Coordinated Induction of VEGF Receptors in Mesenchymal Cell Types During Rat Hepatic Wound Healing

Oncogene (1998) 17, 115 ± 121 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Coordinated induction of VEGF receptors in mesenchymal cell types during rat hepatic wound healing

V Ankoma-Sey1,3, M Matli1, KB Chang1, A Lalazar1, DB Donner2, L Wong1, RS Warren1 and SL Friedman1

1UCSF Liver Center & Departments of Medicine & Surgery, University of California, San Francisco, California; 2Department of Physiology and Biophysics and the Walther Oncology Center, University of Indiana, Indiapolis, Indiana, USA

Homology PCR has been used to identify receptor transition which is similar regardless of the type of tyrosine kinases (RTKs) expressed during activation of liver injury, and which re¯ects a cascade of gene rat hepatic stellate cells, the key ®brogenic mesenchymal induction comprising both regulatory and structural element in the liver. Partial cDNAs encoding several components (Friedman, 1996). RTKs were cloned from stellate cells activated in vivo, An important role in stellate cell activation has including those of Flt-1, Flk-1, c-met, PDGFR, and emerged for receptor tyrosine kinases (RTKs), signal- Tyro10/DDR2. RNAse protection from cells activated in ing molecules which underlie cell growth and differ- vivo demonstrated biphasic induction of ¯t-1 and ¯k-1 entiation in all tissues. Many of the phenotypic features mRNAs, receptors for vascular endothelial growth factor of stellate cell activation are mediated by signaling (VEGF). Culture-activation of stellate cells was asso- through RTKs. For example, stellate cell proliferation ciated with increased [125I]VEGF binding and Flt-1 and in culture and in vivo is preceded by rapid induction of Flk-1 receptor protein. Induction of VEGF binding sites the b subunit of platelet derived growth factor receptor correlated with an 2.5-fold increase in DNA synthesis in (PDGFR) (Wong et al., 1994). Similarly, basic response to VEGF, but only if cells were activated by ®broblast growth factor (bFGF), another RTK growth on collagen I, whereas cells maintained in a ligand, is produced by and stimulates proliferation of quiescent state on a basement membrane-like substratum stellate cells (Hioki et al., 1996; Rosenbaum et al., (EHS matrix) were nonproliferative. In both stellate and 1995). FGF receptors have been identi®ed on stellate endothelial cells VEGF-induced mitogenesis was aug- cells, and bFGF also mediates the mitogenic eect of mented by co-incubation with basic ®broblast growth transforming growth factor b1 (TGFb1) in activated factor (bFGF), a cytokine with known synergy with human stellate cells (Rosenbaum et al., 1995). In vivo, VEGF. These ®ndings suggest that the cellular targets of FGF may stimulate extracellular matrix production VEGF in liver may not be con®ned to sinusoidal and vascularization following liver injury by both endothelial cells, and that VEGF responses re¯ect autorcrine and paracrine mechanisms (Hioki et al., combined eects on both hepatic stellate cells and 1996). sinusoidal endothelium. The general organization of RTKs is well established. All RTKs are composed of three major Keywords: hepatic stellate cells; sinusoidal endothelium; domains: a highly conserved cytoplasmic tyrosine hepatic ®brosis; receptor tyrosine kinases kinase domain, comprising about 200 ± 300 amino acids (Hanks et al., 1988), a single transmembrane spanning domain and a highly diverse extracellular region (reviewed in Fantl et al., 1993). Binding by Introduction ligands elicits a diverse range of cellular responses depending on the cell-type and the speci®c RTK A central event in the pathogenesis of tissue ®brosis is (Schlessinger, 1993). Responsiveness of cells to RTK the activation of resident perivascular mesenchymal ligands may also be regulated by extracellular matrix cells which proliferate and produce extracellular matrix (ECM), which can modulate RTK abundance or in response to injury. In the liver, the hepatic stellate function (Xu and Clark, 1996). Because changes in cell ful®ls this role (Friedman, 1993b). These perisinu- ECM composition are a major component of liver soidal vitamin A-storing cells are widely distributed injury, it is possible that one consequence of ®brosis throughout the normal liver, and transform during may be altered RTK expression and/or signaling. injury into a ®brogenic cell (Friedman et al., 1985; Given the importance of RTKs to wound healing in Gressner, 1995) which is proliferative (Geerts et al., liver, in this study we exploited the technique of 1991) and contractile (Kawada et al., 1992; Rockey et homology PCR to clone RTKs expressed during al., 1993). Stellate cell activation is a phenotypic stellate cell activation in order to de®ne their role in liver injury, and as a model for how RTKs might contribute to tissue repair in other organs. Because this Correspondence: SL Friedman, Box 1123, Division of Liver Diseases, approach yielded the unexpected ®nding of receptors Mount Sinai School of Medicine, 1425 Madison Ave, Room 11-70F, for vascular endothelial growth factor (VEGF) in New York, NY 10029, USA stellate cells, we have speci®cally explored the 3Current address: Division of Gastroenterology, University of Texas potential role of VEGF signaling in mediating features at Houston Medical School, 6431 Fannin, MSB 4.23, Houston, Texas of stellate cell activation, and have compared their 77030, USA Received 7 October 1997; revised 17 February 1998; accepted 18 response to sinusoidal endothelial cells, the other February 1998 mesenchymal cell type of the hepatic sinusoid. Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 116 lial cells, which are known to express receptors for Results VEGF (Yamane et al., 1994). Speci®c binding of [125I]VEGF to freshly isolated stellate cells and SECs Homology PCR cloning of RTKs in activated stellate increased with ligand concentration and was saturable. cells The Scatchard plots (Figure 2) demonstrated two To identify cDNAs expressed during activation of classes of high anity binding sites/receptors on each stellate cells, homology cloning was performed accord- cell type (Table 2). ing to the strategy of Raz et al. (1991). The template was a rat stellate cell activation-induced cDNA library VEGF activity and eect of extracellular matrix in derived by subtraction cloning from stellate cells cultured stellate cells isolated 3 h after a single dose of intragastric CCL4 (Lalazar et al., 1997). The PCR cloning strategy We and others (Friedman et al., 1989; Davis, 1988) yielded 64 clones representing six dierent RTK have previously demonstrated that the composition of genes, all containing the expected *200 bp insert (see the extracellular matrix determines whether stellate Table 1). These included receptors for vascular cells will activate in primary culture. Modulation of endothelial growth factor, hepatocyte growth factor, VEGF binding during stellate cell activation was platelet derived growth factor, and an orphan receptor explored by allowing the cells to activate through (Tie-1). Tie-1 is expressed on human hematopoietic growth on type I collagen, an established model of progenitor cells, endothelial cells and megakaryocytes culture activation (Friedman et al., 1989). [125I]VEGF (Batard et al., 1996; Rodewald and Sato, 1996) and binding sites/cell were determined in stellate cells and plays an important role in vasculogenesis (Dumont et SECs after 1 and 10 days in cultlure on collagen I. al., 1993; Sato et al., 1995). In addition to these known Maximal VEGF binding by stellate cells occurred RTKs. a novel RTK, clone `RTK40' (Genbank following 10 days in culture associated with expression Accession no. AF016247), was recovered whose of Flt-1 and Flk-1 by Western blot (Figure 3); the tyrosine kinase sequence assigns it to a new subfamily which have a lectin-like domain in their extracellular regions (other members include Tyro 10/DDR2) (Johnson et al., 1993; Lai and Lemke, 1994). a

Characterization of 125I VEGF binding sites in stellate cells The cloning of VEGF receptor cDNAs from stellate cells was surprising given its presumed restriction to endothelial cells. To con®rm expression of ¯k-1 and ¯t- 1 mRNA by stellate cells we analysed their expression in pure cell isolates after activation in vivo by

administration of CCl4. There was biphasic expression of the two VEGF receptors in stellate cells during injury: ¯t-1 mRNA was modestly induced beginning

3 h after CCl4 (Figure 1) whereas ¯k-1 mRNA was down-regulated initially. However, both transcripts were markedly induced in vivo beginning 96 h after a

single dose of CCl4. This time course of induction diers from that of the b-PDGF receptor which is induced within 1 h under identical conditions (Wong et al., 1994 ). b To further characterize VEGF binding sites on stellate cells, saturation binding studies were performed and compared to those of sinusoidal endothe-

Table 1 Receptor tyrosine kinase cDNAs identi®ed in activated hepatic stellate cells by homology PCR Clone % nt homology name Homologies to (Genbank #) Ligand RTK 17 Flt-1,4 95 to mouse (L07297) VEGF RTK 19 Tie-1 95 to mouse (X73960) Unknown RTK 40 Tyro 10/DDR2 91 to mouse (X76505) Collagens RTK 47 c-met 89 to mouse (S52036) HGF Figure 1 Flt-1 and Flk-1 mRNA in hepatic stellate cells after RTK 48 PDGF-R 85 to mouse (X04367) PDGF CCl4-induced liver injury. (a) ¯t-1 and ¯k-1 mRNA expression RTK 61 Flk-1 89 to mouse (X59397) VEGF were analysed by RNAse protection assay in 10 mg total RNA Six distinct 200 bp cDNAs were represented among 64 clones from puri®ed rat hepatic stellate cells at intervals after a single analysed. GenBank accession numbers listed refer to closest dose of CCl4. S14, a ribosomal protein mRNA, is shown as an nucleotide homologies based on FASTA sequence search. The clone internal control. (b) Normalized relative expression of ¯t-1 and RTK40 sequence has been entered into GenBank, accession number ¯k-1 mRNAs from a, expressed as fold increase compared to cells AF016247 from normal animals Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 