Fibroblast Growth Factor (FGF) Receptor 1-IIIb Is a Naturally Occurring Functional Receptor for FGFs That Is Preferentially Expressed in the Skin and the Brain*

Fibroblast growth factors (FGFs) transmit their signals through four transmembrane receptors that are designated FGFR1–4. Alternative splicing in the extracellular region of FGFR1–3 generates receptor variants with different ligand binding affinities. Thus two types of transmembrane receptors (IIIb and IIIc isoforms) have been identified for FGFR2 and FGFR3, and the existence of analogous variants has been postulated for FGFR1 based on its genomic structure. However, only a single full-length transmembrane FGFR1 variant (FGFR1-IIIc) has been identified so far. Here we describe the cloning of a full-length cDNA encoding FGFR1-IIIb from a mouse skin wound cDNA library. This receptor isoform was expressed at the highest levels in a subset of sebaceous glands of the skin and in neurons of the hippocampus and the cerebellum. FGFR1-IIIb was expressed in L6 rat skeletal muscle myoblasts and used in cross-linking and receptor binding studies. FGF-1 was found to bind the receptor with high affinity, whereas FGF-2, -10, and -7 bound with significantly lower affinities. Despite their apparently similar but low affinities, FGF-10 but not FGF-7 induced the activation of p44/42 mitogen-activated protein kinase in FGFR1-IIIb-expressing L6 myoblasts and stimulated mitogenesis in these cells, demonstrating that this new receptor variant is a functional transmembrane receptor for FGF-10.

Fibroblast growth factors (FGFs) 1 comprise a growing family of structurally conserved mitogens, which currently includes 19 different members (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). With the exception of FGF-7, which appears to act specifically on epithelial cells (11), most members of the FGF family are broad-spectrum mitogens, and several also stimulate various processes such as cell migration, angiogenesis, neurite outgrowth, and neuroprotection (12). FGFs interact with three different types of binding proteins. These include a cystein-rich receptor of as yet unknown function, which binds various types of FGFs with high affinity (13), and heparan sulfate proteoglycans, which bind FGFs with low affinities, although this interaction is essential for the binding of these factors to their high-affinity receptors (12). There are four different genes encoding high-affinity transmembrane receptors. These receptors, designated as FGF receptors 1 to 4 (FGFR1-4), are characterized by the presence of two or three immunoglobulin (Ig)-like domains in the extracellular region and a tyrosine kinase domain in the intracellular region of the receptor (14). They are responsible for FGF-mediated signal transduction (14). Additional diversity in the FGF receptor family is generated by alternative splicing. Of particular importance are splice variants that differ in the second half of the third Ig-like domain, since these variants differ in their ligand binding specificity (15)(16)(17). Thus two transmembrane receptor variants have been identified for FGFR2 that were designated FGFR2-IIIb and FGFR2-IIIc. Whereas the IIIb variant of FGFR2 binds FGF-1, -3, -7, and -10 with appreciable affinity (15, 19 -21), the IIIc variant is a high-affinity receptor for FGF-1, -2, -4, -6, and -9 (15,20). By analogy to FGFR2, IIIb and IIIc splice variants have also been identified for FGFR3 (17), whereas this type of alternative splicing does not occur with the FGFR4 transcripts (22). Interestingly, three different exons (designated IIIa, IIIb and IIIc) that might encode possible alternatives within the third Ig-like domain are present in both the human and murine FGFR1 genes, and receptors including the IIIa and the IIIc exon have been identified (23). The receptor that contains exclusively the IIIa exon is a soluble variant, which binds FGF-2 with higher affinity than FGF-1 (24). FGFR1-IIIc is a transmembrane receptor that binds FGF-1, -2, -4, -5, and -6 (20). By contrast, a full-length receptor cDNA, which includes sequences encoding the IIIb exon of FGFR1 has as yet not been isolated. However, mRNAs encoding the FGFR1-IIIb exon have been found