Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

A Splicing Repressor Domain in Polypyrimidine Tract-binding Protein

2005; Elsevier BV; Volume: 281; Issue: 2 Linguagem: Inglês

10.1074/jbc.m510578200

ISSN

1083-351X

Autores

Fiona Robinson, Christopher W. J. Smith,

Tópico(s)

RNA regulation and disease

Resumo

Polypyrimidine tract-binding protein (PTB) is an hnRNP with four RRM type domains. It plays roles as a repressive alternative splicing regulator of multilple target genes, as well as being involved in pre-mRNA 3′ end processing, mRNA localization, stability, and internal ribosome entry site-mediated translation. Here we have used a tethered function assay, in which a fusion protein of PTB and the bacteriophage MS2 coat protein is recruited to a splicing regulatory site by binding to an artificially inserted MS2 binding site. Deletion mutations of PTB in this system allowed us to identify RRM2 and the following inter-RRM linker region as the minimal region of PTB that can act as splicing repressor domain when recruited to RNA. Splicing repression by the minimal repressor domain remained cell type-specific and dependent upon other defined regulatory elements in the α-tropomyosin test minigene. Our results highlight the fact that splicing repression by PTB can be uncoupled from the mode by which it binds to RNA. Polypyrimidine tract-binding protein (PTB) is an hnRNP with four RRM type domains. It plays roles as a repressive alternative splicing regulator of multilple target genes, as well as being involved in pre-mRNA 3′ end processing, mRNA localization, stability, and internal ribosome entry site-mediated translation. Here we have used a tethered function assay, in which a fusion protein of PTB and the bacteriophage MS2 coat protein is recruited to a splicing regulatory site by binding to an artificially inserted MS2 binding site. Deletion mutations of PTB in this system allowed us to identify RRM2 and the following inter-RRM linker region as the minimal region of PTB that can act as splicing repressor domain when recruited to RNA. Splicing repression by the minimal repressor domain remained cell type-specific and dependent upon other defined regulatory elements in the α-tropomyosin test minigene. Our results highlight the fact that splicing repression by PTB can be uncoupled from the mode by which it binds to RNA. Alternative pre-mRNA splicing is an important mechanism of gene regulation in multicellular eukaryotes. It allows the generation of a number of protein isoforms far in excess of the number of genes, as well as providing quantitative gene control by producing RNAs that are subject to non-sense-mediated decay (1Lareau L.F. Green R.E. Bhatnagar R.S. Brenner S.E. Curr. Opin. Struct. Biol. 2004; 14: 273-282Crossref PubMed Scopus (263) Google Scholar). Much experimental attention has therefore been aimed at understanding the molecular mechanisms of regulated splicing decisions. cis-Acting regulatory elements in the pre-mRNA include not only the essential splice site sequences themselves but also variable auxiliary elements which can act as enhancers or silencers and are located in exons or introns. These elements are referred to as exon splicing enhancers, exon splicing silencers, intron splicing enhancers, and intron splicing silencers (reviewed in Refs. 2Cartegni L. Chew S.L. Krainer A.R. Nat. Rev. Genet. 2002; 3: 285-298Crossref PubMed Scopus (1761) Google Scholar and 3Matlin A.J. Clark F. Smith C.W. Nat. Rev. Mol. Cell. Biol. 2005; 6: 386-398Crossref PubMed Scopus (978) Google Scholar). These elements typically act as binding sites for splicing regulatory proteins that act as repressors or activators. Members of the SR protein family typically act as activators, although in some circumstances they can be repressors. In contrast, proteins of the hnRNP family are most commonly repressors, but some members can be activators under some circumstances (2Cartegni L. Chew S.L. Krainer A.R. Nat. Rev. Genet. 