The Mitogen-inducible Fn14 Gene Encodes a Type I Transmembrane Protein that Modulates Fibroblast Adhesion and Migration*

The binding of polypeptide growth factors to their appropriate cell surface transmembrane receptors triggers numerous biochemical responses, including the transcriptional activation of specific genes. We have used a differential display approach to identify fibroblast growth factor-1-inducible genes in murine NIH 3T3 cells. Here, we report that the fibroblast growth factor-inducible-14 (Fn14) gene is a growth factor-regulated, immediate-early response gene expressed in a developmental stage- and adult tissue-specific manner in vivo. This gene, located on mouse chromosome 17, is predicted to encode an 129-amino acid type Ia membrane protein with no significant sequence similarity to any known protein. We have used two experimental approaches, direct fluorescence microscopy and immunoprecipitation analysis of biotinylated cell surface proteins, to demonstrate that Fn14 is located on the plasma membrane. To examine the biological consequences of constitutive Fn14 expression, we isolated NIH 3T3 cell lines expressing variable levels of epitope-tagged Fn14 and analyzed their phenotypic propertiesin vitro. These experiments revealed that Fn14 expression decreased cellular adhesion to the extracellular matrix proteins fibronectin and vitronectin and also reduced serum-stimulated cell growth and migration. These results indicate that Fn14 is a novel plasma membrane-spanning molecule that may play a role in cell-matrix interactions.

Complex cellular processes such as proliferation, migration, differentiation, and apoptosis are regulated in part by a diverse group of molecules known as polypeptide growth factors. These factors act by binding and thereby activating specific transmembrane receptor tyrosine kinases. The activation of cell surface receptors by polypeptide ligands triggers downstream intracellular events, including the stimulation of protein phos-phorylation cascades and the transcriptional activation of numerous genes (1,2). Many mitogen-inducible genes have been identified, and they encode a diverse group of proteins including transcription factors, protein kinases and phosphatases, cell cycle regulators, and cytoskeletal and extracellular matrix proteins (2,3). A recent study using cDNA microarray technology has demonstrated that Ͼ500 genes are transcriptionally activated after serum stimulation of quiescent human fibroblasts and that a subset of these genes encode proteins implicated in the wound healing process in vivo (3).
Our laboratory has been studying fibroblast growth factor-1 (FGF-1) 1 -regulated gene expression in murine NIH 3T3 cells. FGF-1 (also referred to as acidic FGF) is one of the most extensively characterized members of the FGF family of heparin-binding proteins (4 -6). It is a potent mitogenic, chemotactic, angiogenic, and neurotrophic factor both in vitro and in vivo. These cellular responses are mediated via high affinity binding to a family of related membrane-spanning tyrosine kinase receptors (4 -6). We have shown by Northern blot hybridization analysis that FGF-1 stimulation of quiescent NIH 3T3 cells induces the expression of several previously described serum-regulated genes; e.g. c-fos, c-jun, c-myc, thrombospondin-1, and ornithine decarboxylase (7,8). In addition, we have used a differential display approach to isolate several cDNA clones representing previously unreported mitogen-inducible genes (9). Genes identified to date using this strategy include those encoding an aldose reductase-related protein (10,11), a member of the polo family of serine/threonine protein kinases (12,13), and a member of the transcriptional enhancer factor-1 family of DNA-binding proteins (14).
In this paper, we report that the murine Fn14 gene is a growth factor-and phorbol ester-regulated immediate-early response gene encoding a novel type Ia transmembrane protein, with an amino-terminal signal peptide, a 53-aa ectodomain, a single hydrophobic transmembrane domain, and a 28-aa carboxyl-terminal cytoplasmic domain. Analysis of transfected NIH 3T3 cell lines that constitutively express the Fn14 protein indicate that it may be involved in the regulation of cellular adhesion, growth, and migration.
RNA Isolation and Differential Display-Cells were harvested by trypsin/EDTA treatment, and total RNA was isolated using RNA Stat-60 (Tel-Test) per the manufacturer's instructions. Tissues from newborn or adult FVB/N mice (Taconic Farms) were obtained, and RNA was isolated as above after the samples were initially homogenized using a Tissumizer (Tekmar). RNA concentrations were calculated by measuring UV light absorbance at 260 nm. RNA (1 g) from quiescent or FGF-1/cycloheximide-treated NIH 3T3 cells was converted to cDNA using Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) and random hexamer primers (Roche Molecular Biochemicals) as described (16). PCR assays were performed using a degenerate sense protein kinase domain oligonucleotide primer (12) and a degenerate antisense leucine zipper domain oligonucleotide primer (17). An equivalent aliquot of each amplification mixture was subjected to electrophoresis in a 2% agarose gel, and DNA was visualized by ethidium bromide staining. An ϳ230-bp DNA fragment was excised, recovered using the freeze-squeeze method (18), reamplified, and ligated into the vector pCRII using a T/A cloning kit (Invitrogen).
cDNA Library Screening-A mouse Balb/c 3T3 cell cDNA library (kind gift of T. Lanahan, Johns Hopkins University School of Medicine, Baltimore, MD) was screened with the PCR-derived Fn14 cDNA fragment to obtain longer cDNA clones. The DNA fragment was labeled with [␣-32 P]dCTP (3000 Ci/mmol, Amersham Pharmacia Biotech) using a random primer DNA labeling kit (Roche Molecular Biochemicals). Approximately 2 ϫ 10 5 phage were plated at a density of 2 ϫ 10 4 plaque-forming units/150-mm dish using Escherichia coli C600 Hfl as the host. Duplicate plaque lifts (Colony/Plaque screen, DuPont) were pre-hybridized, hybridized, washed and exposed to Kodak X-Omat AR film as described (14). Five positive phage were plaque-purified and amplified on E. coli C600 Hfl cells. Plasmids were excised from the phage clones using XL-1 Blue cells and R408 helper phage (Stratagene) as described (19) and designated pBluescript/Fn14. 5Ј-RACE Assays-A cDNA fragment representing the 5Ј region of Fn14 mRNA was identified by PCR using mouse heart 5Ј-RACE-Ready cDNA (CLONTECH) per the manufacturer's instructions. The Fn14 antisense oligonucleotide primer used in the PCR amplification step was complementary to nucleotides 361-381 of the Fn14 cDNA sequence. Amplification products were subjected to electrophoresis in a 1.2% agarose gel and visualized by ethidium bromide staining. One DNA fragment was recovered and cloned into the vector pCRII using a T/A cloning kit (Invitrogen).
