Polypyrimidine Tract-binding Protein Is Involved in Vivo in Repression of a Composite Internal/3′ -Terminal Exon of the Xenopus α-Tropomyosin Pre-mRNA*

The Xenopus αfast-tropomyosin gene contains, at its 3′ -end, a composite internal/3′ -terminal exon (exon 9A9′), which is subjected to three different patterns of splicing according to the cell type. Exon 9A9′ is included as a terminal exon in the myotome and as an internal exon in adult striated muscles, whereas it is skipped in nonmuscle cells. We have developed an in vivo model based on transient expression of minigenes encompassing the regulated exon 9A9′ in Xenopus oocytes and embryos. We first show that the different α-tropomyosin minigenes recapitulate the splicing pattern of the endogenous gene and constitute valuable tools to seek regulatory sequences involved in exon 9A9′ usage. A mutational analysis led to the identification of an intronic element that is involved in the repression of exon 9A9′ in nonmuscle cells. This element harbors four polypyrimidine track-binding protein (PTB) binding sites that are essential for the repression of exon 9A9′. We show using UV cross-linking and immunoprecipitation experiments that Xenopus PTB (XPTB) interacts with these PTB binding sites. Finally, we show that depletion of endogenous XPTB in Xenopus embryos using a morpholinobased translational inhibition strategy resulted in exon 9A9′ inclusion in embryonic epidermal cells. These results demonstrate that XPTB is required in vivo to repress the terminal exon 9A9′ and suggest that PTB could be a major actor in the repression of regulated 3′ -terminal exon.

The Xenopus ␣ fast -tropomyosin gene contains, at its 3-end, a composite internal/3-terminal exon (exon 9A9), which is subjected to three different patterns of splicing according to the cell type. Exon 9A9 is included as a terminal exon in the myotome and as an internal exon in adult striated muscles, whereas it is skipped in nonmuscle cells. We have developed an in vivo model based on transient expression of minigenes encompassing the regulated exon 9A9 in Xenopus oocytes and embryos. We first show that the different ␣-tropomyosin minigenes recapitulate the splicing pattern of the endogenous gene and constitute valuable tools to seek regulatory sequences involved in exon 9A9 usage. A mutational analysis led to the identification of an intronic element that is involved in the repression of exon 9A9 in nonmuscle cells. This element harbors four polypyrimidine track-binding protein (PTB) binding sites that are essential for the repression of exon 9A9. We show using UV cross-linking and immunoprecipitation experiments that Xenopus PTB (XPTB) interacts with these PTB binding sites. Finally, we show that depletion of endogenous XPTB in Xenopus embryos using a morpholinobased translational inhibition strategy resulted in exon 9A9 inclusion in embryonic epidermal cells. These results demonstrate that XPTB is required in vivo to repress the terminal exon 9A9 and suggest that PTB could be a major actor in the repression of regulated 3-terminal exon.
Alternative splicing is a widespread mechanism in metazoans, by which multiple mRNAs can be produced from a single gene. This process is often subjected to tissue or developmental control and allows the production of protein isoforms differing in specific domains.
Over the last few years, some general topics have emerged from the intensive studies undertaken to elucidate the mechanism of alternative splicing. It is thus established that regulated exons are generally flanked by suboptimal splice sites whose recognition is modulated by cis-acting regulatory sequences present within exons or introns. In most cases, alternative exons are subject to positive and negative regulation which insures a tight control on their usage (reviewed in Ref. 1). Numerous trans-acting factors that modulate alternative splicing have also been characterized. These regulatory factors belong to two broad families of RNA-binding proteins, designated hnRNP proteins and RS-domain proteins. The latter includes the SR protein subfamily (2,3) that is involved in constitutive or alternative exon recognition through exonic sequence called exonic enhancer (4).
Despite this significant breakthrough, the basis of tissue regulation of alternative splicing is still poorly understood in vertebrates. Indeed, contrary to the paradigm described in Drosophila, where many splicing events are controlled by tissue-specific factors (5), to date the implication of such factors appears limited in vertebrates.
To overcome this lack of tissue specific factors it was proposed that ubiquitously expressed splicing factors could determine a cellular code based on variation of their relative levels or activities in various tissues (6). A current model proposes that hnRNP packaging prevents exon use and that their displacement is a preamble to a splicing complex assembly. For example, hnRNP A1 can antagonize the promotion of alternative exon usage that occurs through the binding of SR proteins to purine-rich exonic enhancer (7,8). The ubiquitously expressed polypyrimidine tract-binding protein (PTB designated also hnRNP I) has been implicated in the repression of several alternative exons including exon 7 of ␤-tropomyosin (9), exon N of GABA␥2 (10), exon 5 of N-methyl-D-aspartate receptor NR1 (11), exons IIIb and IIIc of FGF-R2 (12,13), exon N1 of c-src (14), exon SM of ␣-actinin (15), exon EIIIb of fibronectin (16), exon 3 of ␣-tropomyosin (17), and exon 9 of caspase-2 (18). PTB binds especially to pyrimidine-rich intronic splicing silencer elements present upstream or downstream of the alternative exon. These pyrimidine-rich elements frequently contain several high affinity binding motifs UCUU and (C/U)UCUCU (10,19). The physiological relevance of PTB in alternative exon repression was recently demonstrated by RNAi-mediated PTB depletion that leads to an increase in exon IIIb of FGF-R2 and exon EIIIb of fibronectin inclusion (20). From all of these data, it was proposed that PTB has a general role in exon silencing and preventing exon definition (21).
