SpFGFR, a new member of the fibroblast growth factor receptor family, is developmentally regulated during early sea urchin development.

We describe the cloning of a new fibroblast growth factor receptor, SpFGFR1, that is differentially regulated at the level of transcript abundance during sea urchin embryogenesis. Sequence representing the conserved tyrosine kinase domain was obtained by reverse transcription-polymerase chain reaction using degenerate primers, and the entire open reading frame was obtained by standard cDNA library screening methods. SpFGFR contains a series of domains characteristic of FGFRs: three immunoglobulin-like motifs, an acid box, a transmembrane domain, a relatively long juxtamembrane sequence, a split tyrosine kinase domain, and two conserved intracellular tyrosine residues. Alternative splicing of SpFGFR generates two variants (Ig3L and Ig3S), which differ by insertion in the center of the Ig3 domain of 34 extra amino acids, encoded by an additional exon. Transcripts encoding both variants accumulate when morphogenesis begins with mesenchyme cell ingression and gastrulation. SpFGFR transcripts accumulate in all cell types of the embryo, although in situ hybridization shows that they are somewhat enriched in cells of oral ectoderm and endoderm. Transcripts encoding the Ig3S variant, whose structure resembles more closely that of vertebrate receptors, are enriched in endomesoderm, suggesting that the SpFGFR variants could play distinct roles in the sea urchin embryo.

Cell-cell interactions are important in specifying cell fates in all embryos. Receptor-ligand combinations that activate signal transduction pathways are known mediators of these interactions in some cases and obvious candidates for signaling in others. Cell fate specification in sea urchin embryos is highly dependent on cell-cell interactions (reviewed in Refs. 1 and 2). For example, signaling among different blastomeres is known to begin during early cleavage since the developmental fates of tiers of early blastomeres are altered both by isolation and by transplantation into different signaling environments (3)(4)(5)(6). Maintenance of positional information also requires continued signaling through blastula stage and, for determination of endodermal and secondary mesenchymal lineages, through early gastrula stages (7)(8)(9). In addition to positive inductive influences, signals sent from primary mesenchyme cells prevent pluripotent secondary mesenchyme cells from converting to a skeletogenic fate at the late gastrula stage (10), and unidentified signals repress expression of vegetal markers (11).
The effect of LiCl on sea urchin development suggests the kinds of pathways that might mediate cell-cell interactions required to specify blastomere fates. Animal hemispheres (presumptive ectoderm) (12) and animal mesomere pairs isolated from 16-cell embryos (13) that are cultured in the presence of LiCl form spicules and guts, structures normally derived only from vegetal cells. Increasing LiCl concentration induces correspondingly increased levels of expression of vegetal molecular markers, which mimics the effect of transplantation of vegetal cells to the animal pole of cleaving embryos. LiCl is thought to exert its effect directly on signal transduction pathways by blocking inositol (1,4,5)-trisphosphate metabolism (14) and by altering the activity of G-proteins in membranes, thus affecting diacylglycerol production (15). Levels of inositol (1,4,5)trisphosphate and diacylglycerol are regulated by phospholipase C, several isoforms of which interact with seven-membrane spanning and tyrosine kinase receptors (reviewed in Ref. 16). Thus, the vegetalizing influence of LiCl on animal hemisphere blastomeres supports the idea that induction from vegetal cells may be mediated through one or both of these signaling pathways.
No specific endogenous molecule mediating cell-cell interactions has been identified in the sea urchin embryo. The widespread use of tyrosine kinase receptors (TKRs) 1 -stimulated signal transduction pathways in the development of other embryos (recently reviewed in Refs. 17 and 18) suggests that TKR homologs may also play an important role in sea urchin development. Therefore, we searched for members of this family that are expressed during sea urchin (Strongylocentrotus purpuratus) development. This search identified six partial cDNAs with homology to tyrosine kinases. This report focuses on one with strong homology to the vertebrate fibroblast growth factor receptor family. We have begun to explore the possible developmental role of this receptor in sea urchin embryogenesis by analyzing its sequence and characterizing the temporal and spatial pattern of expression of its gene.

