Alternative splicing of STY, a nuclear dual specificity kinase.

The LAMMER subfamily of kinases has been conserved throughout evolution, and its members are thought to play important roles in the regulation of cellular growth and differentiation programs. STY is a murine LAMMER kinase which has been implicated in the control of PC12 cell differentiation. Multiple transcripts are derived from the Sty gene, and their relative abundance is developmentally regulated. Alternative splicing of the primary Sty transcript generates mRNAs encoding full-length catalytically active (STY) and truncated, kinase-deficient polypeptides. Both STY and its truncated isoform, STY, are localized in the nucleus and are capable of heterodimerizing. We also demonstrate that STY functions as a dual specificity kinase in mammalian cells.

The Sty gene encodes a member of the recently discovered family of dual specificity kinases (1,2). To date, at least 18 distinct genes encoding dual specificity kinases have been identified in the genomes of yeast and mammals (for reviews, see Douville et al. (3) and Lindberg et al. (4)). These kinases, expressed either as bacterial products or isolated from mammalian cells, have the ability to autophosphorylate on serine, threonine, and tyrosine residues. Two of the best studied of these enzymes are the yeast wee1 ϩ kinase, a regulator of progression through the cell cycle (5), and MEK (MAPK/ERK kinase), believed to be a key molecule in mitogen-stimulated signaling pathways (6). We have previously reported the cloning of the murine Sty cDNA which when expressed in bacteria encodes a protein capable of phosphorylating serine, threonine, and tyrosine residues in vitro (1). The STY kinase contains an amino acid motif, LAMMER, found in kinase subdomain X (7), a feature which is shared with at least eight other dual specificity kinases expressed in humans, mice, plants, and insects (7)(8)(9). The LAMMER motif containing protein kinases appears to be conserved throughout evolution suggesting that these enzymes may play important roles in the control of cellular growth and differentiation. Indeed, the Doa kinase, a Drosophila gene product, is critical to the development of the fly embryo and affects eye differentiation in the adult (7,10). The AFC1 kinase gene of Arabidopsis thaliana can complement yeast signal transduction mutants via activation of the transcription factor STE12 (8). In mammalian cells, the physiological function of the LAMMER kinases is largely unknown although overexpression of the STY kinase in PC12 cells appears to initiate their differentiation into neural derivatives possibly through activation of a protein kinase cascade (11). As has been observed with other members of the LAMMER kinase family, the Sty gene appears to express several differentially processed transcripts. To gain a full understanding of the physiological role of the STY kinase, we felt it important to identify and biochemically characterize all its gene products. We report here the cloning and sequencing of three novel Sty cDNAs. Two of these cDNAs are derived from incompletely processed transcripts which accumulate in the nucleus in a developmentally regulated fashion. The third cDNA encodes a truncated polypeptide (STY T ), which like the full-length STY kinase is found in the nucleus and contains domains which facilitate dimerization. Furthermore, we present evidence that the STY kinase is able to phosphorylate serine, threonine, and tyrosine residues when expressed in mammalian cells.

MATERIALS AND METHODS
Construction of cDNA Library-An L1210 leukemia cell line cDNA library was constructed in gt10 from size-selected poly(A) ϩ RNA. Poly(A) ϩ RNA was fractionated through a 2-10% sucrose gradient in 85% formamide, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 0.2% SDS. Gradients were run at 200,000 ϫ g for 20 h at 20°C. Following centrifugation, 0.4-ml fractions were collected, and aliquots were analyzed by Northern blotting using a Sty cDNA probe. Fractions containing RNA larger than 1.8 kb 1 were pooled and precipitated with ethanol. cDNA was generated with 5 g of size-selected poly(A) ϩ RNA using the SuperScript cDNA synthesis system (Life Technologies, Inc.) and was subsequently size-selected for products larger than 1 kb by gel filtration (Sepharose 4B, Pharmacia Biotech Inc.).