2002; 3: 285-298Crossref PubMed Scopus (1761) Google Scholar, 3Matlin A.J. Clark F. Smith C.W. Nat. Rev. Mol. Cell. Biol. 2005; 6: 386-398Crossref PubMed Scopus (978) Google Scholar). One of the best characterized mammalian splicing repressors is polypyrimidine tract binding protein (PTB), 3The abbreviations used are: PTBpolypyrimidine tract-binding proteinRRMRNA recognition motifTMα-tropomyosinRTreverse transcription. which is also known as hnRNP-I (4Wagner E.J. Garcia-Blanco M.A. Mol. Cell. Biol. 2001; 21: 3281-3288Crossref PubMed Scopus (305) Google Scholar). PTB consists of four RNA recognition motif (RRM) type domains, with additional linker regions between the RRMs and both nuclear localization and export sequences at the extreme N terminus (Fig. 1A). PTB is multifunctional, and as well as functioning as a splicing repressor it also has roles in 3′ end processing (5Castelo-Branco P. Furger A. Wollerton M. Smith C. Moreira A. Proudfoot N. Mol. Cell. Biol. 2004; 24: 4174-4183Crossref PubMed Scopus (139) Google Scholar), mRNA localization (6Cote C.A. Gautreau D. Denegre J.M. Kress T.L. Terry N.A. Mowry K.L. Mol. Cell. 1999; 4: 431-437Abstract Full Text Full Text PDF PubMed Scopus (205) Google Scholar), mRNA stability (7Hamilton B.J. Genin A. Cron R.Q. Rigby W.F. Mol. Cell. Biol. 2003; 23: 510-525Crossref PubMed Scopus (71) Google Scholar), and internal ribosome entry site-mediated translation (8Kaminski A. Hunt S.L. Patton J.G. Jackson R.J. RNA (N. Y.). 1995; 1: 924-938PubMed Google Scholar, 9Mitchell S.A. Spriggs K.A. Bushell M. Evans J.R. Stoneley M. Le Quesne J.P. Spriggs R.V. Willis A.E. Genes Dev. 2005; 19: 1556-1571Crossref PubMed Scopus (105) Google Scholar). PTB has been characterized as a splicing repressor in multiple model systems (reviewed in Ref. 4Wagner E.J. Garcia-Blanco M.A. Mol. Cell. Biol. 2001; 21: 3281-3288Crossref PubMed Scopus (305) Google Scholar). The optimal binding site for PTB determined by SELEX is UCUU in a pyrimidine-rich context (10Perez I. Lin C.H. McAfee J.G. Patton J.G. RNA (N. Y.). 1997; 3: 764-778PubMed Google Scholar), and this and related sequences (e.g. CUCUCU) have been shown to act as splicing silencers. Curiously a (CCU)n sequence within a double-stranded context is a high affinity site for PTB and is common in cellular internal ribosome entry sites (9Mitchell S.A. Spriggs K.A. Bushell M. Evans J.R. Stoneley M. Le Quesne J.P. Spriggs R.V. Willis A.E. Genes Dev. 2005; 19: 1556-1571Crossref PubMed Scopus (105) Google Scholar). PTB binds to both intron splicing silencers and exon splicing silencers and most exons that are regulated by PTB have multiple binding sites. Repression of the c-src N1 exon requires cooperative binding of PTB to sites flanking the exon (11Chou M.Y. Underwood J.G. Nikolic J. Luu M.H. Black D.L. Mol. Cell. 2000; 5: 949-957Abstract Full Text Full Text PDF PubMed Scopus (158) Google Scholar). Simple competition with binding of U2AF65 at the polypyrimidine tract is a possible mechanism for PTB action (12Lin C.H. Patton J.G. RNA (N. Y.). 1995; 1: 234-245PubMed Google Scholar, 13Singh R. Valcarcel J. Green M.R. Science. 1995; 268: 1173-1176Crossref PubMed Scopus (466) Google Scholar, 14Ashiya M. Grabowski P.J. RNA (N. Y.). 1997; 3: 996-1015PubMed Google Scholar). However, the facts that multiple PTB binding sites are required and that in some cases none of these sites are within the U2AF65 binding polypyrimidine tract (15Amir-Ahmady B. Boutz P.L. Markovtsov V. Phillips M.L. Black D.L. RNA (N. Y.). 2005; 11: 699-716Crossref PubMed Scopus (94) Google Scholar, 16Shen H. Kan J.L. Ghigna C. Biamonti G. Green M.R. RNA (N. Y.). 2004; 10: 787-794Crossref PubMed Scopus (48) Google Scholar) together suggest that the mechanism of PTB-mediated splicing repression may be more complex. PTB-mediated multimerization to create "zones of silencing," which would be inaccessible to the splicing machinery has been suggested as a mechanism, and is consistent with a requirement for multiple binding sites (4Wagner E.J. Garcia-Blanco M.A. Mol. Cell. Biol. 2001; 21: 3281-3288Crossref PubMed Scopus (305) Google Scholar). Indeed, the solution structure of RRM domains 3 and 4 bound to RNA shows that a loop of at least 15 nucleotides is needed between the two RNA sites that bind to RRM 3 and 4 (17Oberstrass F.C. Auweter S.D. Erat M. Hargous Y. Henning A. Wenter P. Reymond L. Amir-Ahmady B. Pitsch S. Black D.L. Allain F.H. Science. 2005; 309: 2054-2057Crossref PubMed Scopus (336) Google Scholar). This suggests models for splicing repression involving looping of repressed exons or splice sites between RRMs 3 and 4. However, in some cases a single PTB binding exon splicing silencer appears to be sufficient (16Shen H. Kan J.L. Ghigna C. Biamonti G. Green M.R. RNA (N. Y.). 