3Ј-Terminal exons can also be differentially processed (reviewed in Ref. 22). However, selection of alternative 3Ј-terminal exons differs in many aspects from that of internal exon, since it could potentially be regulated at several levels including transcriptional termination and competition between splice sites and cleavage-polyadenylation signals. Use of a proximal 3Ј-terminal exon can be simply controlled by the selection of a proximal polyadenylation signal, since cleavage at this site excludes the distal 3Ј-terminal exon and subsequently any competition with this distal exon. Indeed, to date extensive studies performed with the calatonin/calcium gene-related protein (CT/CGRP) and heavy chain immunoglobulin gene * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY536368. □ S The on-line version of this article (available at http://www.jbc.org) contains three supplemental figures. pre-mRNAs have especially highlighted the importance of the cleavage/polyadenylation reaction in the selection of alternative 3Ј-terminal exons. Thus, the processing of the CT/CGRP pre-mRNA is regulated at the level of polyadenylation through the action of a complex intronic enhancer localized downstream of the proximal 3Ј-terminal exon (23). In the same way, it was demonstrated that the differential processing of the heavy chain immunoglobulin gene during B-cell maturation did not require Ig gene-specific sequences (24) and was regulated primarily by changes in the activity of the general polyadenylation factor CstF-64 (25,26). Despite the importance of the polyadenylation signals in the regulation of 3Ј-terminal exons, it is difficult to explain a complete switch between two 3Ј-terminal alternative exons only by a change in the efficiency of the cleavage-polyadenylation efficiency. It is of particular interest to understand how a proximal exon is excluded when a distal 3Ј-terminal exon is selected. Active repression of the proximal 3Ј-terminal exon, as described for the internal alternative exons, could be a mechanism to ensure a complete skipping of this exon.
In this study, with the aim of investigating the involvement of polyadenylation versus splicing in the selection of 3Ј-terminal regulated exons, we developed an in vivo splicing model based on the differential processing of the Xenopus ␣-tropomyosin pre-mRNA. This pre-mRNA possesses three alternative 3Ј-terminal exons designated exons 9A9Ј, 9B, and 9D, whose differential selection generates three distinct mRNAs (27,28). In this article, we focus our attention on the regulation of exon 9A9Ј that is subjected to a complex regulation. Exon 9A9Ј is skipped in nonmuscle cells, whereas it is used as a terminal or internal exon in somitic and adult striated cells, respectively. For this reason, we designated this exon as a composite internal/3Ј-terminal exon. Using minigenes whose expression is targeted in embryonic epidermal or somitic cells or oocytes, we show that a genomic fragment encompassing exon 9A9Ј contains all of the cis sequences necessary for accurate regulation during development. By a mutagenesis approach, we characterize an intronic element lying between the branch site and exon 9A9Ј involved in the repression of this exon in nonmuscle cells. This element, which is pyrimidine-rich, binds PTB in vitro and contains four high affinity PTB binding sites. We first showed that mutation of these binding sites, which abolishes PTB binding, correlates with the loss of exon 9A9Ј repression in the oocyte and embryonic epidermal cells. To ascertain a physiological role of XPTB in exon 9A9Ј repression, we depleted endogenous XPTB in Xenopus embryos using a morpholinobased translation inhibition strategy. XPTB knockdown resulted in a strong inclusion of exon 9A9Ј in embryonic epidermal cells. This study demonstrates that the composite exon 9A9Ј is repressed in nonmuscle cells as a 3Ј-terminal exon by PTB and suggests that PTB could be a major actor in repression of regulated 3Ј-terminal exon. It also establishes Xenopus oocyte and embryo as valid models to study the tissue-specific regulation of alternative splicing in vivo.

EXPERIMENTAL PROCEDURES
Embryo Culture and Injection-Xenopus laevis eggs were obtained from laboratory-reared females and fertilized artificially. Embryos were staged according to the table of Nieuwkoop and Faber (29). For DNA injection, 250 pg of supercoiled DNA were injected in one blastomere at the two-cell stage. Injections were carried out in F1 medium (30) containing 5% Ficoll, and the embryos were cultured until Stage 26.
Oocyte Culture and Injection-Pieces of X. laevis ovary were dissected and treated with collagenase, and individual Stage VI oocytes were prepared for microinjection as described in (31). Injection and culture of oocytes were carried out in OR2 medium (32). For nuclear injection, oocytes were previously centrifuged to allow the visualization of the germinal vesicle as described in Ref. 33. For minigene injection, 1 ng of supercoiled DNA was microinjected into the nucleus. For RNA competition experiments, different amounts of in vitro transcribed RNA were co-injected with minigenes.
Plasmid Constructions-The plasmids pBS-SV40, pBS-actin, and pBS-keratin, which contain the SV40 early promoter, the Xenopus cardiac actin promoter, and the Xenopus keratin promoter, respectively, have been described previously (34). Wild type minigenes were generated by subcloning the BamHI/SalI-digested X␣g7-9B fragment from the pGEM-X␣g7-9B plasmid into the different expression vectors.
All of the mutated minigenes were produced by primer-directed PCR mutagenesis and cassette substitutions. Two cassettes were used depending on the localization of the mutation. A first cassette localized between SnabI and MluI sites present within the 5Ј region and 3Ј region, respectively, of intron 7-8 was used to substitute all of the mutations introduced in intron 7-8 upstream of the MluI site. A second cassette localized between the MluI restriction site and a BsaBI restriction site present within the 5Ј region of intron 9A9Ј-9B was used to substitute all of the mutations introduced in intron 7-8 downstream of the MluI restriction site or in exon 9A9Ј. For each mutation, the amplified region was cloned into the pGEM-T vector and sequenced. The mutated fragment was then substituted to the wild type sequence using the SnaBI/MluI or the MluI/BsaBI cassettes.
For cross-link experiments, the open reading frame of XPTB was amplified by PCR from the hnRNPI cDNA (35) using an upstream primer containing a BamHI restriction site and a downstream primer designed to remove the stop codon and containing a NotI restriction site. The product was cloned as a BamHI/NotI fragment into the pT7TS-V5 vector in frame with the carboxyl V5 epitope tag. The pT7TS-V5 vector was constructed by cloning the V5 epitope tag into the BglII/SpeI restriction sites of the pT7-TS vector (provided by Paul Krieg, University of Texas).