Isolation of cDNA Clones-The gt10 library was screened using a full-length Sty cDNA probe. From a screen of 200,000 plaques, 18 positives were identified. Six were chosen at random for further analysis. To determine the size of the inserts, PCR was employed to amplify cDNA using gt10 and Sty specific primers. Briefly, pairs of gt10/Sty primers were used to amplify DNA from the primary plaques of the library. Primer pairs were gt10 forward ( 5Ј agcaagttcagcctggttaag 3Ј ) and Sty619F ( 5Ј tccaccacttttccgaaagcac 3Ј ) or gt10 reverse ( 5Ј cttatgagtatttcttccagggta 3Ј ) and Sty619F. 50-l PCR reactions were performed with 50 pmol of each primer, 0.2 mM concentration of each dNTP, 1 clones yielding inserts of approximately 2.6 kb were plaque-purified by standard procedures (12).
Sequencing of cDNA Clones-The cDNA inserts were subcloned into the KpnI site of the plasmid pGEM-4 (Promega). Double-stranded DNA sequencing was carried out using the dideoxy chain termination method (13). Full-length sequence of the cDNA insert was achieved using oligonucleotide primers.
RT-PCR Analysis and Cloning of Sty T -Total RNA from P19 cells or mouse tissues was used to synthesize oligo(dT)-primed first strand cDNA with SuperScript RNase H Ϫ reverse transcriptase (Life Technologies, Inc.) as per the manufacturer's instructions. 1-2 l of a 20-l cDNA synthesis reaction was subsequently used in a PCR amplification. 50-l reactions were performed containing 20 pmol of primer Sty619F, 20 pmol of primer of Sty93R ( 5Ј cggatccatgagacattcaaagagaact 3Ј ), and 1.5 units of Taq DNA polymerase for 30 -40 cycles under the conditions described above. Sty T cDNA was generated by RT-PCR using primer Sty7R ( 5Ј catcgtcgtaatcgtttgc 3Ј ) and primer Sty1539F ( 5Ј cgtatgctttttaagtgg 3Ј ). Amplified products were resolved by electrophoresis through a 1.5% agarose gel and ethidium bromide staining.
Sequencing of RT-PCR Products-Amplified PCR products were gelpurified and cloned directly into the pCRII plasmid (Invitrogen) (7R/ 1539F product) or digested with HindIII and SacI restriction endonucleases (93R/619F product) and subcloned into the plasmid pTZ19R (Pharmacia) following conventional procedures (12). Sequencing was performed as described above.
Identification of Additional Introns-Additional introns were identified within the Sty gene by sequencing of genomic fragments isolated from a D3 embryonic stem cell genomic library (gift of D. A. Gray) and subcloned into pGEM-4 (Promega).
Construction of cDNA Expression Vectors and Tagged Proteins-pECE/Sty was generated by insertion of Sty cDNA into the mammalian expression vector pECE (16). The 1.7-kb Sty fragment was isolated from pTZ19R/Sty by PCR amplification using the following oligonucleotides: 5Ј cggatccatgagacattcaaagagaact 3Ј (including a BamHI site at the 5Ј end) and reverse sequencing primer ( 5Ј agcggataacaatttcacacagga 3Ј ). This fragment was subsequently digested with EcoRI and subcloned into pECE previously digested with SmaI and EcoRI.
The STY polypeptide was tagged at the amino terminus with a six-repeat human Myc epitope by subcloning the Myc epitope isolated from Bluescript KS ϩ MTG (17) (SalI/PstI-blunted fragment) into the pECE/Sty plasmid (KpnI-blunted). The Myc epitope in pECE/M-Sty was confirmed to be in-frame by sequencing.
An amino-terminal deletion mutant, pECE/M-Sty ⌬XhoII , lacking the first 60 amino acids of STY, was created by substitution of the BamHI/ BamHI Sty fragment from pECE/M-Sty with the XhoII/BamHI Sty fragment generated by subsequent digestion of the BamHI/BamHI fragment with XhoII. All mutations were confirmed by sequencing.