in mice and humans using RNase protection assays and reverse transcriptase polymerase chain reaction (16,23). These mRNAs encode the complete extracellular domain of FGFR1-IIIb, but it has as yet not been possible to demonstrate the presence of sequences encoding the intracellular domain. Thus it has been unclear whether FGFR1-IIIb is a soluble or a transmembrane receptor and whether it is a functional receptor for FGFs. To gain some insight into the properties of a putative FGFR1-IIIb, we had previously used the genomic sequence of the human receptor gene to predict the coding potential of the IIIb exon, which was then incorporated in place of the IIIc exon in a FGFR1-IIIc cDNA. The extent of the IIIb exon was based on the strong sequence homology between the IIIb exons of FGFR1 and FGFR2. Expression of this chimeric receptor in rat L6 smooth muscle myoblasts, which lack endogenous FGF receptors, and subsequent binding studies demonstrated that this receptor binds FGF-1 with high affinity and FGF-2 with lower affinity (16). Subsequently, binding of FGF-3 and -10 to this receptor was demonstrated (19,25), but only an extremely weak binding of FGF-7 could be observed (20). Here we describe the isolation of an authentic full-length murine FGFR1-IIIb cDNA from a mouse skin wound cDNA library. We demonstrate that the protein encoded by this cDNA is a functional transmembrane receptor for FGF-1, -2, and -10. FGF-7 also binds the receptor but does not activate mitogen-activated protein (MAP) kinase or induce DNA synthesis.

EXPERIMENTAL PROCEDURES
RNA Isolation and RNase Protection Assay-RNA isolation was performed according to Ref. 26. RNase protection analysis was performed as described previously (27). Samples of total RNA (20 g) were used for all protection assays. As a loading control, 1 g of all RNA samples was loaded on a 1% agarose gel before hybridization and stained with ethidium bromide. All protection assays were carried out in duplicate or triplicate with different sets of RNA. Probe DNAs: a 137-base pair (bp) fragment specific for the murine FGFR1-IIIb exon (16), a 361-bp fragment corresponding to the complete third Ig-like domain of murine FGFR1-IIIc (described in Ref. 16), and a 271-bp fragment corresponding to the third Ig-like domain of murine FGFR2-IIIb (described in Ref. 27).
Isolation of a Full-length Murine FGFR1-IIIb cDNA-To isolate the full-length FGFR1-IIIb cDNA, we screened murine cDNA libraries from normal and wounded mouse dorsal skin (30,31) with the FGFR1-IIIb exon-specific probe described above using standard procedures. The longest insert was completely sequenced in both orientations. In addition, exon IIIb and 100 -200 bp upstream and downstream of this exon were sequenced in four shorter cDNAs.
Generation of L6 Myoblast Cell Lines Expressing Murine FGFR1-IIIb-The full-length FGFR1-IIIb cDNA was cloned into a mammalian expression vector in which the FGFR1 cDNA is driven by the simian virus 40 early promoter. Early passage rat L6 skeletal muscle myoblasts (ATCC CRL 1458) were subsequently co-transfected with this expression plasmid and a vector containing the neomycin resistance gene (pSV2neo) using the Lipofectin transfection system (Life Technologies, Inc.) as described by the manufacturer. Transfected cells were selected in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum, penicillin/streptomycin, and 800 g/ml geneticin (G418; Life Technologies, Inc.). Resistant colonies were isolated and analyzed by Western blot for the expression of FGFR1 using a polyclonal antiserum directed against an intracellular peptide of FGFR1 (15) and an alkaline phosphatase detection system.
Analysis of FGFR1-IIIb Glycosylation-Tunicamycin (Sigma), an inhibitor of N-linked glycosylation (dissolved in Me 2 SO) was added to the cell culture media at a concentration of 5 g/ml. The same volume of Me 2 SO was used as a control. Cells were harvested before and 3, 6, and 24 h after addition of tunicamycin and analyzed by Western blotting.
Growth Factors-Recombinant human FGF-1, -2, and -7 were purchased from Roche Molecular Biochemicals. Recombinant human FGF-10 was expressed in Chinese hamster ovary cells and purified as described previously (32).