117 larger increase in Flk-1 vs Flt-1 further suggests that remain quiescent (Friedman et al., 1989). Mitogenesis Flk-1 may correspond to receptor `B' in Table 2. of HSCs increased 2.5-fold in response to VEGF Whereas cells plated on either uncoated plastic or (50 ng/ml) after 10 days in culture on plastic, and was type I collagen activate progressively, those on a augmented by co-incubation with basic ®broblast basement membrane substratum resembling the nor- growth factor (bFGF), a cytokine with known synergy mal subendoethelial matrix of liver (EHS matrix) with VEGF (Figure 4). VEGF-induced proliferation was more prominent in SECs than in stellate cells: there was an eightfold increase in [3H]thymidine incorporation in SEC cultured for 10 days from a normal liver (50 ng/ml) but only if cells were plated

Figure 2 Speci®c binding of [125I]VEGF to (a) sinusoidal endothelial cells and (b) hepatic stellate cells. All data points represent the mean of triplicates, and similar results were obtained in at least three dierent experiments

Table 2 [125I]VEGF binding to hepatic stellate and sinusoidal endothelial cells in early primary culture Binding Cell type sites per cell Anity (Kd) Stellate cells: A 4334 5 pM Figure 3 Induction of [125I]VEGF binding sites and receptor in B 34 208 314 pM culture activated hepatic stellate and sinusoidal endothelial cells. (a) Day 1 and day 10 primary cultures of the two cell types were Sinusoidal Endothelial cells analysed by [125I]VEGF saturation binding following plating on A 3557 11 pM collagen I. Binding sites/cell is shown for the various culture B 11 094 480 pM conditions. The data represent mean and s.e. of three experiments Cells in 24 well dishes 48 h after plating were analysed by saturation (triplicates were run for each experimental condition). (b) Western binding as described in Materials and methods. Each cell type blot for Flt-1 in stellate cells at days 1, 4 and 10 in primary displayed two distinct populations of binding sites listed as `A' and culture. Both the upper and lower bands represent speci®c signals. `B'. Data represents mean values of three experiments (c) Western blot for Flk-1 in stellate cells at days 1, 4 and 10 Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 118 cells in culture or in vivo, or sinusoidal endothelial cell behavior in vivo has not been explored. The identi®cation of VEGF receptors ¯k-1 and ¯t-1 in stellate cells, combined with earlier studies documenting VEGF expression in hepatocytes (Mo- chida et al., 1996; Suzuki et al., 1996), establishes the components of a potential paracrine loop between hepatocytes and stellate cells. After the onset of liver

injury due to CCl4 there is an early, modest induction of ¯t-1 mRNA in stellate cells; in contrast, ¯k-1 mRNA is slightly downregulated, then both transcripts are markedly induced at a later interval (96 h). This pattern of expression in stellate cells is distinct from b- PDGF receptor mRNA, which is rapidly induced Figure 4 Matrix dependency of VEGF induced hepatic stellate within 1 h (Wong et al., 1994). The dierent behavior cell proliferation. Stellate cells were grown on either collagen type I or basement membrane-like EHS matrix for 6 days in the of ¯t-1/¯k-1 mRNAS compared to b-PDGF receptor presence of 20% FCS, serum starved for 48 h, and then treated underscores the concept that stellate cell activation with VEGF with or without bFGF for 48 h (without serum), comprises a programmed, sequential cascade of gene followed by measurement of [3H]thymidine incorporation. Data expression rather than a generalized transformation represent the mean of two experiments, each performed in triplicate and expressed as percent of control cells without added where many genes are simultaneously induced. More- VEGF; standard error of the mean was less than 10% over, it suggests that angiogenic responses mediated by VEGF may persist beyond the phase of acute tissue repair, since much of the toxic damage is repaired at this later interval. Regardless, it is clear that signaling on collagen I, whereas cells grown on basement associated with VEGF in wound healing extends to cell membrane-like matrix (EHS) were nonproliferative. types apart from endothelial cells. Indeed, ¯t-1 has VEGF and bFGF were also synergistically mitogenic recently been detected in capillary pericytes (Nomura et in SECs, as in HSCs (data not shown). al., 1995) and mesangial cells (Takahashi et al., 1995), cell types which are functionally equivalent to stellate cells. Discussion In addition to its role in tissue repair, angiogenesis is also critical to tumor formation. The recognition that The current study has addressed the potential role of hepatic stellate cells respond to angiogenic stimuli RTKs in the phenotype of activated stellate cells, the could imply a potential role for this cell type in tumor liver's principal ®brogenic element. By utilizing growth, since activated stellate cells have been homology PCR we have successfully identi®ed several identi®ed in the stroma of human liver tumors RTK