Splicing of a myosin phosphatase targeting subunit 1 alternative exon is regulated by intronic cis-elements and a novel bipartite exonic enhancer/silencer element.

Isoforms of the smooth muscle myosin phosphatase targeting subunit 1 (MYPT1) are generated by cassette-type alternative splicing of exons. Tissue-specific expression of these isoforms is thought to determine smooth muscle-relaxant properties and unique responses to signaling pathways. We used mini-gene deletion/mutation constructs to identify cis regulators of splicing of the chicken MYPT1 central alternative exon. Comparisons of alternative exon splicing were made between smooth muscle cells of the fast-phasic contractile phenotype (gizzard), in which the central alternative exon is skipped, and slow tonic contractile phenotype (aorta), in which the alternative exon is included. We demonstrate that splicing of the alternative exon requires a cis-enhancer complex in the vicinity of the alternative exon 5'-splice site. This complex consists of two UCUU motifs in an intronic U-rich sequence (putative PTB (polypyrimidine tract binding) or T cell inhibitor of apoptosis-1 binding sites), an intronic 67-nucleotide enhancer that has similarities with the cardiac Troponin T MSE3 enhancer, and a potentially novel exonic splicing enhancer. The exonic enhancer contains the palindromic sequence UCCUACAUCCU present in many other transcripts where alternative splicing of exons occurs, suggesting that it may be more broadly active. The exonic enhancer is adjacent to a potentially novel exonic silencer element that contains a 13-nucleotide imperfect palindromic sequence. This silencer, in conjunction with a distal intronic silencer, is proposed to mediate the silencing of splicing of the MYPT1 central alternative exon in the fast phasic smooth muscle phenotype.

Isoforms of the smooth muscle myosin phosphatase targeting subunit 1 (MYPT1) are generated by cassettetype alternative splicing of exons. Tissue-specific expression of these isoforms is thought to determine smooth muscle-relaxant properties and unique responses to signaling pathways. We used mini-gene deletion/mutation constructs to identify cis regulators of splicing of the chicken MYPT1 central alternative exon. Comparisons of alternative exon splicing were made between smooth muscle cells of the fast-phasic contractile phenotype (gizzard), in which the central alternative exon is skipped, and slow tonic contractile phenotype (aorta), in which the alternative exon is included. We demonstrate that splicing of the alternative exon requires a cis-enhancer complex in the vicinity of the alternative exon 5-splice site. This complex consists of two UCUU motifs in an intronic U-rich sequence (putative PTB (polypyrimidine tract binding) or T cell inhibitor of apoptosis-1 binding sites), an intronic 67-nucleotide enhancer that has similarities with the cardiac Troponin T MSE3 enhancer, and a potentially novel exonic splicing enhancer. The exonic enhancer contains the palindromic sequence UCCUACAUCCU present in many other transcripts where alternative splicing of exons occurs, suggesting that it may be more broadly active. The exonic enhancer is adjacent to a potentially novel exonic silencer element that contains a 13-nucleotide imperfect palindromic sequence. This silencer, in conjunction with a distal intronic silencer, is proposed to mediate the silencing of splicing of the MYPT1 central alternative exon in the fast phasic smooth muscle phenotype.
Smooth muscle tissues show considerable phenotypic and functional diversity. Much of the phenotypic diversity in smooth muscle tissues is generated by the alternative splicing of exons from single gene transcripts. As many as 50% of vertebrate genes are estimated to have alternative splicing of exons as a mechanism by which multiple protein isoforms with different or modified functions are generated from a single gene (1)(2)(3). The process of splice site selection and exon definition for exons that are constitutively spliced has been well described (4 -7). Binding of U1 small nuclear ribonucleoprotein particle (snRNP) 1 to the consensus 5Ј-splice site sequence and binding of splicing factor 1 (SF1) and U2 snRNP auxiliary factor (U2AF) to the consensus 3Ј-splice site sequences (branchpoint and polypyrimidine tract, respectively) commits the exon to the splicing pathway (7,8). A number of factors are known to contribute to the "weakening" of the splicing reaction so that splicing of an exon may be regulated. These factors include exon and intron size, RNA secondary structure, and the extent to which the 5Ј-splice site, branchpoint, and polypyrimidine tract sequences match the consensus sequences for U1 snRNP, SF1, and U2AF binding, respectively (9 -11).
A number of cis-elements that regulate the splicing of alternative exons have been identified (reviewed in Refs. 10,[12][13][14]. We are interested in defining the cis-regulatory elements that account for the great variety in exon splicing that generates diversity in smooth muscle. This aspect of smooth muscle phenotypic diversity has not been examined. We have selected the MYPT1 subunit of smooth muscle myosin phosphatase as a model gene to address this question for several reasons: 1) MYPT1 isoforms are thought to determine smooth muscle relaxant properties and tissue-specific relaxant responses to signaling pathways (15,16). 2) Isoforms of MYPT1 are generated by cassette-type alternative splicing of single exons in the central and 3Ј-end of the pre-mRNA. Splicing of each exon is tissue-specific and developmentally regulated (15,17). Around the time of hatching (ED21), the MYPT1 isoforms in the chicken gizzard (phasic fast contracting smooth muscle) completely switch from central exon-in and 3Ј-exon-out to central exon-out and 3Ј-exon-in. In contrast, the slow tonic contracting aorta smooth muscle throughout development expresses MYPT1 isoforms in which the central alternative exon is included, and the 3Ј alternative exon is skipped (15,17).
We hypothesized that cis-acting silencer(s) may mediate the silencing of splicing of the MYPT1 central alternative exon that occurs in the phasic gizzard tissue in vivo around the time of * This work was supported in part by National Institutes of Health (NIH) Grants K08HL-03275 and R01 HL-66171 (to S. A. F.). 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  hatching. We have previously shown that smooth muscle phenotypic diversity is rapidly lost under routine cell culture conditions, in which phasic smooth muscle cells (SMCs) revert to the more embryonic tonic phenotype and display tonic/slow contractile properties in vitro (17,23). We therefore developed a novel in vivo (whole animal) plasmid delivery method to identify cis-elements that regulate splicing of the MYPT1 central alternative exon and may be important in the generation of smooth muscle phenotypic diversity.