Receptor Binding Studies-FGF receptor binding studies were performed as described (16). Briefly, FGFR-and vector-transfected L6 cells were seeded into 1-cm wells and incubated with 100,000 cpm of 125 Ilabeled FGF-1 (specific activity, 1179 Ci/mmol; Amersham Pharmacia Biotech) and increasing concentrations of nonlabeled FGF-1, -2, -7, or -10 at 0°C in binding medium (DMEM containing 0.2% bovine serum albumin, 25 mM HEPES (pH 7.1), and 15 U of heparin per ml). After 4 h, the cells were washed twice with cold phosphate-buffered saline (PBS) and twice with PBS containing 2 M NaCl. They were subsequently solubilized, and 125 I was quantified by gamma counting.
Cross-linking of 125 I-FGF-1 to FGFR1-IIIb-transfected L6 Myoblasts-Transfected L6 cells were grown to confluency in 6-cm Petri dishes, washed twice with PBS, and incubated with 750,000 cpm of radiolabeled FGF-1 in the presence or absence of unlabeled competitor (5 and 1 g/ml FGF-10; 150 and 50 ng/ml FGF-1). Binding and washing conditions were as described for binding assays. After washing, the cross-linker bis(sulfosuccinimidyl) suberate (Pierce) was added to a final concentration of 1 mM, and the cells were incubated 30 min longer at 4°C. The cross-linker was then removed, and the cells were solubilized in sample buffer and subjected to SDS-polyacrylamide gel electrophoresis. 125 I-Labeled proteins were detected by autoradiography.
MAP Kinase Activity Assay-To determine mitogen-activated kinase activity in response to FGFs, we used the p44/42 MAP kinase assay kit (New England BioLabs, Beverly, MA). Briefly, FGFR-transfected L6 cells and vector-transfected cells were grown to 70% confluency in 10-cm Petri dishes in DMEM/10% fetal calf serum. After a 20-h incubation in serum-free medium, cells were treated for 7 min with 10 ng/ml FGF-1, -2, -7, or -10, respectively. After washing with PBS on ice, they were lysed and sonicated. Lysates were centrifuged and the supernatant stored at Ϫ80°C. Lysates were immunoprecipitated using immobilized phospho-p44/42 MAP kinase (Thr 202 /Tyr 204 ) monoclonal antibody. The pellet was washed and incubated in the presence of ATP with Elk-1 fusion protein. Phosphorylated fusion protein was detected using Western blotting with an antibody directed against Phospho-Elk-1(Ser 383 ) and an ECL chemiluminescence detection system. The blot was reprobed with an antibody directed against total Elk-1 protein to prove the presence of equal amounts of Elk-1 protein in each reaction mixture.
Proliferation Assay-30,000 cells per well were seeded in 24-well dishes in DMEM/10% fetal bovine serum (FBS). After 2 days the medium was replaced with DMEM/0.5% FBS. 24 h later the medium was changed again and the cells were incubated for 16 h in DMEM/0.5% FBS containing 0.01-100 ng/ml FGF-1, -2, -7, or -10, respectively. DNA synthesis was then measured by labeling cultures with 1 Ci of [methyl- 3 H]thymidine (Amersham Pharmacia Biotech; 5 Ci/mmol)/well for 3 h. At the end of the labeling period, the medium was removed and each well was washed twice with ice-cold 5% trichloroacetic acid and incubated on ice for 15 min in 5% trichloroacetic acid. After aspiration, 750 l of 0.25 N NaOH/0.1% SDS was added per well and the plates were incubated on a rocker for 20 min. 500 l of solubilized cells were counted in a 3 H channel after neutralization by addition of 50 l of 6 N HCl.