RT-PCR
cDNA was synthesized from RNA isolated from embryos as follows. Eleven l of a solution containing 1 M random primers or downstream primer and 1 g RNA in H 2 O were incubated at 65°C for 10 min and then placed on ice. The solutions were brought to a total volume of 20 l containing 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2 , 0.5 mM of each deoxynucleotide triphosphate, 100 M dithiothreitol, and 200 units of either Moloney murine leukemia virus reverse transcriptase or Superscript (Life Technologies, Inc.), incubated at 42°C for 1 h and then placed on ice. Two l of 3 M sodium acetate were added, followed by two extractions with phenol:chloroform:isoamyl alcohol (25:24:1). The cDNA was precipitated by the addition of 2.5 volumes of ethanol, and the precipitate was collected by centrifugation for 10 min at 4°C in a microcentrifuge, washed with 70% ethanol, and dried. The cDNA was brought to a volume of 50 -100 l for PCR, containing the following components: 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 2 mM MgCl 2 , 1 M each of upstream and downstream primers (see below), and 200 M of each deoxynucleotide triphosphate. Amplification was carried out using one of the following cycling programs: cycle 1: 5 min at 94°C/30 -35 ϫ [1 min at 94°C, 1 min at 37°C, and 1 min at 72°C]/10 min at 72°C; cycle 2, 5 min at 94°C/30 -35 ϫ [1 min at 94°C/1 min at 45°C/1 min at 72°C]/10 min at 72°C.
For RT-PCR to obtain partial tyrosine kinase cDNAs, random primers were used to reverse transcribe RNAs isolated from egg, 24-, 36-, and 48-h embryos. Primers 1a, 1b, and c or 2a, 2b, and c (defined in Fig.  1A) were used for PCR and cycling program 1. To detect splice variants, reverse transcription of embryo RNA, followed by PCR with cycling program 2 above, was carried out. Locations of primer sequences in SpFGFR are indicated in Figs. 1 and 3A.

Cloning and Screening of PCR Products
Products obtained using primers containing 5Ј restriction sites were digested with specific restriction endonucleases and inserted into the plasmid Bluescript (pBS) by standard methods. Products obtained with primers lacking restriction sites were cloned into the "TA" cloning vector, pCR (Invitrogen, San Diego, CA). For screening of cDNA clones obtained from PCR products amplified with degenerate primers, bacterial colonies were directly analyzed by PCR using the original forward primer and a new, nested reverse primer (primer c, Fig. 1). Positive plasmid DNAs were sorted into categories by G-ladder sequencing, and one or two clones from each different group were fully sequenced.

Screening cDNA Libraries to Obtain Full-length SpFGFR cDNA
To obtain the complete SpFGFR open reading frame (ORF) sequence, a library of randomly primed cDNA inserts in ZAP representing RNAs from the ectoderm-enriched fraction of 72-h embryos, (20) was screened by hybridization with a 32 P-labeled randomly primed probe representing the partial RT-PCR-derived cDNA. Hybridization was carried out in 4 ϫ SSC, 1% SDS, 10% polyethylene glycol, 25 mM sodium phosphate, and 50 mM Tris-HCl, pH 7.5, at 50°C (melting temperature, Tm Ϫ 35°C) overnight, followed by 20-min washes in 4 ϫ SSC, 2 ϫ SSC, 1 ϫ SSC, and 0.6 ϫ SSC (1 ϫ SSC is 0.15 M NaCl, 0.015 M trisodium citrate), all containing 1% SDS at 69°C (maximum stringency, Tm Ϫ 15°C). Secondary screens yielded one positive cDNA, JG412, containing sequence identical to the original partial cDNA isolated by RT-PCR and extending through the termination codon (ϩ2437 to ϩ2919). JG412 also contains approximately 1.6 kilobase of 3Ј untranslated region, which was only partially sequenced. Approximately 7.5 ϫ 10 5 plaques from a different library containing cDNA randomly primed from 43-h embryo RNA in ZAP (kindly provided by Kathleen Foltz and William Lennarz) (WL library) was differentially screened for sequences extending upstream using randomly primed probes representing either 3Ј sequences from JG412 or the more upstream sequence identified by RT-PCR. After secondary screening, the two longest clones, WL2 and WL10, were sequenced. WL2 contains ORF sequence from ϩ490 to ϩ2453, and WL10 extends upstream of the ORF (Ϫ426 to ϩ2453). The complete ORF was constructed by combining partial cDNA sequences in the plasmid Bluescript (pBS-SpFGFR) by standard methods. Deduced amino acid sequences were analyzed using the software of Genetics Computer Group, University of Wisconsin, Madison, WI (21) and Pearson and Lipman (Ref. 22; FASTA analysis). This sequence has been deposited in Genbank, accession number U17164.

