Oncogene (1997) 15, 2211 ± 2218  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Mouse ®broblast growth factor 10: cDNA cloning, protein characterization, and regulation of mRNA expression Hans-Dietmar Beer2,*, Charles Florence1,*, Johanna Dammeier2, Linda McGuire1, Sabine Werner2 and D Roxanne Duan1 1

Human Genome Sciences, Inc., 9410 Key West Avenue, Rockville, Maryland, 20850, USA; 2Max-Planck Institute of Biochemistry, Am Klopferspitz 18a, D-82152 Martinsried, Germany

Fibroblast growth factor 7 (FGF-7) or keratinocyte growth factor (KGF), is a potent and speci®c mitogen for epithelial cells. We have recently identi®ed a novel human FGF-7 homologue, named FGF-10.To study the expression of this new FGF family member and its regulation in wound repair, we cloned the mouse FGF-10 (mFGF-10) cDNA. The encoded protein is 92% identical to human FGF-10 and 91% identical to rat FGF-10. When expressed in mammalian 293 cells, the mFGF-10 protein was glycosylated but remained cell- or extracellular matrix-associated. Upon addition of heparin, mFGF-10 protein was released into the media. mRNA encoding mFGF-10 was relatively abundant in lung, skin, brain and heart. In the skin, both FGF-7 and mFGF-10 were expressed in the dermal, but not the epidermal compartment. In contrast to FGF-7, mFGF-10 expression was not induced during cutaneous wound repair. In cultured ®broblasts, expression of mFGF-10 was strongly repressed by transforming growth factor b and tumor necrosis factor a, whereas epidermal growth factor and interleukin-1b had no e€ect. These results demonstrate a di€erential regulation of mFGF-10 and FGF-7 expression in vitro and during the wound healing process. Keywords: keratinocyte growth factor; secretion; wound repair; heparin; growth factor; regulation

Introduction The ®broblast growth factor (FGF) family of mitogens comprises at least 14 structurally related growth factors (see reviews by Klagsbrun, 1989; Basilico and Moscatelli, 1992; and Tanaka et al., 1992; Miyamoto et al., 1993; Yamasaki et al., 1996; Smallwood et al., 1996). Individual FGFs play key roles in various physiological and pathological processes, including embryonic development, wound healing, and tumorigenesis (see reviews by Klagsburn, 1989; Basilico and Moscatelli, 1992; Rubin et al., 1995). Whereas most FGFs regulate proliferation and/or di€erentiation of many di€erent cell types, FGF-7 (keratinocyte growth factor) di€ers from other FGF family members by its high speci®city for epithelial cells (Rubin et al., 1989; Finch et al., 1989; Marchese et al., 1990; Miki et al., 1991). Thus the only known high anity receptor for FGF-7, a splice variant of FGF receptor 2 (FGFR2-

*These authors contributed equally to the study Correspondence: DR Duan Received 3 May 1997; revised 27 June 1997; accepted 27 June 1997

IIIb) (Miki et al., 1991; 1992), is only expressed by cells of epithelial origin. Previous studies have suggested an important role of FGF-7 and its receptor in wound repair. Expression of this mitogen is strongly upregulated in dermal ®broblasts after cutaneous injury in mice and humans (Werner et al., 1992; Marchese et al., 1995). This induction is likely to be mediated by serum growth factors and pro-in¯ammatory cytokines which are released at the wound site, since these factors are potent inducers of FGF-7 expression in cultured ®broblasts (Brauchle et al., 1994; Chedid et al., 1994). Most interestingly, induction of FGF-7 expression is signi®cantly reduced in animals su€ering from wound healing abnormalities (Werner et al., 1994a; Brauchle et al., 1995), suggesting that reduced levels of this mitogen are associated with impaired healing. This hypothesis was supported by the bene®cial e€ect of exogenous FGF-7 for wound reepithelialization (Staiano-Coico et al., 1993; Pierce et al., 1994; Wu et al., 1996). The important role of FGF-7 and its receptor in normal and wounded skin was proven in a transgenic mouse model where a truncated FGF-7 receptor (FGFR2-IIIb) mutant was expressed in basal keratinocytes of the epidermis, thereby blocking signal transduction by the endogenous receptor. These transgenic mice had severe abnormalities in nonwounded skin, including epidermal atrophy and hair follicle defects. After cutaneous injury, wound reepithelialization was signi®cantly reduced (Werner et al., 1994b), demonstrating the important role of FGFR2IIIb ligands in this process. Surprisingly, knockout mice which lack FGF-7 had no obvious wound healing defect (Guo et al., 1996). Since the truncated FGFR2IIIb mutant does not only block the action of FGF-7 but also of other FGFs which bind to this receptor (Ueno et al., 1993), the di€erent results obtained with the two transgenic mouse models suggested the existence of other receptor ligands which can compensate for the lack of FGF-7 in the knockout mice. We have recently identi®ed a novel human keratinocyte growth factor, KGF-2, (Jimenez et al., 1997; manuscript submitted) that shared homology with FGF-7. KGF-2 is the human homologue of rat FGF-10 (Yamasaki et al., 1996) for which no function was described. We have demonstrated that FGF-10 was a keratinocyte mitogen in vitro and in vivo. Recombinant FGF-10 protein also promoted skin wound repair in healing-impaired mice (Jimenez et al., 1997; manuscript submitted). Several questions arise from these ®ndings. For example, is FGF-10 expressed by cells of mesenchymal or epithelial origin?

