Cloning of a fibroblast growth factor receptor 1 splice variant from Xenopus embryos that lacks a protein kinase C site important for the regulation of receptor activity.

A cDNA clone, predicted to encode a variant form of the type 1 fibroblast growth factor receptor (FGFR1) containing a dipeptide Val-Thr (VT) deletion at amino acid positions 423 and 424 located within the juxtamembrane region, was isolated from a Xenopus embryo (stage 8 blastula) library. Sequence analysis of genomic DNA encoding a portion of the FGFR1 juxtamembrane region demonstrated that this variant form arises from use of an alternative 5' splice donor site. RNase protection analysis revealed that both VT- and VT+ forms of the FGFR1 were expressed throughout embryonic development, the VT+ being the major form. Amino acid position 424 is located within a consensus sequence for phosphorylation by a number of Ser/Thr kinases. We demonstrate that a VT+ peptide was specifically phosphorylated by protein kinase C (PKC) in vitro, but not by protein kinase A (PKA). A VT- peptide, on the other hand, was not a substrate for either enzyme. Phosphorylation levels of in vitro synthesized FGFR-VT+ protein by PKC were twice that of FGFR-VT- protein. In a functional assay, Xenopus oocytes expressing FGFR-VT- or FGFR-VT+ protein were equally able to mobilize intracellular Ca2+ in response to basic fibroblast growth factor (bFGF). However, pretreatment with phorbol 12-myristate 13-acetate significantly reduced this mobilization in oocytes expressing FGFR-VT+ while having little effect on oocytes expressing FGFR-VT-. These findings demonstrate that alternative splicing of Val423-Thr424 generates isoforms which differ in their ability to be regulated by phosphorylation and thus represents an important mechanism for regulating FGFR activity.

A cDNA clone, predicted to encode a variant form of the type 1 fibroblast growth factor receptor (FGFR1) containing a dipeptide Val-Thr (VT) deletion at amino acid positions 423 and 424 located within the juxtamembrane region, was isolated from a Xenopus embryo (stage 8 blastula) library. Sequence analysis of genomic DNA encoding a portion of the FGFR1 juxtamembrane region demonstrated that this variant form arises from use of an alternative 5 splice donor site. RNase protection analysis revealed that both VT؊ and VT؉ forms of the FGFR1 were expressed throughout embryonic development, the VT؉ being the major form. Amino acid position 424 is located within a consensus sequence for phosphorylation by a number of Ser/Thr kinases. We demonstrate that a VT؉ peptide was specifically phosphorylated by protein kinase C (PKC) in vitro, but not by protein kinase A (PKA). A VT؊ peptide, on the other hand, was not a substrate for either enzyme. Phosphorylation levels of in vitro synthesized FGFR-VT؉ protein by PKC were twice that of FGFR-VT؊ protein. In a functional assay, Xenopus oocytes expressing FGFR-VT؊ or FGFR-VT؉ protein were equally able to mobilize intracellular Ca 2؉ in response to basic fibroblast growth factor (bFGF). However, pretreatment with phorbol 12myristate 13-acetate significantly reduced this mobilization in oocytes expressing FGFR-VT؉ while having little effect on oocytes expressing FGFR-VT؊. These findings demonstrate that alternative splicing of Val 423 -Thr 424 generates isoforms which differ in their ability to be regulated by phosphorylation and thus represents an important mechanism for regulating FGFR activity.
Fibroblast growth factors (FGFs) 1 play a role in a number of cellular responses, including mitogenesis, differentiation, angiogenesis, and transformation (reviewed in Ref. 1). The family of FGFs consists of nine distinct members (2), related by amino acid sequence and their ability to bind heparin, that mediate their response by binding to high affinity cell surface FGF receptors (FGFRs). Functional FGFRs are transmembrane pro-teins composed of an extracellular ligand-binding domain containing two or three immunoglobulin (Ig)-like domains and an intracellular domain consisting of a juxtamembrane region, a split tyrosine kinase domain and a COOH-terminal tail (reviewed in Ref. 3). FGF binding to the extracellular domain of the FGFR results in receptor activation through dimerization and autophosphorylation. The activated receptor can then bind and phosphorylate a number of intracellular substrates, thus altering their catalytic activity and initiating intracellular signal transduction cascades (reviewed in Ref. 3).