Isolation of RNA and RT 1 -PCR Analysis of the Exogenous Transcripts-Total RNA was isolated from oocytes and embryos by Harland and Misher's proteinase K/LiCl method (36). When RNA preparations were used to assay exogenous transcripts, a RQ1 DNase treatment was added in order to remove any residual plasmid DNA template. For RT-PCR, cDNAs were obtained using 10 g of RNA and 1 g of an oligo(dT) anchor primer (5Ј-VTTTTTTTTTTTTTTTTCAGCTGTAGC-TATGCGCACAG-3Ј). The reaction was performed in 25 l of 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2, 10 mM dithiothreitol, and 1 mM dNTPs. Annealing of the oligo(dT) anchor primer was performed for 3 min at 85°C, followed by 5 min at 37°C, after which 100 units of Moloney murine leukemia virus reverse transcriptase (Promega) was added, and the reaction was further incubated at 37°C for 30 min. Samples of 5 l were then used for the PCR amplification that was performed with 32 P-end-labeled sense primers specific of each minigene and an antisense PCR-anchor primer (5Ј-CAGCTGTAGCTATGCG-CACCAG-3Ј). Sense primers were either actin ex7ss oligonucleotide (5Ј-CACAGCACGGGGATCCAGTC-3Ј) or keratin ex7ss oligonucleotide (5Ј-GTCCGGATCCAGTCTCCCAG-3Ј) or sv40ex7ss oligonucleotide (5Ј-GCTCAGATCCAGTCTCCAGAAG-3Ј). Following an initial denaturation a 94°C for 3 min, the reaction was carried out for 22 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 3 s. Labeled PCR products were analyzed on a 4% nondenaturating polyacrylamide gel. Quantification of radioactive PCR products was performed on a "Storm" Amersham Biosciences PhosphorImager using ImageQuant software. In all cases, at least three independent microinjection experiments were carried out and analyzed. For the analysis of mutated minigenes, the wild type minigene was always microinjected in parallel as a control.
Full-length recombinant V5-tagged XPTB was produced in nucleasetreated rabbit reticulocyte lysate according to instructions (Promega).
Antibodies and Western Blot Analysis-XPTB antibodies were raised in rabbits against two separate synthetic peptides encompassing residues 40 -55 (peptide 1, YGSNGNDSKKFKGDGR) and 438 -453 (peptide 2, PREGQEDQGLTKDYST) of XPTB by Eurogentec (Seraing, Belgium). XPTB antibodies used in this study were then purified by immunoaffinity chromatography against the second synthetic peptide. Anti-V5 monoclonal antibody was purchased from Invitrogen. PCNA and ␤-tubulin monoclonal antibodies were purchased from Sigma.
In vitro translated XPTB was immunoprecipitated using Protein G-Sepharose beads coated with monoclonal anti-V5 antibody (Invitrogen). Endogenous XPTB from germinal vesicle extracts was selected using protein A-Sepharose beads coated with our anti-XPTB serum. Immunoselection was performed at 4°C in IP buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Igepal). The beads were subsequently washed three times with IP buffer containing 0.25% Igepal. Immunoselected proteins were eluted with 20 l of SDS-PAGE loading buffer and fractionated on a 10% SDS-PAGE. Cross-linked proteins were visualized on a PhosphorImager or transferred to nitrocellulose membrane for Western blot analysis.
Morpholino Oligonucleotide Injection and XPTB Rescue-Morpholino oligonucleotides (MO) were obtained from Gene Tools, LLC. The XPTB-MO sequence is 5Ј-AACAATTCCTTCCATGGCACACTAA-3Ј, and the control MO sequence is 5Ј-CCTCTTACCTCAGTTACAATT-TATA-3Ј. Oligonucleotides were resuspended in distilled H 2 O to a stock concentration of 10 g/ml, and 20 ng/blastomere was injected into both blastomeres at the two-cell stage. For XPTB rescue, a V5-tagged XPTB mRNA that had third base modification at the 5Ј-end to prevent its hybridization with XPTB-MO was prepared. The sequence of the 5Ј modification is 5Ј-GGGATCCACCATGGAGGGCATAGTCCAAGATA-TAACAG-3Ј. The start codon and modified nucleotides are underlined. 100 pg of V5-tagged XPTB mRNA was microinjected together with 40 ng of XPTB-MO.

Minigenes Driven by Tissue-specific Promoters Recapitulate the Specific Use of Exon 9A9Ј in Muscle Tissue and Its
Repression in Nonmuscle Tissues-The differential processing of the 3Ј-terminal region of the Xenopus ␣ fast -tropomyosin pre-mRNA generates three distinct mRNAs (Fig. 1A). In somites, exon 9A9Ј is used as a terminal exon with the selection of a polyadenylation signal designated ␣e (where "e" represents embryonic), leading to the production of XTM␣7 RNA. In embryonic heart and adult striated muscles, an internal 5Ј splice site present within exon 9A9Ј is selected, and exon 9A is spliced to the terminal exon 9B to produce XTM␣2 RNA (28). In this case, one of the two polyadenylation signals arranged in tandem FIG. 1. Differential processing of the 3-terminal region of the Xenopus ␣-fast tropomyosin pre-mRNA. A, schematic diagram of the 3Ј-terminal region of the Xenopus ␣-fast tropomyosin RNA and its alternative RNA processing pattern in somites, adult striated tissues, and nonmuscle tissues. The boxes represent exons, and horizontal lines represent introns. ␣e, ␣a, and ␣nm represent the polyadenylation signals present within exons 9A9Ј, 9B, and 9D, respectively. B, schematic representation of the minigene constructs. The boxes represent exons, and horizontal lines represent introns. Positions of the anchor primer and the 5Ј-end-radiolabeled primers that hybridize at the junction of the promoter and exon 7 are indicated by the arrows. C, the structure and size of all splicing events are given. An asterisk marks the end of the PCR product that is radiolabeled. present within exon 9B and designated ␣a (where "a" represents adult) is used. Therefore, exon 9A9Ј behaves like a composite internal/terminal exon, depending on the myogenesis stage at which the ␣-tropomyosin gene is expressed. Finally, in nonmuscle cells, the processing reaction skips over exons 9A9Ј and 9B and splices exon 8 directly to the terminal exon 9D to produce XTM05 RNA (27).
