Oncogene (2002) 21, 1791 ± 1799 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc ORIGINAL PAPERS Hepatocyte growth factor promotes hepatocarcinogenesis through c-Met autocrine activation and enhanced angiogenesis in transgenic mice treated with diethylnitrosamine

Norio Horiguchi1, Hisashi Takayama1, Mitsuo Toyoda1, Toshiyuki Otsuka1, Toshio Fukusato2, Glenn Merlino3, Hitoshi Takagi*,1 and Masatomo Mori1

1The First Department of Internal Medicine, Gunma University School of Medicine, Maebashi, Gunma 371-8511, Japan; 2Division of Diagnostic Pathology, Gunma University School of Medicine, Maebashi, Gunma 371-8511, Japan; 3Laboratory of Molecular Biology, National Cancer Institute, Bethesda, Maryland, MD 20892-4255, USA

Hepatocyte growth factor (HGF) is a mitogen for Introduction hepatocytes, but it is not clear whether HGF stimulates or inhibits hepatocarcinogenesis. We previously reported Hepatocyte growth factor (HGF) is a polypeptide that HGF transgenic mice under the metallothionein originally characterized as a highly potent hepatocyte gene promoter developed benign and malignant liver mitogen (Gohda et al., 1988; Nakamura et al., 1989). tumors spontaneously after 17 months of age. To More recent studies have revealed that HGF is a elucidate the role of HGF in hepatocarcinogenesis, multifunctional cytokine and can elicit mitogenic, diethylnitrosamine (DEN) was administered to HGF motogenic, and morphogenic responses in a variety of transgenic mice. HGF overexpression accelerated DEN- cultured epithelial cells expressing the transmembrane induced hepatocarcinogenesis, often accompanied by tyrosine kinase receptor, Met (Brinkmann et al., 1995; abnormal blood vessel formation. In this study, 59% of Zarnegar and Michalopoulos, 1995). HGF is expressed transgenic males (versus 20% of wild-type males) and in a multitude of mesenchymally-derived cells and Met 39% of transgenic females (versus 2% of wild-type expression has been detected in the epithelium of most females) developed either benign or malignant liver tissues, indicating that HGF-Met signal transduction tumors by 48 weeks (P50.005, P50.001, respectively). pathway helps mediate mesenchymal-epithelial interac- Moreover, 33% of males and 23% of female transgenic tions (Rubin et al., 1991; Montesano et al., 1991; mice developed hepatocellular carcinoma (HCC), while Rosen et al., 1994). none of the wild-type mice developed HCC (P50.001, c-met was ®rst cloned as a proto-oncogene through P50.005, respectively). Enhanced kinase activity of the its ability to transform NIH3T3 cells (Park et al., 1986) HGF receptor, Met, was detected in most of these and identi®ed as a HGF receptor (Bottaro et al., 1991). tumors. Expression of vascular endothelial growth factor Robust expression of the c-met proto-oncogene has (VEGF) was up-regulated in parallel with HGF been documented in diverse human and mouse tumors transgene expression. Taken together, our results suggest including hepatocellular carcinomas (HCCs) (Prat et that HGF promotes hepatocarcinogenesis through the al., 1991; Rong et al., 1992, 1993; D'Errico et al., autocrine activation of the HGF-Met signaling pathway 1996), and the c-met gene is ampli®ed in some in association with stimulation of angiogenesis by HGF carcinomas of the gastrointestinal tract (Di Renzo et itself and/or indirectly through VEGF. al., 1995; Kuniyasu et al., 1992). Moreover, activating Oncogene (2002) 21, 1791 ± 1799. DOI: 10.1038/sj/ mutations within the tyrosine kinase domain of c-met onc/1205248 have been reported in human renal papillary carcino- mas and childhood HCCs (Schmidt et al., 1997; Park Keywords: HGF transgenic; hepatocellular carcinoma; et al., 1999). In addition, coexpression of HGF-Met c-Met; diethylnitrosamine; VEGF has been identi®ed in a variety of transformed cultured cells and in some tumors (Bellusci et al., 1994; Tuck et al., 1996; Rahimi et al., 1996). Taken together, these studies suggest that activation of HGF-Met signaling can be intimately associated with neoplastic transfor- mation. Paradoxically, however, HGF has also been reported to inhibit the growth of certain carcinoma cells. With *Correspondence: H Takagi, The First Department of Internal respect to HCC, HGF was ®rst reported to show anti- Medicine, Gunma University School of Medicine, Showa-machi proliferative e€ects on some HCC cell lines, whereas it 3-39-15, Maebashi, Gunma 371-8511, Japan; had a mitogenic e€ect on others (Tajima et al., 1991; E-mail: [email protected] Received 6 September 2001; revised 4 December 2001; accepted 12 Shiota et al., 