Differential Expression of FGFR1-IIIb, FGFR1-IIIc, and FGFR2-IIIb in Adult Mouse
Tissues-In an attempt to isolate the full-length cDNA of murine FGFR1-IIIb, we first determined the optimal source for the corresponding mRNA. For this purpose, we performed RNase protection assays with RNAs from different adult mouse tissues. A riboprobe corresponding to the genomic sequence of the putative IIIb exon was used as a probe. As demonstrated in Fig. 1 (upper panel), FGFR1-IIIb was expressed at the highest levels in the dermis and the epidermis of mouse tail skin as well as in the brain. Furthermore, high levels were detected in dorsal and ventral skin of mice (data not shown). In addition to these tissues, low levels of FGFR1-IIIb mRNA were detected in the kidney, lung, skeletal muscle, heart, testis, and intestine. This expression pattern was strikingly different from that of FGFR1-IIIc, which was not expressed in the epidermis, but at equally high levels in the dermis, brain, kidney, lung, and the heart, and at lower levels in skeletal muscle, testis, liver, and intestine (Fig. 1,  middle panel). FGFR2-IIIb mRNAs were found at particularly high levels in lung and in the epidermis, and at lower levels in intestine and in the dermal compartment of mouse tail skin, where it is present in hair follicle keratinocytes (33). 2 These results demonstrate a differential expression of the two FGFR1 splice variants and of the IIIb variant of FGFR2 and, therefore, suggest that a skin or a brain library is useful for the isolation of the full-length cDNA.
FGFR1-IIIb Is Expressed At Highest Levels in Sebaceous Glands and in Neurons of the Hippocampus and the Cerebellum-To localize the FGFR1-IIIb transcripts in adult skin and brain we performed in situ hybridization with the FGFR1-IIIbspecific probe described above. As shown in Fig. 2, A and B, this type of receptor was expressed at particularly high levels in sebaceous glands. Surprisingly, not all of these glands but only certain lobes expressed this receptor isoform. Thus FGFR1-IIIb expression might reflect a specific functional status of sebaceous glands. In the brain, the highest levels of FGFR1-IIIb transcripts were detected in dentate gyrus neurons of the hippocampus (Fig. 2C) and in the granular cells of the cerebellum (Fig. 2D). As a control, parallel sections were hybridized to other FGF and FGFR probes that gave distinct and different expression patterns (data not shown). These results demonstrate a unique distribution of FGFR1-IIIb, which is strikingly different to that of other FGF receptor variants.
Isolation and Sequencing of a Full-length FGFR1-IIIb cDNA from a Mouse Wound Library-For the isolation of the fulllength FGFR1-IIIb cDNA, we first screened a murine skin cDNA library (30). One clone obtained was identified as a partial cDNA. Upon screening of a mouse cDNA library gener-ated from RNA of 1-and 5-day full thickness excisional wounds (31), additional cDNAs were isolated. Sequencing of the 5Ј-and 3Ј-ends of the inserts revealed that 50% of the cDNAs encoded FGFR1, whereas the others encoded FGFR2. The longest cDNA corresponding to FGFR1 included the published 5Ј-and 3Ј-ends of murine FGFR1 (34). This cDNA was sequenced and shown to include a complete open reading frame, which should encode a full-length transmembrane receptor with two extracellular Iglike domains. The DNA sequence was identical to that published for murine FGFR1-IIIc (34), except for differences in the region encoding the second half of the third Ig-like domain ( Fig.  3; GenBank™ accession number AF176552). This part of the clone was found to be derived from the predicted IIIb exon region, and its sequence was verified in four independent partial cDNAs derived from either skin or wound skin cDNA libraries (data not shown). As shown in Fig. 3, the region of sequence that differs from that of FGFR1-IIIc encompasses the region used as a probe for both the library screen and RNase protection assays and extends further at the 3Ј-end. Therefore, the third Ig-like domain of this receptor is composed of a 5Ј region common to both FGFR1-IIIc and FGFR1-IIIb and a 3Ј region that is different in both variants. Interestingly, the 3Ј-end of the unique IIIb exon encodes one additional amino acid compared with the artificially constructed FGFR1-IIIb (V) and one amino acid that is different in these receptors (K versus A). This means that the IIIb and IIIc exons share the encoded sequence WLTV, but then the IIIb exon ends with six extra amino acids while the IIIc encodes only two. A comparison of the full-length cDNA with the genomic sequence of murine FGFR1 that follows the IIIb exon (16) revealed the use of nucleotides 1195-1202 of this sequence (AG2GTAATG) as a splice site.
