Interactions among SR proteins, an exonic splicing enhancer, and a lentivirus Rev protein regulate alternative splicing.

We examine here the roles of cellular splicing factors and virus regulatory proteins in coordinately regulating alternative splicing of the tat/rev mRNA of equine infectious anemia virus (EIAV). This bicistronic mRNA contains four exons; exons 1 and 2 encode Tat, and exons 3 and 4 encode Rev. In the absence of Rev expression, the four-exon mRNA is synthesized exclusively, but when Rev is expressed, exon 3 is skipped to produce an mRNA that contains only exons 1, 2, and 4. We identify a purine-rich exonic splicing enhancer (ESE) in exon 3 that promotes exon inclusion. Similar to other cellular ESEs that have been identified by other laboratories, the EIAV ESE interacted specifically with SR proteins, a group of serine/arginine-rich splicing factors that function in constitutive and alternative mRNA splicing. Substitution of purines with pyrimidines in the ESE resulted in a switch from exon inclusion to exon skipping in vivo and abolished binding of SR proteins in vitro. Exon skipping was also induced by expression of EIAV Rev. We show that Rev binds to exon 3 RNA in vitro, and while the precise determinants have not been mapped, Rev function in vivo and RNA binding in vitro indicate that the RNA element necessary for Rev responsiveness overlaps or is adjacent to the ESE. We suggest that EIAV Rev promotes exon skipping by interfering with SR protein interactions with RNA or with other splicing factors.     

Analysis of the EIAV Rev-responsive element (RRE) reveals a conserved RNA motif required for high affinity Rev binding in both HIV-1 and EIAV

A cis-acting RNA regulatory element, the Rev-responsive element (RRE), has essential roles in replication of lentiviruses, including human immunodeficiency virus (HIV-1) and equine infection anemia virus (EIAV). The RRE binds the viral trans-acting regulatory protein, Rev, to mediate nucleocytoplasmic transport of incompletely spliced mRNAs encoding viral structural genes and genomic RNA. Because of its potential as a clinical target, RRE-Rev interactions have been well studied in HIV-1; however, detailed molecular structures of Rev-RRE complexes in other lentiviruses are still lacking. In this study, we investigate the secondary structure of the EIAV RRE and interrogate regulatory protein-RNA interactions in EIAV Rev-RRE complexes. Computational prediction and detailed chemical probing and footprinting experiments were used to determine the RNA secondary structure of EIAV RRE-1, a 555 nt region that provides RRE function in vivo. Chemical probing experiments confirmed the presence of several predicted loop and stem-loop structures, which are conserved among 140 EIAV sequence variants. Footprinting experiments revealed that Rev binding induces significant structural rearrangement in two conserved domains characterized by stable stem-loop structures. Rev binding region-1 (RBR-1) corresponds to a genetically-defined Rev binding region that overlaps exon 1 of the EIAV rev gene and contains an exonic splicing enhancer (ESE). RBR-2, characterized for the first time in this study, is required for high affinity binding of EIAV Rev to the RRE. RBR-2 contains an RNA structural motif that is also found within the high affinity Rev binding site in HIV-1 (stem-loop IIB), and within or near mapped RRE regions of four additional lentiviruses. The powerful integration of computational and experimental approaches in this study has generated a validated RNA secondary structure for the EIAV RRE and provided provocative evidence that high affinity Rev binding sites of HIV-1 and EIAV share a conserved RNA structural motif. The presence of this motif in phylogenetically divergent lentiviruses suggests that it may play a role in highly conserved interactions that could be targeted in novel anti-lentiviral therapies.     

Donor site competition is involved in the regulation of alternative splicing of the rat beta-tropomyosin pre-mRNA

The rat beta-tropomyosin (beta-TM) gene encodes both skeletal muscle beta-TM mRNA and nonmuscle TM-1 mRNA via alternative RNA splicing. This gene contains eleven exons: exons 1-5, 8, and 9 are common to both mRNAs; exons 6 and 11 are used in fibroblasts as well as in smooth muscle, whereas exons 7 and 10 are used in skeletal muscle. Previously we demonstrated that utilization of the 3' splice site of exon 7 is blocked in nonmuscle cells. In this study, we use both in vitro and in vivo methods to investigate the regulation of the 5' splice site of exon 7 in nonmuscle cells. The 5' splice site of exon 7 is used efficiently in the absence of flanking sequences, but its utilization is suppressed almost completely when the upstream exon 6 and intron 6 are present. The suppression of the 5' splice site of exon 7 does not result from the sequences at the 3' end of intron 6 that block the use of the 3' splice site of exon 7. However, mutating two conserved nucleotides GU at the 5' splice site of exon 6 results in the efficient use of the 5' splice site of exon 7. In addition, a mutation that changes the 5' splice site of exon 7 to the consensus U1 snRNA binding site strongly stimulates the splicing of exon 7 to the downstream common exon 8. Collectively, these studies demonstrate that 5' splice site competition is responsible, in part, for the suppression of exon 7 usage in nonmuscle cells.     

