Constitutive activation of fibroblast growth factor receptor-2 by a point mutation associated with Crouzon syndrome.

The fibroblast growth factor receptors (FGFRs) are a family of ligand-activated, membrane-spanning tyrosine kinases. Mutations in several human FGFR genes have been identified as playing a role in certain disorders of bone growth and development. One of these, Crouzon syndrome, an autosomal dominant disorder causing craniosynostosis, has been associated with mutations in the human FGFR-2 gene. We report here that microinjection of Xenopus embryos with RNA encoding an FGFR-2 protein bearing a Cys Tyr mutation (FGFR-2CS) found in Crouzon syndrome results in fibroblast growth factor (FGF)-independent induction of mesoderm in animal pole explants. Wild-type FGFR-2 did not induce mesoderm when injected at similar doses. The effects of the mutant receptor were blocked by co-expression of dominant negative mutants of either Raf or Ras. Analysis of the mutant receptor protein expressed in Xenopus oocytes indicates that it forms covalent homodimers, does not bind radiolabeled FGF, and has increased tyrosine phosphorylation. These results indicate that FGFR-2CS forms an intermolecular disulfide bond resulting in receptor dimerization and ligand-independent activation that may play a role in the etiology of Crouzon syndrome.

The fibroblast growth factor receptors (FGFRs) are a family of ligand-activated, membrane-spanning tyrosine kinases. Mutations in several human FGFR genes have been identified as playing a role in certain disorders of bone growth and development. One of these, Crouzon syndrome, an autosomal dominant disorder causing craniosynostosis, has been associated with mutations in the human FGFR-2 gene. We report here that microinjection of Xenopus embryos with RNA encoding an FGFR-2 protein bearing a Cys 332 3 Tyr mutation (FGFR-2CS) found in Crouzon syndrome results in fibroblast growth factor (FGF)-independent induction of mesoderm in animal pole explants. Wild-type FGFR-2 did not induce mesoderm when injected at similar doses. The effects of the mutant receptor were blocked by co-expression of dominant negative mutants of either Raf or Ras. Analysis of the mutant receptor protein expressed in Xenopus oocytes indicates that it forms covalent homodimers, does not bind radiolabeled FGF, and has increased tyrosine phosphorylation. These results indicate that FGFR-2CS forms an intermolecular disulfide bond resulting in receptor dimerization and ligand-independent activation that may play a role in the etiology of Crouzon syndrome.
The fibroblast growth factors (FGFs) 1 are a family of polypeptide mitogens that currently consists of nine members (1,2). The FGFs mediate a variety of biological processes including angiogenesis, wound healing, migration, mitogenesis, neuronal survival, and mesoderm induction (2,3). These biological effects are mediated via binding to four members of a family of high affinity membrane-spanning tyrosine kinase receptors (2,3). The FGFs have also been shown to bind to lower affinity cell surface heparan sulfate proteoglycans (1,2). The prototype FGF receptor (FGFR) is comprised of an extracellular domain made up of three immunoglobulin (Ig)-like domains designated IgI-IgIII, a hydrophobic membrane-spanning region, and a cytoplasmic tyrosine kinase domain (2,3). The amino acid sequences of individual members of the FGFR family are highly conserved among vertebrate species (2). The IgIII domain of FGFR1-3 is encoded by three exons and is generated by alternative splicing of IgIIIa with one of two alternative exons designated IgIIIb and IgIIIc (4,5). This alternative splicing generates receptor isoforms with varying ligand binding specificities (6 -8). Like the FGFs themselves, the FGFRs have unique but overlapping spatiotemporal patterns of expression during vertebrate development (9 -11). The unique patterns of expression of both FGFs and their receptors during vertebrate development suggest that each may have a specialized function. Recent experimental evidence indicates that when FGFR function is disrupted by genetic manipulation, major defects in embryonic development occur (12)(13)(14)(15)(16).
