Fibroblast Growth Factor-10

Fibroblast growth factor (FGF)-10, a homologue of FGF-7, is expressed significantly in normal rat prostate tissue, well differentiated rat prostate tumors with an epithelial and stromal compartment and only in derived prostate stromal cells in culture. Similar to FGF-7, recombinant rat FGF-10 was a specific mitogen for prostate epithelial cells. In contrast to FGF-7 which is widely expressed among stromal cells in tissues, the expression of FGF-10 correlated with the presence of stromal cells of muscle origin. Radioreceptor binding assays and covalent cross-linking analysis revealed that FGF-10 binds with an affinity equal to FGF-7 to resident epithelial cell receptor, FGFR2IIIb, but unlike FGF-7 also binds the IIIb splice variant of FGFR1. Analysis of mRNA expression by RNase protection revealed that, similar to FGF-7, the expression of FGF-10 was responsive to androgen in stromal cells from normal prostate and non-malignant differentiated tumors. Although FGF-10 cDNA exhibits a signal sequence for secretion, cultured stromal cells exhibit strictly a cell-associated FGF-10 antigen that correlates with an alternately translated intracellular isoform. FGF-10 requires 1.4 times higher NaCl for elution from immobilized heparin than does FGF-7 and binds to four times the number of sites on the pericellular matrix of epithelial cells. The results show that prostate stromal cell-derived FGF-10, like FGF-7, exhibits the properties of an andromedin which may indirectly mediate control of epithelial cell growth and function by androgen. Although FGF-10 and FGF-7 bind and activate the same resident epithelial cell receptor (FGFR2IIIb), differences in cell type of origin, compartmentation by alternate translation, the affinity for FGFR1IIIb, and access to FGFR by differential interaction with pericellular matrix heparan sulfate suggest they may play both independent and compensatory roles in prostate homeostasis.

Fibroblast growth factor (FGF)-10, a homologue of FGF-7, is expressed significantly in normal rat prostate tissue, well differentiated rat prostate tumors with an epithelial and stromal compartment and only in derived prostate stromal cells in culture. Similar to FGF-7, recombinant rat FGF-10 was a specific mitogen for prostate epithelial cells. In contrast to FGF-7 which is widely expressed among stromal cells in tissues, the expression of FGF-10 correlated with the presence of stromal cells of muscle origin. Radioreceptor binding assays and covalent cross-linking analysis revealed that FGF-10 binds with an affinity equal to FGF-7 to resident epithelial cell receptor, FGFR2IIIb, but unlike FGF-7 also binds the IIIb splice variant of FGFR1. Analysis of mRNA expression by RNase protection revealed that, similar to FGF-7, the expression of FGF-10 was responsive to androgen in stromal cells from normal prostate and nonmalignant differentiated tumors. Although FGF-10 cDNA exhibits a signal sequence for secretion, cultured stromal cells exhibit strictly a cell-associated FGF-10 antigen that correlates with an alternately translated intracellular isoform. FGF-10 requires 1.4 times higher NaCl for elution from immobilized heparin than does FGF-7 and binds to four times the number of sites on the pericellular matrix of epithelial cells. The results show that prostate stromal cell-derived FGF-10, like FGF-7, exhibits the properties of an andromedin which may indirectly mediate control of epithelial cell growth and function by androgen. Although FGF-10 and FGF-7 bind and activate the same resident epithelial cell receptor (FGFR2IIIb), differences in cell type of origin, compartmentation by alternate translation, the affinity for FGFR1IIIb, and access to FGFR by differential interaction with pericellular matrix heparan sulfate suggest they may play both independent and compensatory roles in prostate homeostasis.