EXPERIMENTAL PROCEDURES
Data Base Search to Identify Candidate Regulatory Sequences-Several strategies were used to identify candidate splicing regulatory sequences. First, the intron sequences (GenBank TM accession numbers AF110177 and AF110176) were compared with sequences in the NCBI data base (BLAST search) to identify sequences that are also present in introns from other genes. Second, the intron sequences were scanned with sequences known to regulate the splicing of other pre-mRNAs using the GeneQuest program (DNASTAR, Madison, WI). Third, the alternative exon sequence was searched for matches to previously characterized ESE sequences, including optimal binding sites for the SR proteins SC35, SF2/ASF, SRp40, and SRp55, as determined by functional systematic evolution of ligands by exponential enrichment (SELEX) experiments (18,19). Finally, the corresponding chicken and rat MYPT1 sequences were compared with each other to identify sequence identities and dissimilarities between them within the alternative exon and flanking introns. This approach identified a number of regions/sequences in the alternative exon and downstream intron that were potentially significant (described under "Results") and facilitated targeted deletion and mutation analyses to determine their functional significance.
Construction of Wild-type and Mutant MYPT1 Mini-genes-A 5.4-kb genomic fragment containing the chicken MYPT1 central alternative exon and flanking introns and exons ( Fig. 2A) was generated by PCR using the Expand Long Template PCR system (Roche Diagnostics). Genomic DNA from chicken liver was amplified using a sense oligonucleotide at ϩ1576 (5Ј-AAAGGATCCAGACTTGCCTATGTTGCACC-3Ј) and an antisense oligonucleotide at ϩ1925 (5Ј-CACGAGCTCAAGTTG-AACCCGTTGTACTC-3Ј) of the chicken MYPT1 M133 cDNA sequence (20). BamHI and SacI restriction enzyme recognition sequences were placed at the 5Ј-ends of the oligonucleotides. The genomic fragment was cloned into the pCR2.1 vector using the Original TA cloning kit (Invitrogen). The exon and splice site sequences in the cloned genomic fragment were confirmed by DNA sequencing to match the sequence as previously determined (17). The entire MYPT1 genomic fragment was removed from the pCR2.1 vector by restriction digestion at BstXI sites in the pCR2.1 vector and cloned into the BstXI site of pRc/RSV, a eukaryotic expression vector (Invitrogen), upstream of the bovine growth hormone polyadenylation signal (Fig. 2B).
Large deletions in the introns were engineered in several steps. First, the alternative exon was deleted along with 2522 and 2169 nucleotides of the upstream and downstream introns, respectively (termed MCS vector). This vector was generated using the Expand Long Template PCR system (Roche Diagnostics) and sense (5Ј-GGTCACCATATGCC-GCGGGCCCCAGAAAGATTCAG-3Ј) and antisense (5Ј-CGGGTCCG-GATCGATGGCCGGCCACAAAACAATGTG-3Ј) oligonucleotides that created a linker region with the following unique restriction sites at the site of the deletion: FseI, ClaI, BspEI, PshAI, BstEII, NdeI, SacII, EcoO109I, and ApaI. Next, a Tsp45I fragment containing the alternative exon surrounded by 267 and 602 nucleotides of the upstream and downstream introns, respectively, was cloned into the BstEII site of the MCS vector, generating the ⌬uiϩdi (upstream intron and downstream intron deleted) vector in Fig. 3 (lane 2). ⌬ui (lane 3) was generated by cloning a HindIII-EcoO109I fragment from the ⌬uiϩdi vector containing the upstream intron deletion into the HindIII and EcoO109I sites of the wild-type MYPT1 mini-gene vector. ⌬di (lane 4) was generated by cloning a EcoNI-SmaI fragment from the ⌬uiϩdi vector containing the downstream intron deletion into the EcoNI and SmaI sites of the wild-type MYPT1 mini-gene vector. Finally, ⌬diϩ1 and ⌬diϩ2 (lanes 5 and 6) were generated by insertion of one and two copies, respectively, of the 1-kb "stuffer" DNA sequence derived from phage lambda DNA (21). This was accomplished by removing one or two copies of the 1-kb stuffer sequence from the vector Bluescript (Stratagene, gift from G. Screaton) by digestion with SacII and ApaI, which were then inserted into the unique SacII and ApaI sites of the vector ⌬di.
A chimeric vector was engineered by an exact replacement of chicken MYPT1 alternative exon with the corresponding exon from the rat MYPT1 gene. This vector was generated using the Expand Long Template PCR system. Briefly, two oligonucleotides (5Ј-CTGTAAATA-AGATTTTGTTTTAGC-3Ј and 5Ј-GTGTGGGATTTTCTCTTGTCTTTT-3Ј), which hybridize immediately upstream and downstream of the chicken alternative exon, were used to PCR-amplify the entire chicken MYPT1 expression vector, minus the chicken alternative exon. The rat MYPT1 exon was PCR-amplified using two phosphorylated oligonucleotides that begin at the first nucleotide of the corresponding rat MYPT1 exon (5ЈP-AGAGTCTTCAAATTTGCGAACA-3Ј) and at the last nucleotide of the rat exon (5ЈP-CTTCTATGGTAAGTAGATCCTTCA-3Ј). These two PCR-amplified fragments were ligated as blunt-ended fragments to generate a chimeric vector containing a precise replacement of the chicken alternative exon sequence with the rat exon sequence. The identity of this chimeric vector was confirmed by DNA sequencing.
All of the remaining deletions and mutations in the MYPT1 minigene (shown in Figs. 4 -7) were generated using the Expand Long Template PCR system with specific oligonucleotides to set the boundaries of the deletion. All of the deletions in the ⌬1, ⌬2, and ⌬3 series (Figs. 4 and 5) contained a unique ApaI restriction enzyme recognition sequence at the site of the deletions so that the deletions could be confirmed by ApaI digestion. Long PCR was also used to make all of the desired point mutations by including the mutation(s) in the oligonucleotides used for PCR. PCR cycle number was limited to minimize the occurrence of second-site mutations, and multiple independent clones of each construct were tested in transfection assays. All clones were sequenced to verify that they had the desired mutations and that the alternative exon and splice site sequences were intact.