cDNA Sequence Analysis-Plasmid DNA was purified using a Magic Miniprep kit (Promega), and both strands of the entire ϳ950-bp Fn14 cDNA clone were sequenced by the dideoxynucleotide chain termination method. Both strands of the Fn14 cDNA fragment isolated by the 5Ј-RACE technique were also completely sequenced. Sequencing was done either automatically using an Applied Biosystems model 373A DNA sequencer and a Dye Terminator Cycle Sequencing kit (Perkin-Elmer) or manually using a Sequenase 2.0 kit (U.S. Biochemical Corp.) and [␣-35 S]dATP (1000 Ci/mmol, Amersham Pharmacia Biotech). The nucleic acid and deduced protein sequences were compared with sequences in the GenBank sequence data base using BLAST search programs accessed through the National Center for Biotechnology Information web site. The predicted Fn14 protein sequence was analyzed using several programs (SignalP, ScanProsite, PSORT II, TMpred, etc.) accessed through the ExPASy Molecular Biology Server.
Interspecific Mouse Backcross Mapping-Interspecific backcross progeny were generated by mating (C57BL/6J ϫ Mus spretus)F 1 females and C57BL/6J males as described (20). A total of 205 N 2 mice were used to map the Fn14 locus (see text for details). DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, Southern blot transfer, and hybridization were performed essentially as described (21). All blots were prepared with Hybond-N ϩ nylon membranes (Amersham Pharmacia Biotech). The probe, an ϳ1.0-kb EcoRI/XhoI cDNA insert from the pBluescript/Fn14 plasmid, was labeled with [␣-32 P]dCTP (3000 Ci/mmol; Amersham Pharmacia Biotech) using a random primer labeling kit (Stratagene). Blots were washed to a final stringency of 0.8ϫ SSCP, 0.1% SDS, 65°C. A fragment of ϳ8.0 kb was detected in EcoRI-digested C57BL/6J DNA, and a fragment of ϳ6.0 kb was detected in EcoRI-digested M. spretus DNA. The presence or absence of the ϳ6.0-kb EcoRI M. spretus-specific fragment was followed in backcross mice. A description of the probes and RFLPs for the loci linked to Fn14 including Mas1, E4f1, and Pim1 has been reported previously (22,23). Recombination distances were calculated using Map Manager, version 2.6.5. Gene order was determined by minimizing the number of recombinant events required to explain the allele distribution patterns.
Northern Blot Analysis-RNA samples (10 g) were denatured and subjected to electrophoresis in 1.2% agarose gels containing 2.2 M formaldehyde. The gels were stained with ethidium bromide to verify that each lane contained similar amounts of undegraded rRNA. RNA was transferred onto Zetabind nylon membranes (Cuno Inc.) by electroblotting and cross-linked to the membrane by UV light irradiation using a Stratalinker (Stratagene). A Northern blot containing 2 g of poly(A) ϩ RNA isolated from mouse embryos at different developmental stages was purchased from CLONTECH. Membrane pre-hybridization, hybridization, and washing conditions were as described (14). The two cDNA hybridization probes were: (a) mouse Fn14, ϳ1.0-kb EcoRI/XhoI fragment of pBluescript/Fn14, and (b) mouse ␣-actin, ϳ1.1-kb EcoRI fragment of pVAA (kind gift of G. Liau, Genetic Therapy Inc., Gaithersburg MD). The probes were radiolabeled with [␣-32 P]dCTP as described above under cDNA library screening.
In Situ Hybridization-A pBluescript/Fn14 plasmid containing the Fn14 cDNA sequence without the 3Ј-untranslated region was constructed and linearized using the appropriate enzymes, and sense or antisense riboprobes were transcribed in vitro using T3 or T7 RNA Polymerase (Roche Molecular Biochemicals), [␣-35 S]UTP (1000 Ci/ mmol, NEN Life Science Products) and reagents included in the SureSite II in situ hybridization kit (Novagen). Riboprobes were subjected to alkaline hydrolysis to yield an average length of ϳ100 nt, as verified by denaturing PAGE. Embryo sections (Novagen) were dewaxed, rehydrated, and deproteinized according to the hybridization kit instructions. Pre-hybridization and hybridization conditions were also per the manufacturer's instructions except that the hybridization step was performed at 65°C. Post-hybridization washes were once with 2ϫ SSC for 5 min at room temperature, once with 2ϫ SSC for 30 min at 50°C, once with 2ϫ SSC containing 20 g/ml RNase A for 30 min at 37°C, once with 2ϫ SSC containing 50% formamide for 30 min at 50°C, twice with 1ϫ SSC for 30 min at 50°C, and once with 0.1ϫ SSC for 30 min at 65°C. Slides were dipped in Kodak NTB-2 emulsion, counterstained with hematoxylin and eosin, and mounted.
Expression of Recombinant Fn14 in Bacterial Cells and Generation of Fn14 Antiserum-Recombinant Fn14 was produced by expression of the cDNA in E. coli as a GST fusion protein. The expression plasmid was constructed by ligation of an EcoRI-XhoI restriction fragment of pBluescript/Fn14 into the same sites located in the polylinker of pGEX-KG (kind gift of R. Friesel, Maine Medical Center Research Institute, Portland ME). DNA sequence analysis was performed to confirm the construct. E. coli HB101 cells (Life Technologies, Inc.) that had been transformed with this plasmid were cultured overnight at room temperature in Luria broth containing 100 g/ml ampicillin (Sigma). Cells were diluted 1:10 in Luria broth, grown for 5 h at room temperature, induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside (Life Technologies, Inc.) for 3 h, and pelleted by centrifugation. Cells were resuspended by repeated pipetting in STE buffer (10 mM Tris/HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA) containing 1 mg/ml lysozyme (Sigma) and incubated for 20 min on ice. Dithiothreitol (Sigma) was added to a final concentration of 5 mM, the lysate was vortexed, Nlaurylsarcosine (Sigma) was added to a final concentration of 1.5%, and again the lysate was vortexed. After mild sonication, the lysate was clarified by centrifugation, and Triton X-100 was added to the supernatant at a 2% final concentration. The supernatant was then incubated with hydrated glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) for 30 min at 4°C with end-over-end mixing. The beads were collected by a brief centrifugation and washed five times with PBS containing 1% Triton X-100. To prepare antigen for rabbit immunization, GST-Fn14 was released from glutathione-Sepharose beads by the addition of 100 mM Tris/HCl, pH 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol and dialyzed against PBS. A New Zealand White rabbit was injected with ϳ0.5 mg of the antigen in complete Freund's adjuvant (Calbiochem) and boosted five times with ϳ0.3 mg of the antigen in incomplete adjuvant (Calbiochem). Crude serum was used for the Western blot experiments.