To study the mechanisms involved in the regulation of the composite exon 9A9Ј during development, the genomic region encompassing exon 7 to exon 9B was cloned by PCR and further analyzed (the sequence and its analysis are available as supplementary Fig. 1). The genomic fragment, designated X␣g7-9B, was then cloned downstream of tissue-specific promoters in order to study the differential use of the terminal exons in muscle and nonmuscle tissues during early development. Three promoters were used: the Xenopus cardiac actin promoter that is activated, around Stage 12, specifically in somites and embryonic heart (37); the Xenopus larval keratin promoter that is specifically activated in epidermis (38); and the SV40 early promoter that is functional in Xenopus oocyte (39). The cardiac actin and keratin-driven minigenes were injected into two-cell stage embryos, which were then cultured until tail bud stage (Stage 26). The mRNA was assayed by labeled 3Ј-rapid amplification of cDNA ends in samples representing equal numbers of whole embryos (Fig. 1B). The size and the composition of each expected PCR product are indicated in Fig. 1C.
The results are presented in Fig. 2A. A major product of 352 nt, containing exon 9A9Ј (␣7-type RNA) was present with the cardiac actin promoter. A faint product of 468 nt, containing exons 9A and 9B (␣2-type RNA) was also present. A minor splicing product of 389 nt was also amplified. Cloning and sequencing of the corresponding fragment showed that it corresponds to a maturation product in which exon 8 was spliced directly to exon 9B, and it was designated ␣2⌬9A. Quantification of the different amplified fragments showed that ␣7-type RNA represents about 87% of the matured transcripts, whereas the minor products ␣2-type and ␣2⌬9A-type RNAs represent altogether Ͻ12% of the matured RNA. On the other hand, with the keratin-driven minigene, Ͼ70% of the mature transcripts correspond to ␣2⌬9A-type RNA. In non-muscle tissues, the endogenous ␣-tropomyosin gene produces RNA containing exon 8 and exon 9D but no ␣2⌬9A-type RNA. Therefore, our results suggest that in the context of the minigene, in which exon 9D is not present, the specific repression of exon 9A9Ј in nonmuscle cells still occurs and that exon 8 is spliced to exon 9B as a default choice.
Splicing of the endogenous ␣-tropomyosin pre-mRNA is also strictly regulated in the oocyte, since only one ␣-tropomyosinderived mRNA (XTM05) is present in such cells (27). Therefore, oocytes could be envisioned as an attractive model to study the ␣-tropomyosin pre-mRNA splicing regulation. Accordingly, we tested the behavior of a minigene in such a context. An SV40driven minigene was injected in the germinal vesicles of Stage VI oocytes, followed by a 6-h incubation. The mRNA was assayed by 3Ј-rapid amplification of cDNA ends. As described in embryonic epidermal cells, Ͼ80% of ␣2⌬9A-type RNA was produced in oocytes (Fig. 2B). Two bands corresponding to unspliced or partially spliced products primed from a poly(A) stretch present within intron 8-9A were also amplified. Their prevalence varies between experiments (data not shown), which could be related to a variation in the transcription efficiency according to the oocyte batch. The presence of these unspliced or partially spliced products suggests that splicing of 7-9B pre-mRNAs was rate-limiting compared with transcription in oocytes.
Altogether, the results observed in the embryo are in accord-ance with the pattern of expression of both promoters and demonstrate that our minigenes can recapitulate the specific use of exon 9A9Ј in somites and its repression in epidermis. Such repression is also observed in oocytes. Therefore, these minigenes constitute valuable tools to study the elements involved in the tissue-specific usage of the composite internal/ terminal exon 9A9Ј. A mutational analysis demonstrated that exon 9A9Ј is defined by suboptimal 3Ј and 5Ј splice sites as well as a weak polyadenylation signal (data available as supplementary Fig. 2). These results indicate that exon 9A9Ј is a FIG. 2. Minigenes injected into Xenopus embryos and oocyte recapitulate the endogenous usage of exon 9A9. A, RNA from minigenes driven by the cardiac actin or keratin promoters injected into two-cell stage embryos was subjected to RT-PCR using an anchor primer and a 5Ј-end-radiolabeled primer that hybridizes at the junction of the promoter and exon 7 sequences. Labeled PCR products were analyzed on a 4% nondenaturating acrylamide gel and autoradiographed. The promoter is indicated at the top of the gel. Relative quantification was performed on a PhosphorImager to yield the proportion of each product given in the table below each panel. The sizes of the PCR products were estimated by comparison with pBS HpaII-digested fragments end-labeled. B, RNA from a minigene driven by the SV40 early promoter injected into the nucleus of Stage VI oocyte was subjected to RT-PCR and analyzed as described in the legend for A.
poorly defined exon whose recognition, in muscle and nonmuscle cells, is modulated by cis acting regulatory sequences present within exons or introns. Since mutations in exon 9A9Ј have not revealed any exonic enhancer (data not shown), we focused our attention to the intron upstream of exon 9A9Ј.