An in Vivo System of Gene Delivery-An in vivo plasmid DNA delivery system to smooth muscle was implemented by adapting methods previously described in striated muscles (22). We optimized expression conditions from an injected plasmid DNA by performing DNA doseresponse curves and time course experiments with a vector containing luciferase as a reporter gene. 1-50 g of plasmid DNA containing either the SV40 early promoter driving transcription of the Sea Panzy (Renilla reniformis) luciferase gene or the RSV promoter driving transcription of the Firefly (Photinus) luciferase gene were injected. White leghorn chickens (Gallus gallus, 1-2 days old, hatched from eggs obtained from Squire Valleevue Farm, Cleveland, OH) were anesthetized with inhaled Halothane (Halocarbon Laboratories) and, after depilation, a longitudinal midline abdominal incision was made to expose the gizzard. Each plasmid in a volume of 20 -30 l of 1ϫ phosphate-buffered saline, 0.015% Evan's Blue dye was injected into each of the two lobes of the gizzard smooth muscle using a 50-l Hamilton syringe. The skin was then sutured and the bird awakened. The birds were sacrificed by decapitation 1-7 days later, and the gizzard smooth muscle was isolated by dissection and homogenized in 0.75-1.75 ml of 1ϫ passive lysis buffer from the Dual-Luciferase Reporter Assay system (Promega). After the lysates were centrifuged, the supernatants were assayed for either Renilla or Firefly luciferase activity as described by the manufacturer, using either a single sample luminometer or a TriLux scintillation counter.
In separate experiments, chicken gizzards were injected with 10 g of the plasmid pDsRed1-N1 (Clontech), which expresses the red fluorescent protein under the control of the cytomegalovirus promoter. Additional chicken gizzards were injected with pUC plasmid as controls. One day later, the gizzards were harvested, frozen in OCT, and cryosectioned. Sections were examined by fluorescence microscopy on the rhodamine channel. Control and experimental samples were photographed with identical exposures using a Spot RT digital camera. Sections were then formalin-fixed, permeabilized, and stained with phalloidin-FITC (Sigma) as per the manufacturer's recommendations. Phalloidin specifically binds to filamentous actin and thus identifies muscle cells in the section. The same approximate region that was photographed on the red channel was rephotographed on the fluorescein channel to identify the phalloidin labeled smooth muscle cells.
The MYPT1 mini-gene expression vectors were injected into gizzard tissue as described above, with the exception that chicks were 1-2 weeks old and 25 g of plasmid DNA was injected in 50 l of 1ϫ phosphate-buffered saline, 0.15-0.3% Evan's Blue dye. One day later the gizzards were removed and the blue-stained portions of the smooth muscle tissue were dissected. The dissected tissues were immediately frozen in liquid nitrogen and processed for total RNA isolation using either Ultraspec II (Biotecx Laboratories, Inc.) or the Absolutely RNA Miniprep kit (Stratagene).
Tissue Culture and Transfections-Smooth muscle cells were isolated from gizzards and aorta of embryonic chicks as previously de-scribed (17). Briefly, gizzard or aorta tissue was removed from ED14 -15 embryos, minced into fine fragments, and digested with repeated rounds of collagenase treatment. Pooled individual cells were cultured on tissue culture plastic with a 50:50 mixture of Dulbecco's modified Eagle's medium and F-12 media in the presence of 10% serum (heat-inactivated fetal bovine serum, HyClone Laboratories). The cells were passaged when sub-confluent with a 1:4 split. Smooth muscle cells were used for transfection after 1-3 passages. These cultures have been previously characterized and shown to be greater than 90% smooth muscle cells as determined by staining for ␣-actin (23). LMH, a chicken hepatocellular carcinoma cell line (ATCC CRL-2117), was cultured on 0.1% gelatin with Waymouth's MB 752/1 medium in the presence of 10% serum. Transfections were performed on 24-well dishes. Cells were plated 1-2 days before each transfection at a density of 1.2 ϫ 10 5 cells/well (LMH) and 1.5 ϫ 10 5 cells/well (SMCs), such that on the day of the transfection they would be 50 -70% confluent. Each mammalian expression vector containing the wild-type or a mutant MYPT1 minigene (0.15 g of plasmid per well) was transfected into these cells using LipofectAMINE (Invitrogen) at a LipofectAMINE to DNA ratio (wt:wt) of 12-18:1 (LMH) and 14:1 (gizzard). Serum-containing media was added to the cells 5 h after transfection. RNA was isolated 24 h after transfection using TRIzol reagent (Invitrogen).
RT-PCR of RNA-The injected and transfected RNA samples were analyzed by RT-PCR for the presence of exon-included and exonskipped MYPT1 mini-gene transcripts. The sense oligonucleotide (5Ј-GGCCAGTACCTCTGACATTGATGA-3Ј) was complementary to chicken MYPT1 cDNA at ϩ1620 (20) and the antisense oligonucleotide (5Ј-GCTGGCAACTAGAAGGCACAGTCG-3Ј) was complementary to the bovine growth hormone polyadenylation signal present in the pRc/ RSV expression vector at position ϩ745 (see Fig. 2B). Thus, this PCR primer pair did not detect the endogenous MYPT1 mRNA (data not shown). Visualization of the RT-PCR products was accomplished by 5Ј-end-labeling the sense PCR oligonucleotide with a Cy5 fluorescent label (IDT and Research Genetics). RT reactions were performed with 0.5 and 5-10 g of input total RNA from transfected and injected RNA samples, respectively. RNA was reverse-transcribed using the antisense primer and AMV Reverse Transcriptase (AMV RT) enzyme (Roche Diagnostics) in a volume of 50 l. The PCR reactions were performed with Taq DNA Polymerase (Roche Molecular Biochemicals) or Taq Bead Hot Start Polymerase (Promega) in a volume of 50 l. The reactions consisted of 23-27 (transfected) and 30 -32 (injected) cycles of 95°C for 45 s, 60°C for 1 min, and 72°C for 1 min, with a final extension step at 72°C for 7 min. RT-PCR products were separated on 6% native polyacrylamide gels. The fluorescence from the Cy5 label was detected using a Storm 860 Imager (Amersham Biosciences). The exonincluded and exon-skipped MYPT1 RT-PCR products were quantified using Amersham Biosciences ImageQuaNT software.