Construction of the Fn14-GFP and Fn14-HA Eukaryotic Expression Plasmids-The plasmid pEGFP-N3/Fn14, which encodes Fn14 with a carboxyl-terminal EGFP tag (24), was constructed as follows. PCR was performed using pBluescript/Fn14 as the template, a sense primer containing a 5Ј BamHI restriction site followed by Fn14 nucleotides Ϫ12 to ϩ12, an antisense primer representing Fn14 nucleotides 380 -399 and Taq polymerase (Life Technologies, Inc.). The DNA product was isolated and then ligated into pCR2.1 using a T/A cloning kit (Invitrogen). A BamHI/EcoRI fragment representing the Fn14 coding region was isolated (18) and subcloned into the BglII and EcoRI cloning sites of the expression vector pEGFP-N3 (CLONTECH). The plasmid pcDNAIneo/Fn14-HA, which encodes Fn14 with a carboxyl-terminal influenza HA epitope tag (25), was constructed by first subcloning an EcoRI/XhoI fragment of pBluescript/Fn14 into the EcoRI and HpaI cloning sites of the expression vector pMEXneo (26). DNA sequence encoding the HA epitope tag was added to the Fn14 cDNA immediately upstream of the stop codon using the PCR overlap extension method with Taq polymerase (27). A NotI/ApaI fragment containing the Fn14-HA coding sequence was filled-in using T4 DNA polymerase and then cloned into the EcoRV site of the expression vector pcDNAIneo (Invitrogen). DNA sequence analysis was performed to confirm the identity of the two expression constructs described above.
Transfection Experiments-NIH 3T3 cells were grown to ϳ60% confluence and transfected with pEGFP-N3, pEGFP-N3/Fn14, pcDNAIneo, or pcDNAIneo/Fn14-HA using the LipofectAMINE PLUS reagent (Life Technologies, Inc.). The first two plasmids were used for transient transfection experiments. To generate stable cell lines, cells were transfected with either pcDNAIneo or pcDNAIneo/Fn14-HA, cultured for 24 h in standard growth medium, and split 1:10. After 24 h, cells were cultured in medium containing 400 g/ml G418 (Life Technologies, Inc.). Individual G418-resistant colonies were visible approximately 10 days later, and they were recovered with glass cloning cylinders. Clones were screened for Fn14-HA expression by Western blot analysis.
Western Blot Analysis-Cells were washed with PBS, collected by scraping with a rubber policeman, and pelleted by centrifugation at 500 ϫ g. TNEN buffer (50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 2.0 mM EDTA, 1% Nonidet P-40, 1ϫ protease inhibitor mixture; PharMingen) was added, and lysis was performed for 10 min at 4°C. Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min at 4°C. Protein concentrations were determined using the BCA assay kit (Pierce). Equivalent amounts of each protein sample were mixed with 2ϫ gel loading buffer (100 mM Tris/HCl, pH 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.2% bromphenol blue), heated at 95°C for 10 min, and subjected to SDS-PAGE using either a 15% or 4 -15% gradient acrylamide gel. For the FGF-2 time course experiment, proteins were transferred to an Immobilon-P SQ membrane (Millipore) by electroblotting. The membrane was stained with Ponceau S (Sigma) to verify that equivalent amounts of protein were present in each gel lane and then blocked overnight at 4°C in TBST (25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.1% Tween 20) containing 5% nonfat dry milk. The membrane was incubated for 1 h at room temperature in TBST containing 5% BSA and a 1:1000 dilution of anti-Fn14 polyclonal serum, washed twice in TBST, and incubated for 1 h in TBST containing 5% nonfat dry milk and a 1:20,000 dilution of goat anti-rabbit Ig-HRP (Bio-Rad). For analysis of transfected cells, proteins were transferred to Protran nitrocellulose membranes (Schleicher & Schuell) by electroblotting. The membrane was blocked overnight at 4°C in TBST containing 5% nonfat dry milk and 1% BSA and incubated for 1 h at room temperature in TBST containing 1% BSA and a 1:1000 dilution of rat anti-HA monoclonal antibody clone 3F10 (Roche Molecular Biochemicals). The membrane was then washed three times in TBST and incubated for 1 h in TBST containing 5% nonfat dry milk, 1% BSA, and a 1:2000 dilution of sheep anti-rat Ig-HRP (Amersham Pharmacia Biotech). For all Western blots, bound secondary antibodies were detected using the ECL system (Amersham Pharmacia Biotech). Autoradiographic signals were quantified by densitometry (VISAGE 4.6I software, BioImage Products).
Fluorescence Microscopy-Cells were plated on fibronectin-coated coverslips and transfected with either pEGFP-N3 or pEGFP-N3/Fn14. One day later, the cells were washed in PBS and fixed to the coverslips with 3% paraformaldehyde in PBS (pH 7.2) for 30 min. The coverslips were washed three times in PBS and mounted on glass slides in 50% glycerol in PBS. The cells were viewed with an Olympus BH-2 fluorescence microscope.