2004; 10: 787-794Crossref PubMed Scopus (48) Google Scholar, 18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar), and U1 snRNP binding to the c-src N1 5′ splice site is not prevented by cooperative binding of PTB to flanking sites (11Chou M.Y. Underwood J.G. Nikolic J. Luu M.H. Black D.L. Mol. Cell. 2000; 5: 949-957Abstract Full Text Full Text PDF PubMed Scopus (158) Google Scholar). Indeed, recent evidence suggests that PTB may interfere indirectly with U2AF65 binding by preventing positive interactions between U2AF65 and splicing factors bound at a 5′ splice site (18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar, 19Sharma S. Falick A.M. Black D.L. Mol. Cell. 2005; 19: 485-496Abstract Full Text Full Text PDF PubMed Scopus (125) Google Scholar). polypyrimidine tract-binding protein RNA recognition motif α-tropomyosin reverse transcription. A number of structure-function analyses have been carried out on PTB (8Kaminski A. Hunt S.L. Patton J.G. Jackson R.J. RNA (N. Y.). 1995; 1: 924-938PubMed Google Scholar, 20Oh Y.L. Hahm B. Kim Y.K. Lee H.K. Lee J.W. Song O. Tsukiyama-Kohara K. Kohara M. Nomoto A. Jang S.K. Biochem. J. 1998; 331: 169-175Crossref PubMed Scopus (84) Google Scholar, 21Perez I. McAfee J.G. Patton J.G. Biochemistry. 1997; 36: 11881-11890Crossref PubMed Scopus (147) Google Scholar, 22Liu H. Zhang W. Reed R.B. Liu W. Grabowski P.J. RNA (N. Y.). 2002; 8: 137-149Crossref PubMed Scopus (44) Google Scholar, 23Conte M.R. Grune T. Ghuman J. Kelly G. Ladas A. Matthews S. Curry S. EMBO J. 2000; 19: 3132-3141Crossref PubMed Scopus (126) Google Scholar, 24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar). Early work suggested that RRMs 3 and 4 were primarily responsible for RNA binding, while RRM2 might be involved in PTB-PTB dimerization (20Oh Y.L. Hahm B. Kim Y.K. Lee H.K. Lee J.W. Song O. Tsukiyama-Kohara K. Kohara M. Nomoto A. Jang S.K. Biochem. J. 1998; 331: 169-175Crossref PubMed Scopus (84) Google Scholar, 21Perez I. McAfee J.G. Patton J.G. Biochemistry. 1997; 36: 11881-11890Crossref PubMed Scopus (147) Google Scholar). More recently, it has become clear that RRMs 1–2 can also bind RNA (17Oberstrass F.C. Auweter S.D. Erat M. Hargous Y. Henning A. Wenter P. Reymond L. Amir-Ahmady B. Pitsch S. Black D.L. Allain F.H. Science. 2005; 309: 2054-2057Crossref PubMed Scopus (336) Google Scholar, 22Liu H. Zhang W. Reed R.B. Liu W. Grabowski P.J. RNA (N. Y.). 2002; 8: 137-149Crossref PubMed Scopus (44) Google Scholar, 24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar, 25Wollerton M.C. Gooding C. Wagner E.J. Garcia-Blanco M.A. Smith C.W. Mol. Cell. 2004; 13: 91-100Abstract Full Text Full Text PDF PubMed Scopus (309) Google Scholar) and that pure PTB does not exist as a stable dimer (15Amir-Ahmady B. Boutz P.L. Markovtsov V. Phillips M.L. Black D.L. RNA (N. Y.). 2005; 11: 699-716Crossref PubMed Scopus (94) Google Scholar, 24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar). RRM4 was found to be critical for splicing repression in the GABAA γ2 system (22Liu H. Zhang W. Reed R.B. Liu W. Grabowski P.J. RNA (N. Y.). 2002; 8: 137-149Crossref PubMed Scopus (44) Google Scholar), even though mutants in RRM4 were still able to bind to RNA. However, in this assay loss of splicing repression could be due to alterations in the mode of PTB binding to RNA or due to direct impairment of a repressor domain that does not itself bind to RNA. Here, we have used a "tethered function" assay to identify a minimal splicing repressor domain of PTB. We previously showed that replacement of an essential PTB binding site in the α-tropomyosin (TM) gene with tandem binding sites for bacteriophage MS2 coat protein resulted in loss of exon 3 skipping. Splicing repression could be restored by cotransfection of PTB-MS2 fusion protein but not by PTB or MS2 alone or a fusion of MS2 with another known splicing repressor, hnRNPA1 (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). Because specific RNA binding is provided by the MS2 coat protein, the tethered function assay provides an ideal opportunity to carry out deletion mutagenesis on PTB. We expect that deletion of domains whose only function is in RNA binding would be without effect upon splicing inhibition in this assay. A process of elimination would allow us to identify a minimal splicing repressor domain(s). Consistent with this expectation we find that PTB RRM2 and the following linker region are together sufficient when fused to MS2 to confer high levels of splicing repression. Tethered repressor activity remains dependent upon other cis-acting regulatory elements and is cell specific. Characterization of the splicing repressor domain should facilitate more detailed analyses of the molecular mechanisms of splicing repression by PTB. Constructs and Cloning—Constructs were prepared by standard