Several controls were performed to verify the specificity and accuracy of the RT-PCR reactions. RT-PCR was performed on RNA from cells that were injected or transfected with a pRc/RSV vector containing no insert. Also, to verify that the RT-PCR products were RNA-dependent, RT-PCR reactions were performed omitting the AMV RT. To verify that the RT-PCR reactions were quantitative, PCR reactions were performed using two different amounts, 0.5 and 2 l, of input cDNA from each RT reaction. Each pair of PCR reactions was judged by the following two criteria: 1) that there would be approximately a 4-fold increase in total signal resulting from the 4-fold difference in input cDNA to verify that the PCR reaction was not saturated and 2) that the ratio of exon inclusion to exon skipping would not change significantly between the two PCR reactions. This control was performed with all of the RNA samples analyzed in this study. Finally, to verify that the injection procedure did not adversely affect the splicing machinery, RT-PCR was performed on a number of injected RNA samples to detect the endogenous MYPT1 transcript. These controls revealed that splicing of endogenous MYPT1 was not affected by the injection procedure (n ϭ 6).
Statistics-Quantification of exon-included and exon-skipped MYPT1 RT-PCR products was obtained as described above. Because the hypothesis to be tested was that cis-elements mediate tissue-specific silencing of the splicing of the alternative exon in phasic smooth muscle, all data are presented as the percentage of exon-skipped/total MYPT1 mRNA. The values shown in this study represent the mean Ϯ S.D. from at least three independent injections or transfections of each mini-gene construct. Data groups were compared with Student's t test or one-way analysis of variance followed by a Bonferroni t test for comparison of pairs of data points as appropriate using Prism version 3.02 (Graph-Pad). p Ͻ 0.05 was considered significant.

An in Vivo System to Identify cis Regulators of Alternative
Splicing-We developed an in vivo (whole animal) mini-gene delivery system modeled after those used in transient gene delivery to skeletal and cardiac muscle (22), taking advantage of the unique property of the gizzard as a solid smooth muscle organ, which could be directly injected with recombinant plasmid DNA. To optimize the experimental system, a plasmid containing the highly sensitive luciferase reporter driven by the viral RSV promoter was used. The naked plasmid DNA was directly injected into the gizzard of 1-to 7-day-old chicks as described under "Experimental Procedures." Dose-response and time course data were generated (data not shown). The highest level of luciferase activity was observed 1 day after injection of 10 g of plasmid DNA, and the activity decreased steadily between days 2 and 7. Injection of a plasmid with a red fluorescent protein (RFP) reporter demonstrated abundant expression of RFP in smooth muscle fibers in the region of injection (Fig. 1A). The smooth muscle fibers are identified by the presence of filamentous actin bound by phalloidin-FITC (Fig.  1B). No red fluorescence was observed in regions remote from the site of injection (not shown), nor in control gizzards injected with pUC plasmid (Fig. 1C). Dose-response experiments demonstrated a linear relationship between amount of injected plasmid DNA (from 1 to 50 g) and luciferase activity (data not shown). Thus, transcripts arising from 25 g of plasmid DNA injected into the gizzard were analyzed 1 day after injection in all subsequent experiments.
Pre-mRNA Splicing from an MYPT1 Mini-gene Construct Is Regulated-We tested an MYPT1 mini-gene expression cassette that contained the chicken central alternative exon and flanking introns and exons (Fig. 2, A and B) in phasic and tonic SMCs and in a non-muscle (liver carcinoma) cell line. In the phasic SMCs (injected gizzard), 27 Ϯ 4% of the spliced products generated from the MYPT1 mini-gene construct lacked the alternative exon (Fig. 2C, lane 1). In contrast, in tonic SMCs (cultured gizzard SMCs), as well as in non-muscle cells (LMH cells), 11 Ϯ 2% of transcripts generated from the MYPT1 mini- gene construct lacked the alternative exon (lanes 2 and 3). Cultured gizzard and aortic SMCs gave identical results for all mutant constructs that were tested in both cell types (data not shown), thus only the results with cultured gizzard SMCs are reported here. This 2.5-fold greater exclusion of the alternative exon in the phasic compared with tonic SMCs is statistically significant (p Ͻ 0.001) and suggests that some of the genetic information necessary for the silencing of the MYPT1 central alternative exon is present within the mini-gene construct.
Role of the Introns in the Regulated Splicing of the Central Alternative Exon-To test the role of the introns in the regulated splicing of the MYPT1 central alternative exon, ϳ2.25 and 1.6 kb of the upstream and downstream introns were deleted, respectively (Fig. 3A). The deletions were designed so as not to alter the splice sites and immediately adjacent sequences. Removal of these large regions from the adjacent introns resulted in virtually complete loss of exon skipping of the MYPT1 mini-gene mRNA in gizzard cells in vivo (phasic), in cultured gizzard cells (tonic), and in cultured non-muscle LMH cells (construct ⌬uiϩdi ; Fig. 3A, lane 2, and Fig. 3B). In addition, removal of either the upstream or downstream regions individually also resulted in significant loss of exon skipping (⌬ui and ⌬di; Fig. 3A, lanes 3 and 4). This suggested either that intron size was crucial to exon skipping and/or that critical regulatory sequences were deleted. To distinguish between these two possibilities, the downstream intron was re-lengthened by the addition of "stuffer" (non-regulatory) DNA sequence derived from phage lambda DNA (21). We focused on the downstream intron, because our sequence analysis had suggested that a number of regulatory elements might be located there (described below). Insertion of one copy of the lambda sequence (⌬diϩ1) resulted in little or no change in the level of exon skipping (compare lanes 4 and 5). Insertion of two copies of the lambda sequence (⌬diϩ2), which increased the size of the intron to 0.4 kb larger than its wild-type length, increased exon skipping in both phasic (lane 6) and tonic SMCs (Fig. 3B) to the level observed with the wild-type mini-gene in tonic SMCs (10 -11%). These results suggest that the relatively large introns surrounding the alternative exon weaken the splicing reaction, allowing for the process to be regulated. The re-lengthening of the introns resulted in significantly less exon skipping as compared with the wild-type construct in phasic SMCs (10.4% versus 27%, p Ͻ 0.001), but not in tonic SMCs (10.3% versus 10.9%, p ϭ NS; Fig. 3B), suggesting that sequences that repress splicing of the alternative exon in the phasic SMCs may be present within the downstream intron.