transcripts expressed during stellate cell activa- (Enzan et al., 1994). Moreover, upregulation of VEGF tion, including those of receptors for VEGF (Flk-1 and and bFGF mRNAs have been documented in primary Flt-1), HGF, PDGF, as well as an RTK (`RTK40') (Mise et al., 1996; Suzuki et al., 1996) and experimental with homology to a new subfamily of receptors metastatic liver tumors (Warren et al., 1995). containing lectin-like domains, as exempli®ed by Tyro In this study as in our previous work, culture models 10 (Lai and Lemke, 1994). Our strategy has uncovered of activation were used to complement in vivo ®ndings. low-abundance RTK transcripts whose expression may Associated with culture-induced activation is a marked have previously been ascribed only to other cell types. increase in VEGF binding sites and receptor on stellate Given PCR's sensitivity, however, it was essential to cells without signi®cantly altering anity, suggesting con®rm that the RTKs identi®ed were truly expressed that receptor number may correlate with stellate cell in stellate cells, as we have done for ¯k-1 and ¯t-1, and activation. We have previously documented a similar RTK40 (not shown). A recent report (Ikeda et al., response in stellate cells for TGFb1 receptors (Fried- 1996) further validates our approach by also identify- man et al., 1994). ing c-met mRNA in stellate cells, which in liver had What are the implications of VEGF acting on been previously thought to be con®ned to hepatocytes. stellate cells? This issue can be addressed by reviewing The liver's response to injury can be viewed as an the known pleiotropic actions of VEGF associated angiogenic one, with new blood vessel formation, with angiogenesis. These activites include increased cell proliferation of mesenchymal elements and extracellu- proliferation, vascular permeability, migration and lar matrix deposition as prominent features. Signaling protease expression (Connolly et al., 1989; Leung et by the RTK for vascular endothelial growth factor al., 1989). Proliferation has been documented in this (VEGF) is a critical component of angiogenesis in study. While it is unclear how VEGF could alter other tissues, but has not been studied in liver injury. sinusoidal permeability via eects on stellate cells, a At least two major vascular endothelial growth factor role in regulating protease release is quite plausible. receptors (VEGFRs) have been cloned and character- For example, VEGF induces the expression of serine ized: Flk-1 (Fetal liver kinase 1) and Flt-1 (Fms-like proteases urokinase-type and tissue-type plasminogen kinase 1) (de Vries et al., 1992; Mustonen and Alitalo, activators (PA) and also PA inhibitor-1 (PAI-1) in 1995; Quinn et al., 1993; Terman et al., 1992). One cultured bovine microvascular endothelial cells and report (Yamane et al., 1994) has suggested that VEGF these components of the plasminogen system have also is proliferative toward cultured sinusoidal endothelial been identi®ed in stellate cells (Pepper et al., 1991). cells, but the importance of VEGF to either stellate Furthermore, VEGF induces the expression of the Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 119 metalloproteinase interstitial collagenase in human (Proctor and Chatamra, 1982). Cells from these animals umbilical vein endothelial cells, a function also were isolated using the same method as for normals, as ascribed to stellate cells (Unemori et al., 1992). Thus, previously described (Wong et al., 1994). Stellate cells our ®nding of VEGF responsiveness in stellate cells maintained in culture were grown on either uncoated points to a similar role in pericytes and related tissue plastic, on plastic coated with a collagen I or a basement membrane-like substratum (ÈHS matrix') as previously myo®broblasts, an issue not previously explored. described (Friedman et al., 1989). Another novel aspect of our work was the demonstration that responsiveness to VEGF in stellate cells was activation-dependent, which in turn is mRNA analysis/RNAse protection assay in¯uenced by the extracellular matrix. This observa- Total RNA was isolated from cell pellets by using tion has direct relevance to matrix remodeling during Trireagent (MRC) according to manufacturer's recommen- liver injury and wound healing. In exploring this issue dations. Aliquots of total RNA were incubated with 0.5 ± we utilized two dierent matrices which model the 1.06106 Cerenkov c.p.m. of labeled probe, denatured at normal and injured liver, respectively: EHS matrix, 788C, and hybridized in solution at 558C overnight, for all which mimics the subendothelial basement-membrane probes as previously described; hybridization to the ribosomal S14 mRNA was used as an internal standard matrix and maintains resident sinusoidal cells in a (Maher and McGuire, 1990). The RNA/probe complexes dierentiated state (Friedman et al., 1989); and were digested with T2 RNAse for 60 min at 308C, and the collagen I, which predominates in the ìnjury' milieu protected probes denatured and resolved on a 5% of the damaged liver. The increased