Hybridization Assays
RNase Protection-For developmental expression studies, RNase protection assays were performed as described previously (23) by hybridizing 10 g of total RNA, isolated as described by Nemer et al. (24), from embryos at selected stages to a molar sequence excess (1 ng) of 32 P-labeled antisense riboprobes for 18 h. These conditions have been shown to achieve kinetic termination (see the legend to Fig. 5 for details.) RNase protection assays using RNAs from tissue fractions were carried out in a similar fashion (see the legend to Fig. 8 for details). Tissue fractionation was carried out as described previously (23).
In Situ Hybridization-Probe preparation and in situ hybridization were carried out according to Angerer and Angerer (25). Antisense and sense riboprobes labeled with [ 33 P]UTP to 2 ϫ 10 8 dpm/g were transcribed in vitro. These probes were hybridized in situ to 5-m sections of embryos of selected stages. Unhybridized probe was removed by a stringent wash (60°C; 50% formamide, 0.3 M NaCl, 20 mM Tris-HCl, and 1 mM EDTA, pH 8.0), and slides were exposed to Kodak NTB-2 liquid track emulsion for 1 week.

Autophosphorylation Assays
To express the tyrosine kinase domain (TKD) and ligand binding domain (LBD) of SpFGFR in Escherichia coli, the following recombinants were constructed in the vector pET15b (Novagen, Madison, WI). For pET-TKD, PCR product was synthesized using primers 20 and 14 (see Fig. 3A for primer location in SpFGFR), each of which contained a 5Ј BamHI restriction site. Amplified fragments were digested with BamHI, treated with Klenow fragment to create blunt ends, and inserted into XhoI-digested, blunted-ended pET15b vector. For pET-LBD, PCR product was synthesized using primers 13 and 12 (see Fig. 3A for primer positions) and SpFGFR template cDNA (WL210). Primers contained 5Ј NdeI and XhoI sites, respectively, and amplified DNAs were directionally inserted between the NdeI and XhoI sites of pET15b.
Lysates of uninduced and induced cultures of E. coli strain BL21 containing plasmid pET-TKD or pET-LBD were prepared according to the manufacturer's instructions (Novagen), fractionated on 7.5% SDS-PAGE gels, blotted to Immobilon (Millipore, Bedford, MA) in buffer containing 25 mM Tris-HCl, 192 mM glycine, pH 8.3, and 20% methanol, and assayed for the presence of phosphotyrosine-containing proteins by immunostaining with monoclonal antiphosphotyrosine antibody 4G10 (UBI Inc., Lake Placid, NY) diluted 1:10,000. Detection of bound antibody was carried out with anti-mouse secondary antibody conjugated to horseradish peroxidase (1:5000) and the Renaissance chemiluminescence detection kit (DuPont NEN).