An element in the 5' common exon of the NCAM alternative splicing unit interacts with SR proteins and modulates 5' splice site selection.

The neural cell adhesion molecule (NCAM) gene contains an 801 nt exon that is included preferentially in neuronal cells. We have set up an in vitro splicing system that mimics the neuro-specific alternative splicing profile of NCAM exon 18. Splicing regulation is observed using model pre-mRNAs that contain competing 5' or 3' splice sites, suggesting that distinct pathways regulate NCAM 5' and 3' splice site selection. While inclusion of exon 18 is the predom-inant choice in neuronal cells, an element in the 5' common exon 17 improves exon 17/exon 19 splicing in a neuronal cell line. A similar behavior is observed in vitro as the element can stimulate the 5' splice site of exon 17 or a heterologous 5' splice site. The minimal 32 nt sequence of the exon 17 enhancer consists of purine stretches and A/C motifs. Mutations in the purine stretches compromise the binding of SR proteins and decreases splicing stimulation in vitro. Mutations in the A/C motifs do not affect SR protein binding but reduce enhancing activity. Our results suggest that the assembly of an enhancer complex containing SR proteins in a 5' common exon ensures that NCAM mRNAs lacking exon 18 are made in neuronal cells.     

The human immunodeficiency virus type 1 Rev protein and the Rev-responsive element counteract the effect of an inhibitory 5' splice site in a 3' untranslated region.

A 5' splice site located in a 3' untranslated region (3'UTR) has been shown previously to inhibit gene expression. Natural examples of inhibitory 5' splice sites have been identified in the late 3'UTRs of papillomaviruses and are thought to inhibit viral late gene expression at early stages of the viral life cycle. In this study, we demonstrate that the interaction of the human immunodeficiency virus type 1 Rev protein with the Rev-responsive element (RRE) overcomes the inhibitory effects of a 5' splice site located within a 3'UTR. This was studied by using both a bovine papillomavirus type 1 L1 cDNA expression vector and a chloramphenicol acetyltransferase expression vector containing a 5' splice site in the 3'UTR. In both systems, coexpression of Rev enhanced cytoplasmic expression from vectors containing the RRE even when the RRE and the inhibitory 5' splice site were separated by up to 1,000 nucleotides. In addition, multiple copies of a 5' splice site in a 3'UTR were shown to act synergistically, and this effect could also be moderated by the interaction of Rev and the RRE. These studies provide additional evidence that at least one mechanism of Rev action is through interactions with the splicing machinery. We have previously shown that base pairing between the U1 small nuclear RNA and a 3'UTR 5' splice site is required for inhibition of gene expression. However, experiments by J. Kjems and P. A. Sharp (J. Virol. 67:4769-4776, 1993) have suggested that Rev acts on spliceosome assembly at a stage after binding of the U1 small nuclear ribonucleoprotein to the 5' splice site. This finding suggests that binding of additional small nuclear ribonucleoproteins, as well as other splicing factors, may be necessary for the inhibitory action of a 3'UTR 5' splice site. These data also suggest that expression of the papillomavirus late genes in terminally differentiated keratinocytes can be regulated by a viral or cellular Rev-like activity.     