FGFR1-IIIb Is a 100-to 130-kDa Glycosylated Protein-To determine whether the new cDNA encodes a full-length transmembrane receptor, we inserted the cDNA into a eukaryotic expression vector in which expression is driven by the SV40 promoter. This construct was subsequently used for the stable transfection of rat L6 skeletal muscle myoblasts, which lack endogenous FGF receptors. By Western blot analysis using total cell lysates with a polyclonal antiserum directed against the intracellular juxtamembrane region of FGFR1, several cell lines were identified that expressed receptor proteins of 100 -130 kDa (Fig. 4A) (16). These proteins were not detected in vector-transfected control cells, indicating that they represent FGFR1-specific proteins. The diffuse character of these bands might result from glycosylation of FGFR proteins. To address 2 S. Werner, unpublished data. this question we incubated these cells in the presence of tunicamycin, an inhibitor of N-linked glycosylation. This treatment resulted in a time-dependent disappearance of the 100-to 130-kDa bands. However, a protein of approximately 85 kDa appeared upon tunicamycin treatment that corresponds to the expected size of the nonglycosylated receptor. This result demonstrates that FGFR1-IIIb is a glycosylated transmembrane protein and not a soluble receptor.
The Endogenous Murine FGFR1-IIIb Is a Functional FGFbinding Protein-To determine if FGFR1-IIIb is able to bind FGF, we incubated the transfected L6 cells with 125 I-FGF-1 in the presence or absence of an excess of FGF-1 or FGF-10. Radiolabeled FGF-1 was chosen because it binds with high affinity to all known FGF receptors (20). A complex of approximately 140 -160 kDa was formed when 125 I-FGF-1 was crosslinked to cells expressing murine FGFR1-IIIb (Fig. 4B). This size correlates with the expected size of a complex between FGF-1 (16 kDa) and FGFR1-IIIb. This band completely disappeared in the presence of an excess of unlabeled FGF-1, whereas only a reduction in the intensity of this band was found in the presence of an excess of FGF-10 (Fig. 4B). This result indicates that FGFR1-IIIb binds FGF-1 with higher affinity than it does FGF-10.
The Endogenous Murine FGFR1-IIIb Binds FGF-1 with High Affinity and FGF-2, -10, and -7 with Lower Affinity-The ligand binding specificity of FGFR1-IIIb was investigated in comparison with the artificially constructed human FGFR1-IIIb (16), human FGFR1-IIIc (35), and human FGFR2-IIIb (36). For this purpose, L6 myoblasts stably transfected with cDNAs encoding the different receptor isoforms were incubated with 125 I-labeled FGF-1 alone or in the presence of increasing concentrations of competitor ligands (FGF-1, -2, -7, and -10). As shown in Fig. 5, FGF-1 and -2 competed well for the binding of 125 I-FGF-1 to FGFR1-IIIc, whereas FGF-7 and -10 (37) did not bind to this receptor. By contrast, FGF-1 and to a lesser extent FGF-10 and -7 competed for the binding of 125 I-FGF-1 to FGFR2-IIIb-transfected cells, whereas only a weak binding of FGF-2 to this receptor was observed. Consistent with our previous data, the artificially generated human FGFR1-IIIb bound FGF-1 with high affinity and FGF-2 with lower affinity. In addition, FGF-7 and -10 competed with 125 I-labeled FGF-1 to bind this receptor, although only at high concentrations. More importantly, L6 cells transfected with the naturally occurring FGFR1-IIIb variant had the same binding specificities as the artificially constructed FGFR1-IIIb. This result was reproduced with three different cell lines expressing the authentic FGFR1-IIIb. These findings demonstrate that the naturally occurring FGFR1-IIIb is a FGF-binding protein that binds FGF-1 preferentially, to a lesser extent FGF-2, and with low affinity FGF-7 and -10. It was surprising that FGF-7 and -10 appeared to compete relatively poorly with 125 I-labeled FGF-1 for binding to FGFR2-IIIb, because these ligands activate the receptor at much lower concentrations in mitogenicity assays. This apparent anomaly probably reflects the different binding characteristics of FGF-7 and -1 for FGFR2-IIIb (see "Discussion"). Therefore, to further clarify the capacity of the different ligands to activate FGFR1-IIIb, we tested the ability of the receptor to activate MAP kinase and to induce DNA synthesis following ligand binding.
