus (169) Google Scholar, 38.Graveley B.R. Hertel K.J. Maniatis T. Curr. Biol. 1999; 9: R6-R7Abstract Full Text Full Text PDF PubMed Google Scholar), an MS2-RS protein stimulated IIIa-MS2E splicing, although very inefficiently in our experimental system (∼2-fold; Fig. 8, lanes 5–8). In contrast, tethering the MS2-ASF/SF2-RBD2 protein to the second exon had essentially no effect on IIIa-MS2E splicing (Fig. 8,lanes 1–4). Interestingly, the MS2-RBD2+RS protein activated IIIa-MS2E splicing slightly better (∼4-fold) compared with the MS2-RS protein, suggesting that the RS domain may be the minimal domain required for the ASF/SF2 splicing enhancer function but that RBD2 augments the enhancer activity of the RS domain. Taken together, our results suggest that the activity of the RBD2 and the RS domains on splicing is opposite and position-dependent. Here we present evidence suggesting that the splicing enhancer and splicing repressor functions of ASF/SF2 can be separated and attributed to distinct domains of the protein. Thus the RS domain has been shown to function as a splicing enhancer domain (24.Graveley B.R. Maniatis T. Mol. Cell. 1998; 1: 765-771Abstract Full Text Full Text PDF PubMed Scopus (169) Google Scholar, 39.Graveley B.R. Hertel K.J. Maniatis T. EMBO J. 1998; 17: 6747-6756Crossref PubMed Scopus (138) Google Scholar), whereas we show that RBD2 encodes for a splicing repressor domain (Figs. 5 and 7). Under our experimental conditions the ASF/SF2-RS domain functioned very inefficiently as a splicing enhancer domain. However, our results indicate that RBD2 may augment the splicing enhancer activity of the RS domain. Thus, MS2-RBD2+RS functioned ∼2-fold better as a splicing enhancer protein compared with MS2-RS (Fig. 8). Furthermore, we show that the effects of the RS and RBD2 domains in splicing are position-dependent and opposite. Thus, RBD2 inhibits splicing when tethered to the IIIa or the β-globin introns (Figs. 5and 7) but has no negative effects when bound to the IIIa second exon (Fig. 8). Conversely, the ASF/SF2-RS domain activates splicing slightly when tethered to the IIIa second exon (Fig. 8) but has no effect when bound to the intron (Fig. 5). Why does ASF/SF2 binding to an intron inhibit splicing? Our previous studies (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar, 23.Petersen-Mahrt S.K. Estmer C. Öhrmalm C. Matthews D.A. Russell W.C. Akusjärvi G. EMBO J. 1999; 18: 1014-1024Crossref PubMed Scopus (141) Google Scholar) have shown that ASF/SF2 binding to the 3RE inhibits IIIa splicing by preventing U2 snRNP recruitment to the spliceosome. Theoretically, this may be accomplished by, at least, three different mechanisms. First, binding of a large protein to a site close to the branch point in a pre-mRNA may not be compatible with the simultaneous binding of U2 snRNP, i.e. a simple steric interference. The results presented here clearly argue against this model, since an MS2-LacZ fusion protein, with a size ∼15% larger than MS2-ASF/SF2, did not inhibit splicing (Fig. 4). Second, protein binding to the so-called anchoring sequence (40.Gozani O. Feld R. Reed R. Genes Dev. 1996; 10: 233-243Crossref PubMed Scopus (196) Google Scholar) may preclude U2 snRNP recruitment to the branch site. The anchoring sequence is the RNA sequence located immediately upstream of the branch site that makes contact with the SF3 component of U2 snRNP (40.Gozani O. Feld R. Reed R. Genes Dev. 1996; 10: 233-243Crossref PubMed Scopus (196) Google Scholar). In the wild type IIIa transcript, the anchoring sequence is contained within the 3RE, likely contributing to the inhibitory effect of the 3RE on IIIa splicing. Such a conclusion is supported by our finding that positioning the highly structured MS2 operator site 15 nucleotides upstream of the IIIa branch site completely ablates IIIa splicing in the absence of any MS2 proteins (Fig. 2B). This finding is compatible with the previous study (40.Gozani O. Feld R. Reed R. Genes Dev. 1996; 10: 233-243Crossref PubMed Scopus (196) Google Scholar), which suggested that the single-stranded nature of the anchoring sequence is required for SF3 interaction with RNA. However, in the IIIa-MS2I transcript, the MS2 operator sequence was moved further upstream to position −53 relative to the IIIa branch site. At this location the MS2 operator does not interfere with basal IIIa splicing. Importantly, at this position MS2-ASF/SF2 inhibited IIIa splicing (Figs. 4 and 6), whereas MS2-ASF/SF2 hybrid proteins lacking RBD2 did not block splicing (Fig. 5), strongly arguing that ASF/SF2 coverage of the anchoring sequence is not the mechanism by which ASF/SF2 inhibits splicing. This conclusion is further supported by the observation that moving the 3RE to position −53 (Fig. 2C), or the MS2 operator site to position −127 (data not shown) relative to the IIIa branch site, still results in SR or MS2-ASF/SF2 protein-mediated inhibition of splicing. Collectively, our data suggest a third model, namely that ASF/SF2-RBD2 functions as a specific splicing repressor domain. Potentially, ASF/SF2-RBD2 interacts directly with a component(s) required for early splice site recognition, or alternatively functions as a nucleation site for recruitment of factors that negatively