Cell Surface Biotinylation/Immunoprecipitation Analysis-Surface biotinylation of the pcDNAIneo V5 and pcDNAIneo/Fn14-HA 10 cell lines was carried out using EZ-Link Sulfo-NHS-LC-Biotin (Pierce) per the manufacturer's instructions. The cells were lysed in modified 1ϫ RIPA buffer (50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 1ϫ protease inhibitor mixture; PharMingen) for 30 min at 4°C and clarified by centrifugation. The cellular lysate was precleared with a 50% slurry of GammaBind Plus-Sepharose (Amersham Pharmacia Biotech) in RIPA buffer by mixing on an orbital rocker for 10 min at 4°C. The anti-HA antibody (described above) and control rat IgG (Sigma) were immobilized by mixing 50 l of a 50% Sepharose bead slurry with 0.5 g of anti-HA antibody or rat IgG for 2 h at 4°C on an orbital rocker. The conjugated beads were washed twice with RIPA buffer, equivalent amounts of cellular lysate were added in a total volume of 1 ml of RIPA, and the beads were incubated overnight at 4°C on an orbital rocker. Immunoprecipitates were washed three times with RIPA buffer, mixed with 2ϫ gel loading buffer (see above), and heated at 95°C for 10 min. Samples were subjected to SDS-PAGE and transferred to a Protran membrane by electroblotting. The membrane was blocked for 1 h at room temperature in TBST containing 3% BSA and incubated in TBST containing 3% BSA and 200 ng/ml streptavidin-HRP (Pierce) for 30 min. Biotinylated proteins were detected using the Amersham Pharmacia Biotech ECL system.
Cell Adhesion Assays-Adhesion assays were performed as described (28,29). After detachment using 5 mM EDTA in PBS, cells were plated in DMEM at 5 ϫ 10 4 cells/well on 96-well collagen I-, collagen IV-, fibronectin-, laminin-or vitronectin-coated CytoMatrix cell adhesion strips (Chemicon). The cells were incubated for 1 h at 37°C, rinsed in PBS, fixed by treatment with 3% paraformaldehyde in PBS for 20 min, rinsed again in PBS, stained with 0.2% crystal violet in 80% methanol for 30 min, and finally rinsed three times in H 2 O. The relative number of adherent cells in each well was evaluated by measuring the absorbance at 590 nm using a Perkin-Elmer HTS 7000 BioAssay plate reader. Additional vitronectin adhesion assays were conducted as above except for the following two modifications. First, the cells were plated on wells that had been coated with various concentrations of vitronectin (Chemicon) for 1 h at 37°C. After washing the coated wells with PBS, nonspecific binding sites were blocked using 1% BSA in PBS. The wells were washed again with PBS before the cells were plated. Second, the crystal violet stain was eluted with 0.1 M sodium citrate (pH 4.2) in 50% ethanol prior to measuring absorbance.
Cell Growth Assays-Cells were plated in standard growth medium at 10 3 cells/well in a 96-well cell culture plate. Cell growth was monitored using the cell proliferation reagent WST-1 (Roche Molecular Biochemicals) per the manufacturer's instructions by measuring absorbance at 450 nm with a 650-nm reference wavelength subtracted.
Cell Migration Assays-Migration assays were performed using 24well Transwell inserts (Costar) with 8.0-m pore polycarbonate membranes per the manufacturer's instructions. After detachment using 5 mM EDTA in PBS, cells were plated in DMEM in the upper compartment of the Transwell insert at 2 ϫ 10 4 cells/insert, and either DMEM alone or DMEM supplemented with 10% calf serum was placed in the lower compartment. The cells were incubated for 18 h at 37°C, and those cells remaining on the upper surface of the filter were removed with a cotton swab. The membranes were rinsed once with PBS, and the cells that had migrated to the lower surface of the membrane were fixed in 25% acetic acid/75% methanol and stained with crystal violet. The stain was eluted, and the relative number of cells on each membrane was determined by measuring absorbance as described above for the vitronectin adhesion assays.

RESULTS
Identification of an FGF-1-inducible Gene by mRNA Differential Display-We have described previously a differential display approach that employs random hexamer-primed cDNA templates, PCR oligonucleotide primers designed to anneal with DNA sequences encoding motifs found in particular protein structural domains, and agarose gel electrophoresis (9). In the present experiments, RNA was isolated from either quiescent NIH 3T3 cells or cells that had been treated with both FGF-1 and cycloheximide for 2 h. This RNA was converted to cDNA using the enzyme reverse transcriptase, and PCR assays were performed using sense protein kinase domain and antisense leucine zipper domain oligonucleotide primers. Amplification products were displayed using agarose gel electrophoresis and ethidium bromide staining. An ϳ230-bp DNA fragment was amplified to a greater degree when cDNA representing the RNA isolated from cells treated with FGF-1 and cycloheximide was used as template (data not shown). This DNA fragment was isolated, cloned, and used as a probe in a preliminary Northern blot hybridization experiment to confirm that it did indeed represent an FGF-1-inducible gene. This gene was named the FGF-inducible 14 (Fn14) gene, because it was predicted to encode an ϳ14-kDa protein (see below).
Fn14 cDNA Sequence Analysis-A mouse fibroblast cDNA library was then screened with the differential display-derived cDNA fragment in order to isolate longer cDNA clones. Both strands of the longest cDNA insert (ϳ950 bp) were sequenced by the dideoxynucleotide chain termination method. The nucleotide sequence contained a 12-nt 5Ј-untranslated region, a 387-nt open-reading frame, and a 550-nt 3Ј-untranslated region with a consensus polyadenylation signal and one copy of an AT-rich sequence motif implicated in rapid mRNA decay (30) (Fig. 1). The presumed initiating ATG codon is flanked by a favorable sequence for translation initiation (31); nevertheless, we used the 5Ј-RACE method to screen for cDNAs containing additional 5Ј sequence information. The longest cDNA we identified had an additional 12 nucleotides that were not present in the original cDNA clone.
The Fn14 gene is predicted to encode a protein of 129 aa with a molecular mass of 13,637 daltons and an isoelectric point of 8.18. The protein is proline-rich (9.3%) and contains eight cysteines but is devoid of asparagine or tyrosine residues. Computer analysis of the predicted Fn14 amino acid sequence using several sequence analysis programs indicated that Fn14 contains two relatively hydrophobic regions: one at the amino terminus that terminates with a signal peptidase cleavage site and another that is predicted to function as a membranespanning domain. This predicted structure indicates that Fn14 is a type Ia transmembrane protein (32) containing a 27-aa signal peptide, a 53-aa ectodomain, a 21-aa transmembrane domain, and a 28-aa cytoplasmic tail. After signal peptidase cleavage, the mature Fn14 polypeptide would be 102 aa in length and have a predicted molecular mass of 10,832 daltons. The possibility that Fn14 is a type Ia integral membrane protein is supported by the presence of a cluster of charged amino acids immediately following the second hydrophobic region, which may function as a stop-transfer signal. The predicted Fn14 protein has no other characteristic sequence motifs that may suggest cellular location or function except for several putative serine or threonine phosphorylation sites and a L-I motif in the cytoplasmic domain that could serve as an endocytosis signal (33)(34)(35).