Identification of a Distal Intronic Silencer of Splicing-A series of overlapping deletions of the downstream intron were generated to locate potential silencer sequences (Fig. 4). Deletion of the proximal 660 nucleotides (⌬1) identified this region as a splicing enhancer and is described below. Deletion of either the mid or distal 600 -700 nucleotide portions of the intron (⌬2 and ⌬3) reduced exon skipping in phasic SMCs (11 and 7% exon skipping, respectively) to a level that was close to the level observed in tonic SMCs (7 and 4%, respectively) and nonmuscle cells (8 and 3%, respectively; Fig. 4, A and B, lanes 3  and 4). Smaller deletions identified a 237-nucleotide sub-region of ⌬2 that had a significant effect on exon skipping (⌬2.2, lane 6), whereas deletion of the remaining 363 nucleotides had no effect (⌬2.1, lane 5). ⌬3 was subdivided into three smaller deletions of 363, 128, and 258 nucleotides (⌬3.1, ⌬3.2, and ⌬3.3, respectively). ⌬3.2 had no effect on splicing of the MYPT1 alternative exon (data not shown), indicating that a 100-nucleotide region that contained several overlapping matches to a number of genes in the NCBI data base (BLAST search) were not functionally significant. Horizontal lines connecting these boxes represent the introns. Sizes of exons and introns are not drawn to scale but are as shown in numbers below each exon and intron. The numbers and letters above the diagram refer to exon and intron labels used in Ref. 17. V-shaped lines above (inclusion of alternative exon) and below (skipping of alternative exon) the diagram represent the endogenous MYPT1 splice variants observed in adult aorta and gizzard. B, schematic diagram of the MYPT1 minigene expression vector used in this study. The genomic fragment described in A was cloned into the pRc/RSV mammalian expression vector in which the viral RSV promoter drives transcription of the MYPT1 mini-gene. The bovine growth hormone (bGH) polyadenylation signal (striped box) is located immediately downstream of the MYPT1 minigene to allow for 3Ј-end formation of the mini-gene mRNA. The arrows indicate the approximate location of the oligonucleotides that were used in all of the RT-PCR reactions in this study. C, representative RT-PCR reactions performed on RNA samples from gizzards that were injected with (Gizz Inj) or smooth muscle and non-muscle cells that were transfected with the wild-type MYPT1 mini-gene (Gizz Trans and LMH). The RT-PCR products were separated, visualized, and quantified as described under "Experimental Procedures." The percentages of exon skipping observed in the injected and transfected RNA samples are shown underneath the appropriate lanes. Labels to the left of the gel refer to exon included (407 nucleotides) and exon-skipped (284 nucleotides) RT-PCR products; n.s., nonspecific RT-PCR products. with lane 4). ⌬3.1 was further subdivided into three smaller deletions of 148, 132, and 105 nucleotides (⌬3.1A, ⌬3.1B, and ⌬3.1C, respectively). ⌬3.1A and ⌬3.1B both caused partial shifts toward exon inclusion compared with ⌬3.1, whereas ⌬3.1C had no effect on splicing (data not shown). These results suggest that the silencer sequence is contained within a continuous region covered by ⌬2.2, ⌬3.1A, and ⌬3.1B. Therefore, a 526-nucleotide deletion covering this region (⌬2.2/⌬3.1B) was constructed that also caused a large shift toward exon inclusion (8% exon skipping in phasic SMCs, lane 9). This deletion eliminated the difference in exon skipping between phasic and tonic SMCs (8% versus 6%, p ϭ NS), suggesting that a silencer of alternative exon splicing that is more active in phasic than tonic SMCs is contained within this region.
Identification of a Proximal Intronic Enhancer of Splicing-Deletion of the proximal 660 nucleotides (⌬1) resulted in an increase in the amount of exon-skipped MYPT1 mRNA (Figs. 4A and 5B, lane 2), suggesting that this sequence contains one or more enhancers of exon inclusion. This sequence contains two UGCAUG motifs, first identified as regulators of splicing of fibronectin alternative exons (24,25), as well as matches to intron L of the chicken embryonic MHC (ceMHC) gene (26) and intron 7 of the chicken interferon alpha/beta receptor 1 (IF-NAR1) gene (27) (Figs. 4A and 5A). Alignment of these MYPT1, ceMHC, and IFNAR1 sequences revealed five blocks of sequence that were highly identical between the three genes, ranging from 83 to 93% identity within these blocks (Fig. 5A). Serial deletions were generated to test the role of each se-quence in enhancing the splicing of the alternative exon. Deletion of 216 nucleotides that contained blocks numbers 1a, 1b, and 2 (⌬1-1 ϩ 2, lane 4) caused a shift toward exon skipping in all cell types (53-71%) that was even more dramatic than the original ⌬1 deletion (38 -48% exon skipping, Fig. 5, B and C, compare lane 4 with lane 2). A 67-nucleotide deletion that removed only block number 1a (⌬1-1a, lane 5) caused essentially the same shift toward exon skipping in all cell types, as did the entire ⌬1-1 ϩ 2 deletion (compare lanes 4 and 5), suggesting that only block number 1a is required for splicing. Consistent with this hypothesis, a 157-nucleotide and a 168nucleotide deletion that removed blocks 1b and 2 (⌬1-1bϩ2) and blocks 4a and b (⌬1-4aϩb), respectively, had little or no affect on splicing of the alternative MYPT1 exon (lanes 6 and  7). Finally, a 167-nucleotide deletion that deletes the two UG-CAUG motifs (⌬1U) had no significant effect on MYPT1 splicing (lane 3), suggesting that these two motifs are not essential for the function of the enhancer contained within ⌬1. Taken together, these results suggest that the ⌬1 region contains a 67-nucleotide enhancer of exon inclusion that is active in phasic, tonic, and non-muscle cells.