The Major Branch Site Used in the Splicing of Exon 8 to Exon 9A9Ј Is Localized 274 nt Upstream of the 3Ј Splice Site of Exon
9A9Ј-The analysis of the intronic region upstream of exon 9A9Ј showed many potential branch sites lying in a region between Ϫ130 and Ϫ274 nt from the 3Ј splice site (see supplementary Fig. 1). To assign experimentally the major branch site used in this region, we mutated individually the adenine residue of each potential branch site to a thymidine or cytosine residue. This approach is often inefficient in vitro, since the mutation of a branch site leads usually to the activation of a cryptic one, but we reasoned that in vivo, for such distant branch sites, some constraint might specifically favor one branch site. The different mutants placed under the control of the cardiac actin promoter were tested in the embryo. As presented in Fig. 3, the mutation of the branch site at Ϫ274 nt (Fig. 3, lane 2) dramatically changed the pattern of splicing to exon 9A9Ј skipping, whereas the other mutants were ineffective (Fig. 3, lanes 3-6). An increase of ␣2-type RNA was also observed, suggesting that this mutation specifically targets the ␣7 pathway and favors the ␣2-type RNA production. The use of the Ϫ274 branch point is not surprising, since it is 100% homologous to the consensus sequence, and it is associated with a strong polypyrimidine stretch composed of 21 consecutive pyrimidines. Additionally, there is no dinucleotide AG between this branch site and the 3Ј splice site. However, it could be envisioned that the mutation Ϫ274 nt disrupts an enhancer element involved in exon 9A9Ј usage. To address this question, an additional mutant designated mutPolyPy was created by shortening the polypyrimidine stretch and subsequently the strength of the Ϫ274 branch site. The 10 last nucleotides of the polypyrimidine stretch (Ϫ253 to Ϫ264) were replaced by the sequence GCACTGCTACGA. This mutant driven by the cardiac actin promoter resulted in a strong decrease in ␣7-type RNA production as the specific mutation of the Ϫ274 branch point (Fig. 3, lane 7; compare with lane 2). Since two different mutations, targeting elements important in the function of a branch site, led to a drastic reduction in ␣7-type RNA produc-tion, we conclude that exon 9A9Ј probably uses a major branch site located 274 nt upstream of the 3Ј splice site to produce ␣7 RNA.
The use of distant branch sites contributes to the weakness of the branch site but also provides space for regulatory sequences. We therefore postulated that the region between this branch site and the AG element could contains cis elements involved in the tissue-specific use of exon 9A9Ј.
PTB Binding Sites within a Pyrimidine-rich Sequence Upstream of Exon 9A9Ј Mediate Its Repression in Embryonic Epidermal Cells-To determine whether sequences lying between the branch site and the 3Ј splice site of exon 9A9Ј are important in the regulation of this exon, we constructed three deletion mutants designated ⌬1, ⌬2, and ⌬3 in which 80, 150, and 230 nt, respectively, were removed upstream of the 3Ј splice site. To avoid any direct effect upon the 3Ј splice site, the deletions maintained the last 14 nt of the intron upstream of exon 9A (Fig. 4A). In nonmuscle cells, the mutant ⌬1 resulted in partial activation of exon 9A9Ј (Fig. 4B, lane 2). The effect was reinforced with the mutants ⌬2 and ⌬3 in which 100% of mature RNA contained exon 9A9Ј (lanes 3 and 4). These results suggested that repressor elements lie in the region deleted in mutant ⌬2. To ensure that the effect of the mutation ⌬2 was not secondary to a shortening of the intron or to a bringing closer of the branch point and the 3Ј splice site, the region was replaced by a 150-nt sequence from pBluescribe (150pBS). This substitution resulted in 77% of exon 9A9Ј usage compared with 100% usage when the element was deleted (lane 5). This result suggests that although the long distance between the branch site and the 3Ј splice site may contribute to the weakness of the 3Ј splice site, specific sequences lying between Ϫ165 and Ϫ15 from the AG border are probably involved in exon 9A9Ј repression in nonmuscle cells. In muscle cells, in which exon 9A9Ј is already used very efficiently, the different deletions did not change the splicing pattern (Fig. 4C, lanes 2-4).
The 150-nt silencer element was notable for its pyrimidinerich composition with four UCUU motifs corresponding to high affinity PTB binding sites (10,19). Since PTB has been described to repress the splicing of some alternative exons, we sought to determine whether these PTB binding sites designated PTB1-PTB4 (Fig. 4A) are involved in the repression of exon 9A9Ј and whether PTB binds to the repressor element. For each PTB site, uridines were mutated to cytidines in order to maintain the overall pyrimidine composition. Each mutant was placed under the control of the keratin promoter and analyzed independently and in combination. Individual mutations and double mutations had no effect (data not shown), whereas mutation of all four PTB binding sites present within the 150-nt silencer element was as effective as the 150pBS substitution (Fig. 4D, lane 2; compare with Fig. 4B, lane 5). This suggests that it is not only the overall pyrimidine composition of the 150PY element that is involved in the repression of exon 9A9Ј but that PTB-specific binding sites are required for the repression of exon 9A9Ј in epidermal cells.
To analyze whether the Xenopus PTB (XPTB) binds to the 150-nt pyrimidine-rich silencer element, a recombinant XPTB protein tagged with a V5 epitope was synthesized in a nuclease-treated rabbit reticulocyte lysate. XPTB programmed lysate and nonprogrammed lysate were then tested by UV crosslinking using probes corresponding to wild type 150PY RNA or mutated 150PY RNA in which the four PTB binding sites were mutated. After immunoprecipitation of the cross-link products with a V5 antibody, a strong signal migrating at about 60 kDa, the expected mobility for XPTB, was detected in the XPTB programmed lysate incubated with the wild type 150PY RNA (Fig. 5, top panel, lane 1). Only a faint signal was observed with FIG. 3. Identification of the major exon 9A9 branch site. The putative adenosine branch sites upstream of exon 9A9Ј were mutated to thymidine or cytidine residues. The 10 last nucleotides of the polypyrimidine stretch of the Ϫ274 branch site were replaced by the sequence GCACTGCTACGA. The mutated minigenes under the control of the cardiac actin promoter were injected into Xenopus embryos. RNA was then subjected to RT-PCR and analyzed as described in the legend to Fig. 2. Quantification was performed on a PhosphorImager to yield the proportion of each product.
the mutated 150PY RNA (lane 3), although equal amounts of the recombinant protein XPTB-V5 had been immunoprecipitated (Fig. 5, bottom panel, compare lanes 1 and 3). Finally, this signal was specific for XPTB-V5 programmed lysate (compare with lanes 2 and 4). Therefore, XPTB can specifically interact with the four high affinity UCUU motifs present within the 150PY element. These data suggest that PTB is a strong candidate for the activity that represses exon 9A9Ј in embryonic epidermal cells.