1992; Miyazaki et al., 1992). Further- December 2001 more, both stimulatory and inhibitory e€ects of HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1792 exogenous administration of HGF on carcinogen- described previously (Takayama et al., 1997a, 2001). treated rats have been reported (Liu et al., 1995; Nonetheless, 39% (12/31) of HGF transgenic females Yaono et al., 1995; Ogasawara et al., 1998). There are developed liver tumors by 40 weeks. In contrast, only also con¯icting reports in HGF transgenic mice. We 2% (1/42) of wild-type females developed liver tumors previously reported that transgenic mice harboring a by 48 weeks (P50.001). Liver tumors in HGF full-length mouse HGF cDNA under the control of the transgenic mice were larger in size and more numerous mouse metallothionein (MT) gene promoter induced compared to those arising in wild-type mice. This liver tumors, which arose spontaneously in six di€erence was extremely evident at 48 weeks male independent transgenic lines after 17 months (Sakata mice. 100% (4/4) of HGF transgenic mice had a et al., 1996). In contrast, overexpression of a human number of (6+1.8 per mouse) frank liver tumors which HGF cDNA under the regulation of the albumin are more than 5 mm in diameter, whereas none (0/5) of promoter in transgenic mice did not induce HCC wild-type developed such tumors (P50.001). (Shiota et al., 1994). Moreover, the HGF transgene With respect to progression to malignancy, 33% (9/ appeared to inhibit hepatocarcinogenesis in bitrans- 27) and 23% (7/31) of male and female HGF genic mice overexpressing c-myc (Santoni-Rugiu et al., transgenic mice developed HCCs, respectively (Table 1996) or transforming growth factor alpha (TGF-a) 2). In contrast, no HCC was found in male or female (Shiota et al., 1995). Therefore, whether HGF wild-type mice (P50.001, P50.005, respectively). participates in hepatocarcinogenesis remains to be Several studies reported that cells expressing both resolved. HGF and c-met show enhanced metastasis in vivo Here we report the promoting e€ect of HGF on (Bellusci et al., 1994; Rong et al., 1994; Je€ers et al., diethylnitrosamine (DEN)-induced hepatocarcinogen- 1996), but in the present study, only one HCC showed esis in MT-HGF transgenic mice, and provide multiple metastasis to the lung by 48 weeks. Prolonged characterization of these liver tumors at histopatholo- observation might be necessary for the study of gical and molecular levels. metastasis in this model. Liver growth over the course of this study was determined by measuring liver weight relative to total body weight. Table 3 shows that transgenic mouse liver weight was much greater than Results that of wild-type mice at each time point. The liver weight in 40-week-old HGF transgenic females with Chemical induction of liver tumors in HGF transgenic multiple tumors was about threefold greater than age- mice matched wild-type. A signi®cant di€erence in the development of liver tumors was observed at 48 weeks of age between HGF Histopathological and immunohistochemical analyses transgenic mice and wild-type mice treated with DEN (Table 1). In males, 59% (16/27) of HGF transgenic Histopathological analysis of livers of treated and mice developed grossly visible liver tumors, although untreated HGF transgenic mice showed striking only 20% (6/30) of wild-type mice had such tumors heterogeneity in hepatocytes and numerous small (P50.005). hepatocytes surrounding portal veins (Sakata et al., Interestingly, this promoting e€ect of HGF on liver 1996). Livers from DEN-treated transgenic mice tumorigenesis was also detected in females, although contained more preneoplastic foci per unit area than females are known to be resistant to HCC in wild-type at 16 weeks, although foci could not be experimental mouse models, including those employing evaluated at later time points because of coalescing chemical carcinogenesis (Farber and Sarma, 1987; tumors (Table 4). This preneoplastic change was also Pressumann, 1988; Takagi et al., 1993). Unfortunately, noted in transgenic females, similar to transgenic it was dicult to monitor HGF transgenic females over males. Notably, neither in¯ammation nor ®brosis long periods because more than half of females died as was found in livers of transgenic and wild-type mice a result of renal failure or intestinal disease, as in this study, in sharp contrast to human. HCCs

Table 1 Incidence of liver tumors in HGF transgenic mice and wild-type mice treated with diethylnitrosamine (DEN) Mouse Incidence of liver tumor (weeks) Sex genotypea 16b 24 32 40 48 Total