The Myc epitope in these constructs is immunoreactive with the monoclonal antibody (mAb) 9E10 in immunoprecipitation/immunoblot analysis (18).
Sty was also subcloned into the bacterial expression vector pGEX-3X (Pharmacia) to generate the fusion protein GST-STY. The 1.7-kb Sty fragment was isolated from pTZ19R/Sty by PCR amplification as described above. This blunt/EcoRI fragment was subcloned into pGEX-3X (BamHI, blunted, followed by EcoRI digestion). Sty was confirmed to be in-frame with GST by sequencing.
COS-1 Cell Culture and Transfection-COS-1 cells were maintained in a minimal essential medium supplemented with 10% calf serum. For transient transfection, COS-1 cells were trypsinized, counted, and resuspended in serum-free ␣-minimal essential medium at 2-3 ϫ 10 6 cells in 0.5 ml. 20 g of the appropriate plasmid DNA was added, and the cells were chilled on ice for 10 min. Cells were electroporated using a Gene Pulser (Bio-Rad) at 220 V and 960 F following which the cells were left at room temperature for 15 min. Transfectants were then plated in ␣-minimal essential medium supplemented with 10% calf serum in the absence or presence of 50 M sodium vanadate. Cells were harvested 24 h later.
GST-STY Binding Columns-GST-STY binding reactions were carried out essentially as described (20). Briefly, bacterially expressed GST or GST-STY was coupled to glutathione-Sepharose 4B beads (Pharmacia) in immunoprecipitation buffer. Approximately 10 g of GST or GST-STY coupled to beads was used in each binding experiment. COS-1 cell lysates containing M-STY, M-STY T , or M-Tik were mixed with GST or GST-STY beads. Incubations were carried out for 2 h at 4°C on a rotating platform. The beads were then washed with immunoprecipitation buffer, and proteins were eluted with sample buffer. Bound proteins were analyzed by SDS-PAGE, transferred to nitrocellulose, and subsequently immunoblotted with anti-Myc mAb.
Nucleotide Sequence Accession Numbers-The GenBank TM accession numbers are U11054 and U21209 for the Sty5. 6 and Sty T cDNAs, respectively.

Sequence Analysis of the Developmentally Induced Sty Transcripts Reveals
Introns-Northern blot analysis revealed that the L1210 cell line expressed three distinct species of Sty mRNA corresponding in size to the developmentally regulated mRNA species identified previously in P19 cells and other cell lines (data not shown, Ref. 1, and Fig. 2). In order to determine if these transcripts encode different isoforms of the STY kinase, we cloned the larger of these mRNA species. Using size-selected mRNA isolated from L1210 cells, we prepared cDNAs from mRNA species larger than 1.8kb (see "Materials and Methods"). Our initial screen of 200,000 recombinants resulted in 18 positive clones, one of which (2.6 kb in size) was sequenced fully. This 2.6-kb cDNA clone contained regions of sequence identity with the Sty 1.8-kb embryonic transcript (hereafter referred to as Sty1.8), including the 5Ј end, interspersed with regions of non-identity. The intervening sequences did not, however, maintain the open reading frame as they contained stop codons in all reading frames (see Fig. 1). Further analysis revealed 5Ј and 3Ј splice sites at the ends of the intervening regions suggesting the presence of intronic sequences within an incompletely spliced transcript. The 2.6-kb cDNA did not contain the entire predicted Sty open reading frame, but rather contained four exons (including an exon with the predicted initiating methionine), three complete introns (with 5Ј and 3Ј splice sites), and one partial intron (with a 5Ј splice site). To determine from which mRNA species the cDNA had been cloned, Northern blots were probed with the cDNA regions encompassing individual introns (data not shown). The results of these experiments are depicted schematically in Fig. 2. Probes derived from intron A (IA), or intron B (IB), hybridized to the 5.6-kb and 3.2-kb transcripts (hereafter called Sty5. 