Within the last year, several mutations have been identified in FGFR genes that appear to be the cause of several human disorders of bone growth and development (2). One of these, Crouzon syndrome, is characterized by craniosynostosis, an abnormality of skull development in which the sutures of the growing bones fuse prematurely (17). A variety of mutations in exons IgIIIa and IgIIIc of FGFR-2 have been identified in Crouzon syndrome (17)(18)(19). These mutations may either directly (Cys 342 3 Tyr/Cys 342 3 Arg/Cys 342 3 Ser/Cys 342 3 Phe; Cys 278 3 Phe; Tyr 328 3 Cys; Ser 347 3 Cys; Ser 354 3 Cys) or indirectly (Ser 267 3 Pro; Gln 289 3 Pro; Tyr 340 3 His) result in the creation of a free cysteine residue that could result in covalent dimerization resulting in ligand-independent activation of the mutant receptor (2). Here we report that mutation of Cys 332 3 Tyr of Xenopus FGFR-2, analogous to the Cys 342 3 Tyr mutation most commonly found in Crouzon syndrome, promotes activation of the mutant receptor in the absence of ligand.

EXPERIMENTAL PROCEDURES
Materials-C4-Raf cDNA and N17-Ras cDNA were gifts of Dr. U. Rapp (National Cancer Institute) and Dr. T. Sargent (NIH), respectively. Affinity-purified rabbit antibodies to FGFR were prepared as described previously (20). A murine monoclonal antibody to phosphotyrosine, PY20, was obtained from Transduction Laboratories. Recombinant FGF-1 was a gift from Dr. W. Burgess (Holland Laboratory).
In Vitro Mutagenesis-A BamHI fragment encoding the entire open reading frame of Xenopus FGFR-2 (10) was subcloned into pTZ19U (Bio-Rad). Mutagenesis of Cys 332 to Tyr in Xenopus FGFR-2 was performed by the method of Kunkel et al. (21) using the mutagenic primer 5Ј-TCCAGCTATATAAGTATATTCCCC-3Ј to yield FGFR-2CS. The presence of the Cys 332 3 Tyr mutation and the absence of other mutations were confirmed by sequence analysis.
Plasmid Construction-All constructs for in vitro transcription were cloned into the BglII site of the SP64T or SP64T3 vectors (gifts of Dr. D. Melton, Harvard University). Synthesis of capped mRNA for microinjection was performed with SP6 RNA polymerase using a Message Machine kit (Ambion).
Embryo Injections-Eggs were collected from Xenopus laevis females and fertilized in vitro as described previously (22). Embryos were dejellied 30 -60 min after fertilization with 2% cysteine, pH 8.0, and maintained at 17°C. At the two-cell stage, embryos were transferred to 1 ϫ MMR (5 mM HEPES, pH 7.8, 100 mM NaCl, 2 mM KCl, 1 mM MgSO 4 , 2 mM CaCl 2 , and 0.1 mM EDTA) containing 50 g/ml gentamicin and 5% Ficoll. Each blastomere of two-cell embryos was injected in the animal pole with 5-10 nl of the indicated amount of RNA.
Animal Cap Assays and RNA Gel Blot Analysis-Animal pole ectoderm (animal caps) was dissected from stage 8 -9 embryos (23) and incubated in 0.5 ϫ MMR containing 1 mg/ml bovine serum albumin (BSA) and 50 g/ml gentamicin in the presence or absence of 200 ng/ml recombinant FGF-1 at 22°C. Animal caps were collected at stage 10.5 or 18 for Xbra mRNA analysis and stage 22-24 for muscle ␣-actin mRNA analysis unless otherwise noted. Total RNA was isolated and analyzed by RNA gel blot as previously described (24 -26). To control for RNA loading, blots were hybridized to a Xenopus 18 S rRNA oligonucleotide probe (26).
Oocyte Injections and Immunoblot Analysis-Oocytes were collected and staged according to established procedures (27). Oocytes were defolliculated by mild collagenase treatment and maintained in 1 ϫ MBS (27) containing 1 mg/ml BSA and 50 g/ml gentamicin at 18°C. Oocytes were injected with 10 -20 nl of RNA at the indicated concentrations and cultured as described above for 1-2 days before immunoblot or crosslinking analysis. Oocytes were extracted in cold lysis buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 1.0 mM EGTA, and 0.1 mM NaVO 4 ) extracted with 1 volume of Freon to remove yolk proteins and centrifuged for 10 min at 10,000 ϫ g. For immunoblot analysis, equal amounts of protein were mixed with an equal volume of 2 ϫ SDS sample buffer and proteins separated by electrophoresis through 6% SDS-polyacrylamide gels. Proteins were electroblotted onto nitrocellulose, and immunoblotting procedures were carried out essentially as described (28). Bound primary antibodies were visualized with horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence procedures (DuPont NEN).