FGFRs are encoded by four genes whose transcripts are alternatively spliced to produce a number of variant forms (reviewed in Ref. 3). Each of the four FGFR types is capable of binding more than one member of the FGF family, the ligand binding specificity being determined not only by the receptor type but by the splicing form. For example, alternative splicing of exons encoding the COOH-terminal half of the third Ig domain of FGFR2 leads to production of FGFRs that no longer recognize FGF-7 (4). In addition, Shi et al. (5) has described an alternatively spliced FGFR isoform that encodes a truncated, kinase-defective receptor which can heterodimerize with fulllength FGFRs and reduce tyrosine kinase activity. Clearly, alternative splicing represents an important mechanism by which FGFR activity can be regulated.
FGFs induce differentiation of mesoderm in Xenopus embryonic tissue (6 -8), and FGFR signaling has been shown to be required for this developmental event (9). Mesoderm induction during embryonic development is precisely regulated in time and space to produce a distinct pattern of mesodermal tissues. In order to investigate the molecular mechanisms involved in regulating this complex developmental process, it is important initially to determine which FGFR genes are involved and how FGFR signaling is regulated. Evidence to date suggests that FGFR1 is likely to be important, since both mRNA (10, 11) and protein (12) for FGFR1 are present in Xenopus blastulae, the stage during which mesoderm induction takes place in the embryo. In addition, we have demonstrated that FGFR1 was activated during FGF-induced mesoderm differentiation in Xenopus (12). Consequently, we decided to focus on the FGFR1 gene and determine which FGFR1 isoforms may be important for mesoderm induction.
Two reports have described FGFR1s cloned from Xenopus, however, neither isolated cDNA from embryos. Musci et al. (10) cloned a three-Ig domain FGFR1 from an oocyte library, whereas Friesel and Dawid (11) cloned both two-and three-Ig forms from a Xenopus cell line (XTC). Accordingly, we prepared and screened a cDNA library from Xenopus blastulae for FGFR1 species. This paper describes a Xenopus FGFR1 isoform which differs in its ability to be regulated by protein kinase C (PKC).

EXPERIMENTAL PROCEDURES
Materials-Xenopus laevis were purchased from Nasco and maintained as described in Ref. 13. Eggs were artificially inseminated, the jelly coats removed, and the embryos cultured as described in Godsave et al. (14). Synthetic peptides corresponding to FGFR-VTϩ (IPLR-RQVTVSGDSS) and FGFR-VTϪ (IPLRRQVSGDSS) were purchased from the Alberta Peptide Institute (Edmonton, Alberta). Recombinant Xenopus bFGF was expressed and purified according to Kimelman et al. (15) then stored at Ϫ20°C. The anti-FGFR1 used for immunoprecipitation in this study was a polyclonal antibody raised against a synthetic COOH-terminal peptide (12).
cDNA Cloning and Sequencing of an FGFR1 Isoform from Xenopus Embryos-A cDNA library was constructed from mRNA isolated from stage 8 Xenopus blastulae using the ZAP II kit (Stratagene) as directed. The library was screened using a 400-bp fragment of the Xenopus FGFR1 cDNA previously cloned by PCR. 2 This 400-bp fragment was amplified from Xenopus stage 17 first strand cDNA using oligonucleotide primers within the FGFR1 tyrosine kinase domain. The 400-bp amplification product was then cloned in the EcoRV site of Bluescript KSϩ and sequenced on both strands, verifying its identity as a fragment of the Xenopus FGFR1. The 400-bp fragment was radiolabeled by random primer labeling (Life Technologies, Inc.) to a specific activity of 5 ϫ 10 8 cpm/g. This probe was used to screen 1.5 ϫ 10 6 recombinant plaques, as described in Wahl and Berger (16). Twelve positive plaques were isolated and the largest of these contained a 3.8-kb insert which was further characterized. The cDNA sequence for both strands of the 3.8-kb insert was determined by the dideoxy chain termination method (17) using the Sequenase system (U. S. Biochemical Corp.).