A Second Potent Enhancer of Alternative Exon Inclusion Is Present in the Intron-Immediately adjacent to the 5Ј-splice site of the MYPT1 alternative exon is a 19-nucleotide pyrimidine-rich sequence containing two UCUU motifs (Fig. 6A), which is a putative TIA-1 (28,29) or PTB (30) binding site. This sequence was not included in the original large intron deletions due to our desire to avoid changing sequences that were near  Fig. 2A. The elongated rectangles in 5 and 6 represent the 1-kb stuffer DNA sequence described under "Results." In this and subsequent figures only pictures of gels from representative RT-PCR reactions from gizzards that were injected with MYPT1 mini-gene constructs are shown below the schematic diagrams to conserve space. Lanes are arranged in the same order as the diagrams of the deletion constructs (numbers 1-6). RT-PCR products were separated, visualized, and quantified as described under "Experimental Procedures" and labeled as described in Fig. 2C. B, bar graph showing quantification of exon skipping for each wild-type and mutant mini-gene in gizzard tissue, cultured SMCs, and LMH cells. ui, upstream intron; di, downstream intron; , 1 kb lambda phage DNA. the splice sites. Deletion of the 19-nucleotide pyrimidine-rich sequence caused a dramatic shift toward exon skipping in phasic, tonic, and non-muscle cells (86, 81, and 84% exon skipping, respectively; Fig. 6B, lane 2). Mutation of both UCUU motifs to CCCC, which would be predicted to abolish binding of both PTB and TIA-1, caused an equivalent shift toward exon skipping in phasic, tonic, and non-muscle cells (92, 84, and 83%; Fig. 6B, lane 3), suggesting that the UCUU sequences were required for inclusion of the alternative exon in all cell types. Interestingly, if only one of the two UCUU motifs sites was mutated (Cmut2, lane 4), the shift to exon skipping was close to that of the mutation of both UCUU motifs in the phasic SMCs (77% versus 92%; Fig. 6A, compare lane 4 with 3) but not in tonic SMCs and non-muscle cells where endogenous exon inclusion normally predominates (34 -39% versus 83-84%; Fig.  6B, compare lane 4 with 3). To distinguish between the possibilities that the specific UCUU motifs were required versus a requirement for U-rich sequence, all three Cs in this polypyrimidine stretch were mutated to Us (Umut1 ϩ 2, lane 5). This mutation had no significant effect on exon splicing in the phasic SMCs and caused a small shift toward exon skipping in the tonic/non-muscle cells (Fig. 6B). This suggests that it is the U-rich nature of the sequence and not specifically the UCUU motif that is necessary for enhancer function.
Role of the Exon in Its Alternative Splicing-To test the role of the exon sequence itself in alternative splicing, we took advantage of the fact that the corresponding exon in the rat MYPT1 gene is similar in size and sequence, but is constitutively spliced (17), and thus may lack regulatory sequences. A chimeric MYPT1 mini-gene was engineered in which the chicken alternative exon sequence was precisely replaced with the corresponding rat exon sequence, preserving the chicken introns and splice site sequences. This replacement results in the mutation and deletion of 36 and 18 nucleotides, respectively, of the 123 nucleotides contained within chicken alternative exon (Fig. 7C). This deletion and mutation of the chicken MYPT1 alternative exon (rat-chicken chimera) resulted in nearly complete exon skipping in phasic, tonic, and non-muscle cells (Fig. 7, A and B, lane 2) and suggested that one or more sequences that enhance splicing is present in the chicken alternative exon. To rule out the possibility that undesired mutations were introduced elsewhere in the mini-gene during the creation of the chimera that crippled its ability to be properly spliced, re-introduction of the chicken alternative exon sequence into this vector restored splicing of the alternative exon (data not shown), verifying the integrity of the chimera construct.
Identification of Specific Regulatory Sequences within the Chicken MYPT1 Central Alternative Exon-Comparison of the chicken and rat exon sequences identified an 11-nucleotide palindromic sequence, UCCUACAUCCU, that is present in the chicken alternative exon but absent from the rat exon sequence (Fig. 7C, sequence B). This palindromic sequence is located within a stretch of sequence where only 6 of 31 nucleotides (19%) are identical (flanking sequences are designated A and C in Fig. 7C), whereas 63 of 92 nucleotides (68%) are identical outside of this region. Deletion of this 31-nucleotide sequence from the chicken MYPT1 mini-gene construct resulted in a significant increase in exon skipping (data not shown), suggesting that this region contained an enhancer of splicing.
Initially, we focused on a potential role for the palindrome sequence in regulating splicing of the MYPT1 alternative exon. Our strategy for mutating the palindrome involved searching the entire MYPT1 alternative exonic sequence, including the 11-nucleotide palindromic sequence, for matches to previously characterized ESEs. The data from this search (see "Experimental Procedures") revealed that the palindrome contains a good match (UACAUC) to the optimal binding site for the SR protein SRp55 (18) that overlapped with less ideal matches for SRp40 and SC35 binding. Mutations were made to each half of the palindrome to improve or destroy the putative SRp55 binding site while also partially or completely disrupting the palindrome. A mutation that disrupts both half-sites of the palindrome and the putative SRp55 binding site (lane 5, Mutpal1 ϩ 2) significantly increased exon skipping in phasic, tonic, and non-muscle cell types (75, 45, and 47%, respectively; Fig. 7, A  and B, lane 5), identifying this 11-nucleotide sequence as a critical ESE of exon inclusion. To determine if the putative SRp55 binding site played a role, this site was mutated to improve the predicted binding of SRp55 while at the same time disrupting the 3Ј-half-site of the palindrome (lane 4, Mutpal2). This mutation increased exon skipping in phasic, tonic, and non-muscle cell types (39,23, and 21%, respectively; Fig. 7, A  and B, lane 4), suggesting that the ESE is not functioning through SRp55. This is supported by the effect of mutation of only the 5Ј-half-site of the palindrome (lane 3, Mutpal1), which destroys the putative SRp55 binding site but has a minimal effect on the splicing of the alternative exon in all cell types (lane 3). The fact that this palindrome does not resemble any of the other known splicing enhancer sequences raises the possibility that it represents a novel ESE.