In Vitro Kinase Assay-Oocyte lysates were prepared as described above, and equal amounts of protein were immunoprecipitated with an FGFR antibody (20). Immune complexes were immobilized onto Protein G-Sepharose beads (Pharmacia Biotech Inc.), washed extensively with lysis buffer and kinase reactions performed as described (20). Samples were separated on 6% SDS-polyacrylamide gels, and tyrosine-phosphorylated proteins were visualized by autoradiography (20).
FGF Receptor Cross-linking-Oocytes were injected with 5 ng of either FGFR-2 or FGFR-2CS RNA and cultured for 2 days prior to cross-linking analysis. Thirty oocytes expressing either FGFR-2 or FGFR-2CS were incubated in 0.5 ml of 1 ϫ MBS containing 1 mg/ml BSA and 10 units/ml heparin (Upjohn) (binding buffer) in the presence of 20 ng/ml 125 I-FGF-1 (8 -10 ϫ 10 4 cpm/ng) prepared as described previously (29). Oocytes were incubated at 4°C for 1 h followed by extensive washing with binding buffer. Ligand-receptor complexes were cross-linked with disuccinimidyl suberate (0.3 mM; Pierce) for 15 min at 4°C as described (29). Oocyte lysates were prepared as described above, and ligand-receptor complexes were partially purified by wheat germ agglutinin-agarose adsorption (30), followed by analysis on 7.5% SDSpolyacrylamide gels. Cross-linked complexes were visualized by autoradiography.

RESULTS AND DISCUSSION
Several mutations have been identified in the IgIII domain of human FGFR-2 in individuals with Crouzon syndrome (17,18,31). The most frequently identified mutation is Cys 342 3 Tyr, resulting in the generation of a free cysteine residue that may then be available to form intermolecular disulfide bonds. To test this possibility we created a mutation in a Xenopus FGFR-2 cDNA (Cys 332 3 Tyr) analogous to the human Cys 342 3 Tyr mutation. Fifty picograms of RNA transcribed from wild-type FGFR-2 or FGFR-2CS plasmids were injected into both blastomeres of two-cell stage Xenopus embryos. At the blastula stage, animal pole ectoderm (animal caps) was dissected, cultured, and assayed for elongation (32). By late neurula stages animal caps from embryos injected with FGFR-2CS RNA (Fig. 1d) elongated in a manner similar to control caps treated with FGF-1 (Fig. 1b). Animal caps injected with similar amounts of wild-type FGFR-2 RNA (Fig. 1c) remained spherical and resembled uninjected control animal caps (Fig. 1a). Injection of greater than 200 pg of FGFR-2CS RNA results in cleavage arrest in blastula stage embryos. 2 FGF-mediated mesoderm induction and animal cap elongation have been shown previously to be blocked by the expression of a dominant negative Raf (C4-Raf) or dominant negative Ras (N17-Ras) (33,34). Consistent with this observation, co-expression of either a dominant negative Raf (Fig. 1e) or dominant negative Ras (Fig.  1f) with FGFR-2CS inhibited FGF-independent animal cap elongation.
We also assessed mesoderm induction by assaying for the expression of mesoderm-specific molecular markers. Xbra, the Xenopus homolog of brachyury, is expressed broadly in the presumptive mesoderm of gastrulating embryos and has been shown to be a useful molecular marker for mesoderm induction by growth factors such as FGF (35). Wild-type FGFR-2 or FGFR-2CS RNAs were injected into the animal pole of both blastomeres of two-cell stage embryos. Animal caps were dissected from blastula stage embryos (stages 8 -9) and cultured until sibling controls reached the indicated stages (stage 10.5 or 18). RNA was isolated and analyzed by RNA gel blot hybridization. Fig. 2A shows that Xbra mRNA is expressed in animal caps in a dose-dependent manner following embryo injection with FGFR-2CS RNA. Injection of as little as 20 pg of FGFR-2CS RNA was sufficient to induce expression of Xbra mRNA, and the level detected was similar to that induced by FGF. No Xbra mRNA expression was detected in caps injected with a similar amount of wild-type FGFR-2 RNA or in uninjected control animal caps. Xbra transcripts were not detected in animal caps co-expressing FGFR-2CS and either a dominant negative Ras or a dominant negative Raf.