FGFR1-IIIb Is a Signaling Receptor-To determine the biological significance of the apparently weak binding of FGF-7 and -10 to FGFR1-IIIb, the ability of this receptor to activate p44/42 MAP kinase following stimulation with these ligands was assessed. This assay was chosen because FGF-2 is a strong stimulator of MAP kinase activity in this cell type (38). Treatment of serum-starved FGFR1-IIIb transfected L6 myoblasts with FGF-1, -2, and -10 resulted in a clear phosphorylation of  4. FGFR1-IIIb encodes a 105-to 130-kDa glycosylated protein that binds FGF. A, rat L6 skeletal muscle myoblasts were stably transfected with a FGFR1-IIIb expression construct and analyzed for the presence of this receptor by Western blot analysis. Tunicamycin treatment of the transfected cells was performed to determine N-linked glycosylation. L6 neo, vector-transfected cells; L6 FGFR1-IIIb, FGFR1-IIIb-expressing L6 cells; tunicamycin, cells treated for the indicated times with 5 g/ml tunicamycin in Me 2 SO; Me 2 SO, cells treated with Me 2 SO alone. B, cross-linking of 125 I-FGF-1-to FGFR1-IIIb-transfected L6 cells. Cells were grown to confluency in 6-cm Petri dishes and incubated with 125 I-FGF-1 in the absence or presence of an excess of unlabeled FGF-1 or -10 as indicated. After cross-linking with bis(succinimidyl) suberate, the cells were solubilized in sample buffer and subjected to SDS-polyacrylamide gel electrophoresis. Radioactively labeled proteins were visualized using autoradiography. the MAP kinase substrate Elk-1 by phospho-p44/42 MAP kinase immunoprecipitates, demonstrating that the receptor is functional (Fig. 6). Despite the strong sequence similarities between FGF-7 and -10, Elk-1 phosphorylation was not observed in response to FGF-7 (Fig. 6). By contrast, FGF-1 and -7 induced a strong response in FGFR2-IIIb-transfected L6 cells, and a weaker phosphorylation of Elk-1 was seen with FGF-10 and -2. No MAP kinase activation was observed in vectortransfected L6 cells upon addition of the different FGFs. These findings clearly demonstrate differences in the signal transduction capacity of FGFR1-IIIb and FGFR2-IIIb in response to FGF-7 and -10. FGF-1, -2, and -10 -To further determine the ligands that elicit biological effects via FGFR1-IIIb, we treated FGFR1-IIIbtransfected L6 cells as well as vector-transfected control cells with different concentrations of FGF-1, -2, -7, and -10. [ 3 H]Thymidine incorporation for 3 h was then measured in triplicate wells. 10% FBS, which was used as a positive control, caused a 5-to 6-fold stimulation of 3 H-thymidine incorporation compared with medium without additional FBS or FGF (Fig. 7). As expected from the results of the MAP kinase assays, FGF-1 was the most potent stimulator of DNA synthesis, and a 3-fold stimulation of [ 3 H]thymidine incorporation was observed upon addition of 1 ng/ml of this factor. Higher concentrations were less effective, possibly as a result of receptor down-regulation. A similar dose-response curve was obtained with FGF-2, although this growth factor was slightly less potent than FGF-1. Interestingly, addition of FGF-10 also stimulated DNA synthesis in FGFR1-IIIb-transfected L6 cells, but 10-to 100-fold higher concentrations were required compared with FGF-1 and -2. In contrast to FGF-10, no stimulation of DNA synthesis was observed with FGF-7. These results were reproduced in an independent experiment. Taken together, these findings as well as the results obtained with the MAP kinase assays demonstrate that FGF-1, -2, and -10, but not FGF-7, elicit biological responses via FGFR1-IIIb. DISCUSSION Fibroblast growth factors comprise a rapidly growing family of growth and differentiation factors that are involved in various physiological and pathological processes. A series of studies have demonstrated that different members of the FGF family transduce a signal to the cell nucleus by binding to, and activating, members of a transmembrane receptor family. There are seven prototype transmembrane receptors encoded by four different receptor genes. This additional receptor complexity is achieved by alternative splicing in the carboxyl-terminal half of the third Ig-like domain, which generates receptor variants with different ligand binding affinities. Such variants have been identified for FGFR2 and FGFR3. In addition, the sequence of the FGFR1 gene predicted the existence of different splice variants for FGFR1 (23). Whereas a transmembrane FGFR1-IIIc variant and a soluble FGFR1-IIIa variant have been identified and cloned (35), a full-length FGFR1-IIIb variant had yet to be identified.