Members of the fibroblast growth factor (FGF) 1 family which is comprised of 20 or more homologous polypeptides and four tyrosine kinase receptor (FGFR) genes play a role in prostate homeostasis and tumor progression (1)(2)(3). FGF-1, FGF-2, FGF-3, FGF-5, and FGF-7 and splice variants of FGFR1 and FGFR2 have been implicated. FGF-1, which acts on FGFR isoforms that are both in epithelium and stroma, is expressed in both prostate compartments and expression increases with malignant progression (4). FGF-2, FGF-3, and FGF-5 are expressed ectopically in malignant epithelial cells (3). FGF-7, whose expression is limited to the stromal cells, is the androgen-responsive signal of a directionally specific paracrine communication system from stroma to epithelium in normal prostate and non-malignant differentiated tumors with a stromal compartment (3). Reception of the stromal signal appears to be mediated by FGFR2IIIb, a mutually exclusive alternate splice variant, which recognizes FGF-7 with high affinity (2,3,5). In progressively malignant prostate tumors, a phenotypic splice switch from expression of the FGFR2IIIb to the FGFR2IIIc isoform occurs with eventual loss of FGFR2 gene expression altogether in some malignant clones (3,6). Restoration of the FGFR2 kinase to malignant tumor cells restores limitations on growth rates by increased differentiation in response to stroma cells expressing FGF-7 (6). Therefore, it is thought that in the epithelial cell context, FGFR2 mediates a homeostatic balance through both positive and negative effects on population growth rates and differentiation. This is in contrast to FGFR1, which is a resident FGFR in stromal cells (6). In this study, we show that, similar to FGF-7, FGF-10 is expressed only in the stroma of normal prostate and differentiated prostate tumors, and that expression is sensitive to androgen. FGF-10 is a mitogen for isolated prostate epithelial cells, but not for stromal cells which express it, and binds specifically to the resident epithelial cell receptor, FGFR2IIIb. FGF-10 differs from FGF-7 by its more restricted expression pattern, the ability to bind FGFR1IIIb in addition to FGFR2IIIb, and in the interaction with heparin and pericellular matrix heparan sulfate. The two factors exhibit potential for both compensatory and specific activities in the prostate.

EXPERIMENTAL PROCEDURES
Construction, Expression, and Purification of Recombinant FGF-10 -Total RNA was isolated from Dunning R3327PAP rat prostate stromal cells by guanidinium thiocyanate/phenol/chloroform extraction and first strand cDNA was prepared using SuperScript II RNase H Reverse Transcriptase (Life Technologies) with random primers. The polymerase chain reaction (PCR) was carried out for 38 cycles at 94°C for 1 min, 55°C for 2 min, and 72°C for 2 min. Full-length FGF-10 cDNA was generated from the template with sense primer p1001 5Ј-CGGGATCCATGTGGAAATGGATACTG-3Ј which included a BamHI restriction site (underlined) and antisense primer p1002 5Ј-GGGAAT-* This work was supported by United States Public Health Service Grants DK40739 and DK35310 from the NIDDK, National Institutes of Health, and NCI, National Institutes of Health, Grant CA59971. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  TCCTATGAGTGGAC-CACCAT-3Ј containing an EcoRI site (underlined). The PCR product was digested by BamHI and EcoRI and then fractionated on a 1.5% agarose gel, recovered by electroelution and cloned into pBluescript II SK vector (Stratagene, La Jolla, CA) at BamHI and EcoRI sites. The cDNA was verified by nucleotide sequencing. After digestion with HinPI and EcoRV and treatment with Klenow enzyme, the cDNA was then cloned in-frame into the pGEX-2T expression vector at the SmaI site and expressed in BL21 (DE3) cells (Novagen) resulting in an expression product fused to glutathione S-transferase at the C terminus to N-terminal amino acid residue alanine 69 of FGF-10. Homogenous FGF-10 with the N terminus at residue serine 75 was generated and recovered by a novel and general method for production of FGF fused at the N terminus with glutathione S-transferase and generation of the most active FGF isoform by modification with serine proteases while immobilized on heparin-agarose (7). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and amino acid sequence verified the homogeneity of FGF-10. FGF-7 beginning at Ser-54 was produced by the same method (7). FGF-1 was from Asn-21 and FGF-2 from Ala-1.