6 and Sty3.2, respectively), while probes from intron C (IC) and D (ID) hybridized to Sty5.6 but not Sty3.2. These data indicated that the 2.6-kb cDNA clone represented a partial 5Ј clone of the Sty5.6 RNA species. Since probes corresponding to introns A and B hybridized to Sty3.2, it is likely that this mRNA species is also an incompletely spliced transcript. Thus, both the larger 5.6-and 3.2-kb transcripts contain intron sequences and arise as a result of incomplete and/or alternative splicing. The intron/exon boundaries of the 2.6-kb cDNA clone are indicated in Table I. Sty 3.2-and Sty 5.6-kb Transcripts Are Nuclear-Although both Sty3.2 and Sty5.6 RNAs contained intron sequences, the possibility remained that these RNA species were capable of being translated into truncated STY polypeptides by the cytoplasmic translation machinery. To test this idea, RNA from NIH 3T3 cells was separated into cytoplasmic and nuclear fractions and analyzed by Northern blotting. All three Sty transcripts were found in the nuclear fraction, whereas only Sty1.8 was found in the cytoplasmic fraction (Fig. 2). This is consistent with the larger transcripts (Sty3.2 and Sty5.6) being nuclear, immature (incompletely spliced) transcripts. Fractionated HeLa cell RNA and L cell RNA showed the same nuclear/ cytoplasmic partitioning of Sty transcripts (data not shown).
To determine whether Sty5. 6 represented the primary transcript from the Sty gene, fragments of the Sty locus were isolated from a D3 embryonic stem cell genomic DASHII library (see "Materials and Methods"). A fragment surrounding the predicted initiating methionine was cloned into the sequencing vector pGEM-4. Portions of this genomic fragment were sequenced using oligonucleotides residing in the Sty cDNA. Two introns were identified in addition to those found in Sty5.6, demonstrating that Sty5.6 is not the primary transcript but is an incompletely spliced RNA. The intron/exon boundaries and relative positions of these introns are indicated in Table I.
Sty Is Alternatively Spliced-Alignment of the amino acid sequence of STY with that of the reported human homologue (hCLK (21)) using the Lipman-Pearson algorithm (22) revealed very high homology (83% identity over 482 amino acid overlap). However, there is a stretch of 30 amino acids contained in STY that is absent within hCLK (Fig. 3A). Sequence analysis indicated that these 30 amino acids are encoded within the second exon (exon B) of the 2.6-kb cDNA clone (see Figs. 1 and 3B). This suggested that exon B may be alternatively spliced. The Sty cDNA originally isolated (1) would encode the isoform containing exon B, whereas the hClk cDNA (21) would encode the isoform lacking this exon. Consistent with this idea, the Sty1.8 transcript often appears as a doublet upon Northern blotting suggesting the presence of a fourth transcript (data not shown, Ref. 1, and Fig. 2).
To investigate this possibility, we performed reverse transcriptase coupled with PCR amplification (RT-PCR) to amplify the entire coding sequence of Sty (nucleotide position 7-1539, data not shown) from P19 cell-derived cDNA. Two DNA species were amplified, the sizes of which were consistent with alternative slicing of exon B (data not shown). To verify the identity of the PCR products, the DNA was isolated, reamplified, and subcloned into the pCRII vector. Sequencing revealed that the larger amplification product corresponded to full-length Sty, whereas the smaller lacked exon B (Fig. 1). This result was observed with cDNA derived from P19 cells (see P19 lane, Fig.  4) suggesting that the alternatively spliced transcript lacking exon B (Sty1.7) arises from a mRNA that co-migrates with the 1.8-kb transcript.
Expression of Sty1.7 and Sty1.8 was assayed by RT-PCR amplification of the region encompassing exon B from cDNA of several mouse tissues (Fig. 4). All tissues tested contained both Sty1.7 and Sty1.8 transcripts. A PCR product of approximately 1 kb in size was also reproducibly amplified from all tissues, the size of which is consistent with that of a splicing interme- diate derived from Sty5.6 (data not shown).