FGFR1-IIIb Induces Mitogenesis of L6 Cells in Response to
In this study we have identified the major sites of expression of FGFR1-IIIb, which in turn facilitated the cloning of a fulllength cDNA encoding a transmembrane form of the receptor. RNase protection analyses demonstrated a low level expression in a wide variety of adult mouse tissues. However, high expression was found in skin and brain, indicating the existence of specific splicing factors in these tissues that recognize the relatively weak FGFR1-IIIb splice site. In the skin, high ex- pression was found to be localized to certain lobes of a subset of sebaceous glands, suggesting a rather specific regulation of expression and splicing requirements, possibly reflecting the functional status of these glands. In the brain, FGFR1-IIIb is particularly abundant in hippocampal dentate gyrus and cerebellar granule neurons. This is noteworthy because FGFs 1 and 2 are known to stimulate neurite outgrowth from both populations of neurons in vitro (39). The detection of FGFR1-IIIb mRNAs in various types of neurons demonstrates that expression of this receptor variant is not restricted to epithelial cells. By contrast, the corresponding splice variant of FGFR2 (FGFR2-IIIb) seems to be specific for various types of epithelial cells in vitro and in mature mouse tissues (40,41).
The particularly high expression of FGFR1-IIIb in the skin and in the brain suggests specific functions of this receptor variant in these tissues. Although recently generated mice, carrying an in-frame stop codon in the IIIb exon, were viable and fertile (42), a detailed analysis of these tissues under normal conditions and in stress situations might reveal unique roles of this receptor variant.
The isolated cDNA encoding a full-length transmembrane receptor was found to be identical to FGFR1-IIIc except for the nucleotides encoding the second half of the third Ig-like domain. The alternative exon sequence, which specifies murine FGFR1-IIIb, is slightly larger than its FGFR1-IIIc counterpart, because it encodes four additional amino acids just downstream of the third Ig-like domain. These additional amino acids have only partially been predicted when a human FGFR1-IIIb cDNA was artificially constructed (16). Therefore, the artificially constructed receptor might have had a different ligand binding specificity than that of the endogenous murine FGFR1-IIIb. This possibility was tested in a receptor binding analysis, which revealed no differences in the binding specificities between the endogenous and the artificially created FGFR1-IIIb. Hence, the differences in the carboxyl terminus of the third Ig-like domain do not overtly influence binding specificity.