Male WT 0/4 (0%) 1/6 (17%) 0/8 (0%) 2/7 (29%) 3/5 (60%) 6/30 (20%) TG 0/4 (0%) 3/6 (50%) 3/6 (50%) 6/7 (86%) 4/4 (100%) 16/27 (59%)f Female WT 0/5 (0%) 1/9 (11%) 0/9 (0%) 0/10 (0%) 0/9 (0%) 1/42 (2%) TG 0/4 (0%) 2/14 (14%)c 7/10 (70%)d 3/3 (100%) N.A.e 12/31 (39%)g

Number of mice per group possessing at least one adenoma or HCC, relative to the total number of mice in that group. Numbers in parentheses represent the occurrence rate of liver tumors. DEN was administered as a single i.p. injection of 5 mg/g body weight at 15 days of age. More than half of the HGF female mice died of renal failure or intestinal disease by 24 weeks. aWT, wilt-type mice; TG, HGF transgenic mice. bSacri®ce interval. c,d Some female HGF mice found moribund or dead before designated time points were included in the number. eNot available. f,g Values are signi®cantly di€erent from wild-type mice based on w2-test. (P50.005, P50.001, respectively)

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1793 Table 2 Incidence of HCCs in HGF transgenic mice and wild-type mice treated with DEN Mouse Incidence of HCCs (weeks) Sex genotypea 16b 24 32 40 48 Total

Male WT 0/4 (0%) 0/6 (0%) 0/8 (0%) 0/7 (0%) 0/5 (0%) 0/30 (0%) TG 0/4 (0%) 1/6 (17%) 2/6 (33%) 3/7 (43%) 3/4 (75%) 9/27 (33%)f Female WT 0/5 (0%) 0/9 (0%) 0/9 (0%) 0/10 (0%) 0/9 (0%) 0/42 (0%) TG 0/4 (0%) 1/14 (14%)c 3/10 (30%)d 3/3 (100%) N.A.e 7/31 (23%)g

Number of mice per group possessing at least one HCC, relative to the total number of mice in that group. Numbers in parentheses represent the occurrence rate of HCCs. a±e See legend of Table 1 for details. f,g Values are signi®cantly di€erent from wild-type mice based on w2-test. (P50.001, P50.005, respectively)

Table 3 Liver weights in HGF transgenic mice and wild-type mice treated with DEN Mouse Liver weight (% of bodyweight) Sex genotypea 16b 24 32 40 48

Male WT 4.1+0.3 (4) 4.1+0.1 (6) 4.1+0.4 (8) 4.0+0.3(7) 5.0+0.5 (5) TG 6.4+0.9 (4)f 6.3+0.7 (6)f 5.3+0.5 (6)f 9.8+8.1 (7)f 7.5+2.7 (4)f Female WT 4.7+0.3 (5) 4.2+0.4 (9) 4.1+0.5 (9) 4.3+0.3 (10) 4.6+0.2 (9) TG 8.8+1.9 (4)f 11.5+1.9 (14)cf 11.2+4.6 (10)d 13.1+3.1 (3)g N.A.e

Numbers represent the percentage of liver weight/body weight (mean+s.d.). Numbers in parentheses represent the number of mice examined. a±e See legend to Table 1 for details. f,g Values are signi®cantly di€erent from age-matched wild-type mice based on student t-test. (P50.005, P50.05, respectively)