Cloning and characterization of mouse FGF-10 H-D Beer et al

TNT

media

mFGF-10

TNT MW

mFGF-10

Iysates

We have recently identi®ed a novel human keratinocyte growth factor, FGF-10 (Jimenez et al., 1997; manuscript submitted) that shared high degree homology with rat FGF-10 (Yamasaki et al., 1996). In order to clone the mouse mFGF-10 cDNA, we utilized identical nucleotide sequences in the human FGF-10 and rat FGF-10 cDNAs to amplify an mFGF-10 cDNA fragment by PCR. This mFGF-10 probe was then used to obtain a longer mFGF-10 cDNA from a mouse lung cDNA library. The mFGF-10 cDNA is

In order to characterize the mFGF-10 protein, we ®rst performed a coupled in vitro transcription/translation (TNT) reaction with an expression construct of mFGF10 containing a carboxyl-terminal tag of the in¯uenza virus hemagglutinin (HA). This epitope tag was used because it can be detected by a monoclonal antibody. The reaction was carried out in the absence of a microsomal fraction in order to obtain the nonprocessed, non-glycosylated primary translation product. As shown in the right lane of Figure 2 (labeled autorad), two proteins of approximately 23 and 25 kD

mFGF-10

Cloning of mFGF-10 cDNA

Characteristics of mFGF-10 proteins expressed in mammalian cells

Vec

Results

2155 bp in length and contains a full length open reading frame for mFGF-10. It is predicted to encode a protein of 209 amino acids (Figure 1), with a calculated molecular mass of 23.6 kD. The amino acid sequence of mFGF-10 is 92% identical to that of human FGF10, and 91% identical to that of rat FGF-10 (Yamasaki et al., 1996). The ®rst 35 amino acids of the mFGF-10 protein are predicted to constitute a signal peptide for secretion (see Figure 1). Among all other FGF family members besides human FGF-10 and rat FGF-10, mFGF-10 is most homologous to FGF-7 (57% identity at the amino acid level). Three cysteine residues in rat FGF-10 and in human FGF-7 are conserved in human FGF-10 and mFGF-10 (Figure 1), including Cys151 of mFGF-10 that is conserved throughout the entire FGF family (see reviews by Klagsbrun; 1989, Basilico and Moscatelli, 1992).

mFGF-10

Is endogenous FGF-10 induced, like FGF-7, during wound repair? Is FGF-10 expression regulated in the same manner as FGF-7 by serum factors and proin¯ammatory cytokines? In order to study the expression and regulation of endogenous FGF-10 during wound repair, we initiated the cloning of the mouse FGF-10 (mFGF-10) cDNA. In this study, we report the identi®cation and the primary structure of mFGF-10. In addition, we characterized the secretion and post-translational modi®cation of mFGF-10 proteins. Finally, we studied the mRNA expression of mFGF-10 in normal mouse tissues and in mouse skin during wound repair. The regulation of mFGF-10 mRNA expression in ®broblasts by serum factors that have been shown to induce FGF-7 expression in these cells was also investigated. Our results suggest that mFGF-10 and mFGF-7 might have overlapping but also distinct functions in various mouse tissues.

Vec

2212

kDa — 66 — 55 — 36

mFGF-10

— 31

mFGF-10 — 21

Western

Figure 1 Amino acid sequence of mFGF-10 in comparison with human FGF-10, rat FGF-10 and human FGF-7. Amino acid sequence of mFGF-10 (top sequence, Genbank accession number U94517) is compared with those of human FGF-10 (hFGF-10) (Jimenez et al., submitted for publication), rat FGF-10 (Yamasaki et al., 1996), and human FGF-7 (hFGF-7) (Finch et al., 1989). Identical sequences to mFGF-10 are boxed. The predicted signal peptide (amino acids 1-35) of mFGF-10 is underlined. Potential N-glycosylation sites in the mFGF-10 protein are indicated with asterics

Autorad

Figure 2 Expression of mFGF-10 protein in mammalian 293 cells. The kidney epithelial cell line 293 was transiently transfected with the pcDNA-mFGF-10-HA plasmid (mFGF-10) or with the pcDNA vector plasmid (vec). mFGF-10 protein in cell culture media (denoted `media') and in cell lysates (denoted `lysates') was immunoprecipitated with an anti-HA-antibody and was analysed by SDS 14% PAGE and Western blotting. mFGF-10 protein expressed from a coupled transcription-and-translation (TNT) reaction in the presence of [35S]methionine and [35C]cysteine was immunoprecipitated and analysed simultaneously. The arrows point to the mFGF-10 protein species. Autoradiography of the TNT product on the Western blot is also shown

Cloning and characterization of mouse FGF-10 H-D Beer et al

tion, whereas the 25 kD and 26 kD proteins might contain O-glycosylation or other types of posttranslational modi®cation. Release of mFGF-10 into the media by heparin Because mFGF-10 has a predicted signal peptide for secretion and because the 30 kD mFGF-10 protein contains N-linked glycosylation, we speculate that mFGF-10 is secreted but remains associated with the cell surface or the extracellular matrix. To test this possibility, cells were grown in the presence or absence of heparin, which was known to compete for binding of FGFs to the extracellular matrix and cell surface proteoglycans. As shown in Figure 4, the 22 kD mFGF-10 protein, which was the least abundant mKGF-2 species in the cell lysate (see also Figure 2), was released into the medium upon addition of heparin. In addition, heparin released some of the 30 kD species into the medium, particularly at a