Muscle-specific ␣-actin mRNA, a late marker for mesoderm formation (36), is also induced in a dose-dependent manner by expression of FGFR-2CS as assessed by RNA gel blot analysis (Fig. 2B). Muscle ␣-actin mRNA is not expressed in uninjected control animal caps (data not shown) or in caps injected with a similar amount of wild-type FGFR-2 RNA. The expression of muscle ␣-actin mRNA is blocked in animal caps co-expressing FGFR-2CS and either a dominant negative Ras or a dominant negative Raf. Thus, by both morphological criteria and by the expression of early and late molecular markers of mesoderm, FGFR-2CS, but not wild-type FGFR-2, has the ability to induce mesoderm in the absence of exogenous FGF.
A noteworthy feature of most Crouzon syndrome mutations identified thus far is the creation or loss of cysteine residues in the IgIII domain of FGFR-2 (2). Both Cys 278 and Cys 342 are predicted to form a disulfide bond essential to formation of the Ig domain. These two cysteine residues are conserved throughout the FGFR family . Mutation of either of these two residues could result in destabilization of the structure of the Ig domain and the creation of a free cysteine residue. The creation of a free cysteine residue either directly or indirectly predicts a possible mechanism for the FGF-independent mesoderm-inducing effects of FGFR-2CS: FGFR dimerization and ligandindependent activation by formation of an intermolecular disulfide bond. To examine this possibility, we microinjected wild-type FGFR-2 or FGFR-2CS RNA into Xenopus oocytes. Lysates from uninjected or injected oocytes were analyzed by reducing or nonreducing SDS-polyacrylamide gel electrophoresis and immunoblotting with an FGFR antibody (20). Under reducing conditions, both FGFR-2 and FGFR-2CS migrated as monomeric forms of 110 and 125 kDa, whereas under nonreducing conditions FGFR-2CS displayed an additional species at ϳ260 -280 kDa (Fig. 3A), consistent with the size of a disulfidelinked homodimer. These same protein samples were then sub-jected to immunoblot analysis with a monoclonal antibody to phosphotyrosine. The pattern of phosphotyrosine immunoreactivity was similar to that observed with the FGFR antibody. However, although similar amounts of receptor protein were present in each sample (Fig. 3A), FGFR-2CS displayed higher amounts of Tyr(P) (Fig. 3B). In addition, under non-reducing conditions, the ϳ260 -280-kDa species seen only in FGFR-2CS lysates displayed significant Tyr(P) immunoreactivity. To confirm these results, immune complex kinase assays were performed. While both FGFR-2 and FGFR-2CS displayed in vitro tyrosine kinase activity, FGFR-2CS consistently displayed higher activity than FGFR-2 (Fig. 3C). When immune complex kinase assay samples were analyzed under non-reducing conditions, a high amount of the kinase activity exists as a ϳ260 -280-kDa species confirming those results seen by immunoblotting with antibodies to FGFR and Tyr(P). These data indicate that a large amount of FGFR-2CS exists as a covalent dimer with elevated tyrosine kinase activity resulting in constitutive activation of the mutant receptor. Such a mechanism has been described for mutagenized epidermal growth factor receptor (37), erythropoietin receptor (38), and for the MEN-2A mutations of the RET receptor-like tyrosine kinase (39).