Table 4 Incidence of preneoplastic lesions in HGF transgenic mice (data not shown). To determine if hepatocarcinogenesis and wild-type mice at 16-week-old in HGF transgenic mice consistently correlated with Foci/cm2 further enhancement in HGF transgene expression, Sex Mouse genotypea DEN (7) DEN (+) HGF expression was analysed in transgenic tumors Male WT 0 0.28+0.03 and adjacent nontumorous tissues. Northern blot TG 0.16+0.03 0.85+0.04b analyses showed that ®ve of six (83%) tumors Female WT 0 0.06+0.02 exhibited HGF transgene transcript levels that were c TG 0 0.66+0.05 higher than those found in adjacent, grossly normal, Numbers of preneoplastic foci/cm2 (mean+s.e.m) found in 16-week- transgenic tissue (Figure 3). Endogenous c-met tran- old and control FVB mouse lines. aWT, wild-type; TG, HGF script levels were detected but no apparent di€erences transgenic mice. b,cValues are signi®cantly di€erent from wild-type were noted between tumors and adjacent liver tissues mice (P50.01, P50.02, respectively) (Figure 3). c-met gene ampli®cation was not found in these tumors (data not shown). developing in HGF transgenic mice were composed of Analysis of Met protein levels and activity trabecular (Figure 1A), solid, and/or pseudoglandular type (data not shown), while none of the wild-type To quantify levels and phosphorylation of Met mice developed HCCs (Figure 1B). Macroscopic and protein, extracts of liver tumors were subjected to microscopic aberrant neovascularization were often immunoprecipitation using an anti-Met antibody, observed in transgenic tumors (Figure 1C), compared followed by Western blot analysis using either an to wild-type tumors (Figure 1D). Some of HCCs in anti-Met antibody or an anti-phosphotyrosine anti- transgenic mice were accompanied peliosis-like change body. Figure 4 shows that levels of Met protein (Figure 1E), which is occasionally observed in human were in general agreement with the c-met transcript liver due to unknown cause (Craig et al., 1989). To data. Met tyrosine phosphorylation was then further quantify this overt increase in blood vessels, determined in extracts obtained from transgenic immunohistological analysis of endothelial cells was tumors and adjacent nontumorous tissue. Met carried out using von - Willebrand Factor (vWF) tyrosine phosphorylation was enhanced in three of antibodies (Figure 1F). The area of blood vessels in four (75%) tumors, which showed elevated HGF liver tumors of DEN-treated transgenic mice was transgene expression relative to adjacent tissue signi®cantly larger than in those of wild-type mice (Figure 4), suggesting that activation of Met can (Figure 2). promote liver tumorigenesis.

Analysis of transcripts of HGF and its receptor (c-met)in Analysis of expression of VEGF in carcinogen-treated or livers of carcinogen-treated transgenic mice -untreated transgenic mice Our preliminary data showed that HGF transgene Because several reports have documented that HGF expression was not elevated by DEN administration can induce VEGF expression in vitro (Van Belle et

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1794

Figure 1 Liver tumors arising in HGF transgenic mice treated with DEN. (A,B) Comparison of liver tumors from 48-week-old male transgenic (A) and wild-type (B). Multiple HCCs were found in transgenic mice and moderately-di€erentiated (thick trabecular) HCC was exhibited as (A). None of HCC was found in wild-type mice and only adenomas (B) were found. (C,D) Comparison of the border between tumor (TU) and adjacent (non-tumor) tissue from transgenic (C) and wild-type (D). Note increased blood vessels especially in tumor lesion in transgenic mice (C) compared to wild-type (D). (E) HCC including peliosis-like change from transgenic mouse. (F) Immunohistochemical detection of endothelial cells using vWF antibody. Note increased and abnormal endothelial proliferation. Magni®cation: A,B 6200; C,E 680; D 6120 (H&E); F 6120

al., 1998; Moriyama et al., 1998; Gille et al., 1998) Discussion and because liver tumors in HGF transgenic mice showed abnormal blood vessel formation, we In the present study, we demonstrated in vivo that analysed VEGF expression in tumor and adjacent HGF overexpression promotes hepatocarcinogenesis in tissue from transgenic and wild-type livers. A 4.2 kb transgenic mice when initiated with DEN, a well- transcript was observed in both mice, and an characterized genetic mutagen, despite the ability of additional 3.7 kb transcript was observed uniquely HGF to inhibit the growth of some HCC cell lines in in HGF transgenic mice livers (Figure 4). Northern vitro (Tajima et al., 1991; Shiota et al., 1992). We also blot analysis showed that ®ve of six (83%) tumors demonstrated that liver tumorigenesis in HGF trans- arising in DEN-treated HGF transgenic mice genic mice is driven by the creation of HGF-Met possessed 3.7 kb VEGF expression levels that were autocrine loops, as evidenced by elevation of both higher than those found in adjacent, non-tumorous, HGF expression and c-Met phosphorylation in tumors transgenic tissue. Moreover, 3.7 kb VEGF transcript relative to adjacent tissues. Moreover, our data was elevated in parallel with HGF transgene indicated that this promotion may be mediated expression, suggesting that 3.7 kb VEGF overexpres- through enhanced angiogenesis provided by HGF sion is upregulated by HGF overexpression in these itself, and/or indirectly through the induction of tumors (Figure 4). VEGF.