Sequencing of a Genomic Fragment Containing the VT Region of the FGFR1-A 1.2-kb genomic fragment containing the VT region was amplified from a Xenopus genomic library (Stratagene). The amplification was performed with oligonucleotide primers 5Ј(GGGCTGCTTTT-GTGTCCGCAAT) and 3Ј(GCCATGACTACTTGCC) bracketing the VT region (see Fig. 1 for location of primers), for 30 cycles consisting of denaturation at 94°C for 1 min, annealing at 50°C for 1 min and extension at 72°C for 2 min. PCR products were separated on 1% agarose, the 1.2-kb band cut out and the DNA extracted from the agarose gel with Qiaex (Qiagen) according to the manufacturer's directions. Sequencing was performed as above, using the PCR primers.
RNase Protection Analysis-RNA was extracted and purified from whole embryos using the LiCl/urea protocol described in Goldin (18). RNase protections were performed as in Paterno et al. (7). The RNA antisense probe was prepared from a BstEII-HgaI cDNA fragment of XFGFR-A2 (gift from Dr. Robert Friesel, American Red Cross) cloned into the EcoRV site of pBluescript KSϩ (Stratagene). Transcription from the T7 promoter yielded a 261-base probe which protected a fragment of 162 bases for the VTϩ isoform and two fragments of 107 and 49 bases for the VTϪ isoform.
PKC and PKA Phosphorylation Assays-Phosphorylation by PKC of both the peptides (see "Materials" for sequence) and full-length FGFR1 protein was measured using a PKC assay kit (Life Technologies, Inc.), 15 ng of purified PKC enzyme (Upstate Biotechnology, Inc.), and 10 Ci [␥-32 P]ATP (Amersham Corp.) per reaction. The FGFR1 proteins used in this assay were synthesized in vitro from the FGFR-VTϪpcDNAIneo or FGFR-VTϩpcDNAIneo plasmids (plasmid construction described below) using a coupled transcription/translation system (Promega) as described in Ryan and Gillespie (12). The assays were performed according the manufacturer's directions with the exception that the 5 ϫ substrate solution supplied with the kit was replaced with one lacking the control peptide substrate. The substrate was then added separately to each reaction: 25 M peptide or in vitro synthesized FGFR1-VTϪ or -VTϩ protein that had been purified by immunoprecipitation from equal inputs of trichloroacetic acid-precipitable counts/min. The control substrate was a gift from Dr. J. Reynolds (Memorial University) and consisted of a synthetic peptide (CNPLLRMFSFKAPT) corresponding to amino acids 336 -348 of the ␥2L subunit of the ␥-aminobutyric acid receptor, which contains a Ser that is phosphorylated by PKC (19).
Plasmid Construction-An FGFR-VTϪ receptor construct lacking both 5Ј-and 3Ј-untranslated regions was generated by subcloning a BseAI-RcaI cDNA fragment (Fig. 1), which encodes most of the open reading frame (amino acids 3-789) of the FGFR-VTϪ cDNA, into the same sites of a pcDNAIneo mammalian expression vector containing the FGFR-A2 cDNA. The FGFR-A2pcDNAIneo plasmid contains the coding region of an FGFR1 isoform lacking the first Ig domain (11) inserted into the BamHI site of pcDNAIneo (Invitrogen). The FGFR-VTϩ receptor construct was generated by subcloning a BstEII-RcaI cDNA fragment (Fig. 1) of the FGFR-A2pcDNAIneo, encoding the transmembrane and intracellular domains, into the same sites of FGFR-VTϪpcDNAIneo plasmid. The FGFRSP64T constructs used for expression in Xenopus oocytes were generated by subcloning a BamHI fragment, containing the entire FGFR coding region, from the FGFR-VTϪpcDNAIneo or FGFR-VTϩpcDNAIneo constructs in the BglII site of the SP64T vector (20).