affect splice site recognition. Currently we have no data that discriminate between these or other alternative mechanisms. However, our data clearly show that ASF/SF2-RBD2 indeed functions as a splicing repressor domain when tethered to an intronic position in the IIIa or the β-globin pre-mRNAs. Obviously, characterization of the mechanism by which ASF/SF2-RBD2 inhibits pre-mRNA splicing is a major goal in our future research. Our initial characterization of ASF/SF2-RBD2 mutant proteins provides some interesting clues that will help in this characterization. Thus, we show that mutating the SWQDLKD motif in the putative α-helix 1 or deletion of the β4strand of RBD2 destroys the repressor activity of ASF/SF2-RBD2, whereas mutations that affect the RNP1 motif or α-helix 2 has no effect, or only a minor a effect, on the splicing repressor activity of RBD2. Since neither α-helix 1 nor the β4 strand is expected to be involved directly in RNA binding, they may provide an essential surface for ASF/SF2-RBD2 interaction with splicing components mediating splicing repression. In retrospect the finding that the ASF/SF2-RBD2 is sufficient for repression of IIIa splicing may provide a logical explanation to our previous observation (20.Kanopka A. Mühlemann O. Akusjärvi G. Nature. 1996; 381: 535-538Crossref PubMed Scopus (209) Google Scholar) that all tested SR proteins, except SRp20, inhibited IIIa splicing efficiently. Thus, SRp20, which is the shortest SR protein, encodes for only one RBD. All other tested SR proteins (SRp75, SRp55, SRp40, and SRp30) encode for two RBDs. The SRp30 fraction, used in this study, contained a mixture of SR proteins with one RBD (SC35 and 9G8) or two RBDs (ASF/SF2 and SRp30c) (1.Zahler A.M. Lane W.S. Stolk J.A. Roth M.B. Genes Dev. 1992; 6: 837-847Crossref PubMed Scopus (632) Google Scholar, 41.Cavaloc Y. Popielarz M. Fuchs J.P. Gattoni R. Stevenin J. EMBO J. 1994; 13: 2639-2649Crossref PubMed Scopus (134) Google Scholar, 42.Screaton G.R. Caceres J.F. Mayeda A. Bell M.V. Plebanski M. Jackson D.G. Bell J.I. Krainer A.R. EMBO J. 1995; 14: 4336-4349Crossref PubMed Scopus (248) Google Scholar). Potentially the RBD2 in other SR proteins may similarly be splicing repressor domains. From this point it is interesting to note that the SWQDLKD motif here shown to be required for ASF/SF2-RBD2 repression of splicing is conserved in the RBD2 of all SR proteins containing two RBDs (37.Lutzelberger M. Gross T. Kaufer N.F. Nucleic Acids Res. 1999; 27: 2618-2626Crossref PubMed Scopus (30) Google Scholar). The significance of the SWQDLKD motif rather than the predicted structure of ASF/SF2-RBD2 is indicated by the observation that MS2-RBD1 generates distinct complexes in a gel retardation assay (Fig. 3B, lane 3), whereas MS2-RBD2, which is of the same size as MS2-RBD1 (Fig. 1B), generates smaller complexes with a fuzzy appearance (Fig. 3B, lane 4). These results may be interpreted to indicate that ASF/SF2-RBD2 does not adopt the predicted structure of an RNP-type RBD. The finding that ASF/SF2 encodes for separate domains, functioning as splicing enhancer and splicing repressor domains, adds another level by which ASF/SF2 may partake in constitutive and alternative RNA splicing. Potentially, this modular structure of ASF/SF2 may contribute to the splice site switching activity of this protein. Previous studies (12.Caceres J.F. Krainer A.R. EMBO J. 1993; 12: 4715-4726Crossref PubMed Scopus (222) Google Scholar,13.Zuo P. Manley J.L. EMBO J. 1993; 12: 4727-4737Crossref PubMed Scopus (161) Google Scholar, 43.Bai Y. Lee D. Yu T. Chasin L.A. Nucleic Acids Res. 1999; 27: 1126-1134Crossref PubMed Scopus (60) Google Scholar) have shown that ASF/SF2 promotes proximal 5′ and 3′ splice site selection in model transcripts. Also, ASF/SF2-RBD1+RBD2 was confirmed to be sufficient to promote proximal 5′ splice site selection both in vitro and in vivo (12.Caceres J.F. Krainer A.R. EMBO J. 1993; 12: 4715-4726Crossref PubMed Scopus (222) Google Scholar, 13.Zuo P. Manley J.L. EMBO J. 1993; 12: 4727-4737Crossref PubMed Scopus (161) Google Scholar). In a recent study (44.van Der Houven Van Oordt W. Newton K. Screaton G.R. Caceres J.F. Nucleic Acids Res. 2000; 28: 4822-4831Crossref PubMed Scopus (38) Google Scholar) it was demonstrated that RBD2 of ASF/SF2 plays a dominant role in determining the alternative splicing specificity of chimeric SR proteins. This result may be related to our finding that ASF/SF-RBD2 has a distinct function as a repressor domain in splicing regulation. In fact, our recent studies 2V. Dauksaite and G. Akusjärvi, manuscript in preparation.have shown that, under our experimental conditions, ASF/SF2-RBD2 is sufficient to induce proximal 5′ splice site selection in the adenovirus E1A pre-mRNA. However, the effect of ASF/SF2-RBD2 on alternative 3′ splice site selection appears more complex. Preliminary data 3V. Dauksaite and G. Akusjärvi, unpublished observations. suggest that RBD2 or RBD2+RS does not reproduce the effect of the wild type ASF/SF2 protein on proximal 3′ splice site selection. We thank L. Liljas for providing the phage MS2, J.-P. Kreivi for the clone encoding SRPK1, and M. Lützelberger and J.-P. Kreivi for much help and critical comments on the manuscript.

Referência(s)