Isolation of cDNA Encoding a SpFGFR and Characterization
of Its Sequence-We used RT-PCR to identify receptor-type tyrosine kinase domains (TKRs) expressed during sea urchin embryogenesis. Two sets of degenerate oligonucleotide primers were used that represent conserved sequences in these domains (24) (Fig. 1). Set 1, generously provided by Dr. John Casnellie (University of Rochester), was designed to discriminate against nonreceptor tyrosine kinases, whereas Set 2 was designed to select for TKRs typified by the epidermal growth factor receptor. Randomly primed cDNAs representing RNAs isolated from eggs and 12-, 24-, or 48-h embryos were amplified with downstream primers 1a or 2a and upstream primer 1b or 2b and yielded products of predicted length (ϳ220 bp). These products were cloned and screened for the presence of a third sequence diagnostic of TKRs by amplifying with the original upstream primer and a nested downstream primer (primer c). Sequencing of the positive clones obtained from this additional screen identified cDNAs encoding six different partial TK domains. One of these was chosen for further study because it showed high similarity to FGFRs, which have been shown to play key roles in the early development of both invertebrate and vertebrate embryos (26 -32).
The sequence of cDNA encompassing a complete ORF was assembled from overlapping cDNAs obtained from several libraries by screening with the PCR-derived TKD probe. The inferred amino acid sequence is presented in Fig. 2. Although the longest cDNA contains nine nonproductive translation initiation codons at its 5Ј end (data not shown), the one assigned position ϩ1 is the likely initiation codon because it is the first ATG of the longest ORF and is surrounded by sequence corre-sponding to a translation initiation site as defined by Kozak (33,34). FASTA searches of the data bases using the deduced amino acid sequence of the 2916-nucleotide ORF show that the 13 most similar sequences all are members of the FGFR family. This new member of the FGFR family has been named SpF-GFR. The structure of the predicted peptide, which contains all of the characteristic structural features of these receptors (reviewed in Ref. 35), is illustrated in Fig. 2. Following the putative translation initiation codon, the extracellular domain contains a cluster of hydrophobic and charged amino acids characteristic of a signal sequence (Ref. 36; underlined), then three Ig-like repeats (shaded) with the motif, C-(X 11-12 )-W-(X 50 -60 )-D-XGXYXC (W, D, G, and Ys are boxed). A stretch of acidic residues ("acid box" motif), which distinguishes FGFRs from other receptor tyrosine kinases, is found between the first and second Ig-like repeats (circled), although it is shorter than the acid boxes of vertebrate FGFRs and contains glutamate. The extracellular domain contains 11 potential sites for Nlinked glycosylation (circled), and tunicamycin-sensitive modification of this protein has been observed in vivo. 2 Following the transmembrane domain (underlined) are two other features characteristic of this class of receptor tyrosine kinases: 1) an unusually long (82-amino acid) juxtamembrane domain; and 2) a kinase domain split by a 15-amino acid insert (KI). The ATP binding motif, GXGXXG (amino acids 646 -651), as well as the catalytic motifs DLXXXN, DFG, P(VI)(KR)W(MT)APE, and DVW(SA)(FY)G, found in other protein kinases (37), are present in the C-terminal half of the TK domain (Fig. 2). The sequence motifs DLAARN (amino acids 781-786) and PVKW-MAPE (amino acids 820 -827) are diagnostic for protein kinases with tyrosine specificity (37). Although the most divergent portions of the cytoplasmic domain are the kinase insert and the C terminus, their lengths are conserved, as are two Tyr residues (Tyr 812 and Tyr 924; triangles), which, in vertebrate receptors, have been shown to be important for receptor function (38,39).