Analysis of Expression of FGF-7 and FGF-10 mRNA by Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)-FGF-10 mRNA was analyzed in the PCR using oligonucleotide primers p1001 and p1002 which flank the coding sequence as described above. Since mouse FGF-10 C-terminal cDNA sequence is different from that of rat, another pair of primers, sense p1001 and antisense p1004 5Ј-GCCGTTGTGCT-GCCAGTT-3Ј was used to analyze mouse FGF-10 mRNA expression. Sense primer p701 5Ј-CCGCCCG-GGATGGCTTGCAATGACATGA-3Ј and antisense primer p702 5Ј-CCAATTCTTCTCTGC-ATGCTTCTT-3Ј which flank a 367-base pair coding sequence of rat FGF-7 were used for analysis of FGF-7 mRNA expression. Total RNA (4 g) from tissues and cells indicated in the text was converted to cDNA. FGF-10 was analyzed using the same conditions described above for cloning the cDNA. Optimized conditions determined by experiment for analysis of FGF-7 was 35-40 cycles at 94°C for 1 min, 57°C for 2 min, and 72°C for 2 min. The RT-PCR products were analyzed in 1.5% agarose gel. Purified cDNA templates for FGF-7 and FGF-10 and cDNA from total RNA of lung tissue were used for positive controls. Analysis of ␤-actin was used as an internal mRNA level control.
Analysis of mRNA Expression by Ribonuclease (RNase) Protection-A fragment spanning base pairs 49 to 207 between NciI and BstUI restriction sites in rat FGF-10 cDNA was treated with Klenow enzyme and subcloned into pBluescript SK vector at EcoRV site. The resulting plasmid template was linearized with EcoRI and the antisense RNA probes were transcribed using T7 RNA polymerase and labeled to a specific activity of about 10 8 cpm/mg of linearized cDNA template using the Ambion MAXIscript kit. Rat FGF-7 mRNA was determined by protection of a 205-base antisense BamHI-KpnI fragment from base pairs 23 to 224 as described previously (3,8). About 20 g of total RNA from tissues or cells was hybridized with 1 ϫ 10 5 cpm of labeled antisense RNA probe. Hybrid duplexes were digested with RNase A/T1 for 30 min at 37°C. Protection fragments were analyzed on 5% polyacrylamide gels containing 8 M urea followed by autoradiography. Yeast tRNA was used as a control for nonspecific hybridization. All analyses were performed using the HybSpeed ribonuclease protection kit (Ambion). To quantitate the relative expression of FGF-7 and FGF-10, 1 ϫ 10 5 cpm each of FGF-7, FGF-10, and ␤-actin probes (2) were added in the same RNase protection reaction. All three protection probes were carefully standardized to the same specific activity and the relative expression was quantitated by densitometry of protected bands. ␤-Actin was used as the standard to normalize total RNA loads across lanes from the different cell and tissue sources.
Expression of FGF-10 Protein in Cultured Dunning Tumor Stromal Cells-DT-S of DT-E cells were cultured in 1:1 RPMI medium 1640 and Dulbecco's modified Eagle's medium (RD medium) supplemented with 2% fetal bovine serum in 40 ϫ T75 cell culture flasks until subconfluent, and the cells were then placed in serum-free RD medium. After culture overnight, the medium was collected, and the cells were exposed to 1.5 M sodium chloride in 1% Triton. The medium (300 ml) from about 1.5 ϫ 10 8 cells was concentrated 100 times to 12 mg/ml total protein and the cell extract was concentrated 30 times from 100 ml of cell lysate to 67 mg/ml of total protein. The extracts were treated with rabbit anti-rat FGF-10 antibody and protein A beads followed by analysis by SDS-PAGE and immunoblotting. Homogenous recombinant rat FGF-10 (75-214) was prepared from bacteria as described (7). This pure antigen was used to immunize rabbits (Cocalico Biologicals, Inc.) which produced antiserum that reached a titer of 1:1600 in immunoblot analysis.