Consistent with previous data (16), FGFR1-IIIb bound FGF-1 with high affinity, FGF-2 with lower affinity, and FGF-7 and -10 weakly. To determine whether the apparently weak affinity of FGF-7 and -10 is of biological significance, the effectiveness of FGFR1-IIIb to activate MAP kinase was tested using these FGFs. The results showed that phospho-p44/42 MAP kinase immunoprecipitates from FGF-10-treated but not from FGF-7-treated cells were able to induce the phosphorylation of a MAP kinase substrate. Consistent with these results, [ 3 H]thymidine incorporation studies revealed a mitogenic effect of FGF-10 but not of FGF-7 for FGFR1-IIIb-transfected L6 cells. In contrast to FGFR1-IIIb, FGFR2-IIIb was strongly activated by FGF-7 and to a lesser extent by FGF-10, demonstrating a clear difference in the signaling capacities of these two receptors. FGF-7 and -10 are expressed at high levels in brain and skin, but it is unclear how the different ligand/receptor combinations function in vivo.
The data also show that FGFR1-IIIb and FGFR2-IIIb responded more strongly in the MAP kinase and mitogenesis assays to FGF-10 and -7, respectively, than would have been predicted from the results of the receptor binding competition assays. This is most easily explained by differences in binding site preference (16,43,44). For example, Cheon et al. (43) showed that the second Ig-like domain of FGFR2-IIIb bound FGF-1 strongly but did not bind FGF-7, whereas the third Ig-like domain bound FGF-7. Similarly, chimeras composed of FGFR2-IIIb and FGFR3 demonstrated that the third Ig-like domain of FGFR2-IIIb was sufficient to confer binding of FGF-7 to FGFR3 (16), although this was not true for FGFR1/FGFR2 chimeras (44). Recently, Chellaiah et al. (45) provided evidence for the existence of two binding sites on a single FGF receptor, whereby FGF-1 can bind to both sites. Therefore, it is likely that FGF-7 and -10 would compete poorly for FGF-1 if it preferentially associates with the second Ig-like domain of the receptor or if it associates with two independent domains of which only one binds FGF-7 and -10. Nevertheless, the MAP kinase results as well as the mitogenesis assay clearly show that FGFR1-IIIb is activated by FGF-10. Vector-transfected L6 myoblasts as well as L6 cells transfected with murine FGFR1-IIIb or human FGFR2-IIIb were rendered quiescent and subsequently stimulated for 7 min with 10 ng/ml recombinant FGF-1, -2, -7, or -10 as indicated. Lysates of nonstimulated (control) and stimulated cells were immunoprecipitated using immobilized phospho-p44/42 MAP kinase (Thr 202 /Tyr 204 ) monoclonal antibody. Precipitates were incubated in the presence of ATP with Elk-1 fusion protein. Phosphorylated fusion protein was detected by Western blotting using an antibody directed against Phospho-Elk-1 (Ser 383 ) and by using the ECL chemiluminescence detection system (Amersham Pharmacia Biotech). The blot was reprobed with an antibody against total Elk-1 protein to determine the presence of equal Elk-1 levels in each reaction mixture. The major receptor for FGF-7 and -10 appears to be FGFR2-IIIb (15,21), which is expressed on epithelial cells. Therefore, these ligands have been regarded as being important for epithelial cell proliferation/differentiation. However, FGFR1-IIIb is expressed by various types of fibroblasts (16, 23, and unpublished data). This finding in conjunction with the ability of FGF-10 but not FGF-7 to activate MAP kinase and to stimulate mitogenesis of FGFR1-IIIb-transfected L6 cells could explain the mitogenic effect of high concentrations of FGF-10 on NIH-3T3 fibroblasts (21). Moreover, as NIH-3T3 cells do not express FGFR2-IIIb, FGF-7 is not mitogenic for these cells (21). These findings, together with the results presented here, suggest that FGF-7 acts specifically on epithelial cells, whereas FGF-10 seems to have a broader cell type specificity. This is further supported by the demonstration that FGF-10 is mitogenic for primary rat preadipocytes (46). The mitogenicity of FGF-10 for mesenchymal cells in vivo is unclear, because fairly high concentrations of the ligand are required for this effect. However, the stimulatory effect of recombinant FGF-10 on granulation tissue formation and collagen deposition during cutaneous wound healing (32,47) suggests that mesenchymal cells are directly affected by exogenously added growth factor and perhaps also by endogenous FGF-10 under conditions where this factor is highly expressed.