Multiple sequence alignments and phylogenetic analysis us-ing the Clustal program (40) show that the FGFR sequences from four groups, vertebrates, sea urchins, Drosophila, and Caenorhabditis elegans, are similarly diverged from each other in all pairwise combinations. Thus, SpFGFR has approximately the same degree of similarity to the vertebrate FGFRs as it has to those recently isolated from Drosophila (41) and C. elegans (30). Although sea urchins are invertebrates, their embryos are bilaterally symmetric enterocoelous deuterostomes and share a closest common ancestor with vertebrates (42). Therefore, we will compare the structural features and degree of similarity of different domains of SpFGFR to those of the vertebrate receptors.  substitutions) at the amino acid level, respectively. In contrast, the sequence just upstream of the Ig2 loop is divergent in SpFGFR. In vertebrate receptors, this region includes the motif, KMEKKLHAVPAAKTVK, which functions in heparin binding and is required for high affinity ligand binding (46 and references cited therein). Its functional importance has been implied by the observation that receptors mutant in this sequence are inactive in dominant-negative interference assays (47). Presence or absence of a heparin binding site may reflect an important difference in the way FGFRs function in vertebrates and invertebrates since we note this motif is not present in the Drosophila and C. elegans receptors (30,41).
The extracellular region of SpFGFR differs from that of its vertebrate counterparts in that it contains additional amino acid residues. Approximately 100 are found between the signal sequence and the first Ig domain in a region that is not critical for ligand binding by the vertebrate receptors. As described below, two isoforms of SpFGFR, Ig3L and Ig3S, have been identified (Fig. 2). The former contains 34 amino acids inserted in the middle of the Ig3 domain. The remaining 15 extra amino acids are scattered throughout the LBD.
Two Alternatively Spliced Transcripts Are Produced from the SpFGFR Gene-Because vertebrate FGFR Ig3 variants generated by alternative splicing have different ligand affinities and specificities and because SpFGFR contains extra amino acids within this region, we asked whether Ig3 variant transcripts exist by using RT-PCR with total RNA templates isolated from blastula-and/or gastrula-stage embryos, as diagrammed in Fig. 3A. Using primers 6 and 12, two products were obtained whose sequences are shown in Fig. 3B, top two lines. Neither sequence represents a cloning artifact since both forms can be detected in embryo RNA by RNase protection and RT-PCR (see below). Aside from several nucleotide differences attributable to allelic variation, the shared amino acid sequences of the longer (Ig3L) and shorter (Ig3S) Ig3 domains are identical (Fig.  3B), but Ig3S represents a different transcript from which 102 nucleotides/34 amino acids are deleted. The extra sequence in Ig3L is inserted between the Ig3 constant and variable regions of the vertebrate receptors, very close to the intron/exon border in Ig3 domains of vertebrate receptors (Fig. 3B, 1). It is probable that these variants arise by alternative splicing of transcripts from a single gene. The extra nucleotides in the Ig3L form do not constitute an unspliced intron since 5Ј and 3Ј splice signals are not found at the borders of the insertion. Thus, the simplest model is that an additional exon of 102 nucleotides is incorporated in the Ig3L variant. This contrasts with alternative splicing in the Ig3 domains of vertebrate FGFRs in which variants are generated by splicing in any one of three alternate exons encoding the Ig3v region (48). The amino acid sequence of the SpFGFR Ig3S C-terminal half is most similar to that of exon IIIc (38% identity) and less closely related to exons IIIb (29% identity) and IIIa (9% identity).
Consistent with the hypothesis that Ig3S and L variant sequences reside within transcripts from the same gene, RT-PCR using primers and templates diagrammed in Fig. 3A demonstrates that Ig3S is linked to the same sequences as Ig3L. With the primer pair (6 and 11) that amplifies sequences from upstream of Ig3 extending to the C terminus of the ORF, two fragments were obtained corresponding to the expected sizes (Fig. 3C, lane 1, 1637 and 1535 bp, indicated by lines at the right side of the panel). This result was confirmed using a second downstream primer located in the TKD domain (primer 3), which yielded fragments of 972 and 870 bp as expected (Fig.  3C, lane 3), each of which contained an XhoI site at the correct position (Fig. 3C, lane 4). Similar analysis of upstream sequences shows that the Ig3S form extends at least through the insertion point, it is not expected to function with an Ig3L template under the amplification conditions used. Consistent with this, products were observed using sea urchin embryo cDNA but not with a plasmid bearing the Ig3L sequence (data not shown). Conclusive data on this point will require isolation and sequencing of additional cDNAs representing the Ig3S transcript variant. We have tested for the presence of other transcript variants corresponding to those demonstrated for vertebrate receptors, i.e. ones lacking the Ig1 domain (49) or truncated in the intracellular domain (50), but none was detected by RT-PCR using multiple primer combinations. Thus, our analysis suggests that only two variants of the reading frame are expressed in sea urchin embryos.