2 µg/ml

5 µg/ml

hr 0 4 8 16 24 0 4 8 16 24 0 4 8 16 24

Vec

0 µg/ml

Tunicamycin

Ig

mFGF-10

Figure 3 E€ects of tunicamycin treatment on the expression of mFGF-10 protein. After transfection with the pcDNA-mFGF-10HA plasmid (mFGF-10) or the pcDNA vector plasmid (Vec), 293 cells were treated with 2 or 5 mg/ml tunicamycin for 4, 8, 16, and 24 h before harvesting. mFGF-10 proteins in the cell lysates were immunoprecipitated and analysed by SDS 13% PAGE and Western blotting. The arrows point to mFGF-10 protein species

mFGF-10 (0.05)

mFGF-10 (0.02)

mFGF-10 (0.01)

mFGF-10 (0.005)

mFGF-10 (0)

MW

Vec

mFGF-10 (0.05)

Lysates mFGF-10 (0.02)

mFGF-10 (0.01)

mFGF-10 (0.005)

mFGF-10 (0)

Media

Vec

were obtained. In a westernblot analysis the larger protein (25 kD) was detected by the HA antibody, demonstrating that the complete carboxy-terminus was present. Thus this protein is likely to represent the complete primary translation product including the signal peptide and the HA tag. By contrast, the smaller protein was not recognized by the HA antibody, suggesting that it is truncated at the carboxylterminus. After immunoprecipitation and westernblot analysis of the TNT products, a 31 kD protein was also observed. However, this protein is unlikely to represent an mFGF-10 species, since (i) it was not seen in the autoradiogram and (ii) it was also present after immunoprecipitation of vector-transfected cells. Thus, it is likely to represent an ubiquitous cellular protein which cross-reacts with the HA-antibody. Because mFGF-10 contains a predicted signal peptide for secretion, we analysed whether the mFGF-10 protein is indeed secreted by mammalian cells and whether it is post-translationally modi®ed. For this purpose the human kidney epithelial cell line 293 was transiently transfected with the mFGF-10-HA plasmid. mFGF-10 proteins were then immunoprecipitated with the monoclonal anti-HA antibody from either the cell lysates or the cell culture media. Surprisingly, the mFGF-10-HA protein appeared in the cell lysates but not in the media (Figure 2). Therefore, mFGF-10 protein could be either inside the cells, associated with the cell surface, or associated with the extracellular matrix. We could not rule out the possibility that mFGF-10 secreted into the media had a cleaved carboxyl-terminus so that it was not immunoprecipitable by the anti-HA antibody. In the cell lysates, the carboxyterminal HA-tagged mFGF-10 appeared as four bands on SDS ± PAGE, migrating at approximately 22, 25, 26 and 30 kD (Figure 2). Since the 22 kD protein was recognized by the HA antibody, it seems to have a truncation at the amino-terminus and not at the carboxyl-terminus. Thus it might represent the primary translation product without the signal peptide. We cannot rule out the possibility that the amino-terminal truncation extends beyond the predicted signal peptide. The observed higher molecular weight species from 293 cells could represent mFGF-10 proteins with posttranslational modi®cation such as glycosylation, although the 25 kD protein could also represent the primary translation product including the signal sequence.

kDa

- 66 - 55

N-linked glycosylation of mFGF-10 protein There are two potential N-linked glycosylation sites (Asn 50 and Asn 197; indicated with an asterics in Figure 1) in mFGF-10. In addition to the fact that the observed molecular weight of three mFGF-10 species was higher than the predicted one, glycosylation of mFGF-10 protein is supported by an experiment in which 293 cells were treated with tunicamycin, an inhibitor of N-linked glycosylation. Even a short incubation (4 h) in tunicamycin-containing medium (2 or 5 mg/ml) resulted in a drastic reduction of the 30 kD mFGF-10 species. By contrast, the levels of the 25 and the 26 kD species increased (Figure 3). Therefore, the cell- or extracellular matrix-associated 30 kD mFGF-10 protein contains N-linked glycosyla-

- 36 - 31

- 21 Figure 4 Release of mFGF-10 into cell culture media by heparin. After transfection with the pcDNA-mFGF-10-HA expression construct (mFGF-10) or pcDNA alone (Vec), 293 cells were treated with various concentrations of heparin for 24 h (the numbers in parenthesis indicate the heparin concentration, unit/ml). mFGF-10 protein in the cell lysates or in cell culture media was immunoprecipitated and analysed by SDS 13% PAGE and Western blotting. The arrows point to mFGF-10 protein species. mFGF-10 protein released into the media by heparin has a molecular weight of approximately 22 kD