Oocyte Injections and Protein Analysis-cRNA was transcribed using the SP6 Ribomax system (Promega) from the FGFRSP64T constructs (described above) that had been linearized with XbaI. 4.6 nl containing 500 pg of cRNA was microinjected into stage VI Xenopus oocytes prepared and cultured as described in Amaya et al. (9). Injected oocytes were metabolically labeled by culturing for 24 h at 22°C in medium containing 1 mCi/ml [ 35 S]methionine (1000 Ci/mmol; DuPont NEN). After extensive washing, the oocytes were solubilized and the FGFR immunoprecipitated as described in Ryan and Gillespie (12). The immunoprecipitates were analyzed by 8% SDS-polyacrylamide gel electrophoresis followed by autoradiography.
45 Ca 2ϩ Release Assays-Microinjected oocytes were maintained at 22°C for 24 h, washed extensively in Ca 2ϩ -free medium, then loaded for 3 h with 45 Ca 2ϩ (10 Ci/g; DuPont NEN) at a final concentration of 100 Ci/ml. 45 Ca 2ϩ release was measured from groups of 10 oocytes as described in Amaya et al. (9). Phorbol 12-myristate 13-acetate (PMA; Life Technologies, Inc.) and Xenopus bFGF were added at the indicated times to a final concentration of 250 nM and 100 ng/ml, respectively.

RESULTS
Mesoderm induction takes place during blastula stages of Xenopus development. In our efforts to understand the role of the FGFR in this induction event, we set out to identify FGFR1 isoforms that are expressed during blastula stages. We prepared a cDNA library from mid-blastula (stage 8) Xenopus embryos and screened it for FGFR1. A positive plaque containing a 3.8-kb insert was purified and sequenced. The cDNA consisted of an open reading frame of 2.4 kb bracketed by a 183-bp 5Ј-untranslated region and 1.3-kb 3Ј-untranslated region. The amino acid of our clone was compared with previously cloned Xenopus FGFR1s: XFGFR (10) and XFGFR-A2 (11) (Fig.  1). Our clone and XFGFR encode FGFRs containing three Ig domains in the extracellular region while XFGFR-A2 contains only two; this is a common variation of the FGFR1 that has been extensively studied in other species (reviewed in Ref. 3). Our clone was most similar to XFGFR-A2 in the remaining sequence, with only four amino acid changes as opposed to eight for XFGFR. Examination of these amino acid changes revealed one common difference between our clone and the other two: the deletion of Val 423 -Thr 424 (VT) in the juxtamembrane region of our FGFR1 cDNA. We have therefore named our clone FGFR-VTϪ.
To investigate the possibility that this deletion is generated by alternative splicing, we sequenced a genomic fragment containing the VT region (Fig. 2). By comparing the genomic DNA sequence to the cDNA sequence, the amino acid sequence and 5Ј and 3Ј consensus splice sequences (5Ј: (C/A)AG/GU(G/A)AG; 3Ј: ((C/U)) n NCAG/G; reviewed in Refs. 21 and 22), we were able to examine a number of possible origins for these two isoforms, including alternative exons and alternative 5Ј and/or 3Ј splice sites. We concluded that the most likely mechanism for the production of the two receptor forms is the use of alternative 5Ј splice donor sites (Fig. 2). Splicing to produce FGFR-VTϪ would make use of an excellent consensus 5Ј splice donor site, whereas the splice site to produce FGFR-A2 or XFGFR lacks three of the eight consensus nucleotides. Therefore, one would predict that the major splicing product would be FGFR-VTϪ mRNA. Interestingly, RNase protection of total RNA from embryos at various developmental stages revealed that in fact VTϩ mRNA was the major form (Fig. 3). In addition, there appeared to be little change in ratio of the VTϩ/VTϪ isoforms at the developmental stages examined.