SpFGFR Is a Functional Tyrosine Kinase-The ability of SpFGFR to autophosphorylate tyrosine residues was tested in E. coli, which lack endogenous tyrosine kinase activity. Lysates of uninduced or induced cultures expressing either a kinase domain fusion protein or a ligand binding domain fusion protein were analyzed on Western blots probed with an antibody specific for phosphotyrosine. A peptide of the expected molecular mass was detected by the antibody in cultures induced to express the kinase domain fusion protein, whereas no peptide was detectable in uninduced cultures or in cultures expressing ligand binding domain fusion proteins (Fig. 4). Therefore, SpF-GFR is a functional tyrosine kinase.
SpFGFR Transcript Accumulation Is Developmentally Regulated-SpFGFR transcript abundance is developmentally regulated as shown by RNase protection assays (Fig. 5). Total RNA from embryos of selected developmental stages was hybridized with a probe to the ligand binding domain sequence that detects both Ig3L and Ig3S transcript variants, yielding protected fragments of 320 and 183 nucleotides, respectively. A third expected fragment of 36 nucleotides is not retained on the gel in this experiment. Both transcripts increase in abundance in parallel between late cleavage (12 h) and early gastrula stage (36 h) and persist through pluteus stage (72 h), with the longer variant RNA being severalfold more abundant. Ig3L and Ig3S transcripts are also present at similar relative abundance in total RNA from an adult male sea urchin (data not shown). In contrast, only Ig3L transcripts are detectable in eggs and during early cleavage stages (through 12 h, ϳ150-cell stage), suggesting that splicing of maternal transcripts is regulated during oogenesis. From comparison of SpFGFR signal intensities to those obtained in similar assays using probes for other mRNAs of known abundance, we estimate that there are, on average, ϳ10 -20 SpFGFR transcripts/cell at peak abundance at gastrula (48 h) and pluteus (72 h) stages. Thus, SpFGFR transcript concentrations are relatively low, being only about 10-fold more abundant than typical rare messenger RNAs in the sea urchin embryo (51).
SpFGFR Transcripts Accumulate in Most, If Not All, Different Cell Types during Embryogenesis-Because RNase protection assays showed that SpFGFR transcripts are relatively rare, we used riboprobes corresponding to the whole ORF and 58 nucleotides of 5Ј untranslated region that were labeled with 33 P to carry out sensitive in situ hybridization assays. This probe detects both Ig3L and Ig3S-encoding RNAs. These experiments showed that SpFGFR transcripts accumulate in all major regions of the embryo (Fig. 6). Control hybridizations with sense strand probes of the same specific activity gave signals similar to emulsion background (data not shown). The magnitude of in situ signals was in agreement with those obtained by RNase protection assays. In blastulae, higher transcript concentrations are frequently observed near the animal and vegetal poles (Fig. 6A). Transcripts are detectable throughout embryos at later stages, with severalfold higher levels in archentera of gastrulae (Fig. 6B) and in ciliary bands, guts (Fig. 6C), and vertices (data not shown) of pluteus larvae. We conclude that there are only relatively small differences in SpFGFR transcript levels among most different cell types.