The Sty1.7 mRNA, generated by alternative splicing of exon B, should encode a protein homologous to the hCLK protein product. The protein product from the murine Sty1.7 transcript is predicted to be truncated and catalytically inactive (we have termed this product STY T for truncated STY) (see Fig. 3B).
However, the protein product of hClk, which lacks exon B, is predicted to be catalytically active. This apparent contradiction suggested that either the sequence encompassing exon B is absent in the human genome and that in its absence the gene has evolved to maintain a catalytically active molecule or, alternatively, the hClk cDNA is the homologue of Sty1.7 and TABLE I Intron and exon boundaries found in the Sty5.6 cDNA The coding sequence is capitalized, while the noncoding sequence is in small letters. The numbering of nucleotide positions is relative to that of Sty1.8 (1). Intron/exon boundaries indicated by an asterisk were determined by sequencing genomic DNA from the Sty locus. ND, not determined. FIG. 3. Schematic representation of the alternative splicing of exon B. A, comparison of the amino acid sequence surrounding exon B of STY (mouse) with that of the human homologue (hCLK). The sequence encoded by exon B in mouse which is absent in the human cDNA is denoted by the dashed line. Amino acid positions indicated are relative to those predicted for STY (1) and hCLK (17). Identical residues are identified by bars, and conservative substitutions by dots. B, the nucleotide sequence of the intron/exon boundaries surrounding the alternatively spliced exon is shown between the predicted splicing products. Splicing in of exon B would result in production of full-length STY. Splicing out of exon B results in a frameshift which introduces a premature stop codon (tag, indicated by a period). The resulting protein (STY T ) is truncated prior to the catalytic domain.
should encode a truncated molecule. To investigate this further, a human liver cDNA library (gift of Dr. D. A. Gray) was screened with a full-length Sty cDNA probe. From an initial screening of 200,000 recombinants, 30 clones were identified. Of these, 10 were picked at random and analyzed by PCR to determine if they contained the exon A-exon C region. One positive clone was analyzed further, and the sequence of this boundary indicated that it was the human homologue of Sty1.7 and would encode a truncated protein product (data not shown). The human cDNA, as published by Johnson and Smith (21), contains an additional G residue at this boundary which would maintain the open reading frame through the catalytic domain. The absence of this residue in our mouse and human cDNAs causes a frameshift to occur which introduces a stop codon (Fig. 3B).
Expression of STY Protein in Mammalian Cells-The Sty cDNA encodes a 57-kDa polypeptide, whereas the Sty T cDNA encodes a 16.3-kDa truncated polypeptide lacking the catalytic domain. In order to detect these protein products, we tagged the STY and STY T polypeptides with a six-repeat human Myc epitope and expressed these fusion proteins in COS-1 cells. The epitope tag added 89 amino acids to the amino-terminal end,  (Fig. 5A, lanes 2 and 4) corresponding to fulllength and truncated STY kinases.