Binding of equine infectious anemia virus rev to an exon splicing enhancer mediates alternative splicing and nuclear export of viral mRNAs

In addition to facilitating the nuclear export of incompletely spliced viral mRNAs, equine infectious anemia virus (EIAV) Rev regulates alternative splicing of the third exon of the tat/rev mRNA. In the presence of Rev, this exon of the bicistronic RNA is skipped in a fraction of the spliced mRNAs. In this report, the cis-acting requirements for exon 3 usage were correlated with sequences necessary for Rev binding and transport of incompletely spliced RNA. The presence of a purine-rich exon splicing enhancer (ESE) was required for exon 3 recognition, and the addition of Rev inhibited exon 3 splicing. Glutathione-S-transferase (GST)-Rev bound to probes containing the ESE, and mutation of GAA repeats to GCA within the ESE inhibited both exon 3 recognition in RNA splicing experiments and GST-Rev binding in vitro. These results suggest that Rev regulates alternative splicing by binding at or near the ESE to block SR protein-ESE interactions. A 57-nucleotide sequence containing the ESE was sufficient to mediate Rev-dependent nuclear export of incompletely spliced RNAs. Rev export activity was significantly inhibited by mutation of the ESE or by trans-complementation with SF2/ASF. These results indicate that the ESE functions as a Rev-responsive element and demonstrate that EIAV Rev mediates exon 3 exclusion through protein-RNA interactions required for efficient export of incompletely spliced viral RNAs.     

A 5' splice site-proximal enhancer binds SF1 and activates exon bridging of a microexon

Internal exon size in vertebrates occurs over a narrow size range. Experimentally, exons shorter than 50 nucleotides are poorly included in mRNA unless accompanied by strengthened splice sites or accessory sequences that act as splicing enhancers, suggesting steric interference between snRNPs and other splicing factors binding simultaneously to the 3' and 5' splice sites of microexons. Despite these problems, very small naturally occurring exons exist. Here we studied the factors and mechanism involved in recognizing a constitutively included six-nucleotide exon from the cardiac troponin T gene. Inclusion of this exon is dependent on an enhancer located downstream of the 5' splice site. This enhancer contains six copies of the simple sequence GGGGCUG. The enhancer activates heterologous microexons and will work when located either upstream or downstream of the target exon, suggesting an ability to bind factors that bridge splicing units. A single copy of this sequence is sufficient for in vivo exon inclusion and is the binding site for the known bridging mammalian splicing factor 1 (SF1). The enhancer and its bound SF1 act to increase recognition of the upstream exon during exon definition, such that competition of in vitro reactions with RNAs containing the GGGGCUG repeated sequence depress splicing of the upstream intron, assembly of the spliceosome on the 3' splice site of the exon, and cross-linking of SF1. These results suggest a model in which SF1 bridges the small exon during initial assembly, thereby effectively extending the domain of the exon.     

Ser/Arg-rich protein-mediated communication between U1 and U2 small nuclear ribonucleoprotein particles

Previous work demonstrated that U1 small nuclear ribonucleoprotein particle (snRNP), bound to a downstream 5' splice site, can positively influence utilization of an upstream 3' splice site via exon definition in both trans- and cis-splicing systems. Although exon definition results in the enhancement of splicing of an upstream intron, the nature of the factors involved has remained elusive. We assayed the interaction of U1 snRNP as well as the positive effect of a downstream 5' splice site on trans-splicing in nematode extracts containing either inactive (early in development) or active (later in development) serine/arginine-rich splicing factors (SR proteins). We have determined that U1 snRNP interacts with the 5' splice site in the downstream exon even in the absence of active SR proteins. In addition, we determined that U1 snRNP-directed loading of U2 snRNP onto the branch site as well as efficient trans-splicing in these inactive extracts could be rescued upon the addition of active SR proteins. Identical results were obtained when we examined the interaction of U1 snRNP as well as the requirement for SR proteins in communication across a cis-spliced intron. Weakening of the 3' splice site uncovered distinct differences, however, in the ability of U1 snRNP to promote U2 addition, dependent upon its position relative to the branch site. These results demonstrate that SR proteins are required for communication between U1 and U2 snRNPs whether this interaction is across introns or exons.     

An intron element modulating 5' splice site selection in the hnRNP A1 pre-mRNA interacts with hnRNP A1.