PTB Represses Exon 9A9Ј Usage in Oocyte through Binding to Four High Affinity PTB Binding Sites-The Xenopus oocyte offers a unique possibility to study alternative splicing regulation. We therefore wished to study whether repression of exon 9A9Ј in this nonmuscle cell was controlled in a similar way as in embryonic epidermal cells. For this purpose, the mutants ⌬2 and mutPTB 1-4 placed under the control of the SV40 promoter were tested in the oocyte. In agreement with its behavior in embryonic epidermal cells, the mutation ⌬2 gave rise to an almost complete derepression of exon 9A9Ј in oocyte (Fig. 6A,   lane 2). Similarly, the mutation mutPTB 1-4 resulted in a strong activation of exon 9A9Ј usage (lane 3), indicating that mutation of the four high affinity PTB binding sites was also sufficient to derepress exon 9A9Ј in the oocyte. These results suggested that PTB was involved in the repression of exon 9A9Ј in the oocyte through binding to the four high affinity binding sites. To confirm this assertion, we tested whether competitor RNA corresponding to the sequence that had been removed in mutant ⌬2 (150PY RNA) and that binds PTB could activate exon 9A9Ј splicing in the oocyte. Increasing amounts of competitor RNA corresponding to the 150-nt pyrimidine-rich sequence were microinjected into oocyte nuclei along with the wild type minigene. The 150PY competitor RNA caused a complete derepression of exon 9A9Ј (Fig. 6B, lanes 2-4), indicating that 150PY has titrated out factors that are important in the repression of exon 9A9Ј. It is important to notice that the partial spliced products primed from intron 8 are not affected by the 150PY RNA, demonstrating that constitutive splicing is not altered by the 150PY RNA. Competitor RNA corresponding FIG. 4. PTB motifs within an intronic silencer element present upstream of exon 9A9 mediates its silencing in embryonic epidermal cells. A, nucleotide sequence of deletion mutants in intron 8-9A. The partial nucleotide sequence of intron 8-9A is presented. The major adenosine branch site is shown in boldface capital letters, whereas the putative adenosine branch sites are indicated in boldface lowercase letters. The deletions ⌬1, ⌬2, and ⌬3 extended from vertical mark I to vertical marks II, III, and IV, respectively. They maintained the 14 last nucleotides of intron 8-9A and extended over 80, 150, and 220 nt, respectively. A 150-nt sequence derived from pBS was substituted to the sequence deleted in the mutant ⌬2 to produce the mutant ⌬2 150pBS. The four UCUU PTB motifs designated PTB1-PTB4 are framed. B and C, the mutated minigenes comprising the different deletions and under the control of the keratin (B) or cardiac actin (C) promoters were injected into Xenopus embryos. D, a mutant minigene, driven by the keratin promoter, in which the four PTB UCUU motifs were mutated to CCCC, was microinjected into Xenopus embryos. B-D, RNA was then subjected to RT-PCR and analyzed as described in the legend to Fig. 2. Quantification was performed on a PhosphorImager to yield the proportion of each product.
to the 150pBS sequence that is unable to restore exon 9A9Ј repression was ineffective in activating exon 9A9Ј splicing (Fig.  6B, lanes 5-7).
The efficient binding of the splicing repressor XPTB to the 150PY RNA is consistent with the hypothesis that the 150PY competitor RNA traps XPTB, which is required for exon 9A9Ј repression. To assess the ability of XPTB to repress exon 9A9Ј splicing in the oocyte, we first studied its expression in this cell. Western blot analysis was performed on oocytes with affinitypurified polyclonal antibodies raised against polypeptides. Stage VI oocyte extracts as well as purified nuclear and cytoplasmic fractions (Fig. 7A, lanes 1-4) were analyzed. The membrane was also probed with PCNA and ␤-tubulin antibodies to check the purity of the cytoplasmic and nuclear fractions. A protein with an apparent molecular mass of 60 kDa is detected in both cytoplasmic and nuclear fractions (lanes 2-4). This protein is not detected with the preimmune serum (data not shown). We conclude therefore that XPTB is present in the nucleus and cytoplasm of Stage VI oocyte.
To directly test whether XPTB present in oocyte nuclei can bind to the 150PY RNA, UV cross-linking of oocyte nuclei protein extracts was performed with the wild type 150PY RNA and the mutated 150PY RNA. In vitro translated radiolabeled XPTB protein was run in parallel for comparison. A band at ϳ60 kDa, the expected mobility for XPTB, was present for the wild type RNA (Fig. 7B, lane 3). This band co-migrates with the in vitro translated XPTB (compare lanes 3 and 5). A faint signal was also detected for the mutated RNA (lane 1), suggesting that in the context of a nuclear extract, a low level of XPTB binding to the 150PY RNA occurs outside of the four high affinity binding sites. To confirm that the cross-linked band was XPTB, an immunoprecipitation was then performed. The 60-kDa band was immunoprecipitated specifically by anti-XPTB serum (Fig. 7C, lane 2) but not by preimmune serum (lane 1), confirming that nuclear XPTB can bind to the 150PY repressor element. Our data strongly suggest that PTB represses exon 9A9Ј usage in the oocyte through binding to PTB 1-4 high affinity binding sites.