Fn14 Amino Acid Sequence Comparisons-A search of the available sequence data bases using the Fn14 deduced amino acid sequence revealed no significant degree of sequence identity between Fn14 and other known proteins. However, we have noted that the Fn14 transmembrane domain has some sequence similarity to the transmembrane region found in the syndecan family of integral plasma membrane proteins. Specifically, alignment of the Fn14 transmembrane sequence to the transmembrane sequences of the mouse, rat, and human syndecan proteins revealed that four amino acids are conserved between Fn14 and all the syndecans, while five more amino acids are conserved between Fn14 and one or more of the syndecans. These nine conserved Of special interest is conservation of the double glycine, since the transmembrane region of the syndecans contains this motif and in general, glycine residues are found infrequently in transmembrane sequences (36). Furthermore, the Fn14 pattern of bulky and small side chain residues in this region is similar to the conserved pattern found in the syndecans (37).
Fn14 Chromosomal Location-The chromosomal location of Fn14 was determined by interspecific backcross analysis using progeny derived from matings of (C57BL/6J ϫ M. spretus)F1 ϫ C57BL/6J mice. This interspecific backcross mapping panel has been typed for over 2800 loci that are well distributed among all the autosomes as well as the X chromosome (20). C57BL/6J and M. spretus DNAs were digested with several restriction enzymes and analyzed by Southern blot hybridization for informative RFLPs using a Fn14 cDNA probe. The ϳ6.0-kb EcoRI M. spretus RFLP (see "Experimental Procedures") was used to follow the segregation of the Fn14 locus in backcross mice. The mapping results indicated that Fn14 is located in the proximal region of mouse chromosome 17 linked to Mas1, E4fl, and Pim1 (Fig. 2). Although 127 mice were analyzed for every marker and are shown in the segregation analysis, up to 189 mice were typed for some pairs of markers. Numbers to the left refer to the first amino acids on the lines, and the numbers to the right refer to the last nucleotides on the lines. The nucleotide sequence obtained from a 5Ј-RACE-derived cDNA clone is in boldface type. The solid line below amino acids 1-27 indicates the predicted signal peptide sequence, and the boxed amino acid stretch indicates the predicted transmembrane domain. In the 3Ј-untranslated region, the stop codon is denoted by an asterisk, a putative mRNA destabilization sequence motif is underlined, and the polyadenylation signal is boxed.
Each locus was analyzed in pairwise combinations for recombination frequencies using the additional data. The ratios of the total number of mice exhibiting recombinant chromosomes to the total number of mice analyzed for each pair of loci and the most likely gene order are: centromere, Mas1-2/189-Fn14-0/156-E4fl-1/132-Pim1. The recombination frequencies (expressed as genetic distances in centimorgans Ϯ the standard error) are: Mas1-1.1 Ϯ 0.7-(Fn14, E4fl)-0.8 Ϯ 0.8-Pim1. No recombinants were detected between Fn14 and E4fl in 156 animals typed in common, suggesting that the two loci are within 1.9 centimorgans of each other (upper 95% confidence limit).
Regulation of Fn14 mRNA Expression in NIH 3T3 Cells-We first investigated the kinetics of Fn14 mRNA accumulation following FGF-1 stimulation of NIH 3T3 cell growth by Northern blot hybridization analysis. A single Fn14 transcript of ϳ1.2-kb in size was detected in FGF-1-treated cells (Fig. 3A). Increased Fn14 mRNA levels were first evident at 1 h after FGF-1 addition, and maximal levels were present at 4 h. The effect of the RNA synthesis inhibitor actinomycin D and the protein synthesis inhibitor cycloheximide on FGF-1 induction of Fn14 mRNA levels was also examined. Actinomycin D cotreatment inhibited FGF-1 induction of Fn14 mRNA (Fig. 3B); in contrast, cycloheximide co-treatment resulted in Fn14 mRNA superinduction (Fig. 3C). Cycloheximide treatment alone also increased Fn14 mRNA levels. Taken together, these results indicate that Fn14 is an FGF-1-inducible, transcription-ally activated immediate-early response gene.
Regulation of Fn14 Protein Expression in NIH 3T3 Cells-We next investigated whether mitogenic stimulation of quiescent NIH 3T3 cells resulted in elevated Fn14 expression. We first obtained Fn14 polyclonal antiserum by immunizing rabbits with recombinant GST-Fn14 fusion protein purified from bacterial cultures. Initial Western blot experiments indicated that this antiserum specifically recognized recombinant Fn14 protein expressed in either bacterial or insect cell systems (data not shown). To analyze Fn14 expression levels in growth factor-stimulated cells, serum-starved cells were either RNA was isolated, and equivalent amounts of each sample were analyzed as described above. C, serum-starved cells were either left untreated or treated with FGF-1, FGF-1 and cycloheximide (Chx), or cycloheximide alone for 4 h. RNA was isolated, and equivalent amounts of each sample were analyzed as described above. D, serum-starved cells were either left untreated or treated with FGF-1, FGF-2, PDGF-BB, TGF-␤1, EGF, IGF-1, calf serum (CS), or PMA for 4 h. RNA was isolated, and equivalent amounts of each sample were analyzed as described above.
left untreated or treated with FGF-2 for different lengths of time and Western blot analysis was conducted using the Fn14 antiserum. A major immunoreactive protein of ϳ22 kDa was detected in FGF-2-treated cells (Fig. 4). FGF-2 stimulation increased the expression level of this protein, with maximal levels present at 12 h after mitogen addition. The apparent molecular mass of this immunoreactive protein is approximately twice the predicted size of the mature Fn14 protein (minus signal peptide); however, we have concluded that this protein is Fn14 based on several experimental findings (see "Discussion").

Fn14 mRNA Expression in Mouse Embryos and Tissues-
The pattern of Fn14 mRNA expression during mouse development was investigated. A blot containing RNA isolated from four different developmental time-points was obtained, and Fn14 mRNA levels were examined by Northern hybridization analysis. Fn14 mRNA expression was detected at all of the time points examined (Fig. 5A). Maximal levels of expression were detected at 7.5 days post-coitum; however, this RNA sample was isolated from both the developing embryo and the surrounding extra-embryonic and maternal tissues. 2 To identify the precise sites of Fn14 mRNA expression, we performed in situ hybridization analysis on serial sections from 8.5-day post-coitum mouse embryos sectioned in utero. Fn14 transcripts were detected primarily in the maternal decidual tissue nearest the ectoplacental (mesometrial) pole (Fig. 5B).