The hnRNP A1 pre-mRNA is alternatively spliced to yield the A1 and A1b mRNAs, which encode proteins differing in their ability to modulate 5' splice site selection. Sequencing a genomic portion of the murine A1 gene revealed that the intron separating exon 7 and the alternative exon 7B is highly conserved between mouse and human. In vitro splicing assays indicate that a conserved element (CE1) from the central portion of the intron shifts selection toward the distal donor site when positioned in between the 5' splice sites of exon 7 and 7B. In vivo, the CE1 element promotes exon 7B skipping. A 17-nucleotide sequence within CE1 (CE1a) is sufficient to activate the distal 5' splice site. RNase T1 protection/immunoprecipitation assays indicate that hnRNP A1 binds to CE1a, which contains the sequence UAGAGU, a close match to the reported optimal A1 binding site, UAGGGU. Replacing CE1a by different oligonucleotides carrying the sequence UAGAGU or UAGGGU maintains the preference for the distal 5' splice site. In contrast, mutations in the AUGAGU sequence activate the proximal 5' splice site. In support of a direct role of the A1-CE1 interaction in 5'-splice-site selection, we observed that the amplitude of the shift correlates with the efficiency of A1 binding. Whereas addition of SR proteins abrogates the effect of CE1, the presence of CE1 does not modify U1 snRNP binding to competing 5' splice sites, as judged by oligonucleotide-targeted RNase H protection assays. Our results suggest that hnRNP A1 modulates splice site selection on its own pre-mRNA without changing the binding of U1 snRNP to competing 5' splice sites.     

Novel exploitation of a nuclear function by influenza virus: the cellular SF2/ASF splicing factor controls the amount of the essential viral M2 ion channel protein in infected cells.

We show that a cellular nuclear protein, the SR splicing factor SF2/ASF, controls the level of production of an essential influenza virus protein, the M2 ion channel protein. The M2 mRNA that encodes the ion channel protein is produced by alternative splicing of another viral mRNA, M1 mRNA. The production of M2 mRNA is controlled in two ways. First, a distal (stronger) 5' splice site in M1 mRNA is blocked by the complex of viral polymerase proteins synthesized during infection, allowing the cellular splicing machinery to switch to the proximal (weaker) M2 5' splice site. Second, utilization of the weak M2 5' splice site requires its activation by the cellular SF2/ASF protein. This activation is mediated by the binding of the SF2/ASF protein to a purine-rich splicing enhancer sequence that is located in the 3' exon of M1 mRNA. We demonstrate that activation of the M2 5' splice site is controlled by the SF2/ASF protein in vivo during influenza virus infection. Utilizing four cell lines that differ in their levels of production of the SF2/ASF protein, we show that during virus infection of these cell lines both M2 mRNA and the M2 ion channel protein are produced in amounts that are proportional to the different expression levels of the SF2/ASF protein.     

Exonic splicing enhancers contribute to the use of both 3' and 5' splice site usage of rat beta-tropomyosin pre-mRNA

The rat beta-tropomyosin gene encodes two tissue-specific isoforms that contain the internal, mutually exclusive exons 6 (nonmuscle/smooth muscle) and 7 (skeletal muscle). We previously demonstrated that the 3' splice site of exon 6 can be activated by introducing a 9-nt polyuridine tract at its 3' splice site, or by strengthening the 5' splice site to a U1 consensus binding site, or by joining exon 6 to the downstream common exon 8. Examination of sequences within exons 6 and 8 revealed the presence of two purine-rich motifs in exon 6 and three purine-rich motifs in exon 8 that could potentially represent exonic splicing enhancers (ESEs). In this report we carried out substitution mutagenesis of these elements and show that some of them play a critical role in the splice site usage of exon 6 in vitro and in vivo. Using UV crosslinking, we have identified SF2/ASF as one of the cellular factors that binds to these motifs. Furthermore, we show that substrates that have mutated ESEs are blocked prior to A-complex formation, supporting a role for SF2/ASF binding to the ESEs during the commitment step in splicing. Using pre-mRNA substrates containing exons 5 through 8, we show that the ESEs within exon 6 also play a role in cooperation between the 3' and 5' splice sites flanking this exon. The splicing of exon 6 to 8 (i.e., 5' splice site usage of exon 6) was enhanced with pre-mRNAs containing either the polyuridine tract in the 3' splice site or consensus sequence in the 5' splice site around exon 6. We show that the ESEs in exon 6 are required for this effect. However, the ESEs are not required when both the polyuridine and consensus splice site sequences around exon 6 were present in the same pre-mRNA. These results support and extend the exon-definition hypothesis and demonstrate that sequences at the 3' splice site can facilitate use of a downstream 5' splice site. In addition, the data support the hypothesis that ESEs can compensate for weak splice sites, such as those found in alternatively spliced exons, thereby providing a target for regulation.     