Expression of Recombinant FGFR Isoforms in Insect Cells-Expression of all isoforms except FGFR1IIIb and FGFR3IIIb in baculovirus-infected insect cells have been described previously (9). The human FGFR1IIIb and the murine FGFR3IIIb cDNA were kindly provided by Dr. David M. Ornitz (5,10). FGFR1IIIb from pBluescript KS(ϩ) vector (Stratagene, La Jolla, CA) was digested with ApaI followed by Klenow enzyme treatment, then digested with BamHI. The resulting 2.3-kilobase fragment was cloned into the BamHI and SmaI sites of baculovirus transfer vector pVL1393. FGFR3IIIb from MomFR3SV vector was first digested with HindIII and XhoI, ligated to pBluescript-SK vector, then digested with BamHI and SalI or digested with KpnI (treated with Klenow enzyme) and SalI, respectively, the two resulting fragments were ligated into pBluescript-SK vector at BamHI and SmaI sites. Finally, this construction was digested with EcoRV and ligated to pVL1392 vector at the SmaI site; the sense clone was confirmed by restriction enzyme digestion. The expression vectors bearing FGFR1IIIb and FGFR3IIIb were then co-infected with BaculoGold viral DNA into Spodoptera frugiperda (Sf9) insect cells, respectively, using the Lipofectin method (9).

Expression of FGF-10 in Prostate
Stromal Cells-A screen of 18 FGF mRNAs using the RT-PCR with multiple sets of paired primers revealed the presence of FGF-10 mRNA in normal rat prostate (NP) and the transplantable non-malignant well differentiated Dunning R3327PAP tumor (DT) (Fig. 1A) in addition to adult lung tissue, which was used as a positive standard (12). The FGF-10 signal parallels that of FGF-7 in the normal prostate tissue (NP), both primary cultures and lines of stromal cells from the DT tissue (DT-S), non-malignant, differentiated tumors (SE) reconstituted from isolated DT epithelial (DT-E) and stromal (DT-S) cells (2,3). Similar to FGF-7, the FGF-10 mRNA could not be detected in isolated epithelial cells from the DT tumors (DT-E). As reported previously, FGF-7 can be detected in E tumors derived from isolated DT-E cells as they progress to malignancy, malignant Dunning R3327AT3 tumors (AT3) and derived cells (AT3-C) (2) (Fig. 1A). However, the FGF-10 transcript was undetectable in these samples when analyzed under the same conditions.
Since expression of FGF-10 mRNA was limited to the stromal cells of normal prostate and differentiated prostate tumors, we compared expression of the two mRNAs in lung, skin, and muscle and mesenchymal cells derived from them in which FGF-7 is expressed. Although FGF-10 is expressed in adult lung tissue (12), FGF-10 could not be detected in isolated lung fibroblasts from fetal tissue relative to FGF-7 (Fig. 1B). Compared with FGF-7, the FGF-10 signal was weak in mouse 3T3 cells, which are also thought to be fibroblast-like cells. Both FGF-7 and FGF-10 were undetectable in human umbilical vein endothelial cells. The FGF-7 mRNA was barely detectable in subcutaneous skin samples taken from tumor implantation sites in rat hosts and FGF-10 was undetectable. Relative to FGF-7, FGF-10 mRNA was undetectable in normal human fibroblasts isolated from neonatal foreskin. In contrast to the fibroblast-like cells, the embryonic rat skeletal muscle myoblast cell line L6, and a smooth muscle cell line from normal rat prostate (PS-1) (13), exhibited the strongest positive signals of FGF-10 relative to FGF-7 mRNA. In contrast to skin, both rat skeletal muscle tissue and small intestine tissue, which are rich in smooth muscle cells, exhibited FGF-10 signals equal to or greater than FGF-7. Both FGF-7 and FGF-10 were absent in mouse keratinocytes, human hepatoblastoma cells (HepG2), and the human prostate tumor epithelial cell line (PC3). These results show that, similar to FGF-7, the expression of FGF-10 is associated with the stromal or mesenchymal compartment of prostate and a number of other tissues. However, the expression of FGF-10 is much more restricted than that of FGF-7 and may partition with muscle cell lineages relative to the general stromal or mesenchymal cell population in the adult and derived tumors.