The tissue distribution of Fn14 mRNA was evaluated by Northern blot analysis using RNA isolated from tissues obtained from either newborn or adult mice. In the newborn animals, Fn14 transcripts were expressed at a relatively high level in all six tissues examined (Fig. 6A). In adult mice, Fn14 mRNA was expressed at the highest level in heart and ovary and at an intermediate level in kidney, lung and skin (Fig. 6B). Fn14 mRNA was expressed at a relatively low level in the other seven adult tissues examined. Thus, the Fn14 gene is expressed in a developmental stage-and tissue-specific manner in vivo.
Fn14 Subcellular Localization Studies-Fn14 was transiently expressed as a EGFP fusion protein in NIH 3T3 cells in order to determine its subcellular location. When EGFP itself is expressed in these cells, it is primarily found in the cytoplasm (Fig. 7A). In contrast, the Fn14-EGFP fusion protein is localized to the plasma membrane and to the trans-Golgi network near the nuclear membrane (Fig. 7, B and C). At the cell surface, Fn14-EGFP is especially prominent in thin membrane extensions resembling microspikes or filopodia (Fig. 7C). In additional experiments, Fn14 containing distinct amino-and carboxyl-terminal epitope tags was expressed in transfected NIH 3T3 cells and localization was assayed by immunofluores-cence microscopy of permeabilized or non-permeabilized cells. We found that Fn14 was oriented with its amino terminus located outside of the cell (data not shown), and this result is consistent with the orientation predicted by computer analysis of the Fn14 primary sequence.
We confirmed that Fn14 was present on the plasma membrane using a biochemical approach. First, the expression plasmid pcDNAIneo/Fn14-HA, which encodes a Fn14 protein containing a carboxyl-terminal HA epitope tag (25), was con-2 CLONTECH, personal communication.  structed. Second, NIH 3T3 cells were transfected with either the pcDNAIneo vector or the pcDNAIneo/Fn14-HA plasmid, stable cell lines were isolated by drug selection with G418, and Fn14-HA expression levels were assayed by Western blot analysis using an anti-HA antibody. One control cell line (V5) and three experimental cell lines expressing different levels of Fn14-HA (lines 3, 11, and 10) were chosen for subsequent biochemical and phenotypic studies. The major immunoreactive protein detected in the Fn14-HA-expressing cells was ϳ23 kDa in size; however, a ϳ16-kDa protein was also detected in lysates prepared from the cell line with the highest level of Fn14-HA production (Fig. 8, A and B). Third, we harvested an equivalent number of pcDNAIneo V5 and pcDNAIneo/Fn14-HA 10 cells and biotinylated the surface proteins using a cell membrane-impermeable biotinylation reagent. Cell lysates were prepared, immunoprecipitation was performed using either anti-HA IgG or control IgG, and the biotinylated cell surface proteins were detected by Western blot analysis using a streptavadin-HRP conjugate. Several biotinylated proteins were detected, but a ϳ23-kDa protein was immunoprecipitated specifically by anti-HA IgG (Fig. 8C). Thus, taken together, the results described above indicate that Fn14 is a plasma membrane-anchored protein.
Biological Effects of Fn14-HA Expression in Transfected NIH 3T3 Cells-We compared the properties of the control (vectortransfected) or Fn14-HA-expressing clonal cell lines described above using several different assays in order to gain insight into the possible biological functions of the Fn14 protein. First, we determined whether constitutive Fn14 expression had an effect on NIH 3T3 cell adhesion to various extracellular matrix proteins. Neither the control nor the three experimental cell lines were able to adhere to collagen I or collagen IV; furthermore, these lines displayed only weak adherence to laminin (Fig. 9A). In contrast, the control and Fn14-HA-expressing cell lines did adhere to fibronectin or vitronectin. In comparison to the control V5 line, the two experimental lines expressing the highest levels of the Fn14-HA protein (lines 11 and 10) exhib-ited decreased adhesion to fibronectin, while all three experimental cell lines (lines 3, 11, and 10) exhibited decreased adhesion to vitronectin. Adhesion of the Fn14-HA 10 cell line on fibronectin and vitronectin was reduced by ϳ42% and ϳ74%, respectively. We subsequently performed an additional set of adhesion assays in which the control V5 and the Fn14-HA 10 cell lines were plated onto increasing concentrations of vitronectin. For both cell lines, the extent of cell adhesion increased in a dose-dependent manner; however, in comparison to the control cell line, the Fn14-HA-expressing cell line exhibited decreased adhesion at all of the vitronectin concentrations tested (Fig. 9B). We have also observed that when the V5 and Fn14-HA 10 cells are plated on vitronectin and examined ϳ4 h later, the V5 cells are spread out and display a typical fibroblast morphology while the Fn14-HA-expressing cells are more spherical in shape (data not shown).