Identification and characterization of a nuclear factor that specifically binds to the Rev response element (RRE) of human immunodeficiency virus type 1 (HIV-1)

The Rev protein of human immunodeficiency virus type 1 (HIV-1) differentially transactivates the expression of viral structural proteins by allowing the accumulation of unspliced and singly spliced viral mRNA in the cytoplasm. The cis-acting RNA target sequence for the Rev protein, termed the Rev response element (RRE), is present in the env gene and is predicted to form a highly ordered RNA secondary structure. Recent data indicate that Rev directly binds to RRE and, further, that this binding can be mapped to a 90-nucleotide subfragment at the 5' end of RRE. We now report that RRE also binds specifically and predominantly to a nuclear factor of approximately 56 kD. Mapping of the binding site reveals that the same subfragment that binds Rev also binds this nuclear factor. We designate this protein as NFRRE for nuclear factor, RRE binding. Rev and NFRRE appear to bind simultaneously to RRE. NFRRE is widely distributed in various mammalian cells. We speculate that this factor plays an important role in Rev-mediated transactivation and is likely to be involved in the processing or transport of cellular mRNA.     

Exon definition may facilitate splice site selection in RNAs with multiple exons.

Interactions at the 3' end of the intron initiate spliceosome assembly and splice site selection in vertebrate pre-mRNAs. Multiple factors, including U1 small nuclear ribonucleoproteins (snRNPs), are involved in initial recognition at the 3' end of the intron. Experiments were designed to test the possibility that U1 snRNP interaction at the 3' end of the intron during early assembly functions to recognize and define the downstream exon and its resident 5' splice site. Splicing precursor RNAs constructed to have elongated second exons lacking 5' splice sites were deficient in spliceosome assembly and splicing activity in vitro. Similar substrates including a 5' splice site at the end of exon 2 assembled and spliced normally as long as the second exon was less than 300 nucleotides long. U2 snRNPs were required for protection of the 5' splice site terminating exon 2, suggesting direct communication during early assembly between factors binding the 3' and 5' splice sites bordering an exon. We suggest that exons are recognized and defined as units during early assembly by binding of factors to the 3' end of the intron, followed by a search for a downstream 5' splice site. In this view, only the presence of both a 3' and a 5' splice site in the correct orientation and within 300 nucleotides of one another will stable exon complexes be formed. Concerted recognition of exons may help explain the 300-nucleotide-length maximum of vertebrate internal exons, the mechanism whereby the splicing machinery ignores cryptic sites within introns, the mechanism whereby exon skipping is normally avoided, and the phenotypes of 5' splice site mutations that inhibit splicing of neighboring introns.     

Computational modeling suggests dimerization of equine infectious anemia virus Rev is required for RNA binding

Background

The lentiviral Rev protein mediates nuclear export of intron-containing viral RNAs that encode structural proteins or serve as the viral genome. Following translation, HIV-1 Rev localizes to the nucleus and binds its cognate sequence, termed the Rev-responsive element (RRE), in incompletely spliced viral RNA. Rev subsequently multimerizes along the viral RNA and associates with the cellular Crm1 export machinery to translocate the RNA-protein complex to the cytoplasm. Equine infectious anemia virus (EIAV) Rev is functionally homologous to HIV-1 Rev, but shares very little sequence similarity and differs in domain organization. EIAV Rev also contains a bipartite RNA binding domain comprising two short arginine-rich motifs (designated ARM-1 and ARM-2) spaced 79 residues apart in the amino acid sequence. To gain insight into the topology of the bipartite RNA binding domain, a computational approach was used to model the tertiary structure of EIAV Rev.

Results

The tertiary structure of EIAV Rev was modeled using several protein structure prediction and model quality assessment servers. Two types of structures were predicted: an elongated structure with an extended central alpha helix, and a globular structure with a central bundle of helices. Assessment of models on the basis of biophysical properties indicated they were of average quality. In almost all models, ARM-1 and ARM-2 were spatially separated by >15 Å, suggesting that they do not form a single RNA binding interface on the monomer. A highly conserved canonical coiled-coil motif was identified in the central region of EIAV Rev, suggesting that an RNA binding interface could be formed through dimerization of Rev and juxtaposition of ARM-1 and ARM-2. In support of this, purified Rev protein migrated as a dimer in Blue native gels, and mutation of a residue predicted to form a key coiled-coil contact disrupted dimerization and abrogated RNA binding. In contrast, mutation of residues outside the predicted coiled-coil interface had no effect on dimerization or RNA binding.