To confirm that the RT-PCR results reflected authentic and more than trace levels of FGF-7 and FGF-10 mRNAs, the mRNA levels for the two factors in prostate tissues and derived cells were further examined by ribonuclease protection (RPA) ( Fig. 2 and Table I). The results generally confirmed the RT-PCR analysis. The more reliable quantitative analysis suggested that the ratio of expression of FGF-7 to FGF-10 mRNA in NP tissue was about 2.5, but that levels of FGF-10 increased near 4-fold in the DT tumors and derived DT-S cells resulting in an expression ratio of near 1:1 for the two homologues. As tumors and cells progress to malignancy, expression of FGF-10 is lost while FGF-7 remains detectable, although at reduced levels ( Fig. 2A and Table I). Although an FGF-7 signal could be detected by PCR in tissue at the subcutaneous site of injection (Fig. 1B, Skin) of the transplantable tumors in control hosts, expression of both FGF-7 and FGF-10 was below the limits  of detection by the ribonuclease protection method ( Fig. 2A, Table I).
Expression of FGF-10 in Prostate Stromal Cells Is Androgenresponsive-To determine whether the expression of FGF-10 mRNA was responsive to androgen as previously reported for FGF-7, primary cultured stromal cells from both normal prostate and the androgen-responsive Dunning R3327PAP tumor were isolated and analyzed as described previously (2,3). Cultures were maintained in serum-free medium in the absence or presence of 10 Ϫ8 M dihydrotestosterone for 48 h and total RNA was isolated and assayed by nuclease protection (Fig. 2B). Expression of FGF-10 mRNA in primary cultures of stromal cells from normal rat prostate (NP-S) was completely dependent on androgen. Densitometric analysis of protected autoradiographic bands indicated that dihydrotestosterone stimulated the expression of FGF-10 mRNA by about 4-fold in both primary (DT-SP) and serially cultured lines (DT-SL) of stromal cells from DT tumors. As reported previously for FGF-7, tumorderived stromal cells were characterized by an elevated level of baseline androgen-independent expression of FGF-10, which rises, with extent of serial culture. No effect on the ␤-actin control mRNA could be detected in this time frame (14).
To determine whether FGF-10 mRNA gave rise to FGF-10 secretory product, the medium of the cultured DT tumor stromal cells (DT-S) was harvested and analyzed by immunoprecipitation and immunoblot with rabbit polyclonal antiserum prepared against recombinant FGF-10 from bacteria. FGF-10 antigen from medium exposed overnight to 1.5 ϫ 10 8 stromal cells from the DT tumors was below the limits of detection. Subsequent extraction and analysis of the cell monolayers with 1.5 M NaCl and detergent revealed a single major band at 20.2 kDa estimated from electrophoretic standards that could not be detected in equivalent amounts of the DT epithelial cells (Fig.  2C). This species is smaller than the bacterial-derived recombinant FGF-10 (37-214) band at apparent mass 21.4 kDa (predicted mass ϭ 19.8 kDa), which was engineered to begin at Ala-37. A shorter bacterial recombinant isoform, FGF-10 (75-214), at apparent mass 15.5 kDa (predicted ϭ 15.4 kDa), which started at Ser-75 (7) which arises by proteolysis of heparinbound precursor is also shown.

FGF-10 Supports Growth of and Binds to Receptors on Specifically the Prostate Epithelial Cells-Recombinant rat FGF-10
was generated by a novel method for production and recovery of FGF polypeptides in bacteria developed in our laboratory (7). Fig. 3 shows that FGF-10 stimulates DNA synthesis at 50% of maximum at 1 ng/ml in non-malignant rat prostate tumor epithelial cells (DT-E) which was slightly less than the potency of FGF-7. Neither FGF-10 nor FGF-7 support DNA synthesis of stromal cells (DT-S) from the same tumors in which both factors are expressed.
Competition binding assays using radiolabeled FGF-1 which stimulated and bound to receptors on both DT-E and DT-S cells indicated that FGF-7 and FGF-10 compete to near equal extent with FGF-1 bound to the DT-E epithelial cells, but both failed to compete with FGF-1 bound to stromal DT-S cells (Fig. 4). To ensure that the mitogenic activity and competition of FGF-10 with the binding of FGF-1 reflected authentic binding to DT-E cell surface receptors, FGF-10 was radiolabeled, covalently cross-linked with the bifunctional cross-linker, DSS, and the  resultant radiolabeled complexes analyzed by SDS-PAGE and autoradiography (Fig. 5). Radiolabeled FGF-1, FGF-7, and FGF-10 yielded expected bands at 140 and 75 kDa from DT-E cells which represent the native FGFR and a truncated product resulting from proteolysis (15). Neither FGF-7 nor FGF-10 cross-linked to FGFR on the stromal cells (DT-S), even at 5 times higher radiolabeled ligand and after long exposure of the analytical gels.