responsiveness to polyacrylamide sequencing gel. The gel was dried and VEGF during prolonged culture likely re¯ects, at least exposed overnight to X-ray ®lm, and speci®c bands were in part, the eects of a changing microenvironment on quantitated by densitometry. Each assay included a sample cell activation. Similar ®ndings are now emerging in of probe incubated with yeast tRNA as a negative control. related cell types. In microvascular endothelial cells for example, the eect of TFGb1 on cell proliferation is DNA sequencing and analysis matrix-dependent (Sankar et al., 1996). The synergy of VEGF by bFGF in stellate cells on Sequencing was performed by the UCSF Biomolecular collagen I is also compatible with recent insights into Resource Center using a Taq polymerase based system and how extracellular matrix can modulate cell function. an automated sequencer by Applied BioSciences. cDNA inserts in pGEM4 vector in water were used for sequencing bFGF is normally sequestered by the ECM, and during from the SP6 and T7 polymerase sites. For high quality injury it is released available to the adjoining cells (Levi sequencing prior to data entry into GenBank we con®rmed et al., 1996; Vlodavsky et al., 1996). Thus, matrix the automated sequence using a commercial kit according degradation in liver may present additional bFGF to to manufacturer's speci®cations (Sequenase 2.0, US stellate and endothelial cells coincident with increased Biochemical). Dried gels were exposed overnight to X-ray responsiveness via VEGF receptor induction. Since ®lm and manually read. The GenBank FASTA program stellate cells are the primary source of bFGF during was used for homology searches. liver injury (Hioki et al., 1996) an autocrine loop for bFGF activity may be established during liver injury. Polymerase chain reaction Moreover, VEGF is known to increase production of For reverse transcriptase PCR (RT ± PCR), ®rst strand bFGF (Stavri et al., 1995). Collectively, these ®ndings cDNA was generated with avian myeloblastosis virus suggest stellate cell mitogenesis likely results from both reverse transcriptase using 2 mg mRNA and oligo-dT autocrine and paracrine interactions. primer (Gibco BRL). This template cDNA was incubated In summary, our ®ndings demonstrate for the ®rst with Taq polymerase in the presence of 5 mM nucleotides, time that VEGF receptors are induced during stellate magnesium-containing reaction buer (2 mM), and primers cell activation and that reponsiveness to VEGF is containing restriction sequences (XbaI and EcoRI) to matrix-dependent. This observation points toward the facilitate subcloning of ampli®ed products. The regions of possibility of neutralizing RTK activity as a therapy interest were ampli®ed for 30 cycles in a thermal cycler for hepatic ®brosis associated with injury or neoplasia. (Ericomp) with annealing temperatures ranging from 42 ± 588C, determined empirically, and extension time from 1 ± 3 min. Ampli®ed products were visualized on a 1 ± 1.5% low-melt agarose gel and the excised bands of interest puri®ed using a commercial kit (Promega Magic PCR). All experiments contained a negative control in which template Materials and methods was omitted. Positive control template DNA (i.e., known to contain the sequence of interest) were included as Isolation and primary culture of pure cell populations from appropriate. normal and injured rat liver Hepatic stellate cells were isolated from retired breeder Sprague-Dawley rats (450 ± 600 g) by in situ portal vein Homology PCR of receptor tyrosine kinases (RTKs) induced perfusion with pronase and collagenase, followed by during stellate cell activation in vivo density gradient centrifugation with arabinogalactan Activation-induced genes in stellate cells were isolated by (Friedman, 1993a). Sinusoidal endothelial cells (SECs) subtraction cloning from cells activated in vivo by were further puri®ed from the lower layers of the density intragastric administration of CCl4 as previously described gradient by centrifugal elutriation (Jarnagin et al., 1994). (Lalazar et al., 1997). In brief, cDNA libraries were All cells were cultured on plastic dishes in the presence of constructed from both freshly isolated stellate cells from 20% serum (10% calf : 10% horse). For experiments normal rats (i.e. quiescent cells) and from rats treated 3 h requiring the isolation of cells from injured liver, animals earlier with the hepatotoxin CCl4 (i.e. àctivated cells'). were administered 0.5 ml carbon tetrachloride (CCl4) per mRNA was extracted directly from quiescent and activated gram body weight by gavage, as described by Proctor stellate cells using a commercial kit (Invitrogen). Following Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 120 biotinylation of the quiescent stellate cells-cDNA library Western blot and the addition of linkers speci®c for each population, the Stellate cells in primary culture for the indicated intervals two cDNA populations were exhaustively hybridized to were harvested in lysis buer (20 mM Tris-HCl, pH 7.9; one another. Duplexes containing quiescent-speci®c 1% Triton X-100, 137 mN NaCl; 10% glycerol; 1 mM cDNAs were removed by binding to streptavidin, and PMSF; 1 mg/ml aprotinin; 1 mg/ml leupeptin) and the activation-speci®c (i.e. unhybridized) cDNAs were selec- protein content was determined by commercial assay tively ampli®ed by polymerase chain reaction. The (Coomassie Plus protein assay reagent, Pierce Chemical, activation-speci®c clones were packaged using a commer- Rockford, IL). For Flt-1, 20 mg protein per lane was cial packaging mix (Packagene) and transfected into LE separated on 6% SDS ± PAGE under reducing conditions, 392 bacteria; packaging eciency was calculated at transferred to nitrocellulose and incubated overnight at *76106 p.f.u. per mg of DNA insert. 