In situ hybridization signals were low, reflecting the low concentration of SpFGFR transcripts. Therefore, to verify widespread accumulation of these transcripts throughout the embryo, we used an independent assay in which transcript levels were compared in embryo fractions enriched for different cell types (52); RNA isolated from ectoderm and endomesoderm fractions of either early (32-h) or late (47-h) gastrulae was assayed for the presence of SpFGFR transcripts using a probe from the 5Ј end of the cDNA (Fig. 7A, left). As a control to demonstrate the quality of the tissue fractionation, a probe for an endoderm-specific mRNA, Endo-16 (53), was used (Fig. 7A,  right). Endo 16 signals were much higher with RNA from endomesoderm quantitated with a phosphorimager, demonstrating effective separation of cell types. In contrast, SpFGFR signals were approximately equal for both cell populations. This result confirms the in situ hybridization results, indicating that SpFGFR transcripts are distributed throughout the embryo.
The expression pattern revealed by in situ hybridization represents the sum of contributions of Ig3S and Ig3L variants. Because detection of SpFGFR transcripts by in situ hybridization requires probes of much higher sequence complexity than could be provided by the Ig3L-specific sequence, we used RNase protection as described above (Fig. 5)  Endo 16 mRNA (5-6-fold enrichment), Ig3S transcripts are not confined to the endomesoderm fraction. However, this observation does not exclude the possibility that a subset of cell types within the endomesoderm fraction may express only this form of the receptor. Cell types in the endomesoderm fraction include skeletogenic mesenchyme, endoderm, and secondary mesenchyme cells, which give rise to coelomic pouches, pigment, muscle, and other blastocoelar cells (54). DISCUSSION The SpFGFR gene encodes a new member of the family of FGFR tyrosine kinases. We have shown that the predicted peptide possesses all of the structural characteristics of such receptors and that its tyrosine kinase domain is functional. Furthermore, the peptide contains highly conserved tyrosine residues which, in vertebrate receptors, mediate activation of downstream signaling cascades (55).
SpFGFR is not more closely related to any one of the human FGFRs than to any of the others. The vertebrate FGFRs are less closely related to each other, within a species, than are the individual FGFR among different species, implying that the vertebrate FGFR gene family is the result of gene duplications that occurred before divergence of the lines leading to the vertebrates sampled (35). Sea urchin embryos are bilaterally symmetric, enterocoelous deuterostomes in the lineage leading to vertebrates. It is not known whether additional FGFR genes exist in the sea urchin genome. At a level of stringency that would have allowed cross reaction among the vertebrate receptors, genomic DNA blots have not identified additional sea urchin receptor genes. Based on these considerations, it is conceivable that SpFGFR is homologous to the ancestral vertebrate FGFR gene. In addition, the existence of FGFRs in Drosophila and C. elegans shows that FGFR is an ancient gene that existed prior to the divergence of the protostomes and deuterostomes.
The ligand(s) that activate SpFGFR have not yet been identified. FGF1, FGF2, FGF4, and FGF6 homologs were not detected by RT-PCR, and SpFGFR expressed in COS cells does not bind FGF2 or FGF7. 2 Similarly, the ligands for FGFRs expressed in other invertebrate embryos have not been identified. Although there is considerable conservation of sequence within most regions of the Ig2, Ig2-Ig3 interloop and Ig3 portions of SpFGFR, which regulate ligand specificity and affinity in vertebrate receptors (43,44), there is at least one probable key difference; sequence conserved among the vertebrate receptors at the N terminus of Ig2 that mediates heparin interaction (46) required for binding of at least several FGF ligands (56) is not found in SpFGFR or in the other invertebrate receptors. These observations, and the recent demonstration that Xenopus FGFRs bind novel ligands completely unrelated to the FGF family (57), underscore the need to screen for ligand(s) for SpFGFR using a ligand-dependent functional receptor assay.