STY Protein Exhibits Dual Specificity Kinase Activity in Vitro and in Vivo-STY expressed in bacteria has dual specificity kinase activity (1). To determine if STY expressed in mammalian cells has similar dual specificity kinase activity, we used an immune complex kinase assay. Cells transfected with the M-Sty plasmid expressed a 75-kDa phosphoprotein as detected in an anti-Myc mAb immune complex kinase assay (Fig. 5B, lane 2). Phosphoamino acid analysis of the M-STY phosphoprotein revealed substantial phosphoserine, phosphothreonine, and phosphotyrosine (Tyr(P)) (Fig. 5C). It remained possible that a co-precipitating kinase was responsible for phosphorylation of some or all of the sites on M-STY. To address this problem, a catalytically inactive mutant of M-STY, M-STY K190R was generated. Oligonucleotide-directed PCR mutagenesis was used to change the invariant lysine residue 190 in subdomain II of the catalytic domain to an arginine residue (K190R). While expression of this mutant could be readily detected by immunoblotting COS-1 cell extracts (Fig. 5A, lane  3), M-STY K190R showed no kinase activity or evidence of phosphorylation (Fig. 5B, lane 3). This result demonstrates that STY protein expressed in mammalian cells exhibits dual specificity kinase activity. To test its in vivo phosphorylation state, M-Sty was transfected into COS-1 cells, immunoprecipitated with either anti-Myc mAb or anti-Tyr(P) mAb, and then subsequently immunoblotted with anti-Myc mAb (Fig. 6). Cells grown in the absence or presence of vanadate, a phosphotyrosine phosphatase inhibitor, showed similar levels of anti-Myc immunoprecipitable M-STY (Fig. 6, lanes 1 and 2). In the absence of vanadate, only a fraction of the M-STY could be immunoprecipitated with anti-Tyr(P) mAb (Fig. 6, lane 3). In the presence of vanadate, however, similar amounts of M-STY could be immunoprecipitated with anti-Tyr(P) mAb (Fig. 6,  compare lanes 2 and 4). As expected, the M-STY K190R mutant did not react with the antibody directed against phosphotyrosine (data not shown). These results were corroborated by incubating transfected COS-1 cells with [ 32 P]orthophosphate and performing phosphoamino acid analysis on labeled M-STY protein. As with the in vitro kinase assay, all three hydroxyamino acids were phosphorylated when M-STY was labeled in vivo (data not shown).
Nuclear expressed in COS-1 cells was nuclear with undetectable cytoplasmic staining (Fig. 7b). Deletion of the first 60 amino acids of STY (M-STY ⌬XhoII ), which contain the predicted nuclear localization signal, directed widespread cytoplasmic expression of this polypeptide (Fig. 7c). Nuclear staining was not eliminated, suggesting that there may be multiple nuclear localization signals in STY. M-STY ⌬XhoII expressed in COS-1 cells as assessed by immunoblotting, retained in vitro catalytic activity (Fig. 5A, lane 5, Fig. 5B, lane 5, respectively). The mutants, M-STY T and M-STY K190R , were also targetted to the nucleus ( Fig. 7d and data not shown) demonstrating that catalytic activity is not required for nuclear compartmentalization of STY.
STY Protein Can Dimerize-Since both STY and its truncated counterpart, STY T , are found in the nucleus, we wished to determine if the two isoforms could physically interact with one another. M-STY expressed in COS-1 cells was incubated in vitro with bacterially expressed glutathione S-transferase (GST) or a GST-STY fusion protein coupled to glutathione-Sepharose 4B beads. M-STY from COS-1 cells did not bind to beads alone or to GST-coupled beads (Fig. 8, lanes 2 and 3). However, M-STY was able to bind to immobilized GST-STY (Fig. 8, lane 4). To rule out the possibility that the Myc epitope tag was mediating this interaction, we tested the ability of another epitope-tagged dual specificity kinase, M-Tik, 2 to bind to the GST-STY column. As can be seen in Fig. 8 (lanes 5-8), the M-Tik kinase did not bind to either GST or GST-STY columns demonstrating that the Myc epitope was not responsible for the observed binding of M-STY to GST-STY.
Truncated STY (M-STY T ) also formed heterodimers with full-length STY (Fig. 8, lanes 9 -11) demonstrating that the STY dimerization domain is contained in the amino-terminal portion of the molecule and is not dependent upon catalytic activity of the kinase. These results, however, do not differentiate between a direct association of STY with itself or an association mediated through an intermediary molecule. DISCUSSION Isoforms of several protein kinases have been predicted and identified based on cDNA cloning of multiple RNA transcripts (23)(24)(25). For instance, six transcripts, coding for at least two protein isoforms, are derived from the gene encoding the trkB neurogenic tyrosine kinase receptor (23). These trkB proteins, gp145 trkB and gp95 trkB , have identical extracellular and transmembrane domains, but only gp145 trkB contains the cytoplasmic kinase domain. Similarly, the fibroblast growth factor tyrosine kinase receptor 1 gene generates several transcripts which encode receptor variants (24). One of these, a receptorlike molecule that lacks the transmembrane and kinase domains, is secreted and catalytically inactive.