48c in TBS-Tween containing a polyclonal antibody to Using this activation-speci®c cDNA library as template, human Flt-1 (1 : 1000) (Santa Cruz Biotechnology, catalog PCR was performed with RTK-speci®c primers based on a #sc-316). Western blot for Flk-1 was performed in 200 bp region within the tyrosine kinase domain which is separate stellate cell isolates at the same plating conserved in the RTK superfamily. Primers were derived from intervals, using 10 mg protein per lane (1 : 500) (Santa domains VI and IX of tyrosine kinases (Hanks et al., 1988) and Cruz Biotechnology, catalog #sc-504). Following three were nearly identical to those previously described by Raz et washes, the blots were incubated in secondary antibody, al. (1991), but with slightly more codon degeneracy and anti-rabbit Ig-HRP-linked whole antibody from donkey dierent restriction sites (XbaI/EcoRI). The sequences of (1 : 1000) (Amersham) for 1 h at room temperature, and primers used were: (i) TK6 (forward): 5'-TCTAGA- proteins visualized using a commercial chemiluminiscence CA[TC][CA]GIGA[TC][CT]TIGC-3' (underlined bases de- kit (SuperSignal chemiluminescent substrate, Pierce note XbaI sequence; [ ] represents a mix of nucleotides; I Chemical). represents inosine); (ii) TK9 (reverse): 5'-GAATTCCC[AG]- [AT]A[GA]-[GC]ACCA[GC]AC[AG]T-3' (underlined bases denote EcoRI sequence; [ ] represents a mix of nucleotides). Cell proliferation Ampli®cation conditions were identical to those described To measure cell proliferation, DNA replication was by Raz et al. (1991) employing 35 cycles with 3 min measured by [3H]thymidine incorporation as previously elongation at 658C. The PCR products were puri®ed using described (Friedman et al., 1989). Brie¯y, cells were seeded a commercial kit (Promega Magic PCR), directionally cloned in triplicate wells at a density of 26104 cells/well of an into pGEM-4 at XbaI/EcoRI sites and used to transform uncoated 24-well dish in standard growth medium and competent JM 109 bacteria. After overnight growth, 5 ml cultured at 378C. After 6 ± 10 days of incubation, cells were liquid cultures were inoculated with individual colonies, from serum starved for 48 h and then received recombinant which DNA was puri®ed using a commercial kit (Promega VEGF (0 ± 100 ng/ml) (a gift from Dr N Ferrara, Miniprep). Inserts were identi®ed by restriction with XbaI/ Genentech) or 20% FCS for an additional 72 h. EcoRI and sequenced as described. [3H]thymidine (Amersham) was added to cells during the last 4 h of incubation at a ®nal concentration of 1 mCi/ml. Saturation binding assay Cells were washed in PBS and ®xed with ice-cold 10% (w/ v) trichloracetic acid for 15 min. The resulting precipitates Binding assays for VEGF were performed as previously were solubilized by mixing with 500 ml/well of 0.33 M KOH described (Myoken et al., 1991). Puri®ed VEGF was at 378C for 45 min followed by neutralization with 0.3 M iodinated to a speci®c activity of 156 Ci/g, (1.7861078 M) perchloric acid. [3H]-radioactivity was measured by liquid using Iodogen beads. Cells were seeded at 2.56104 cells/ scintillation counting. well in 24-well tissue culture plates and allowed to grow for 1, 4 and 10 days. The cells were washed with 0.5 ml of binding medium (Hanks' BSS containing 1% bovine serum albumin, 25 mM HEPES (pH 7.4)) and then incubated in Note added in proof 0.5 ml binding medium containing [125I]VEGF for 6 h at Two recent reports have identi®ed triple-helical collagens 48C. After two washes in binding medium, the cells were as ligands for the DDR receptors (Shrivastava et al., 1997; solubilized with 0.5 ml of 0.1 M NaOH. Triplicate samples Vogel et al., 1997). of each condition were obtained for each experiment. Nonspeci®c binding was determined in the presence of a 500-fold molar excess of unlabeled VEGF to the binding Acknowledgements medium of parallel cultures and was subtracted from all This work was supported by grants from the NIH samples. The soluble lysates were counted in a gamma (DK37340 (to SLF), DK 07007, DK26743, CA67891, counter. The data were analysed as described by Mori et R29 CA52863 (to RSW), the American Digestive Health al. (1991). The number of cells/well was assessed by Foundation (to VAS) and the UCSF Academic Senate (to trypinization and direct counting of duplicate wells. SLF).