FGF-10 Binds to Complexes of Heparin and Recombinant FGFR2IIIb and FGFR1IIIb-FGFR1IIIb and IIIc, FGFR2IIIb and IIIc, FGFR3IIIb and IIIc, and FGFR4 were expressed on the surface of baculoviral-infected insect cells and the ability of FGF-10 to compete with the binding of radiolabeled FGF-1 which binds to all isoforms equally in the presence of heparin. FGF-10 competed effectively with 125 I-FGF-1 bound to Sf9 insect cells expressing recombinant FGFR1IIIb and FGFR2IIIb with an ID 50 (dose required to displace 50% of radiolabeled FGF-1) of 33 and 1.3 ng, respectively (Fig. 6). No significant competition was observed with FGF-1 bound to cells expressing FGFR1IIIc, FGFR2IIIc, FGFR3IIIb, FGFR3IIIc, and FGFR4. Covalent affinity cross-linking analysis revealed that the binding to cells expressing FGFR2IIIb and FGFR1IIIb was to molecular species that corresponded to the recombinant FGFR (Fig. 6, inset). Both FGF-7 and FGF-10 competed with radiolabeled FGF-1 for binding to recombinant FGFR2IIIb expressed in Sf9 cells at nearly identical levels (Fig. 7A) and each competed similarly with the other when one or the other was the radiolabeled ligand (Fig. 7, B and C). These results suggest that when all other conditions are equal, FGF-10 and FGF-7 exhibit near equal affinities for FGFR2IIIb.
Differential Interaction of FGF-7 and FGF-10 with Heparin and the Pericellular Matrix-Similar to other members of the FGF polypeptide family, FGF-10 exhibits affinity for heparin, which can be utilized in its isolation, stabilization, and recovery, and cellular heparan sulfate sites. In contrast to FGF-7, which elutes from immobilized heparin at 0.8 M NaCl, FGF-10 elutes at 1.15 M NaCl, which is similar to FGF-1 (Fig. 8A). This is still below the 1.55 M that is required to elute FGF-2 from the same column. Comparative Scatchard analysis of the binding of FGF-7 and FGF-10 to pericellular matrix sites, which are presumed to be heparan sulfate sites displayed on DT3 epithelial cells, revealed that FGF-10 binds to 1 ϫ 10 5 sites per cell, while FGF-7 binds to 2.3 ϫ 10 4 sites per cell (Fig. 8B). Both factors exhibit a similar affinity (K d ϭ about 2.2 nM) to the pericellular matrix sites.

FGF-10 Is a Second Candidate Prostate Andromedin-Here
we show that, similar to FGF-7, expression of FGF-10 is restricted to stromal cells of normal adult prostate and androgenresponsive differentiated non-malignant prostate tumors. FGF-10 mRNA levels were dependent on androgen in cells from normal prostate and responsive in stromal cells from a nonmalignant, well differentiated transplantable tumor. In addition, recombinant FGF-10 specifically stimulates epithelial cells and binds specifically to the resident epithelial cell FGFR splice variant, FGFR2IIIb, with similar affinity to FGF-7. Thus FGF-10 constitutes a second FGF polypeptide that is potentially involved in a directional paracrine signaling system from prostate stroma to epithelium. Stromal cells exhibit no response to either FGF-7 or FGF-10. As proposed previously for FGF-7, FGF-10 is also a candidate involved in relay of the action of androgen in the stroma to epithelial cells (2,3).