The IgIII domain of FGFR-2 has been shown to be important for ligand binding (6,7). The Cys 332 3 Tyr mutation of FGFR-2CS, in addition to generating a free cysteine residue capable of forming intermolecular disulfide bonds, may also disrupt the tertiary structure of IgIII resulting in diminished ligand binding. To test this possibility, ligand binding and cross-linking experiments were performed on Xenopus oocytes microinjected with either wild-type FGFR-2 or FGFR-2CS RNA. Oocytes injected with wild-type FGFR-2 RNA exhibited a major 125 Ilabeled cross-linked band of 150 kDa (Fig. 4). Oocytes injected with FGFR-2CS RNA exhibited no increase in 125 I-FGF-1 binding relative to control oocytes (Fig. 4). These data indicate that mutation of Cys 332 results in a loss of ligand binding, and thus activation of the mutant receptor is indeed ligand-independent. These data are consistent with mutation of an equivalent residue in FGFR-1, which has also been shown to abrogate ligand binding (40).

FIG. 2. Induction of molecular markers of mesoderm formation by FGFR-2CS.
Embryos at the two-cell stage were injected with either FGFR-2 or FGFR-2CS RNA in the indicated amounts. For experiments involving dominant negative Raf and Ras mutants, 50 pg of FGFR-2CS RNA was co-injected with 100 pg of C4-Raf RNA or N17-Ras RNA. A, animal caps were dissected at stage 8 -9 and harvested at either stage 10.5 or 18 as indicated. RNA was isolated and Xbra mRNA expression analyzed by RNA gel blot hybridization. The blot was rehybridized with an 18 S rRNA oligonucleotide probe to serve as an RNA loading control. B, embryos were injected and animal caps dissected as described in A. Animal caps were harvested when sibling control embryos reached stage 22-24, and RNA was isolated and analyzed by RNA gel blot hybridization for muscle ␣-actin mRNA expression (arrow). Cytoskeletal actin transcripts (upper two bands) serve as an internal control for RNA loading.

FIG. 3. Biochemical analysis of FGFR-2 and FGFR-2CS expressed in Xenopus oocytes.
A, oocytes were either left uninjected or injected with FGFR-2 RNA or FGFR-2CS RNA, and lysates were prepared and subjected to SDS-polyacrylamide gel electrophoresis under reducing and nonreducing conditions as described under "Materials and Methods." Expression of FGFR-2 and FGFR-2CS was analyzed by immunoblot analysis with a polyclonal antibody against FGFR. B, samples were prepared as described above for A and subjected to immunoblotting with a monoclonal antibody to phosphotyrosine (pTyr, PY20). C, lysates from injected oocytes were subjected to immunoprecipitation with an anti-FGFR antibody, and in vitro kinase assays were performed as described under "Materials and Methods." Electrophoresis conditions are indicated beneath each panel. The molecular mass markers (in kilodaltons) are shown to the right.
These data establish that creation of a free, exposed cysteine residue in the IgIII domain of FGFR-2 results in the formation of an intermolecular disulfide bond and ligand-independent activation of this receptor. These results indicate a potential mechanism for the dominant phenotypic effects observed in Crouzon syndrome and other craniosynostoses involving mutations in FGFR genes (2). It is likely that most of these FGFR mutations are activating ones, and the heterogeneity of clinical features observed in these syndromes may reflect the degree to which individual mutations activate the receptor. Alternatively, the variation in phenotype among individuals with identical point mutations may indicate that other genes may be involved that modify the effects of mutated FGFRs. A systematic analysis of each FGFR mutation, as described here, may shed light on subtle functional differences between different mutations, which may account for phenotypic variability.
The Xenopus system has proven to be a very useful model for the functional analysis of mutant signal transduction molecules in vivo (41). In particular, much has been learned about FGF receptor structure and function employing this system (12,30,42). Although the induction of mesoderm in Xenopus animal caps by a constitutively activated FGFR-2 does not explain all of the events that lead to craniosynostosis, it does provide an excellent assay system to determine the functional consequences of the mutations associated with these syndromes. Particularly advantageous is the ability to demonstrate a dose-response relationship between the amount of RNA injected and the observed biological effects (see Fig. 2). This type of analysis may be particularly important since all of the FGFR mutations identified thus far are heterozygous, indicating that homozygous mutants may be lethal, and thus level of expression may be linked to severity of disease. While the full mechanistic details on the etiology of Crouzon syndrome remain to be established, our results demonstrate that a mutation associated with this syndrome results in constitutive activation of FGFR-2 with potentially adverse biological consequences for vertebrate development.