References

Batard P, Sansilvestri P, Scheinecker C, Knapp W, Debili N, Dumont DJ, Gradwohl GJ, Fong GH, Auerbach R and Vainchenker W, Buhring HJ, Monier MN, Kukk E, Breitman ML. (1993). Oncogene, 8, 1293 ± 1301. Partanen J, Matikainen MT, Alitalo R, Hatzfeld J and Enzan H, Himeno H, Iwamura S, Onishi S, Saibara T, Alitalo K. (1996). Blood, 87, 2212 ± 2220. Yamamoto Y and Hara H. (1994). Hepatology, 19, 895 ± Connolly DT, Heuvelman DM, Nelson R, Olander JV, 903. Eppley BL, Del®no JJ, Siegel NR, Leimgruber RM and Fantl WJ, Johnson DE and Williams LT. (1993). Annu. Rev. Feder J. (1989). J. Clin. Invest., 84, 1470 ± 1478. Biochem., 62, 453 ± 481. Davis BH. (1988). J. Cell Physiol., 136, 547 ± 553. Friedman S. (1993a). In: In vitro biological systems. Vol. 1: de Vries, C., Escobedo JA, Ueno H, Houck K, Ferrara N and Methods in Toxicology. Tyson C and Frazier J (eds). San Williams LT. (1992). Science, 255, 989 ± 991. Diego: Academic Press, CA; pp. 292 ± 310. Tyrosine kinases in rat hepatic wound healing V Ankoma-Sey et al 121 Friedman SL. (1993b). N. Engl. J. Med., 328, 1828 ± 1835. Pepper MS, Ferrara N, Orci L and Montesano R. (1991). Friedman SL. (1996). Progress in Liver Disease, 14, 101 ± Biochem. Biophys. Res. Commun., 181, 902 ± 906. 130. Proctor E and Chatamra K. (1982). Gastroenterology, 83, Friedman SL, Roll FJ, Boyles J, Arenson DM and Bissell 1183 ± 1190. DM. (1989). J. Biol. Chem., 264, 10756 ± 10762. Quinn TP, Peters KG, de Vries C, Ferrara N and Williams Friedman SL, Roll FJ, Boyles J and Bissell DM. (1985). LT. (1993). Proc. Natl. Acad. Sci. USA, 90, 7533 ± 7537. Proc. Natl. Acad. Sci. USA, 82, 8681 ± 8685. Raz V, Kelman Z, Avivi A, Neufeld G, Givol D and Yarden Friedman SL, Yamasaki G and Wong L. (1994). J. Biol. Y. (1991). Oncogene, 6, 753 ± 760. Chem., 269, 10551 ± 10558. Rockey DC, Housset CN and Friedman SL. (1993). J. Clin. Geerts A, Lazou JM, De Bleser P and Wisse E. (1991). Invest., 92, 1795 ± 1804. Hepatology, 13, 1193 ± 1202. Rodewald HR and Sato TN. (1996). Oncogene, 12, 397 ± 404. Gressner AM. (1995). J. Hepatol., 22, 28 ± 36. Rosenbaum J, Blazejewski S, Preaux AM, Mallat A, Hanks SK, Quinn AM and Hunter T. (1988). Science, 241, Dhumeaux D and Mavier P. (1995). Gastroenterology, 42 ± 52. 109, 1986 ± 1996. Hioki O, Minemura M, Shimizu Y, Kasii Y, Nishimori H, Sankar S, Mahooti-Brooks N, Bensen L, McCarthy TL, Takahara T, Higuchi K, Yoshitake Y, Nishikawa K and Centrella M and Madri JA. (1996). J. Clin. Invest., 97, Watanabe A. (1996). J. Hepatol., 24, 217 ± 224. 