The full-length FGF-10 cDNA predicts a translation product of 23.8 kDa which exhibits a candidate secretory signal sequence, the removal of which, in the absence of post-translational modification, would result in a 19.8-kDa product (12). In vitro translation of the full-length cDNA yields two products (12,16). In agreement with a report which analyzed epitopetagged FGF-10 overexpressed by transfection in mammalian cells (16), FGF-10 antigen could not be detected in the medium of cultured prostate stromal cells from differentiated non-malignant tumors which naturally express the FGF-10 mRNA. Immunoanalysis with an antibody against bacterial recombinant antigen revealed a single band with apparent mass of about 20 kDa in extracts of the stromal cells that could not be detected in an equivalent amount of extract from epithelial cells. Without compensating for losses during recovery, the analysis suggested that about 1 g of FGF-10 antigen is associated with 10 8 cells. This single product was detectably smaller than the FGF-10 (37-214), which is the mature unmodified product predicted to arise by removal of the secretory signal sequence. Although the single FGF-10 species may arise by an efficient proteolytic modification from a higher molecular weight glycosylated form, which has been suggested to exist in addition to unglycosylated species (16), it also may be an unmodified, alternately translated product beginning at Met-42 with predicted molecular mass of 19.2 kDa by amino acid content. An alternately translated intracellular product is consistent with the lower molecular weight product observed from in vitro translation (12,16), the presence of a significant portion of FGF-10 antigen in cell extracts and the slightly lower molecular mass caused by six less amino acid residues of the major prostate stromal cell product. FGF-3, which exhibits similar receptor specificity to FGF-10 (7), also exhibits secreted and intracellular isoforms due to alternate translational initiation (17). These observations raise the possibility that disposition of FGF-10 product is also regulated at the level of alternate initiation, and production of the secretory form from prostate stromal cells requires the in vivo environment and signals from neighboring epithelial cells.
The Different Expression Pattern of FGF-7 and FGF-10 -Despite the apparent similarity in function in vitro, the expression of FGF-10 appears much more restricted than FGF-7, which is expressed broadly in the stromal or mesenchymal compartment of most embryonic and adult tissue (18). FGF-10 was first identified in E14 rat embryos using PCR primers homologous to FGF-3 and FGF-7, and reported to be expressed only in adult lung and heart tissue (19). Subsequent reports showed that FGF-10 was expressed at low levels in specific areas of the brain (20) and prominently in the prospective limb mesoderm in the embryo (20). We confirmed that expression of the FGF-10 mRNA is much more restricted than FGF-7 in tissues and cells employed in the current study. Although the FGF-10 to FGF-7 ratio is very low in fibroblast-like cells, expression of FGF-10 was equal to or greater than that of FGF-7 in prostate stromal cells which are thought to consist of predominantly smooth muscle cells. The presence of significant quantities of FGF-10 relative to FGF-7 in rat prostate smooth muscle cells (PS-1), myoblasts, and tissues rich in smooth muscle cells relative to skin suggest that the FGF-10-expressing stromal cells from the non-malignant, transplantable Dunning R3327PAP tumors (4,21) likely derive from the transplanted tumor stroma rather than invading stroma from the subcutaneous environment in the host animals (4,6). Although we cannot exclude the possibility that FGF-10 expression is induced in surrounding host tissue by presence of the tumor or host stroma that has invaded the tumor, FGF-10 expression was very low or undetectable in skin samples from subcutane- ous pre-implantation sites. From these results we propose that FGF-10 might be a useful marker to distinguish smooth muscle cells in mixed stromal cell populations.
In addition, similar to normal prostate that has a well defined epithelial and stromal compartment, FGF-10 is detectable only in tumors that exhibit both compartments. This includes both the parent pre-malignant DT tumor and SE tumors which were reconstituted with isolated epithelial and stromal cells derived from the DT tumors. SE tumors, which are at first slow growing and predominantly differentiated epithelial cells surrounded by stroma, but also exhibit islands of undifferentiated cells (3), express a significantly reduced ratio of FGF-7 to FGF-10. Only FGF-7 was observed in E tumors derived from cloned DT tumor epithelial cells which are devoid of stroma within the tumor and the fully malignant AT3 tumors that emerge from them. Neither E nor AT3 tumors, which exhibit FGF-7 mRNA, have a distinct stromal compartment. The fact that isolated and cloned cells (AT3-C) derived from the malignant AT3 tumors after many generations in culture still exhibit FGF-7 mRNA suggests that FGF-7, but not FGF-10, may be abnormally activated in epithelial cells during the evolution to malignancy.