Alternative splicing of the C-terminal half of the vertebrate FGF receptors 1-3 generates multiple isoforms with altered ligand specificities (58). Interestingly, alternative splicing of SpFGFR transcripts in this domain is distinctly different since an extra exon is either retained (SpFGFR-Ig3L) or not retained (SpFGFR-Ig3S). This extra sequence does not correspond to an unspliced intron since it lacks splice consensus sequences. The facts that this optional exon encodes a contiguous ORF segment and is inserted in a region of the ligand binding domain known to be critical for specificity and affinity suggest that the Ig3L and Ig3S variants of SpFGFR may have distinct functions.
SpFGFR gene expression is temporally regulated at a critical developmental interval, blastula through gastrula stages, when differentiation of certain cell types, particularly those derived from the vegetal anlagen of primary and secondary FIG. 7. Sp FGFR transcripts are present at similar levels in ectoderm and endo-mesoderm. A, RNase protection assays in which 10 g of RNA isolated from ectoderm (Ec) and endomesoderm (E/M) fractions of early (32 h; left two lanes for each set) or late (47 h; right two lanes for each set) gastrulae were hybridized to a SpFGFR riboprobe (4 ϫ 10 8 dpm/g) or an endo-16 probe (2 ϫ 10 8 dpm/g). Two protected bands evident in the SpFGFR 47-h gastrula lanes are attributable to the presence of two different alleles in the embryos from which this RNA was isolated. B, as in A, except that the endo-16 probe specific activity was 1 ϫ 10 8 dpm/g, and the amount of 47-h gastrula RNA was 5.5 g. The SpFGFR probe in this case spanned the Ig3 domain to monitor the relative concentrations of Ig3L-and S-transcript variants. Ig3L-and Ig3S-specific transcripts are indicated by arrows, and the signals in these fragments were quantitated by phosphorimagery. Additional bands correspond to polymorphism between the probe sequence specific to the Ig3L form and that in the gastrula RNA, which are derived from different individuals. Markers in this experiment were 32 P end-labeled X174 single-stranded DNAs. For both experiments in A and B, P indicates the position of unhybridized probe, and C is a control in which probe was incubated with yeast RNA.

FIG. 6. Distribution of SpFGFR transcripts in embryos.
Antisense riboprobes, labeled to 2 ϫ 10 8 dpm/g with [ 33 P]UTP and representing the entire SpFGFR ORF and 58 nucleotides of 5Ј untranslated region (Ϫ58 to ϩ2916), were hybridized at saturating probe concentration in situ to 5-m sections of embryos of selected stages. Exposures were for 1 week. A, blastula with higher signals at the animal (a) and vegetal (v) poles. B, gastrula with higher signals in the archenteron (ar). C, pluteus larva cut parallel to the oral/aboral axis of the embryo (oe, oral ectoderm; aoe, aboral ectoderm), with higher signals in ciliary band (cb) and gut (g). The bar in A represents 10 m. mesenchyme and endoderm, commences. This period also marks the onset of morphogenesis of these tissues, involving migrations and cell shape changes that include the ingression of mesenchyme, invagination of endoderm, and emigration of cells from the tip of the archenteron to give rise to pigment cells, the coelomic pouches, and the circumesophageal musculature. The timing of SpFGFR expression in sea urchin embryos suggests potential roles for this receptor analogous to those demonstrated in mesodermal patterning and cell migration in embryos of both vertebrates (26 -28) and invertebrates (30 -32).
The accumulation of SpFGFR transcripts, however, is not confined to vegetal lineages that yield endoderm and mesenchyme. In situ hybridization and RNase protection assays independently show that the SpFGFR gene is transcribed in most, if not all, cell types and clearly in ectoderm, which is derived from animal blastomeres. Nevertheless, some spatial regulation of isoform type does occur since endomesoderm fractions are enriched in transcripts encoding the Ig3S variant, the structure of which more closely resembles that of vertebrate receptors. In fact, this form could be the only variant expressed in a subset of endomesodermal cells. To examine the distribution of SpFGFR proteins and to asses their developmental roles, we are generating appropriate immunological reagents and dominant-negative versions of the receptor to perturb corresponding signaling pathways.