Second, since cell-extracellular matrix interactions play a key role in cellular proliferation and migration, we determined whether constitutive Fn14 expression altered serum-stimulated NIH 3T3 cell growth or motility. When the growth prop-   3, 11, and 10) were harvested, and cell lysates were prepared. Equivalent amounts of protein were subjected to SDS-PAGE and Western blot analysis using anti-HA monoclonal antibodies. Molecular masses of protein size standards (in kDa) are shown on the left. B, the Fn14-HA expression data shown in panel A were quantified by densitometry (only the 23-kDa species) and plotted as fold increase over the background signal found in vector-transfected cells. C, the pcDNAIneo V5 and pcDNAIneo/Fn14-HA 10 cell lines were surface biotinylated, cell lysates were prepared, and immunoprecipitation analysis was performed using either control IgG or anti-HA IgG. Biotinylated plasma membrane proteins were visualized using HRP-conjugated streptavidin. Molecular masses of protein size standards (in kDa) are shown on the left. erties of the control V5 cell line and the Fn14-HA 10 cell line were compared in serum-containing medium using a colorimetric assay, we observed that the Fn14-HA-expressing cell line had a statistically significant decrease in growth rate. In comparison to the V5 cells, growth of the Fn14-HA 10 cells was reduced by ϳ60% after 4 days of culture (Fig. 10A). Similar results were obtained when cellular proliferation was measured by directly counting viable cells (data not shown). These same two cell lines were then used in cellular migration assays. Each cell line was resuspended in serum-free medium and plated in the upper compartment of Transwell inserts, and their migration through a porous membrane in response to either serum-free medium or serum-containing medium was measured. When serum-free medium was present in both the upper and lower compartments, no cellular migration occurred (data not shown). When serum-containing medium was placed in the lower compartment, both the control and experimental cells exhibited chemotactic migration; however, migration of the Fn14-HA-expressing cell line was ϳ40% of that observed with the control cell line (Fig. 10B). DISCUSSION The addition of serum or polypeptide growth factors to quiescent fibroblast cultures initiates an intracellular signal transduction cascade that promotes the transcriptional activation of numerous cellular genes (1-3). These genes are generally classified as either immediate-early, delayed-early, or late response genes, and they encode proteins with diverse functions (2, 3). We have used a differential display approach to identify FGF-1-inducible genes in NIH 3T3 cells (9), and here we describe the cloning, chromosomal location, and expression pattern of the Fn14 gene. We also report the biological effects of constitutive Fn14 expression in transfected cells.
The original Fn14 cDNA was isolated by differential display using degenerate sense and antisense oligonucleotide primers designed to anneal to DNA sequences encoding conserved amino acid motifs found in protein kinase domains and leucine zipper domains, respectively. We found that neither of these structural domains were present in the predicted Fn14 sequence. This result was not unexpected since comparison of the PCR primer sequences with the corresponding regions of the Fn14 cDNA sequence indicated that both primers annealed to regions of relatively low sequence identity. The Fn14 gene is predicted to encode a type Ia transmembrane protein containing a 27-aa signal peptide, a 53-aa extracellular domain, a 21-aa transmembrane domain, and a 28-aa cytoplasmic domain. The mature 102-aa Fn14 polypeptide is predicted to have a molecular mass of ϳ10.8 kDa. Sequence motifs present in the Fn14 sequence include a cluster of charged amino acids following the predicted membrane-spanning domain and a L-I dipeptide in the cytoplasmic tail. The first motif may function as a stop-transfer signal during membrane insertion, and the second motif could serve as an endosomal targeting signal (33)(34)(35).
The deduced Fn14 protein sequence was compared with the GenBank sequence data base, and no significant sequence identity was found between Fn14 and known proteins. However, a pattern search using Fn14 transmembrane domain residues revealed some sequence similarity to the single transmembrane domain found in the syndecan family of heparan sulfate proteoglycans (38,39). The highest overall amino acid sequence identity in this domain, ϳ33%, was between Fn14 and murine syndecan-1. Of particular interest is that (i) Fn14 and all four syndecan family members contain a G-G dipeptide motif in their transmembrane domains, even though glycine residues are not generally found in membrane-spanning sequences (36) and (ii) Fn14 contains a pattern of bulky and small side chain residues that is similar, but not identical, to the pattern found in the syndecan transmembrane domains (37). It has been demonstrated by site-specific mutagenesis that this pattern is critical for the formation of noncovalent, SDS-resistant syndecan-3 dimers and oligomers (37). It is presently unknown whether the Fn14 protein, like the syndecans, can selfassociate, but it does not migrate at its predicted molecular weight when analyzed by SDS-PAGE (see below).
The Fn14 gene is located on mouse chromosome 17, in the middle of the T-locus. We have compared our interspecific map of chromosome 17 with a composite mouse linkage map that reports the map location of many uncloned mouse mutations (Mouse Genome Data Base). Fn14 mapped in a region of the composite map that lacks mouse mutations with a phenotype that might be expected for an alteration in this locus. The proximal region of mouse chromosome 17 shares a region of homology with human chromosomes 6 and 16 (summarized in Fig. 2). In particular, E4fl has been assigned to human 16p13.3-p13.2. The tight linkage between E4fl and Fn14 in mouse suggests that the human homolog of Fn14 will map to this same chromosomal location. Consistent with this possibility, a recent search of the GenBank data base using the murine Fn14 nucleotide sequence revealed a significant degree of sequence relatedness between Fn14 and human chromosome 16p13.3 genomic DNA (GenBank accession no. AC004643).
The Fn14 gene can be classified as a growth factor-inducible, immediate-early response gene in vitro that is expressed in a developmental stage-and tissue-specific manner in vivo. FGF-1 stimulation of quiescent fibroblasts rapidly increases Fn14 mRNA levels, with peak expression detected at 4 h. It is likely that this response is due, at least in part, to transcriptional activation of the Fn14 gene since Fn14 mRNA accumulation does not occur in the presence of an RNA synthesis inhibitor. Fn14 mRNA levels remain elevated for a significant period of time after mitogen addition, suggesting that Fn14 gene transcription is somewhat sustained and/or Fn14 mRNA is relatively stable. In regard to this second possibility, we have noted that the Fn14 mRNA 3Ј-untranslated region contains one copy of an AU-rich motif implicated in rapid mRNA decay (30), and therefore this transcript may have a short half-life. However, nuclear run-on assays and mRNA stability measurements are required in order to determine the precise molecular basis for the Fn14 mRNA temporal expression pattern that is observed. FGF-1 induces Fn14 mRNA levels in the presence of a protein synthesis inhibitor; thus, Fn14 gene activation does not require the de novo synthesis of intermediary proteins. The simultaneous addition of FGF-1 and cycloheximide promoted Fn14 mRNA superinduction and, in addition, cycloheximide treatment alone induced Fn14 mRNA levels. These cycloheximide effects are likely to occur because the drug is preventing the synthesis of labile proteins required for Fn14 transcriptional repression and/or Fn14 mRNA decay. We found that, like the majority of the immediate-early genes identified to date, the Fn14 gene can be induced by various polypeptide growth factors as well as by PMA, a tumor-promoting phorbol ester that activates protein kinase C. Finally, Fn14 transcripts are expressed in mice at a relatively high level in the maternal decidual tissue of the developing embryo, in many of the major organs of newborn animals, and in the adult heart, kidney, lung, ovary, and skin.