1436 ± 1446. Ikeda H, Nagoshi S and Fujiwara K. (1996). Hepatology, 24, Sato TN, Tozawa Y, Deutsch U, Wolburg-Buchholz K, 460A. Fujiwara Y, Gendron-Maguire M, Gridley T, Wolburg H, Jarnagin WR, Rockey DC, Koteliansky VE, Wang SS and Risau W and Qin Y. (1995). Nature, 376, 70 ± 74. Bissell DM. (1994). J. Cell Biol., 127, 2037 ± 2048. Schlessinger J. (1993). Harvey Lect., 89, 105 ± 123. Johnson JD, Edman JC and Rutter WJ. (1993). Proc. Natl. Shrivastava A, Radziejewski C, Campbell E, Kovac L, Acad. Sci. USA, 90, 5677 ± 5681. McGlynn M, Ryan TE, Davis S, Goldfarb M, Glass DJ, Kawada N, Klein H and Decker K. (1992). Biochem. J., 285, Lemke G and Yancopoulos GD. (1997). Molecular Cell, 1, 367 ± 371. 25 ± 34. Lai C and Lemke G. (1994). Oncogene, 9, 877 ± 883. Stavri GT, Zachary IC, Baskerville PA, Martin JF and Lalazar A, Wong L, Yamasaki G and Friedman SL. (1997). Erusalimsky JD. (1995). Circulation, 92, 11 ± 14. Gene, 195, 235 ± 243. Suzuki K, Hayashi N, Miyamoto Y, Yamamoto M, Ohkawa Leung DW, Cachianes G, Kuang WJ, Goeddel DV and K, Ito Y, Sasaki Y, Yamaguchi Y, Nakase H, Noda K, Ferrara N. (1989). Science, 246, 1306 ± 1309. Enomoto N, Arai K, Yamada Y, Yoshihara H, Tujimura Levi E, Fridman R, Miao HQ, Ma YS, Yayon A and T, Kawano K, Yoshikawa K and Kamada T. (1996). Vlodavsky I. (1996). Proc. Natl. Acad. Sci. USA, 93, Cancer Res., 56, 3004 ± 3009. 7069 ± 7074. Takahashi Y, Kitadai Y, Bucana CD, Cleary KR and Ellis Maher JJ and McGuire RF. (1990). J. Clin. Invest., 86, LM. (1995). Cancer Res., 55, 3964 ± 3968. 1641 ± 1648. Terman BI, Dougher-Vermazen M, Carrion ME, Dimitrov Mise M, Arii S, Higashituji H, Furutani M, Niwano M, D, Armellino DC, Gospodarowicz D and Bohlen P. Harada T, Ishigami S, Toda Y, Nakayama H, Fukumoto (1992). Biochem. Biophys. Res. Commun., 187, 1579 ± 1586. M, Fujita J and Imamura M. (1996). Hepatology, 23, 455 ± Unemori EN, Ferrara N, Bauer EA and Amento EP. (1992). 464. J. Cell. Physiol., 153, 557 ± 562. Mochida S, Ishikawa K, Inao M, Shibuya M and Fujiwara Vlodavsky I, Miao HQ, Medalion B, Danagher P and Ron K. (1996). Biochem. Biophys. Res. Commun., 226, 176 ± D. (1996). Cancer Metastasis Rev., 15, 177 ± 186. 179. Vogel W, Gish GD, Alves F and Pawson T. (1997). Mori S, Claesson-Welsh L and Heldin CH. (1991). J. Biol. Molecular Cell, 1, 13 ± 23. Chem., 266, 21158 ± 21164. Warren RS, Yuan H, Matli MR, Gillett NA and Ferrara N. Mustonen T and Alitalo K. (1995). J. Cell. Biol., 129, 895 ± (1995). J. Clin. Invest., 95, 1789 ± 1797. 898. Wong L, Yamasaki G, Johnson RJ and Friedman SL. (1994). Myoken Y, Kayada Y, Okamoto T, Kan M, Sato GH and J. Clin. Invest., 94, 1563 ± 1569. Sato JD. (1991). Proc. Natl. Acad. Sci. USA, 88, 5819 ± Xu J and Clark RA. (1996). J. Cell. Biol., 132, 239 ± 249. 5823. Yamane A, Seetharam L, Yamaguchi S, Gotoh N, Nomura M, Yamagishi S, Harada S, Hayashi Y, Yamashima Takahashi T, Neufeld G and Shibuya M. (1994). T, Yamashita J and Yamamoto H. (1995). J. Biol. Chem., Oncogene, 9, 2683 ± 2690. 270, 28316 ± 28324.