The expression of the Sty gene is developmentally regulated at the level of transcript processing. Embryonic stem cells express two major transcripts (1.7 and 1.8 kb), whereas differentiated cells express two additional partially spliced mRNAs (3.2 and 5.6 kb). While the two larger mRNAs are sequestered in the nucleus and are unavailable for protein translation, the 1.7-and 1.8-kb mRNAs are capable of directing synthesis of a truncated and full-length STY protein, respectively. In general, unspliced primary transcripts are relatively short-lived with 2 N. Abraham and J. C. Bell, unpublished observations.  half-lives on the order of minutes (26). However, several examples of developmentally regulated mRNAs exist which reflect developmental changes in expression of factors involved in the splicing process. In Drosophila, differentially expressed splicing factors control the expression of the sex lethal and transformer RNAs during sexual determination (27). In the developing pituitary, increased transcription of the proopiomelanocortin (POMC) gene results in accumulation of immature transcripts, possibly due to limiting amounts of specific splicing factors (28). Immature T cell antigen mRNA species appear to accumulate at specific stages of T cell development due to the presence of a cycloheximide-sensitive splicing inhibitor protein (29). It is unknown at this time which, if any, of these mechanisms accounts for the accumulation of Sty mRNA precursors. It is of interest to note, however, that two leukemic cell lines, P388 (pre-B) and SP10 (myeloma), like embryonic stem cells, do not accumulate these larger transcripts (1). Thus the 1.7/ 1.8-kb embryonic transcripts may be favored in rapidly proliferating cells such as embryonic or tumor cells.
Amino acid comparison of STY with its human homologue, hCLK, revealed a 30-amino-acid insertion in STY. Isolation and sequencing of the Sty5.6 partial cDNA clone revealed that this 30-amino acid segment was contained within a single exon. Alternative splicing of this exon would generate either the full-length product, STY (containing the exon), or a truncated polypeptide (STY T , lacking this exon) due to a frameshift which introduces a stop codon (see Fig. 3B). Johnson and Smith (21) suggested that the hClk cDNA, which lacks the exon, coded for an active kinase. We have isolated a human Clk cDNA and sequenced it in this region. Our sequence indicated that the G residue at position 546 in the hClk cDNA is not present, and, as a result, an mRNA with this 91-bp exon absent would encode a truncated protein as predicted by the mouse sequence. One simple interpretation of these data is that the original human cDNA contained a sequencing error and that STY T and STY isoforms are encoded in both the human and mouse genomes. A similar conclusion has been reached by Hanes et al. (9) who have recently reported the sequence of two new human kinase cDNAs with significant sequence homology to both Sty and hClk. As we have shown here for Sty, the primary transcripts of these two kinases (hClk2 and hClk3) undergo alternative splicing to generate mRNAs predicted to encode full-length and truncated, catalytically inactive kinases (9). This striking conservation of amino acid sequence and splicing patterns for three separate kinase genes suggests that they may represent a subfamily of enzymes with related physiological targets. It is of interest to note that the Doa and AFC1 LAMMER kinases have been implicated in the control of transcription. The nuclear localization of STY is consistent with its having a role in the regulation of transcription and/or RNA processing. With respect to this latter idea, we and others have recently found that STY and STY T can interact with members of the SR (serine-arginine-rich) family of splicing factors. 3 We are testing the possibility that hCLK2, hCLK3, and their truncated derivatives also associate with splicing factors.
One commonly accepted paradigm of kinase regulation is that dimerization is required for kinase activation and subsequent signal transduction (30,31). Evidence for this idea includes the demonstration that catalytically inactive kinase mutants dominantly suppress wild type kinase activity (32). STY and STY T show identical patterns of expression and subcellular localization and can form heterodimers in vitro. We suggest that a dynamic interaction between STY and STY T may be involved in the regulation of the biological properties of STY and related kinases.