2213

Cloning and characterization of mouse FGF-10 H-D Beer et al

mRNA expression of mFGF-10 during wound repair A remarkable feature of FGF-7 is the strong induction of its expression after cutaneous injury in mice and humans (Werner et al., 1992; Marchese et al., 1995). Therefore we analysed the expression of mFGF-10 during the healing process of full-thickness excisional wounds in mice. As expected from the strong expression of mFGF-10 in the dermal compartment of mouse tail skin, expression of this factor was also high in normal mouse back skin where the wounds were introduced (Figure 6). Unlike FGF-7, however, no signi®cant induction of FGF-10 expression could be detected during wound healing and expression levels of this growth factor even declined after skin injury (Figure 6). Regulation of mFGF-10 expression in cultured ®broblasts

13d wound

7d wound

5d wound

3d wound

1d wound

Due to the lack of induction of mFGF-10 expression after skin injury, we speculated about a di€erential regulation of mFGF-7 and mFGF-10 in skin-derived cells. To address this question, we ®rst determined the

skin

epidermis

To determine potential sites of FGF-10 action, we analysed the expression pattern of mFGF-10 in adult mouse tissues by Northern blot and RNase protection assays. Using a mouse multiple tissue Northern blot, we observed four mRNA species of 1.7, 4.2, 4.4. and 4.8 kb which were present at di€erent levels in various mouse tissues and organs (data not shown). To determine if these transcripts are indeed derived from the mFGF-10 gene and also to detect low amounts of mFGF-10 mRNA, we used a highly sensitive RNase protection assay to further elucidate the expression pattern of mFGF-10. Under the chosen conditions, every single mismatch is recognized by the RNases. Therefore, transcripts derived from a di€erent FGF gene cannot be detected by this assay. As shown in Figure 5, mFGF-10 was highly expressed in lung, brain, and in the dermal compartment of mouse tail skin. This is di€erent from the human situation, where FGF-10 mRNA was only present at extremely low levels in the skin (Jimenez et al., submitted for publication; S Werner, unpublished data). In addition to these tissues, high levels of mFGF-10 mRNA were also found in the mammary gland (data not shown)

tRNA

Tissue-speci®c expression of mFGF-10

and moderate levels were seen in heart and skeletal muscle. By contrast, mFGF-10 expression could not be detected in intestine, kidney, liver, testis or in the epidermal compartment of tail skin. The mRNA expression pattern of mFGF-10 was similar to that of mFGF-7 in general, although the latter was also expressed in the small and large intestine of normal adult mice (Figure 5). Thus, mFGF-7 and mFGF-10 have mostly overlapping but also distinct sites of expression.

probe

concentration of 0.01 units/ml (0.055 (mg/ml). Furthermore, heparin appeared to stabilize mFGF-10 proteins in the cells, since the levels of mFGF-10 protein were higher in heparin-treated cells compared to non-treated cells. These data suggest that the 22 kD protein and at least some of the 30 kD protein are bound to the extracellular matrix and/or cell surface, whereas the other mFGF-10 species mostly remain inside the cell.

probe tRNA small int. large int. liver testis heart sk. muscle lung kidney brain dermis

2214

mFGF-10

mFGF-7

Figure 5 RNase protection analysis of mFGF-10 and mFGF-7 mRNA in adult mouse tissues. First panel: 20 mg of total cellular RNA from various adult mouse tissues was analysed by RNase protection assay for the presence of mFGF-10 mRNA. Hybridization was performed under high-stringency conditions to avoid cross-hybridization with other FGF mRNAs. The 322 bp fragment described in Materials and methods was used as a probe. 1000 c.p.m. of the hybridization probe were loaded in the lane labeled `probe' and used as a size marker. Twenty mg of tRNA was used as a negative control. The mFGF-10-speci®c band is indicated with an arrow. The additional bands result from self-hybridization of the probe, since they are also present in the tRNA lane. Second panel: The experiment shown above was repeated with a murine FGF-7 probe using the same batch of RNAs. The FGF-7-speci®c band is indicated with an arrow. An ethidium bromide stain of 1 mg of the same batch of RNAs is shown below

Figure 6 mRNA expression of mFGF-10 in normal and wounded skin. Total cellular RNA (20 mg) from normal and wounded mouse back skin was analysed by RNase protection assay for the presence of mFGF-10 mRNA. The 178 bp fragment described in Materials and methods was used as a probe. 1000 c.p.m. of the hybridization probe were loaded in the lane labeled `probe' and served as a size marker. Twenty mg of tRNA was used as a control. An ethidium bromide stain of the same batch of RNAs is shown below

Cloning and characterization of mouse FGF-10 H-D Beer et al

types of cells in the skin which express mFGF-10 in vitro. Similar to FGF-7, mFGF-10 was expressed by cultured ®broblasts (Balb/c 3T3 cell line) but not by keratinocytes or macrophages (data not shown). Since FGF-7 expression in ®broblasts is strongly upregulated by the serum growth factors EGF and PDGF-BB and by the pro-in¯ammatory cytokines IL-1b and TNF-a, but not by TGF-b1, we analysed the e€ect of these factors on mFGF-10 expression. As shown in Figure 7, EGF and IL-1b had no e€ect on mFGF-10 expression in Balb/c 3T3 cells. By contrast, a strong repression of mFGF-10 was observed within 1.5 and 5 h after addition of TGF-b1 or TNF-a and a minor repression was obtained with PDGF-BB. This repression was reversible and mFGF-10 mRNA levels increased again after 8 h of growth factor or cytokine treatment. These data demonstrate an opposite e€ect of various growth factors and cytokines on FGF-7 and mFGF-10 expression in ®broblasts. Discussion