cloning procedures. Constructs for MS2 fusion proteins were prepared from a parental vector supplied by Richard Breathnach (27Del Gatto-Konczak F. Olive M. Gesnel M.C. Breathnach R. Mol. Cell. Biol. 1999; 19: 251-260Crossref PubMed Scopus (199) Google Scholar), which contained in N- to C-terminal order: an N-terminal FLAG tag, a nuclear localization signal, and MS2 coat protein. Full-length PTB, or defined subregions (Fig. 1A), were cloned between the FLAG tag and nuclear localization signal as previously described (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). All constructs were verified by sequencing. Transfections—PAC-1 smooth muscle cells, HeLa cells, and L cells were cultured under standard conditions in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Transfections were carried out in 35 mm wells using Lipofectamine (Invitrogen) (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar, 28Wollerton M.C. Gooding C. Robinson F. Brown E.C. Jackson R.J. Smith C.W. RNA (N.Y.). 2001; 7: 819-832Crossref PubMed Scopus (116) Google Scholar). Unless otherwise stated, transfections contained 200 ng of reporter vector and 800 ng of effector vector (e.g. PTB-MS2). When lower levels of effector were used, including negative cotransfection controls, the total DNA concentration was maintained using pGem4Z. RNA Analysis—RNA was harvested and analyzed as described previously (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar, 28Wollerton M.C. Gooding C. Robinson F. Brown E.C. Jackson R.J. Smith C.W. RNA (N.Y.). 2001; 7: 819-832Crossref PubMed Scopus (116) Google Scholar). Harvesting was with either TRIReagent (Sigma) as described previously or PURESCRIPT (Gentra) solutions, according to supplier's instructions. Reverse transcription was carried out using AMV reverse transcriptase and a specific RT primer: 5′ GCA AAC TCA GCC ACA GGT 3′. PCR was carried out with forward primer SV5′2 (5′ GGA GGC CTA GGC TTT TGC AAA AAG 3′) and 32P 5′-end labeled SV3′1 (5′ ACT CAC TGC GTT CCA GGC AAT GCT 3′), with an 80 °C hot-start and 30 PCR cycles with the following cycling parameters: 94 °C, 30 s, 62 °C, 30 s, 72 °C, 60 s. Products were fractionated on denaturing (8 m urea) polyacrylamide gels, and quantitation was carried out using a Amersham Biosciences Storm phosphorimager and ImageQuant software, by line analysis and area integration. This procedure produces reproducible output of the ratio of exon inclusion to exon skipping (28Wollerton M.C. Gooding C. Robinson F. Brown E.C. Jackson R.J. Smith C.W. RNA (N.Y.). 2001; 7: 819-832Crossref PubMed Scopus (116) Google Scholar, 29Gooding C. Roberts G.C. Moreau G. Nadal-Ginard B. Smith C.W. EMBO J. 1994; 13: 3861-3872Crossref PubMed Scopus (92) Google Scholar, 30Gooding C. Roberts G.C. Smith C.W. RNA (N. Y.). 1998; 4: 85-100PubMed Google Scholar, 31Gromak N. Smith C.W. Nucleic Acids Res. 2002; 30: 3548-3557Crossref PubMed Scopus (20) Google Scholar). Results are reported as the percentage of exon skipping with respect to total spliced RNA (exon 3 inclusion + exon 3 skipping). Histograms show mean and standard deviations resulting from at least three repeats of the experiment. Protein Harvest and Analysis—Cells were washed twice with phosphate-buffered saline, dissolved directly in 150 μl of SDS loading buffer, and then subjected to five cycles of 90 °C, 5 min, freezing on dry ice, and slowly thawing to room temperature. 10–20-μl samples were electrophoresed on 15% polyacrylamide SDS gels. Proteins were transferred to polyvinylidene difluoride membrane using a semidry blotting apparatus at 0.5 mA for 75 min. Detection was by enhanced chemiluminescence using anti-FLAG (Sigma) monoclonal and donkey anti-mouse antibody conjugated to horseradish peroxidase (Jackson ImmunoResearch Laboratories). We designed an initial series of six PTB domain deletions based upon structurally defined domain boundaries (23Conte M.R. Grune T. Ghuman J. Kelly G. Ladas A. Matthews S. Curry S. EMBO J. 2000; 19: 3132-3141Crossref PubMed Scopus (126) Google Scholar, 24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar) (Fig. 1A). These were all based upon PTB4, which is a more potent repressor in the TM system (28Wollerton M.C. Gooding C. Robinson F. Brown E.C. Jackson R.J. Smith C.W. RNA (N.Y.). 