Interaction of FGF-10 with FGFR1IIIb in Addition to FGFR2IIIb-The specificity of FGF-7 for the resident epithelial cell FGFR2IIIb isoform remains the most specific of the 18 FGF polypeptides tested to date. We show here that FGF-7 and FGF-10 specifically stimulate DNA synthesis and compete with the resident FGFR isoforms displayed on prostate epithelial cells with equal efficacy, while no effect on the counterpart prostate stromal cells which do not express FGFR2IIIb could be detected at concentrations up to 100 ng/ml. This is in contrast to a recent report which showed that FGF-10 was only 50% as effective as FGF-7 in stimulation of DNA synthesis in mouse keratinocytes and that FGF-10 stimulated DNA synthesis in mouse 3T3 fibroblasts at concentrations greater than 5 nM (100 ng/ml) (22). Although FGF-10 exhibited an equal affinity for recombinant FGFR2IIIb, in contrast to FGF-7, it bound the FGFR1IIIb isoform, but was 25 times less effective in competing FGF-1 from FGFR1IIIb than from FGFR2IIIb. In this respect, FGF-10 appears to be similar to FGF-3 which elicits a mitogenic response in transfected cells bearing the FGFR1IIIb as well as the FGFR2IIIb ectodomain (5). A detailed analysis of expression pattern of FGFR1IIIb relative to FGFR1IIIc in the prostate tissue and cells utilized in this study is under investigation. Previously we have demonstrated that the reduction of resident FGFR2 gene products and the ectopic appearance of FGFR1 in premalignant prostate tumor epithelial cells correlate with progression to malignancy (3,6). If the FGFR1IIIb isoform is part of the ectopic activation of the FGFR1 gene in epithelial cells, then stromal-derived FGF-10, which otherwise contributes to homeostasis of the epithelial cell population through resident FGFR2IIIb, may contribute to malignant progression. Preliminary analysis of FGFR1IIIb expression in our model rat prostate tissues indicates that expression of the isoform is elevated in the premalignant differentiated tumors with well defined stromal compartments, but not detectable in fully malignant tumors with no clear stroma. 2 Identification of the cell type expressing the FGFR1IIIb variant is under investigation.
FGF-10 Differs from FGF-7 in the Interaction with Heparin and Pericellular Matrix-Heparin and transmembrane or pericellular matrix heparan sulfate proteoglycans play roles in stabilization and control of access of FGF to FGFR (1) and an integral role as a subunit of the FGFR complex (23,24). In contrast to FGF-1, the mitogenic activity of FGF-7 on epithelial cells is conditionally inhibited by soluble heparin (22,25,26), although the binding of FGF-7 to isolated cell-free FGFR2IIIb is dependent on heparin similar to FGF-1 (26). In the mitogenic assays and conditions utilized in this study, we observed no significant effect of external heparin on the mitogenic activity of FGF-1, FGF-7, or FGF-10 for prostate epithelial cells at concentrations up to 100 g/ml. However, as reported previously for FGF-7 and FGFR2IIIb (26), heparin was required for the binding of all three ligands to recombinant FGFR expressed on the surface of insect cells or isolated and immobilized cellfree FGFR (results not shown). However, our results show that FGF-10 has a higher affinity for immobilized heparin than does FGF-7 and that the pericellular matrix of epithelial cells exhibits 4 times more binding sites for FGF-10 than for FGF-7. Although this difference is not reflected as a differential mitogenic or binding activity under the assay conditions employed, the results suggest that heparan sulfate may differentially regulate access or affinity of the two FGFs for FGFR2IIIb under specific conditions present in the tissue microenvironment. We have recently demonstrated that the heparan sulfate subunit of the FGFR kinase-cellular heparan sulfate duplex discriminates between FGF ligands that can bind to the duplex in contrast to heparin, which is an artificial mimic of pericellular matrix heparan sulfate (27).
In conclusion, although FGF-7 and FGF-10 are potentially redundant in respect to stromal origin, responsiveness to androgen and the binding and activation of epithelial cells through FGFR2IIIb, the differential expression among stromal cell subtypes, the differential interaction with pericellular heparan sulfate and the affinity of FGF-10 for FGFR1IIIb may confer unique roles on the two FGFs in a physiological context.