A similar deletion of Thr-Val was reported for a FGFR1 cDNA cloned from a human hepatoma cell line (23). These authors suggested that this location may represent a possible site for phosphorylation by a Ser/Thr kinase. Comparison with consensus sequences for various Ser/Thr kinases (24) revealed that amino acid position 424 was located within a consensus sequence for phosphorylation by PKC and PKA; in FGFR-VTϪ, a Ser is in this position, whereas in the VTϩ isoform, a Thr is in this location. We decided to examine whether this Ser or Thr could be phosphorylated by PKC or PKA. Two peptides, corresponding to amino acids 417-428 of FGFR-VTϪ or amino acids FIG. 2. Genomic fragment spanning the VT region. Partial sequence of the genomic fragment, amplified by PCR using primers (shown in Fig. 1) that bracket the VT region, is shown with the predicted amino acid sequence listed underneath. Predicted exon and intron sequences are shown in upper-and lowercase, respectively, with the sequence encoding Val 423 and Thr 424 shown in bold. Alternative 5Ј splice donor sites used to generate the VTϪ or VTϩ isoforms are indicated by arrows. 417-430 of the VTϩ isoform, were synthesized and used in in vitro kinase assays. As can be seen in Fig. 4A, neither peptide was phosphorylated by PKA. PKC, on the other hand, selectively phosphorylated the VTϩ peptide. We also examined the ability of PKC to phosphorylate the full-length proteins. For this purpose, we constructed an FGFR1 that contains 3 Ig domains and Val 423 -Thr 424 , thus differing from FGFR-VTϪ only by the presence of Val 423 -Thr 424 . We refer to this construct as FGFR-VTϩ. The substrates in this PKC assay were FGFR-VTϪ or FGFR-VTϩ protein isolated by immunoprecipitation from in vitro transcription/translation reactions. Both proteins were phosphorylated by PKC (Fig. 4B); however, twice as much [ 32 P]PO 4 was incorporated into FGFR-VTϩ. This demonstrates that the full-length proteins were substrates for PKC and that presence of the VT increased the degree of phosphorylation. The fact that FGFR-VTϪ protein, but not the peptide, was phosphorylated by PKC suggests that there are additional phosphorylation sites in the protein.
One of the questions that remained was whether differential phosphorylation of these two isoforms by PKC affects receptor function. To examine this question, we measured mobilization of intracellular Ca 2ϩ stimulated by FGF in oocytes expressing either form of the FGFR1. Mobilization of intracellular Ca 2ϩ , as measured by 45 Ca 2ϩ efflux from oocytes, is commonly employed as a functional assay of FGFR activity (9,10,25). Xenopus oocytes were microinjected with H 2 O (control) or mRNA encoding either FGFR-VTϩ or FGFR-VTϪ. After a 24-h incubation period to allow for expression of FGFR protein, oocytes were loaded with 45 Ca 2ϩ in calcium-free medium. 45 Ca 2ϩ release into the medium was measured in response to addition of 100 ng/ml Xenopus bFGF (XbFGF) to oocytes; parallel samples were pretreated for 20 min with 250 nM PMA, a phorbol ester that activates PKC, before addition of XbFGF. H 2 O-injected oocytes showed no response to XbFGF (Fig. 5A). Oocytes expressing either FGFR isoform exhibited a similar response to XbFGF treatment alone but not when stimulated with XbFGF in the presence of PMA (Fig. 5, B and C). Pretreatment with PMA resulted in a slight reduction in the magnitude of the 45 Ca 2ϩ release by oocytes expressing FGFR-VTϪ (Fig. 5B), whereas the 45 Ca 2ϩ release by oocytes expressing FGFR-VTϩ

FIG. 4. Phosphorylation of the VT؉ and VT؊ isoforms by Ser/ Thr kinases in vitro.
A, incorporation of [ 32 P]PO 4 into VTϩ and VTϪ peptides by PKC or PKA in vitro. Assays were performed as described under "Experimental Procedures." PKC assays were carried out in the presence or absence of an PKC-specific inhibitor peptide. PKC-specific phosphorylation was calculated by subtracting the counts/min incorporated in the presence of inhibitor from that incorporated in the absence of inhibitor. PKA assays were performed in the presence or absence of peptide and the difference used to calculate PKA-specific incorporation. The average and standard deviation of three separate experiments is shown. 32 P incorporation into the PKC control substrate peptide was 50,825 cpm/nmol and that for the PKA control substrate peptide was 672,506 cpm/nmol. B, incorporation of 32 P into FGFR-VTϩ and FGFR-VTϪ protein by PKC. The substrate in each case was in vitro synthesized protein isolated by immunoprecipitation. Assays were performed as in A. The average and standard deviation of three separate experiments is shown.