2001; 7: 819-832Crossref PubMed Scopus (116) Google Scholar) and which differs from PTB1 by the presence of a 26-amino acid peptide insert between RRMs 2 and 3 (Fig. 1A). The mutants are named according to the RRMs remaining in the fusion protein, and an "L" indicates whether the adjacent linker region was also present in the fusion protein. For example, 12L-MS2 contains the region of PTB from the N terminus up to and including the linker following RRM2 (amino acids 1–360). All constructs contained an N-terminal FLAG epitope allowing us to demonstrate equivalent expression levels and a strong nuclear localization signal between the PTB and MS2 moieties. Therefore, any observed differences in repressor activity are not due to differential expression levels or localization (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). Unless otherwise stated, all transfections were carried out in PAC-1 cells, a partially differentiated smooth muscle cell line that supports regulated skipping of TM exon 3 (29Gooding C. Roberts G.C. Moreau G. Nadal-Ginard B. Smith C.W. EMBO J. 1994; 13: 3861-3872Crossref PubMed Scopus (92) Google Scholar, 30Gooding C. Roberts G.C. Smith C.W. RNA (N. Y.). 1998; 4: 85-100PubMed Google Scholar, 31Gromak N. Smith C.W. Nucleic Acids Res. 2002; 30: 3548-3557Crossref PubMed Scopus (20) Google Scholar). The construct TM-2MS2 has the PTB binding sites downstream of TM exon 3 replaced by tandem MS2 sites (Fig. 1B) and when transfected alone into PAC-1 cells gives low basal levels of exon skipping (Fig. 2, A and B, left-hand lane). Cotransfection of PTB4-MS2 produced high levels of exon skipping. In contrast, cotransfection of MS2, PTB4, or hnRNPA1-MS2 had little or no effect (data not shown and Ref. 26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). The six mutant constructs had widely different effects (Fig. 2, A and B) despite being expressed at equivalent levels (Fig. 2C). Three of the initial deletion mutants (12L-MS2, 123L-MS2, and L234-MS2) had repressor activity equivalent to full-length PTB4-MS2 (∼70–80% exon skipping). In contrast, 12-MS2, 34-MS2, and L34-MS2 had little or no activity. The region common to all the active fusion proteins is L2L. The inactive proteins all had deletions in one or more parts of this region. The next series of four deletions all contained RRM2 and RRM3 and the intervening linker but varied by the presence or absence of the flanking linker regions (L23L-MS2, L23-MS2, 23L-MS2, and 23-MS2). All four mutants had equivalent repressor activity to full-length PTB4-MS2 and 12L-MS2 (Fig. 3, A and B). The PTB region common to all of the active repressors is therefore 2L. We next tested three further mutants; L2L-MS2, 2L-MS2, and L23-MS2 (Fig. 4A and B, subscript 23 indicates that this is the linker between RRMs 2 and 3). While L2L-MS2 and 2L-MS2 had substantial repressor activity, L23-MS2 was inactive, despite being expressed to equivalent levels (Fig. 4C). Taken together with the preceding data (Figs. 2, 3, 4) this indicates that the minimal repressor domain of PTB comprises RRM2 and the following linker region. Comparison of data from multiple independent experiments (Fig. 4B) indicated that this minimal repressor region retained ∼60% of the full activity of PTB4-MS2 and is the smallest fragment of PTB4 to retain substantial activity. The minimal repressor region retains the 26-amino acid peptide insert by which PTB4 differs from PTB1. The two isoforms have differential activity in the TM system, with PTB4 being more repressive (28Wollerton M.C. Gooding C. Robinson F. Brown E.C. Jackson R.J. Smith C.W. RNA (N.Y.). 2001; 7: 819-832Crossref PubMed Scopus (116) Google Scholar). We therefore compared the activity of 2L-MS2 constructs derived from either PTB4 or PTB1. Strikingly, the PTB1 version of 2L-MS2 had a much lower level of activity than the PTB4 version (Fig. 4, D and E), despite being expressed to equivalent levels (Fig. 4F).FIGURE 4PTB4 RRM2-L is the minimal repressor domain. A, RT-PCR analysis of TM-2MS2 splicing in PAC-1 smooth muscle cells after cotransfection with the various effectors (800 ng) indicated above. 4Z is a negative control in which pGem4Z was cotransfected. Positions of 1-3-4 and 1-4 spliced products are indicated to the left, and the percent exon skipping for this gel is shown below the relevant lanes. B, histogram summarizing the results of multiple (≥3) independent experiments. In addition to the constructs shown in A, full-length PTB4-MS2 and 12L-MS2 are also shown for comparison. C, Western blot using anti-FLAG antibodies. B and C are aligned so that the effector label in B also applies to C. D, RT-PCR of TM-MS2 splicing after cotransfection with PTB4 and PTB1 versions of RRM2L. 4Z is a negative control cotransfection. E, histogram summarizing the results of multiple (≥3) independent experiments corresponding to the experiment of D. F, Western blot using anti-FLAG antibodies. The asterisk indicates a cross-reacting band present in all lanes. E and F are aligned so that the effector label in E refers to both panels. Black lines between the lanes in C and D indicate where lanes have been cropped from the original gel. All lanes shown are from the same gel.