XPTB Is Required to Repress Exon 9A9Ј in Embryonic Epi-dermal Cells-Morpholino-based translational inhibition is a powerful tool to specifically knock down the level of an endogenous protein during Xenopus development (40,41). An MO directed against the 5Ј region of XPTB mRNA or a control nonspecific morpholino oligomer was microinjected into both blastomeres of Xenopus embryos at the two-cell stage. Microinjection of embryos with the XPTB-MO but not with the nonspecific MO resulted in a strong inhibition of the translation of XPTB. Indeed, 100% of Stage 26 embryos (n ϭ 30) injected with MO-XPTB showed a 7-fold reduction of endogenous XPTB as assayed by Western blot quantification (Fig 8A, lanes 5-9;  compare with lanes 1 and 2). To further address the effect of XPTB depletion on the inclusion of exon 9A9Ј in nonmuscle cells, XPTB-MO were microinjected along with the minigene encompassing exons 7-9B and driven by the Xenopus keratin promoter that targets expression to epidermis. Embryos were fixed at Stage 26, and extracted RNA and proteins were analyzed by RT-PCR analysis and Western blot, respectively. As previously described, a strong skipping of exon 9A9Ј was observed in the morpholino-uninjected control embryos (Fig. 8B, lane 1). XPTB-MO-mediated knockdown of XPTB resulted in an almost complete inclusion of exon 9A9Ј (lanes 3 and 4), whereas the MO control did not change the processing pattern (lane 2). To confirm that the XPTB-MO-induced phenotype was generated by a specific knockdown of XPTB, in vitro transcribed V5-tagged XPTB A, the ⌬2 and mutPTB 1-4 mutants driven by the SV40 promoter were microinjected into the oocyte nucleus. B, the wild type minigene was injected, alone or with increasing amounts (13,65, and 325 fmol) of in vitro transcribed 150PY and 150pBS RNAs, into the oocyte nucleus. A and B, RNA was subjected to RT-PCR and analyzed as described in the legend to Fig. 2. Quantification was performed on a PhosphorImager to yield the proportion of each product. mRNAs that have a third-base modification in the sequence spanning the translation start site were injected together with XPTB-MO. Rescued expression of V5-tagged XPTB in endogenous XPTB-depleted embryos restored exon 9A9Ј skipping (lanes 5 and 6), demonstrating that XPTB can directly repress exon 9A9Ј. These results demonstrate that endogenous XPTB is re-quired in vivo to repress exon 9A9Ј in embryonic epidermal cells in the context of a minigene construct. DISCUSSION In the present study, we have analyzed the differential usage of the composite internal/terminal exon 9A9Ј of the Xenopus ␣-fast tropomyosin pre-mRNA. This exon was chosen because it appears to be a good model to study the contribution of splicing and polyadenylation in the regulation of 3Ј-terminal exons.
To study the mechanisms involved in the regulation of exon 9A9Ј, we first cloned the genomic region encompassing exons 7-9B and constructed minigenes driven by tissue-specific promoters. Our results show that the pre-mRNAs derived from these minigenes are spliced according to the tissue-specific promoters. In muscle cells, exon 9A9Ј is used as a terminal exon to produce ␣7-type RNA, whereas it is skipped in nonmuscle cells to produce ␣2⌬9A-type RNA. Although no endogenous isoform corresponding to a direct splicing of exon 8 to exon 9B was reported in Xenopus, the generation of ␣2⌬9A-type RNA from our minigenes indicates that there is no intrinsic blockage of the splicing of both exons. Indeed, in rat, a brain-specific isoform designated TMBr2 is synthesized by direct splicing of exon 8 to exon 9B, resulting in the inclusion of a specific 30-amino acid carboxyl sequence (42). This suggests that re- FIG. 7. XPTB expression in the oocyte and UV cross-linking of oocyte nuclear extracts with the silencer element. A, proteins from one Stage VI oocyte (O), from the cytoplasmic fraction equivalent to one oocyte (C), and from the nuclear fraction equivalent to one (N1) or four (N4) oocytes were analyzed by Western blot using an affinitypurified XPTB antibody. To check the purity of the nuclear and cytoplasmic fractions, the membrane was also probed with PCNA and ␤-tubulin antibodies. B, wild type 150PY RNA (WT-150PY) and 150PY RNA mutated at the four high affinity PTB binding sites (Mut-150PY) were incubated in Stage VI oocyte nuclear extract (N.E.). Samples were UV-irradiated and treated with RNase T1, and cross-linking products were resolved onto a 10% polyacrylamide gel in the presence of SDS and visualized by autoradiography. 35 S in vitro translated XPTB protein was run in parallel. The protein size markers are indicated to the left. C, cross-linking reactions of the 150PY RNA were immunoprecipitated with XPTB antiserum (ϩ) or preimmune serum (Ϫ). The immunoprecipitated proteins were resolved onto a 10% polyacrylamide gel in presence of SDS and visualized by autoradiography. The 60-kDa crosslinked XPTB protein is indicated. The protein size markers are indicated to the left. Each lane represents an individual embryo. B, the wild type minigene driven by the keratin promoter was injected alone or together with control morpholino or XPTB-MO. To check the specific effect of XPTB-MO, a synthetic 5Ј-modified V5-tagged XPTB RNA was injected together with XPTB-MO. XPTB depletion and V5-tagged XPTB expression were verified by Western blot using the XPTB antiserum and a monoclonal anti-V5 antibody, respectively. The XPTB antiserum recognizes a nonspecific band that appears in all four lanes and is indicated with a star. The band corresponding to XPTB is indicated by an arrow. RNA extracted from a batch of three embryos was subjected to RT-PCR and analyzed as described in the legend to Fig. 2. Quantification was performed on a PhosphorImager to yield the proportion of each product. pression of exon 9A in nonmuscle cells can be controlled independently of exon 9B selection.