8h

5h

1.5 h

0h

probe

We have recently identi®ed a novel human FGF, keratinocyte growth factor 2, or human FGF-10. In order to characterize the regulation of FGF-10 expression in vivo, we chose to clone the mouse

EGF

lL-1β

PDGF BB

TGF-β1

TNF-α

Figure 7 Regulation of mFGF-10 mRNA expression in ®broblasts by puri®ed growth factors and pro-in¯ammatory cytokines. Balb/c 3T3 ®broblasts were rendered quiescent by serum starvation. They were stimulated with 20 ng/ml EGF, 100 U/ml IL-1b, 10 ng/ml PDGF-BB, 1 ng/ml TGF-b1, or 300 U/ml TNF-a for 1.5, 5 or 8 h as indicated. Twenty mg of total cellular RNA from these cells was analysed by RNase protection assay for the presence of mFGF-10 mRNA. The 178 bp fragment described in Materials and methods was used as a probe

FGF-10 (mFGF-10) cDNA. Although the mFGF-10 cDNA is nearly identical to human FGF-10 and rat FGF-10, its sequence diverges from the other two in a serine-repeat region near the amino-terminus and in the last few amino acids of the carboxyl-terminus (see Figure 1). mFGF-10 and human FGF-10 both have shorter serine stretches than rat FGF-10. The function of this serine-repeat is not clear and it is not found in FGF-7 or other FGFs. The ®rst 35 amino acids of the mFGF-10 protein consist of hydrophobic residues that are predicted to be a signal peptide for secretion. FGF-7 also has a signal peptide (Finch et al., 1989) and is eciently secreted from mammalian cells into the culture media (Brauchle et al., 1994; Chedid et al., 1994). Surprisingly, unlike FGF-7, mFGF-10 proteins from transfected epithelial 293 cells appeared in the cell lysates but not in the tissue culture media. Therefore, mFGF-10 protein could be inside the cells, cell surface-associated, or extracellular matrix-associated. The hypothesis that mFGF-10 enters the secretory pathway but remains cell- or matrix-associated is supported by two ®ndings: (1) at least one form of mFGF-10 protein is Nglycosylated (see Figure 3), and N-glycosylation occurs in the endoplasmic reticulum and the golgi apparatus but not in the cytosol. (2) heparin releases 22 kD and 30 kD mFGF-10 proteins into the media (see Figure 4). Although the 22 kD species was the least abundant mFGF-10 variant present in the cell lysate and was hardly detectable after immunoprecipitation and Western blot analysis of total cell lysate, it was abundant and clearly visible after analysis of the conditioned medium of heparin-treated cells. This might be due to stabilization of the 22 kD protein by heparin and its accumulation in the media over time. The secretion properties of mFGF-10 are reminiscent to those observed with another FGF family member, FGF-3. This type of FGF has a predicted signal peptide (Dickson and Peters, 1987, Brookes et al., 1989) but the Xenopus and the Zebra®sh FGF-3 proteins were shown to remain associated with the extracellular matrix (Kiefer et al., 1993; Kiefer et al., 1996) and were released by heparin in a dosedependent manner (Kiefer et al., 1993; Kiefer et al., 1996). A unique feature of FGF-7 is the paracrine mechanism of action, whereby the ligand is predominantly produced by mesenchymal cells such as ®broblasts and endothelial cells (Finch et al., 1989; Smola et al., 1993) but not by epithelial cells (Finch et al., 1989). By contrast, the receptor for FGF-7, FGFR2-IIIb, is exclusively expressed on epithelial cells (Miki et al., 1992; Werner et al., 1992). Several results suggest that FGF-10 could also act in a paracrine manner: (1) mFGF-10 mRNA was detected in the dermis but not in the epidermis of mouse tail skin; (2) cultured ®broblasts but not keratinocytes expressed mFGF-10 and (3) human FGF-10 has recently been shown to bind to FGFR2-IIIb (Jimenez et al., submitted for publication). The latter result strongly suggests that both FGF-7 and FGF-10 stimulate the same cell types. Although the cell-type speci®c expression of both FGFR2-IIIb ligands is obviously similar, their regulation in ®broblasts was found to be very di€erent. Whereas FGF-7 expression is strongly upregulated by