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Artifical Recruitment of PTB Does Not Bypass Normal Regulatory Requirements—The preceding experiments indicated that artificial recruitment of full-length PTB4 or of the RRM2L region was sufficient to confer skipping of TM exon 3 in PAC-1 cells. We next proceeded to test whether this form of exon skipping is related to the normal mechanism of regulated splicing. We first compared a titration of PTB4-MS2 and 2L-MS2 in cotransfections either with the TM-2MS2 (Fig. 5A) reporter or with a second reporter, TM-Sxl (Fig. 5B), in which the original PTB binding site has been replaced with an unrelated pyrimidine tract that does not bind PTB or MS2 (Fig. 1B) (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar, 30Gooding C. Roberts G.C. Smith C.W. RNA (N. Y.). 1998; 4: 85-100PubMed Google Scholar). Both PTB4-MS2 and 2L-MS2 showed specificity for conferring exon skipping dependent on the presence of the MS2 binding sites (Fig. 5C). Indeed 2L-MS2 showed a higher degree of substrate specificity compared with full-length PTB4-MS2 (compare 1 and 800 ng lanes). This could be because the smaller fusion protein lacking three of the RRMs is less likely to bind non-specifically away from the MS2 site. We next tested the effects of a series of the PTB4-MS2 deletion constructs in two additional cell lines, HeLa (human epithelial) and L (mouse fibroblast) cells. Both of these cell lines have been used as convenient "non-smooth muscle" lines that give lower levels of TM exon 3 skipping than PAC-1 (29Gooding C. Roberts G.C. Moreau G. Nadal-Ginard B. Smith C.W. EMBO J. 1994; 13: 3861-3872Crossref PubMed Scopus (92) Google Scholar, 30Gooding C. Roberts G.C. Smith C.W. RNA (N. Y.). 1998; 4: 85-100PubMed Google Scholar, 31Gromak N. Smith C.W. Nucleic Acids Res. 2002; 30: 3548-3557Crossref PubMed Scopus (20) Google Scholar). All of the fusion proteins were less active in L and HeLa cells than in PAC-1 (Fig. 6). Full-length PTB4-MS2 and the constructs containing three RRMs (123L-MS2 and L234-MS2) produced relatively high levels of exon skipping in L cells but much lower levels in HeLa cells. Significantly, the minimal active fusion constructs (L2L-MS2 and 2L-MS2) showed a higher degree of smooth muscle specificity, with substantial levels of exon skipping in PAC-1 cells but much lower levels in both L and HeLa cells. Therefore the artificial recruitment of the minimal repressor domains does not bypass the normal mechanisms of cell specific exon skipping. Finally, we tested the dependence of the artificially recruited repressor function upon other essential cis-acting regulatory elements. PTB4-MS2 and 2L-MS2 were cotransfected with three additional reporter constructs in which the two essential regulatory elements upstream of TM exon 3 have been mutated, in addition to the replacement of the DY element by MS2 sites (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). In TM-2MS2ΔP3, point mutations have been made in the three UCUU motifs within the polypyrimidine tract; TM-2MS2ΔURE has a 15-nucleotide deletion of the UGC motif containing URE; TM-2MS2ΔP3ΔURE contains both mutations. All three mutant constructs showed a reduced response to both PTB4-MS2 and 2L-MS2 compared with TM-2MS2 (Fig. 7). This confirms that splicing repression mediated by artificially recruited PTB repressor domains remains dependent upon the other essential cis-acting regulatory elements and does not bypass the normal mechanisms of control. The tethered function assay has allowed us to narrow down a minimal PTB repressor domain that restores skipping of exon 3 without bypassing the usual cell-specific (Fig. 6) and cis-acting (Fig. 7) regulatory requirements. Since recruitment to the RNA substrate is directed by the MS2 coat protein this assay allowed us to delete domains of PTB whose principal function is in RNA binding and to identify domains that have a direct effector role. Consistent with this approach, RRMs 3 and 4, which have been attributed the major role in RNA binding in many structure-function analyses (8Kaminski A. Hunt S.L. Patton J.G. Jackson R.J. RNA (N. Y.). 