Conclusions

Our results suggest that EIAV Rev binding to the RRE requires dimerization via a coiled-coil motif to juxtapose two RNA binding motifs, ARM-1 and ARM-2.

     

Characterization of functional domains of equine infectious anemia virus Rev suggests a bipartite RNA-binding domain

Equine infectious anemia virus (EIAV) Rev is an essential regulatory protein that facilitates expression of viral mRNAs encoding structural proteins and genomic RNA and regulates alternative splicing of the bicistronic tat/rev mRNA. EIAV Rev is characterized by a high rate of genetic variation in vivo, and changes in Rev genotype and phenotype have been shown to coincide with changes in clinical disease. To better understand how genetic variation alters Rev phenotype, we undertook deletion and mutational analyses to map functional domains and to identify specific motifs that are essential for EIAV Rev activity. All functional domains are contained within the second exon of EIAV Rev. The overall organization of domains within Rev exon 2 includes a nuclear export signal, a large central region required for RNA binding, a nonessential region, and a C-terminal region required for both nuclear localization and RNA binding. Subcellular localization of green fluorescent protein-Rev mutants indicated that basic residues within the KRRRK motif in the C-terminal region of Rev are necessary for targeting of Rev to the nucleus. Two separate regions of Rev were necessary for RNA binding: a central region encompassing residues 57 to 130 and a C-terminal region spanning residues 144 to 165. Within these regions were two distinct, short arginine-rich motifs essential for RNA binding, including an RRDRW motif in the central region and the KRRRK motif near the C terminus. These findings suggest that EIAV Rev utilizes a bipartite RNA-binding domain.     

TIA proteins are necessary but not sufficient for the tissue-specific splicing of the myosin phosphatase targeting subunit 1

We are using the tissue-specific splicing of myosin phosphatase targeting subunit (MYPT1) as a model to investigate smooth muscle phenotypic diversity. We previously identified a U-rich intronic enhancer flanking the 5' splice site (IE1), and a bipartite exonic enhancer/suppressor, that regulate splicing of the MYPT1 central alternative exon. Here we show that T-cell inhibitor of apoptosis (TIA-1) and T-cell inhibitor of apoptosis-related (TIAR) proteins bind to the IE1. Co-transfection of TIA expression vectors with a MYPT1 mini-gene construct increase splicing of the central alternative exon. TIA proteins do not enhance splicing when the palindromic exonic splicing enhancer (ESE) is mutated, indicating that TIAs are necessary but not sufficient for splicing. The ESE specifically binds SRp55 and SRp20 proteins, supporting a model in which both SR and TIA proteins binding to their cis-elements are required for the recruitment of the splicing complex to a weak 5' splice site. Inactivation of TIA proteins in the DT40 cell line (TIA-1(-/-)TIAR(+/-)) reduced the splicing of the central alternative exon of the endogenous MYPT1 as well as stably transfected MYPT1 minigene constructs. Splicing of the MYPT1 3' alternative exon and the MLC(17) alternative exon were unaffected, suggesting that TIA proteins regulate a subset of smooth muscle/nonmuscle alternative splicing reactions. Finally, reduced RNA binding and reduced expression of the TIA and SR proteins in phasic (gizzard) smooth muscle around hatching coincided with the switch from exon inclusion to exon skipping, suggesting that loss of TIA and SR enhancer activity may play a role in the developmental switch in MYPT1 splicing.     

Regulation of SRp20 exon 4 splicing

SR proteins are essential splicing factors involved in the use of both constitutive and alternative exons. We previously showed that the SR proteins SRp20 and ASF/SF2 have antagonistic activities on SRp20 pre-mRNA splicing. SRp20 activates exon 4 recognition in its pre-mRNA, whereas ASF/SF2 inhibits this recognition. In experiments aimed at testing the specificity of SRp20 and ASF/SF2 for exon 4 splicing regulation, we show here that this specificity lies in the RNA binding domains of SRp20 and ASF/SF2 and not in the RS domains. Surprisingly, a deletion of 14 amino acids at the end of ASF/SF2-RBD2 converts ASF/SF2 from an inhibitor to an activator of exon 4 splicing. We found that ASF3 also inhibits exon 4 recognition, thus acting similarly to ASF/SF2, while SC35 activates a cryptic 5' splice site downstream of exon 3 and, in doing so, represses exon 4 use. In contrast, Tra2 and the SR proteins 9G8 and SRp40 do not appear to affect exon 4 splicing.