2215

Cloning and characterization of mouse FGF-10 H-D Beer et al

2216

various serum growth factors and pro-in¯ammatory cytokines (Brauchle et al., 1994; Chedid et al., 1994), most of these factors had no e€ect on mFGF-10 expression (this study). To our surprise, TNF-a and TGF-b treatment of ®broblasts did even result in a substantial reduction of the levels of mFGF-10 mRNA, suggesting that FGF-7 and mFGF-10 could be oppositely regulated under conditions where TGFb1 and TNF-a are present at high levels such as during cutaneous wound repair or in other types of in¯ammatory conditions. This hypothesis is supported by the lack of induction of mFGF-10 expression after skin injury. The rapid decrease in the levels of mFGF10 mRNA after TNF-a or TGF-b treatment indicates a short half-life of the mRNA, a phenomenon which is characteristic for strongly regulated genes. Our detailed expression study demonstrated the presence of mFGF-10 mRNA in various mouse tissues. This expression pattern is very similar to that of FGF-7, although the latter is also expressed in the intestine. The detection of particularly high levels of mFGF-10 mRNA in skin and lung is of particular interest with respect to its potential in vivo function. Thus the overexpression of a truncated FGFR2-IIIb muntant in these organs had caused severe phenotypic abnormalities. In the skin, the targeted expression of a dominant-negative FGFR2-IIIb in basal keratinocytes of transgenic mice caused epidermal atrophy, hair follicle abnormalities, and a severe delay in wound reepithelialization (Werner et al., 1994b). In the developing lung, inhibition of FGFR2IIIb signaling inhibited lung branching morphogenesis and alveolar di€erentiation (Peters et al., 1994). These results are in contrast to the normal histologic features of skin and lung in mice lacking FGF-7 (Guo et al., 1996), suggesting the existence of other FGFR2-IIIb ligands which compensate for the lack of FGF-7. The identi®cation of mFGF-10 and its high expression in skin and lung could now provide an explanation for the lack of abnormalities in FGF-7 knockout mice. By contrast, a truncated receptor blocks the action of all receptor ligands, including FGF-7 and mFGF-10, resulting in a complete blockade of the action of endogenous receptors. This could explain the severe phenotype in mice expressing dominant-negative receptors. One of the most striking features of FGF-7 is its strong upregulation in cutaneous wound repair (Werner et al., 1992; Marchese et al., 1995). This result suggested an important role of FGF-7 in the healing process, a hypothesis which was strongly supported by the defect in reepithelialization of fullthickness excisional wounds in transgenic mice expressing a dominant-negative FGFR2-IIIb in the epidermis (Werner et al., 1994b). Surprisingly, the healing process of incisional wounds appeared normal in FGF-7 knockout mice (Guo et al., 1996). Although it remains to be elucidated whether the much more extensive reepithelialization of excisional wounds is also normal in these animals, this result demonstrates that at least incisional wounds can heal in the absence of this mitogen. The identi®cation of FGF-10 and its strong expression in mouse skin suggested that this novel FGFR2-IIIb ligand could also compensate for the lack of FGF-7 during wound healing in FGF-7 null mice. In this case one might expect upregulation of mFGF10 expression after skin injury. To our surprise,

however, expression levels of mFGF-10 mRNA did not change signi®cantly after wounding in normal mice. It will be interesting to determine if upregulation occurs in FGF-7 knockout mice. It might also be possible that lower levels of mFGF-10 are sucient to stimulate FGFR2-IIIb in wounds. This hypothesis is supported by the potent stimulatory e€ect of exogenous human FGF-10 protein on the wound repair process in mice (Jimenez et al., submitted for publication). Finally, our ®nding that mFGF-10 remains cell-associated suggests that this factor is stored in its producer cells and might be released upon cell damage, e.g. after cutaneous injury. This could lead to elevated levels of mFGF-10 protein at the wound site in spite of the lack of mFGF-10 mRNA induction. In summary, we have identi®ed and characterized a novel mouse FGF-7 homologue which di€ers from FGF-7 in its secretion properties and its regulation during wound healing and in cultured ®broblasts. However, both factors are co-expressed in various mouse tissues, suggesting that they might have overlapping but also distinct in vivo functions. Material and methods Cloning of mFGF-10 cDNA The mFGF-10 DNA probe used for screening cDNA libraries was generated by PCR from a mouse brain cDNA library (Stratagene) using nested primers. The mFGF-10speci®c primer sequences were derived from identical sequences in human FGF-10 and rat FGF-10 (Yamasaki et al., 1996): ®rst round PCR primer: 5'-GTTTCC CCT TCT TGT TCA TGG C-3'; second round PCR primer: 5'-CTT GTT CAT GGC TAA GTA ATA GTT G3'. To generate the probe, two rounds of standard PCR reactions were performed using gene-speci®c primers described above and vector-speci®c primers designed from the Bluescript vector. An mFGF-10 cDNA fragment was obtained and used as a probe for screening a mouse lung cDNA library (Stratagene) according to standard methods for cDNA library screening (Sambrook et al., 1989). Transfection and immunoprecipitation experiments For transfection of mammalian cells, the mFGF-10 open reading frame without 5'- or 3'-untranslated regions was subcloned into the pcDNA3 expression vector (Invitrogen) with an in-frame epitope tag from the in¯uenza virus hemagglutinin (HA) (YPYDVPDYA) at the carboxylterminus. Human kidney epithelial 293 cells (ATCC), grown in Dulbecco's modi®ed Eagles medium supplemented with 10% fetal calf serum, were plated in 10-cm plates at 80 % con¯uency and transfected with pcDNA-mFGF10-HA using the Lipofectamine method (Life Technology) according to manufacturer's instructions. Cell culture media were collected 24 h post-transfection. Cells were harvested at the same time as the culture media and were lysed in NP-40 lysis bu€er (20 mM Tris, pH 8.0, 150 mM NaCl, 1% NP-40, 10% Glycerol, 1 mM PMSF, 1 mg/ml Leupeptin, 1 mg/ml Pepstatin, 1 mM EDTA). FGF-10 proteins in the cell culture media and cell lysates were immunoprecipitated with a monoclonal anti-HA antibody (Boehringer Mannheim, 2 mg/10 ml for cell culture media and 1 mg/ml for cell lysates) and protein A Sepharose 4B (Sigma) for 3 h at 48C. The protein A beads were washed three times with ice cold NP-40 lysis bu€er, and immunoprecipitates were analysed on SDS ± PAGE. Western blots were