We generated Fn14 polyclonal antiserum and then performed Western blot analysis to investigate whether the mitogenic stimulation of NIH 3T3 cells increased Fn14 protein expression. FGF-2 treatment of quiescent cell cultures induced the accumulation of a ϳ22-kDa immunoreactive protein, and the temporal expression kinetics observed were consistent with the kinetics of Fn14 mRNA expression in FGF-1-treated cells. Although this immunoreactive protein migrated at an apparent molecular mass twice the predicted mass of mature Fn14 (ϳ10.8 kDa), we believe it is likely to be Fn14 for two reasons. First, when Fn14 cDNA is expressed in baculovirus-infected insect cells, the 129-aa protein is processed and mature Fn14 migrates an apparent molecular mass of ϳ23 kDa (data not shown). Second, when an HA epitope-tagged Fn14 protein predicted to be ϳ11.9 kDa in size is expressed in transfected NIH 3T3 cells, the major Fn14-HA protein species detected is ϳ23 kDa in size (see Fig. 8). Fn14 may be migrating at a higher apparent molecular mass when analyzed by SDS-PAGE due to post-translational modifications, self-association into reducing agent/SDS-resistant dimers or an amino acid composition (e.g. high proline content) that causes the protein to resist denaturation or to bind SDS poorly. In regard to the first possibility, this explanation is unlikely because (i) there are no consensus sequence motifs for N-glycosylation or glycosaminoglycan attachment in the predicted Fn14 protein, (ii) the major Fn14 protein synthesized in an in vitro transcription/translation system also migrates at ϳ22 kDa (data not shown), and (iii) full-length Fn14 expressed in bacterial cells also migrates at a higher apparent molecular mass than predicted from the cDNA sequence when it is analyzed by SDS-PAGE (data not shown). In regard to the second possibility, although it is possible that Fn14 could be forming dimers, especially when one considers that the Fn14 transmembrane domain has some sequence identity to the region of the transmembrane domain implicated in syndecan-3 self-association (37), we have never detected Fn14 migrating at a molecular mass indicative of a monomeric molecule or of oligomeric complexes. Thus, we presently favor the third possibility, anomalous migration due to amino acid composition, as the likely explanation for the higher than predicted apparent molecular mass of the Fn14 protein.
Two independent experimental approaches were used to demonstrate that Fn14 is a plasma membrane protein. Fluorescence microscopy analysis of Fn14-EGFP localization in transfected NIH 3T3 cells indicated that Fn14 was concentrated in the trans-Golgi network (probably due to inefficient transport of overexpressed protein through the secretory pathway) and in areas along the cell periphery that could represent regions of cell-substratum attachment; however, additional colocalization experiments are required to confirm that these are in fact focal adhesion sites. Fn14 was also present on the cell surface in thin membrane protrusions resembling actin microspikes or filopodia, which are believed to act as sensory structures that play a role in the control of cell growth and migration (40 -42). Fn14 was also localized to the plasma membrane of a Fn14-HA-expressing cell line using cell surface biotinylation followed by immunoprecipitation and Western blot analysis.
NIH 3T3 stable cell lines expressing HA epitope-tagged Fn14 were isolated in order to examine the biological consequences of constitutive Fn14 expression. We are aware that a limitation of this experimental approach is that Fn14 may exhibit abnormal behavior when overexpressed in cells; therefore, we have analyzed several independent cell lines with varying levels of expression. Since Fn14 was present on the cell surface, we initially assayed cellular adhesiveness to several immobilized extracellular matrix proteins. Cellular adhesion to the extracellular matrix is mediated primarily by the integrin family of heterodimeric transmembrane proteins (43,44). Neither the vector control nor the three Fn14-HA-expressing cell lines adhered significantly above background levels when plated on collagen I, collagen IV, or laminin. This result is consistent with previous reports (45)(46)(47) and our results (data not shown) demonstrating that the ␣ 1 and ␣ 2 integrin subunits (which represent two of the major ␣-integrin subunits which bind collagen and/or laminin) are expressed at relatively low levels in NIH 3T3 fibroblasts. Both the control and the Fn14-HAexpressing cell lines could adhere to fibronectin or vitronectin, consistent with previous studies (45)(46)(47)(48) and our results (data not shown) demonstrating ␣ 3 /␤ 1 , ␣ 5 /␤ 1 , and ␣ V /␤ 1 expression in NIH 3T3 fibroblasts. In comparison to the control V5 line, the two experimental cell lines expressing the highest Fn14-HA levels showed decreased attachment to fibronectin while all three Fn14-HA-expressing lines showed decreased attachment to vitronectin. In these assays, the level of Fn14-HA expression did not strictly correlate with the loss of cellular adhesiveness. This may reflect a saturation of the inhibitory effect at intermediate (for fibronectin binding) or low (for vitronectin binding) levels of ectopic expression. Additional experiments using only the control cell line and the experimental cell line expressing the highest level of Fn14-HA revealed that constitutive Fn14 expression inhibited cellular adhesion to various concentrations of immobilized vitronectin.
Nontransformed cells in culture are dependent on interactions with an adhesive surface for both cellular proliferation (49 -52) and migration (42,53,54). Therefore, we performed serum-stimulated growth and chemotactic migration assays using the control cell line and the experimental cell line expressing the highest level of Fn14-HA. In these growth and migration experiments, it is likely that serum-derived vitronectin is the major attachment factor adsorbed to the tissue culture plastic or Transwell nucleopore membranes, respectively (55). We found that constitutive Fn14 expression inhibited both serum-stimulated NIH 3T3 cell growth and motility.
In conclusion, we have identified a growth factor-and tumor promoter-inducible immediate-early response gene on mouse chromosome 17, designated Fn14, which encodes a novel plasma membrane protein. Constitutive Fn14 expression in transfected NIH 3T3 cells reduces cellular adhesion to fibronectin and vitronectin and also inhibits serum-stimulated cell proliferation and migration. The molecular basis for these effects is presently under investigation, but we propose that Fn14 could be altering integrin subunit expression, ligand binding, or signal transduction. This could occur by direct binding of Fn14 to integrin subunits; alternatively, Fn14 could interact with other transmembrane proteins that mediate integrin function; for example, integrin-associated protein CD47 (56), calveolin (57), syndecan-2 or -4 (58,59), or members of the tetraspanin family (60). Finally, it is possible that the transient expression of this protein in growth factor-stimulated cells may be important for the "de-adhesion" process associated with cell division and motility (61).