As already described for several mutually exclusive exons, the 3Ј splice region of exon 9A9Ј is characterized by a distant branch site. Thus, five potential branch sites are present 149 -274 nt upstream of the 3Ј splice site of exon 9A9Ј. Although, we have not experimentally mapped the branch point by an in vitro assay, the dramatic effect of the mutations to the Ϫ274 nt branch point or to the polypyrimidine-associated sequence strongly suggests that Ϫ274 is the major branch point used in embryonic muscle cells. Whereas a mutational approach is often inappropriate to determine a branch point in vitro because of the activation of cryptic sites, this result suggests that the usage of the Ϫ274 nt branch point could be highly regulated in vivo. Indeed, the increase in ␣2 RNA production with the Ϫ274 mutation indicates that branch sites other than Ϫ274 are functional. Our current explanation is that the other functional branch sites are unavailable in embryonic muscle cells. In this respect, it is interesting to note that we have identified an enhancer element necessary for exon 9A9Ј usage in embryonic muscle cells that spans the Ϫ196, Ϫ158, and Ϫ149 potential branch sites. 2 Whereas the long distance between the branch site and the 3Ј splice site is a prerequisite for exon 9A9Ј repression in nonmuscle cells, splicing also could be activated by the removal of specific sequences between the branch site and the 3Ј splice site. Competitor RNAs corresponding to these specific sequences activated splicing in the oocyte, which suggests that they bind and sequester factors that normally repress exon 9A9Ј. The intronic sequence and the corresponding competitor RNA contain four high affinity PTB binding sites. UV crosslinking experiments and immunoprecipitation demonstrated that PTB binds to the competitor RNAs. Involvement of PTB as a repressor was strengthened by the observation that specific mutation of the four high affinity PTB binding sites that strongly reduced PTB binding also reduced exon 9A9Ј repression in epidermal cells and the oocyte. To give definitive evidence for XPTB involvement in exon 9A9Ј repression, we specifically depleted XPTB in embryos using a morpholino-based translation inhibition strategy. As expected, the knockdown of XPTB caused a strong derepression of exon 9A9Ј in embryonic epidermal cells. These data demonstrate not only that XPTB is required for exon 9A9Ј exclusion in nonmuscle cells but that this exclusion is based mainly or exclusively on XPTB-mediated repression. Importantly, knockdown of XPTB resulted in a stronger increase of exon 9A9Ј inclusion than the mutation of the four high affinity PTB binding sites, suggesting that additional PTB binding sites are involved in exon 9A9Ј repression.
PTB has been implicated in the repression of alternative exons of many genes including ␣-fast tropomyosin, ␣-actinin, ␤-tropomyosin, and c-src (9,14,15,17). In most cases, PTB seems to reinforce the default splicing pattern through the repression of the tissue-specific exon inclusion. Several nonexclusive models have been proposed for PTB action. In some cases, PTB is able to compete the binding of U2AF65 to the polypyrimidine tract (43,44). Interestingly, an additional high affinity binding site is present within the polypyrimidine tract associated with the Ϫ274 branch point. However, it is difficult to test in vivo the relevance of this site by a mutational approach, since its destruction weakens the polypyrimidine tract and therefore prevents exon 9A9Ј splicing. 3 Numerous exons repressed by PTB also have PTB binding sites upstream and downstream of the exon. Given that PTB can multimerize, it was proposed that PTB can interact across the exon and se-quester it (45). Such a model could also be proposed for the repression of exon 9A9Ј, since in addition to the four high affinity binding sites characterized in this study and present upstream of exon 9A9Ј, two UCUU motifs are present downstream of exon 9A9Ј. Interestingly, one of these downstream sites comprises the uridine-rich element of the ␣e poly(A) site, and its mutation abolished ␣7 RNA production in somitic cells. 3 This feature does not allow testing by a mutational approach to know whether this downstream site is implicated in the repression of exon 9A9Ј, but it suggests a new model of regulation in which binding of PTB to the UCUU motif present within the uridine-rich element region could preclude CstF-64 binding. This observation raises the possibility that PTB in addition to repressing splicing could also repress polyadenylation. This activity could be the basis for a more general role of XPTB in repression of regulated proximal 3Ј-terminal exon. Finally, it was also shown that PTB binding sites can overlap with enhancer elements. In this case, PTB can compete the binding of activator factors, preventing the formation of an activator complex (46,47). In accord with this model, we identified a complex intronic enhancer element that overlaps with the repressor region. 2 We therefore predict a model where the different interspersed binding sites for PTB are involved in the formation of a strong complex repressor in which the branch site Ϫ274 and the ␣e poly(A) site are inaccessible to the splicing and polyadenylation machineries. In this context, the four high affinity binding sites present within the repressor region and overlapping with the activator sequences could serve as a core element for the binding of PTB that could then bind synergistically to the additional motifs through multimerization leading to a complete sequestration of exon 9A9Ј.
In contrast to this general repressor function, PTB has also been implicated in the activation of 3Ј-terminal exons. In vitro experiments demonstrated that PTB could activate the poly(A) signal of the C2 complement gene by binding to a U-rich sequence (designated USE) immediately upstream to the hexanucleotide AAUAAA (48). The mechanical basis of such a stimulatory effect is unknown and remains to be established. In vivo, PTB has also been shown to activate the proximal terminal exon 4 of the calatonin/calcium gene-related protein pre-mRNA by binding to a downstream intronic enhancer (49). It was proposed that it was acting by counteracting U2AF binding that is inhibitory for the enhancer activity. Surprisingly, it was not studied whether this effect was partially mediated by exon 5 silencing. Indeed, two PTB high affinity motifs are present within the polypyrimidine tract upstream of exon 5.
Our results establish the Xenopus embryo and oocyte as an attractive model to study alternative splicing regulation in vivo. Our data provide further insights into PTB function as a splicing repressor and support a model in which PTB could prevent the definition of a regulated proximal 3Ј-terminal exon. The mechanistic explanation of how PTB inhibits such a definition is under investigation using our model system. It should provide new insights into the relationship between polyadenylation and splicing.