1995; 1: 924-938PubMed Google Scholar, 20Oh Y.L. Hahm B. Kim Y.K. Lee H.K. Lee J.W. Song O. Tsukiyama-Kohara K. Kohara M. Nomoto A. Jang S.K. Biochem. J. 1998; 331: 169-175Crossref PubMed Scopus (84) Google Scholar, 21Perez I. McAfee J.G. Patton J.G. Biochemistry. 1997; 36: 11881-11890Crossref PubMed Scopus (147) Google Scholar), were entirely dispensable for repressor activity (Fig. 2). In contrast, the minimal repressor domain requires RRM2, which has been suggested to mediate PTB-PTB dimerization (20Oh Y.L. Hahm B. Kim Y.K. Lee H.K. Lee J.W. Song O. Tsukiyama-Kohara K. Kohara M. Nomoto A. Jang S.K. Biochem. J. 1998; 331: 169-175Crossref PubMed Scopus (84) Google Scholar, 21Perez I. McAfee J.G. Patton J.G. Biochemistry. 1997; 36: 11881-11890Crossref PubMed Scopus (147) Google Scholar). The effector domain of PTB might act, at least in part, by mediating PTB-PTB interactions. This could lead to cooperative binding of PTB to the upstream P3 sites, similar to the situation in the c-src N1 exon (11Chou M.Y. Underwood J.G. Nikolic J. Luu M.H. Black D.L. Mol. Cell. 2000; 5: 949-957Abstract Full Text Full Text PDF PubMed Scopus (158) Google Scholar). A similar model was proposed for hnRNPA1; MS2 recruitment of just the glycine-rich domain that mediates hnRNPA1-hnRNPA1 interactions was sufficient to induce exon splicing silencer-dependent exon skipping (27Del Gatto-Konczak F. Olive M. Gesnel M.C. Breathnach R. Mol. Cell. Biol. 1999; 19: 251-260Crossref PubMed Scopus (199) Google Scholar). The preceding model may require some re-evaluation for PTB. Various biophysical analyses suggest that pure PTB is an extended monomer (15Amir-Ahmady B. Boutz P.L. Markovtsov V. Phillips M.L. Black D.L. RNA (N. Y.). 2005; 11: 699-716Crossref PubMed Scopus (94) Google Scholar, 24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar), rather than a dimer or higher order oligomer as suggested previously (20Oh Y.L. Hahm B. Kim Y.K. Lee H.K. Lee J.W. Song O. Tsukiyama-Kohara K. Kohara M. Nomoto A. Jang S.K. Biochem. J. 1998; 331: 169-175Crossref PubMed Scopus (84) Google Scholar, 21Perez I. McAfee J.G. Patton J.G. Biochemistry. 1997; 36: 11881-11890Crossref PubMed Scopus (147) Google Scholar). This suggests that any PTB-PTB interactions mediated by RRM2 are not stable in the absence of additional protein-RNA interactions, but it does not preclude a role for such interactions in cooperative binding of two or more PTB monomers to RNA. Protein fragments containing RRM2 and the preceding linker had a tendency for nonspecific self-association in vitro (24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar), but this is unlikely to be connected with the minimal repressor function that we have defined, as this linker region is not part of the minimal 2L-MS2 construct. Another potentially important target for the PTB minimal repressor domain is raver1 (32Huttelmaier S. Illenberger S. Grosheva I. Rudiger M. Singer R.H. Jockusch B.M. J. Cell Biol. 2001; 155: 775-786Crossref PubMed Scopus (94) Google Scholar), which can act as a corepressor with PTB in the TM system (26Gromak N. Rideau A. Southby J. Scadden A.D. Gooding C. Huttelmaier S. Singer R.H. Smith C.W. EMBO J. 2003; 22: 6356-6364Crossref PubMed Scopus (91) Google Scholar). We have characterized an interaction between PTB and a peptide motif repeated within raver1. The peptide ([S/G][I/L]L-GXXP) occurs four times in raver1 and is sufficient to interact with PTB. NMR analysis using isotopically labeled fragments of PTB (24Simpson P.J. Monie T.P. Szendroi A. Davydova N. Tyzack J.K. Conte M.R. Read C.M. Cary P.D. Svergun D.I. Konarev P.V. Curry S. Matthews S. Structure (Camb.). 2004; 12: 1631-1643Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar) indicates that the raver1 peptide interacts specifically with RRM2. 4A. P. Rideau, C. Gooding, P. J. Simpson, T. P. Monie, M. Lorenz, S. Hüttelmaier, R. H. Singer, S. Curry, S. Matthews, and C. W. J. Smith, manuscript in preparation. The peptide does not interact with the linker following PTB RRM2, even though this is required for the splicing repressor activity, as shown here. This suggests that even if the conserved raver1 peptide is a molecular target of RRM2 there are also additional targets of the minimal PTB repressor domain. These targets could be other parts of the raver1 protein, other co-repressor proteins (e.g. the UGC motif binding factors), or components of the core splicing machinery. The identification of RRM2-L as the minimal repressor domain might be taken to call into question the recently proposed models for PTB-mediated repression in which looping of RNA between RRMs 3 and 4 plays a critical role (17Oberstrass F.C. Auweter S.D. Erat M. Hargous Y. Henning A. Wenter P. Reymond L. Amir-Ahmady B. Pitsch S. Black D.L. Allain F.H. Science. 