Cloning and characterization of mouse FGF-10 H-D Beer et al

probed with the same anti-HA antibody and a goat-antimouse IgG-alkaline phosphatase conjugate (Promega). Coupled transcription-and-translation (TNT) reactions were carried out with the pcDNA-mFGF-10-HA plasmid (1 mg per reaction) using T7 RNA polymerase in the TNT kit (Promega) according to manufacturer's speci®cations. The reaction mixture was then diluted with 1 ml of NP-40 lysis bu€er and immunoprecipitated with the anti-HA antibody. Tunicamycin (Sigma) was added to cell culture media after transfection to ®nal concentrations of 0, 2, or 5 m g/ml. Cells were incubated in tunicamycin-containing medium for 4, 8, 16, or 24 h before harvesting. Heparin (Sigma) (0.005, 0.01, 0.02, or 0.05 units per ml of culture media) was added to 293 cells immediately after transfection. Cell culture media and cell lysates were collected 24 h after addition of heparin. Analysis of mFGF-10 regulation To determine the regulation of mFGF-10 expression in ®broblasts, the Balb/c 3T3 ®broblast cell line was used. Cells were cultured in Dulbecco`s modi®ed Eagles medium (DMEM) supplemented with 10% newborn calf serum (NCS), grown to con¯uence in 10 cm culture dishes, and rendered quiescent by a 16 h incubation in DMEM/ 1% NCS. They were subsequently incubated for varying periods in fresh DMEM containing 20 ng/ml epidermal growth factor (EGF), 10 ng/ml platelet-derived growth factor BB (PDGF-BB), 1 ng/ml transforming growth factor b1 (TGF-b1), 100 U/ml interleukin-1b (IL-1b), or 300 U/ ml tumor necrosis factor alpha (TGF-a). Aliquots of cells were harvested before and at di€erent time points after treatment with these reagents and used for RNA isolation. NCS and DMEM were purchased from Gibco Life Technologies, Inc., growth factors and cytokines were from Boehringer Mannheim Biochemicals. Each experiment was repeated at least twice. Northern blot analysis The DNA probe for Northern blot analysis was made by random primer labeling according to manufacturer's speci®cation (Rediprime kit, Amersham). The probes were puri®ed by passing through NucTrap probe puri®cation columns (Stratagene) to remove unincorporated nucleotides. Multiple tissue Northern blots were prehybridized in Hybrisol I solution (Oncor) for 3 h at 408C. Probe DNA was denatured and added to hybridization solution at 106 c.p.m/ml of solution. Hybridization was carried out at 428C overnight. Blots were washed twice for 15 min with 26SSC/0.1% SDS at room temperature, once in 0.26 SSC/0.1% SDS at 558C, once in 0.26SSC/0.1% SDS at 658C and autoradiographed.

RNA isolation and RNase protection assay Total cellular RNA was isolated from various tissues and organs of adult Balb/c mice (3-months-old), or from cultured cells as described by Chomczynski and Sacchi, 1987. RNase protection assays were carried out as recently described (Werner et al., 1992). Brie¯y, DNA probes were cloned into the transcription vector pBluescript KSII (+) (Stratagene) and linearized. An antisense transcript was synthesized in vitro using T3 or T7 RNA polymerases and 32P-UTP (800 Ci/mmol). RNA samples were hybridized at 428C overnight with 100 000 c.p.m of the labeled antisense transcript. As a loading control, 1 mg of the same batch of RNAs were loaded on a 1% agarose gel and stained with ethidium bromide. Hybrids were digested with RNases A and T1 for 1 h at 308C. Protected fragments were separated on 5% acrylamide/8 M urea gels and analysed by autoradiography. All protection assays were repeated with a di€erent set of RNAs from independent experiments, using two di€erent mFGF-10 templates. The following templates were used: (i) a 178 bp fragment corresponding to nt 947-1125 of the mouse FGF-10 cDNA, (ii) a 322 bp fragment corresponding to nt 7951117 of the mouse FGF-10 cDNA (both fragments are located within the coding region), and (iii) a 201 bp fragment corresponding to nucleotides 23-224 of the murine FGF-7 cDNA (Mason et al., 1994). Wounding and Preparation of Wound Tissues Wounding and preparation of wound tissues was performed as recently described (Werner et al., 1994a). Brie¯y, 6 mm full-thickness excisional wounds were generated on the back of female Balb/c 3T3 mice by excising skin and panniculus carnosus. Wounds were left uncovered without dressings. At di€erent time points after injury, the complete wounds of four mice (N=24 wounds), including 2 mm of the epithelial margins, were excised and immediately frozen in liquid nitrogen. All experiments with animals were carried out with permission from the local government of Bavaria. Acknowledgements We thank the sequencing group in the Exploratory Research Department for sequencing the mFGF-10 cDNA, Dr Clive Dickson for providing murine mammary gland RNA, and Frederique Torterotot for excellent technical assistance. This work was partly supported by a grant from the Human Frontier Science Program (to SW), the Deutsche Forschungsgemeinschaft (WE 1983/3-1) and by a Hermann-and-Lilly-Schilling award (to SW).