was significantly reduced (Fig. 5C). To verify that, in these experiments, the oocytes expressed equal amounts of FGFR-VTϪ or VTϩ protein, FGFRs were immunoprecipitated from oocytes labeled with [ 35 S]methionine and the precipitates analyzed by SDS-polyacrylamide gel electrophoresis. The inset in Fig. 5C shows that there was no difference in the synthesis of VTϪ and VTϩ FGFR proteins. DISCUSSION FGFs are known to mediate a number of diverse and complex cellular responses (reviewed in Ref. 1). The existence of nine different FGFs, four FGFR genes with a number of alternative spliced forms may in part explain the pleiotropic effects of the FGF family. Thus, it will be important to investigate the biological activity of the different FGFR gene products, in response to different FGF members, in order to elucidate the signal transduction pathways leading to these varied responses. We have isolated an FGFR1 cDNA from Xenopus blastulae that differs from previously cloned Xenopus FGFR1s by a Val-Thr deletion in the juxtamembrane region. Although similar isoforms have been cloned from a human hepatoma cell line (23) and from rat brain (26), their biological activity was not characterized. We show here that Thr 424 can be phosphorylated by PKC and in an in vivo functional assay, we demonstrate that the biological activity of the FGFR1 containing this Thr was significantly reduced by activation of PKC.
Our data shows that, as in the human FGFR1 gene (27), the nucleotides encoding the Val-Thr are located at an exon-intron boundary, indicating that this isoform is generated by the use of an alternative 5Ј splice site. Both FGFR-VTϪ and -VTϩ mRNA were expressed in Xenopus embryos at various stages of development and contrary to what one would predict from comparison to 5Ј splice site consensus sequences, FGFR-VTϪ was the minor form. However, it has been suggested that identity of consensus sequences at the 5Ј splice site is not the sole determinant in site selection but that there must be other sequence elements or factors that contribute to the choice of 5Ј splice site (22).
We have shown that Thr 424 can be phosphorylated by PKC. In the VTϪ peptide, a Ser is in position 424, but was not a substrate for PKC. Since PKC requires basic residues in the Ϫ3 to ϩ3 region of the phosphoacceptor site (24), one possible explanation for this discrepancy is the presence of an acidic residue (Asp) in the ϩ2 position. Alternatively, deletion of Val-Thr may change the secondary structure in this region, modifying recognition by PKC.
Members of the FGF family induce mesoderm differentiation in explanted tissue from Xenopus embryos (6 -8) and are thought to play a role in mesodermal patterning in the developing embryo. Convincing evidence for this comes from experiments with a dominant negative mutant construct of the FGFR1 which inhibited wild-type receptor activity (9). These authors showed that expression of mutant FGFR1 in Xenopus embryos resulted in deficiencies in organized mesodermal tissue, suggesting a specific role for FGF in differentiation of presumptive mesodermal tissue. However, FGFRs are present on the surface of all cells in the embryo during blastula stages (28), making it was unclear how FGF induction might be limited to presumptive mesoderm. In further studies, we demonstrated that PKC was activated during mesoderm induction by FGF in explants (29). Our data suggested that PKC was involved in the negative regulation of FGFR activity, since pretreatment of explants with PMA inhibited FGF induction in this tissue. These data suggest there may be an autocrine regulation of FGFR activity whose extent may depend upon the proportion of VTϩ and VTϪ forms expressed by individual cells or tissues. Certainly, the tissue-specific expression pattern of the two isoforms in the adult rat suggests that the VTϪ isoform plays an important role in mediating FGF responses in the brain (26). Although we observed no change in the temporal expression pattern of mRNA encoding the two isoforms in the whole embryo, differential expression may occur over shorter time periods than those examined or FGFR-VTϪ mRNA may be selectively expressed in a subpopulation of cells within the embryo. We are currently investigating which FGFR1 isoforms are expressed in different tissues of the Xenopus blastula and determining the biological role of these two isoforms in the developing embryo.