2005; 309: 2054-2057Crossref PubMed Scopus (336) Google Scholar). However, we have only tested the PTB domain requirement at the downstream DY element. Exon skipping mediated by the RRM2L repressor domain remained dependent upon PTB binding sites in the P3 pyrimidine tract (Fig. 7). It is possible that RNA looping between PTB RRMs 34 may play a role at the P3 sites. It would be particularly useful to be able to carry out a similar analysis of PTB domain requirements at the P3 polypyrimidine tract. As one of the consensus elements for splicing of exon 3, it is likely that the P3 pyrimidine tract could be the site at which PTB has its important effects, while the DY site may play an auxiliary role. However, artificial recruitment at the polypyrimidine tract would be technically far more difficult as substitution with MS2 sites at this location would likely interfere with both U2AF65 binding to the polypyrimidine tract and with step 2 of splicing. In the future we will investigate whether it is possible to complement mutations in the P3 PTB binding sites by artificially tethering PTB downstream of P3 or upstream of the branch point. If we were able to use a second tethering system (e.g. Box B, λ N peptide (33De Gregorio E. Preiss T. Hentze M.W. EMBO J. 1999; 18: 4865-4874Crossref PubMed Scopus (136) Google Scholar)), this would allow us to dissect whether there are different minimal PTB domain requirements on each side of the exon. A question related to the direct molecular targets of the PTB repressor domain is "what is the mechanism by which PTB interferes with splicing complex assembly?" The earliest models for PTB-mediated repression involved simple binding competition with U2AF65 at the polypyrimidine tract (12Lin C.H. Patton J.G. RNA (N. Y.). 1995; 1: 234-245PubMed Google Scholar, 13Singh R. Valcarcel J. Green M.R. Science. 1995; 268: 1173-1176Crossref PubMed Scopus (466) Google Scholar). The requirement for additional PTB binding sites remote from the polypyrimidine tract does not rule out this mechanism; cooperative binding of PTB might be required in order for effective competition with U2AF. However, some PTB-regulated exons do not require PTB binding at the polypyrimidine tract (15Amir-Ahmady B. Boutz P.L. Markovtsov V. Phillips M.L. Black D.L. RNA (N. Y.). 2005; 11: 699-716Crossref PubMed Scopus (94) Google Scholar, 16Shen H. Kan J.L. Ghigna C. Biamonti G. Green M.R. RNA (N. Y.). 2004; 10: 787-794Crossref PubMed Scopus (48) Google Scholar, 18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar). Indeed, two recent reports suggest that the mechanism of PTB-mediated repression involves interference with U2AF65 binding to the polypyrimidine tract not by direct steric competition but by preventing positive cross-exon or cross-intron interactions with U1 snRNP bound at a 5′ splice site (18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar, 19Sharma S. Falick A.M. Black D.L. Mol. Cell. 2005; 19: 485-496Abstract Full Text Full Text PDF PubMed Scopus (125) Google Scholar). This is an inherently more interesting mechanism for splicing regulation. As suggested by Valcarcel and colleagues (18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar), this activity of PTB may allow insights into the poorly understood network of interactions involved in exon definition. In particular, identification of the molecular targets of the minimal repressor domain of PTB (and raver1) may reveal some of the crucial players in exon definition. Finally, artificial recruitment of PTB by MS2 has also been demonstrated to restore exon skipping in FGFR2 (34Wagner E.J. Garcia-Blanco M.A. Mol. Cell. 2002; 10: 943-949Abstract Full Text Full Text PDF PubMed Scopus (125) Google Scholar) and Fas (18Izquierdo J.M. Majos N. Bonnal S. Martinez C. Castelo R. Guigo R. Bilbao D. Valcarcel J. Mol. Cell. 2005; 19: 475-484Abstract Full Text Full Text PDF PubMed Scopus (272) Google Scholar) model systems. It would be interesting to analyze the PTB-MS2 deletion mutants in these systems to determine whether the same minimal repressor domain operates in all three systems. We thank Stephen Curry and Steve Matthews for information on PTB domain boundaries prior to publication.

Referência(s)