References Basilico C and Moscatelli D. (1992). Adv. Cancer Res., 59, 115 ± 165. Brauchle M, Angermeyer K, HuÈbner G and Werner S. (1994). Oncogene, 9, 3199 ± 3204. Brauchle M, FaÈssler R and Werner S. (1995). J. Invest. Dermatol., 105, 597 ± 584. Brookes S, Smith R, Casey G, Dickson C and Peters G. (1989). Oncogene, 4, 429 ± 436. Chedid M, Rubin JS, Csaky KG and Aaronson SA. (1994). J. Biol. Chem., 269, 19753 ± 10757. Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, 156 ± 159. Dickson C and Peters G. (1987). Nature, 326, 833. Finch PW, Rubin JS, Miki T, Ron D and Aaronson SA. (1989). Science, 245, 752 ± 755.

Guo L, Degenstein L and Fuchs E. (1996). Genes and Dev., 10, 165 ± 175. Jimenez PA, Gruber JR, Liu B, Feng P, Florence C, Blunt A, Huddleston KA, Teliska M, Alfonso P, Coleman TA, Ornitz DM, Dillon PJ and Duan DR. (1997). Manuscript Submitted. Kiefer P, Mathieu M, Close MJ, Peters G and Dickson C. (1993). EMBO J., 12, 4159 ± 4168. Kiefer P, Mathieu M, Mason I and Dickson C. (1996). Oncogene, 12, 1503 ± 1511. Klagsbrun M. (1989). Prog. Growth Factor Res., 1, 207 ± 235. Marchese C, Rubin J, Ron D, Faggioni A, Torrisi MR, Messina A, Frati, L and Aaronson SA. (1990). J. Cell. Physiol., 144, 326 ± 32.

2217

Cloning and characterization of mouse FGF-10 H-D Beer et al

2218

Marchese C, Chedid M, Dirsch OR, Csaky KG, Santanelli F, Latini C, LaRochelle WJ, Torrisi MR and Aaronson SA. (1995). J. Exp. Med., 182, 1369 ± 1376. Mason IJ, Fuller-Pace F, Smith R and Dickson C. (1994). Mech. Dev., 45, 15 ± 30. Miki T, Fleming TP, Bottaro DP, Rubin JS, Ron D and Aaronson SA. (1991). Science, 251, 72 ± 75. Miki T, Bottaro DP, Fleming TP, Smith CL, Burgess WH, Chan A M-L and Aaronson SA. (1992). Proc. Natl. Acad. Sci. USA, 89, 246 ± 250. Miyamoto M, Naruo K, Seko C, Matsumoto S, Kondo T and Kurokawa T. (1993). Mol. Cell. Biol., 13, 4251 ± 4259. Peters KG, Werner S, Liao X, Wert X, Whitsett J and Williams LT. (1994). EMBO J., 13, 3296 ± 3301. Pierce GF, Yanagihara D, Klopchin K, Danilenko DM, Hsu E, Kenney WC and Morris CF. (1994). J. Exp. Med., 179, 3, 831 ± 840. Rubin JS, Osada H, Finch PW, Taylor WG, Rudiko€ S and Aaronson SA. (1989). Proc. Natl. Acad. Sci. USA, 86, 802 ± 806. Rubin JS, Bottaro DP, Chedid M, Miki T, Ron D, Cheon G, Taylor WG, Fortney E, Sakata H, Finch PW and LaRochelle WJ. (1995). Cell Biol. International, 19, 399 ± 411. Sambrook J, Fritsch EF and Maniatis T. (1989).Molecular cloning: A Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Smallwood PM, Munoz-Sanjuan I, Tong P, Macke JP, Hendry SH, Gilbert DJ, Copeland NG, Jenkins NA and Nathans J. (1996). Proc. Natl. Acad. Sci. USA, 93, 9850 ± 9857.

Smola H, Thiekotter G and Fusenig NE. (1993). J. Cell. Biol., 122, 417 ± 429. Staiano-Coico L, Krueger JG, Rubin JS, D'limi S, Vallat VP, Valentino L, Fahey T 3d, Hawes A, Kingston G, Madden MR, Mathwich M, Gottlieb AB and Aaronson SA. (1993). J. Exp. Med., 178, 865 ± 878. Tanaka A, Miyamoto K, Minamino N, Takeda M, Sato B, Matsuo H and Matsumoto K. (1992). Proc. Natl. Acad. Sci. USA, 89, 8928 ± 8932. Ueno H, Escobedo JA and Williams LT. (1993). J. Biol. Chem., 268, 22814 ± 22819. Werner S, Peters KG, Longaker MT, Fuller-Pace F, Banda MJ and Williams LT. (1992). Proc. Natl. Acad. Sci. USA, 89, 6896 ± 6900. Werner S, Breeden M, HuÈbner G, Greenhalgh DG and Longaker MT. (1994a). J. Invest. Dermatol., 103, 469 ± 473. Werner S, Smola H, Liao X, Longaker MT, Krieg T, Hofschneider PH and Williams LT. (1994b). Science, 266, 819 ± 822. Wu LC, Pierce GF, Galiano RD and Mustoe TA. (1996). Arch. Surg., 131, 660 ± 666. Yamasaki M, Miyake A, Tagashira S and Itoh N. (1996). J. Biol. Chem., 271, 15918 ± 15921.