The mutation of the palindromic sequence significantly reduced the splicing of the alternative exon but not to the level observed with the rat/chick chimera, suggesting that additional enhancer sequences may be present in the exon. We thus turned our attention to the 10-and 9-nucleotide sequences flanking the palindrome and contained within the 31-nucleotide sequence (Fig. 7C). Our strategy for mutating these two sequences was to replace them with the corresponding sequences in the rat exon (Fig. 7A), because the rat exon sequence did not splice in the context of the rat/chick chimera and thus presumably lacked enhancer sequences. Mutation of the upstream 10-nucleotide sequence to the corresponding rat sequence (lane 6) significantly increased exon skipping in phasic, tonic, and non-muscle cell types (83, 33, and 50%; Fig. 7, A and B, lane 6) to a degree similar to the mutation of the palindromic sequence. Simultaneous mutation of both the 10-nucleotide sequence and the adjacent palindrome (lane 7) produced a significant additive effect toward exon skipping in the tonic/non-muscle cells where the alternative exon is normally included but a minimal additive effect in the phasic SMCs (lane 7). Of note, the level of exon skipping in the tonic SMCs after mutation of the two putative enhancer sequences is not equivalent to that of the rat/chick chimera (compare Fig. 7B, lane 7 with 2) suggesting that an additional enhancer element may be present.
The 9-nucleotide sequence immediately downstream of the 11-nucleotide perfect palindrome is part of a 13-nucleotide imperfect palindrome (UUAAACACAAGUU, Fig. 7C), in which the first U is also common to the perfect palindrome described above. Mutation of six nucleotides of this sequence to the corresponding rat sequence (Fig. 7A, lane 8) nearly abolished exon skipping in phasic SMCs (2%; Fig. 7, A and B, lane 8) and abolished the difference in levels of exon skipping between phasic, tonic, and non-muscle cells ( Fig. 7B; p ϭ NS), suggesting that this sequence mediates the greater silencing of alternative exon splicing in the phasic SMCs compared with tonic SMCs. Thus, the chicken MYPT1 central alternative exon contains a bipartite regulatory region consisting of a 22-nucleotide ESE and a 9-nucleotide exonic splicing silencer (ESS).

DISCUSSION
In this study, we have used deletion and mutation analysis of a MYPT1 mini-gene to identify cis-regulators of splicing of the central alternative exon. This study was performed using cultured SMCs and non-muscle cells as well as a novel in vivo injection system. Our data support a model in which multiple cis-enhancers clustered near the alternative exon 5Ј-splice site are required for its inclusion in MYPT1 mRNA in slow tonic smooth muscle and non-muscle cells (Fig. 8A). The splicing enhancers appear to be countered by two silencer elements, one in the exon and one distal in the downstream nucleotides of the chicken MYPT1 alternative exon and the corresponding exon in rat, respectively. The remaining sequences represent the six nucleotides that are at the 5Ј-end of the downstream intron. The small capital letters (construct number 2) represent the nucleotides in the rat exon that differ from the chicken exon. The underlined nucleotides represent the 11-nucleotide palindrome discussed in the text. The nucleotides surrounded by small boxes represent the nucleotides that were changed in each of the mutant MYPT1 mini-genes. Representative RT-PCR reactions are shown, labeled and analyzed as in Fig. 3A. B, bar graphs showing quantification of exon skipping as in Fig. 3B. C, alignment of the chicken alternative exon and the corresponding rat exon. Dots, identical nucleotides in chicken and rat sequences; dashes, gaps required for optimal alignment. The nucleotides that are marked by vertical lines represent the three regions that were mutated in A. B indicates the 12-nucleotide sequence that is missing in the rat exon. This sequence contains the 11-nucleotide perfect palindrome referred to in the text. A and C refer to the 10nucleotide and 9-nucleotide sequences that are upstream and downstream, respectively, of the palindrome. The six underlined nucleotides represent the optimal SRp55 binding site described in the text.
intron. We propose that these function to silence the splicing of the alternative exon in the fast-phasic smooth muscle phenotype.
Intronic Enhancer Elements-We identified an enhancer complex that is necessary for splicing of the alternative exon in all cell types tested. The requirement of cis-enhancer sequences for the alternative exon to be efficiently spliced is consistent with the general rule that the sequence and/or structure of alternative exons are unfavorable for recognition and splicing by the basal splicing machinery (9 -11). Indeed, the MYPT1 central alternative exon 5Ј-splice site deviates from consensus (AAG:GUGUGG versus CAG:GURAGU), the alternative exon 3Ј-splice site contains a poor polypyrimidine tract, and the relatively large flanking introns (2.3-2.7 kb in chicken) have a negative effect on exon splicing in the mini-gene context (Figs. 2 and 6) (17). One of the enhancers present in the intron is a 19-nucleotide U-rich sequence containing two UCUU motifs located immediately downstream of the alternative exon 5Јsplice site. Deletion of the U-rich sequence or mutation of six residues from U to C led to a dramatic shift toward exon skipping (Fig. 6A), suggesting that these two sites function as splicing enhancers. Based on their presence in other transcripts that undergo alternative splicing, this regulatory motif may be considered a putative TIA-1 or a PTB binding site (28 -30). TIA-1 binds to U-rich sequences immediately downstream of the 5Ј-splice site and enhances splicing, perhaps by stabilizing binding of U1 snRNP to a weak 5Ј-splice site (29). On the other hand, PTB binds to UCUU motifs and functions in most instances as a negative regulator of the splicing of alternative exons (14). The identification of the trans-acting factors that bind to this site in the MYPT1 transcript and how they might influence splicing of the exon will require further study.
A 67-nucleotide sequence required for inclusion of the alternative exon was identified in the intron ϳ190 nucleotides downstream of the alternative exon. The specific nucleotides required for the function of this second putative enhancer remain to be determined. However, within the 32-nucleotide region that is highly identical between MYPT1, ceMHC, and IFNAR1 (Fig. 5A) is the sequence UUUCCCUG, which is also present in the central portion of MSE3, a muscle-specific enhancer of exon inclusion of chicken cardiac troponin T (31,32). This suggests the possibility that this is an MSE3-like enhancer, perhaps functioning through a CUG-binding protein.
Deletion of this enhancer sequence, like the deletion or mutation of the other two enhancer sequences identified in this study, resulted in significant reduction of a exon splicing in tonic and phasic SMCs and non-muscle cells. This suggests that these enhancers function to enhance the splicing of a weakly spliced (alternative) exon but are not involved in the tissue-specific regulation of the splicing of this exon.
Exonic Enhancer Element-The third identified enhancer sequence resides within the alternative exon itself. It consists of an 11-nucleotide perfect palindrome (UCCUACAUCCU) that is absent from the corresponding constitutively spliced rat exon, along with the 10 nucleotides immediately upstream of the palindrome. As was observed with the U-rich intronic enhancer, mutation of both sequences produces additive effects in cells where the alternative exon is normally spliced (i.e. tonic SMCs and non-muscle cells). In the cell type where the endogenous exon is excluded (i.e. phasic SMCs) a single mutation results in nearly complete exon exclusion, and there is little additive effect of the double mutations. This suggests a model in which the enhancers of alternative exon splicing are more susceptible to silencing in the phasic SMCs than in the tonic SMCs, as would be expected based on the splicing of the endogenous exon.
Interestingly, this exact UCCUACAUCCU palindromic sequence is present in or near the alternatively spliced exons of many other transcripts, including the human DSCAM, mouse and rat CamK1-␤, and chicken c-ski (NCBI data base search, not shown). This suggests that this specific 11-nucleotide sequence may function as a novel enhancer of alternative exon splicing in a number of cell types. Alternatively, it was possible that it would function through the well-described ESEs that bind SR proteins. We scanned the exon sequence against a SELEX-generated data base of SR binding sites and identified several sequences where SR proteins might bind (data not shown). A particularly good match to the consensus binding sequence for the SR protein SRp55 (18) was contained within the palindromic sequence (UACAUC) that overlapped with less ideal matches for SRp40 and SC35 binding. However, mutations of the palindromic sequence to sequences that would be predicted to abrogate (CAAAUC) or improve (UACGUA) binding of the SR proteins did not alter splicing of the exon in the former case and actually reduced exon splicing in the latter FIG. 8. Proposed model for MYPT1 alternative splicing. A, inclusion of the alternative exon appears to require at least three distinct splicing enhancers. The putative factors that recognize these enhancers are represented by the open ovals. Because removal of any one of these enhancers significantly reduces exon inclusion in all cell types tested, we propose that the splicing factors that recognize these sequences form an enhancer complex that is required for basal splicing of the exon. The location of this enhancer complex suggests that it may be involved in stabilizing binding of U1 snRNP to the 5Ј-splice site. The two solid ovals represent silencers that appear to preferentially inactivate the enhancer complex in the phasic SMCs. B, putative stem-loop RNA structure showing potential base-pairing between the 11-nucleotide perfect palindrome (ESE) and the 5Ј donor sequence, whereas the 13-nucleotide imperfect palindrome (ESS) forms part of a 20-nucleotide loop. The two palindromes overlap each other by one nucleotide suggesting a steric interference model for the ESE and ESS. The arrow indicates the 5Ј-splice site (exon-intron junction). case. It is therefore proposed that the 11-nucleotide palindrome represents a novel ESE.
Silencer Elements-Two silencers of exon inclusion were identified in the MYPT1 mini-gene construct. One of these silencer sequences is located within the distal portion of the downstream intron, whereas the other silencer is found within the alternative exon (ESS). An interesting feature of these silencers is that, unlike the enhancers, mutation of either one of them completely abolishes the difference in splicing observed between phasic and tonic SMCs, suggesting that the silencers play an important role in determining smooth muscle phenotypic diversity. The ESS is found adjacent to the ESE and just upstream of the alternative exon 5Ј-splice site. Mutation analysis identified six nucleotides in this sequence that are critical for exon skipping, but the exact boundaries of the ESS remain to be defined. The mutated sequence within the ESS is AC-rich and thus resembles A/C-rich exon enhancers (ACE elements) (33,34). However, in this case the sequence appears to function as a cis-silencer. This is similar to several other alternatively spliced exons where bipartite elements containing ESEs and ESSs have been described (35)(36)(37)(38)(39)(40), but the MYPT1 ESE/ESS sequences appear to be different from these elements.
An interesting feature of the MYPT1 ESS is that it is part of an imperfect 13-nucleotide palindrome sequence, where only a G residue in the 11th position interrupts the palindrome (UUAAACACAAGUU). This imperfect palindrome overlaps by one nucleotide the upstream palindrome enhancer sequence. In the modeling of the RNA secondary structure (41), this 13nucleotide sequence forms part of a 20-nucleotide loop in the stem-loop where the 5Ј-splice site and the 11-nucleotide perfect palindrome are base-paired to one another (Fig. 8B). This potential structure suggests a model in which the stem-loop structure prevents U1 snRNP from binding to the 5Ј-splice site and directing splicing unless a trans-acting factor bound to the 11-nucleotide palindrome disrupts the stem-loop. The ESS would function in this model by sterically interfering with the binding of a protein to the ESE. This stem-loop model is similar to that proposed for the regulated splicing of the human tau alternative exon 10 associated with inherited dementia (42) and the chicken ␤-tropomyosin alternative exons (43,44). In each instance, both formation of a stem-loop structure that masks the 5Ј-splice site as well as specific regulatory sequences are proposed to regulate splicing. The fact that deletion of the 11-nucleotide perfect palindrome causes an effect that is the opposite of mutation of the palindrome, i.e. a shift toward exon inclusion (data not shown) is consistent with a potential role for secondary structure in MYPT1 pre-mRNA splicing. However, proof of the importance of the proposed secondary structure will require an examination of the effects of compensatory mutations that alter the sequence but maintain the proposed base-pairing scheme.