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1795

Figure 4 Met protein levels and activity in HGF-induced liver tumors by Western blotting. Extracts were prepared from four pairs of liver tumors. Five hundred mg of each, except where marked with an asterisk (250 mg), were subjected to immunopre- Figure 2 Quanti®cation of blood vessel areas in tumors from cipitation (IP) and then immunoblotting (IB) with an anti-Met HGF transgenic and wild-type mice. Sixteen randomly selected antibody or anti-phosphotyrosine (PY) antibody. The expected ®elds from eight tumors in each group were analysed using bands of Mr 170 000 and Mr 140 000 were observed. Note National Institutes of Health image software and area of blood enhanced kinase activity relative to nontumorous tissue in three vessels stained with vWF antibody was estimated. The area of of four liver tumors. (#No. is mouse ID to distinguish each blood vessels in HGF transgenic mouse tumors was signi®cantly mouse) larger compared to wild-type tumors (*P50.05, **P50.0001)

growth (Yeh et al., 1987; Suzuki et al., 1996). However, the role of HGF-Met signaling in hepatocarcinogenesis has not been clearly established because of somewhat con¯icting reports in the literature (Tajima et al., 1991; Shiota et al., 1992; Miyazaki et al., 1992; Liu et al., 1995; Yaono et al., 1995; Ogasawara et al., 1998). The involvement of HGF-Met autocrine loops in tumorigen- esis in other cell types has been clearly established both in vitro and in vivo (Bellusci et al., 1994; Tuck et al., 1996). A multitude of human cell lines and tumors, particularly sarcomas, overexpress both Met and its ligand (Rong et al., 1993; Tuck et al., 1996). In fact, diverse tumors developed in our HGF transgenic mice, including melanoma and rhabdomyosarcoma, demonstrating the formation of HGF-Met autocrine loops through forced expression of the transgene and overexpression of endogenous receptor (Takayama et al., 1997b). In liver tumors in this model, Met kinase activity was enhanced Figure 3 Analysis of transgenic HGF, native c-met, and VEGF in tumorous tissue, but not in adjacent tissue, in transcripts in liver tumors derived from DEN-treated HGF accordance with HGF transgene expression. These transgenic mice or wild-type by Northern blot hybridization. a, results suggest that hepatocarcinogenesis requires the adjacent nontumorous liver tissue; t, tumor tissue. Adjacent and selection of cells capable of expressing the HGF tumor tissues from the same mouse liver are presented as pairs. Note that increased level of 3.7 kb VEGF expression in tumor transgene in a strong, constitutive fashion, providing a tissue correlated with elevated HGF transgene expression in the mechanism for autonomous cellular proliferation same tumor shown at the top panel. No. 47, 16 and 42 are pairs through autocrine signal transduction. This is reminis- of HCCs and nontumorous tissues, and the others are pairs of cent of the behavior of NIH3T3 and C127 cells liver adenomas and nontumorous tissues (#No. is mouse ID to distinguish each mouse) cotransfected with constructs expressing both HGF and Met, which demonstrated a greatly strengthened transformed phenotype (Rong et al., 1994; Je€ers et al., 1996). The acquisition of autonomous growth has been These ®ndings on the consequences of HGF-Met strongly implicated in the development and progression autocrine signaling in the liver are similar to those of HCCs (Farber and Sarma, 1987). Several growth associated with perturbation of signaling by other factor signals, including TGF-a-epidermal growth factor receptor tyrosine kinases. A number of studies have receptor (EGFR), HGF-Met, and VEGF-VEGF recep- shown that TGF-a-EGFR signaling is involved in liver tor, are reported to participate in this autonomous tumorigenesis (Yeh et al., 1987; Hsia et al., 1991). In

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1796 fact, robust overexpression of human TGF-a in abnormal blood vessel formation and intrahepatic transgenic mouse (line MT42) induced liver tumor hemorrhage in some HCCs. These results suggest that development (Jhappan et al., 1990; Takagi et al., 1992). HGF can directly stimulate tumor progression through Strong expression of TGF-a in transgenic mouse enhancement of angiogenesis (Matsumoto and Naka- hepatocytes activates in autocrine growth stimulatory mura, 1996; Rosen et al., 1997), and/or indirectly loop that permits autonomous replication in culture through induction of VEGF. (Wu et al., 1994a,b). Interestingly, these liver tumors There are con¯icting reports on liver tumorigenesis in also showed elevated TGF-a transgene transcript levels HGF transgenic mouse models. Overexpression of a relative to adjacent tissue (Jhappan et al., 1990). It is human HGF cDNA under the regulation of the albumin intriguing to speculate that the similar tumor promot- promoter in transgenic mice, on the FVB strain, did not ing e€ect observed in the two distinct autocrine-loops, induce HCC (Shiota et al., 1994). The level of expression HGF-Met and TGF-a-EGFR, might represent the achieved through the use of either the albumin or MT activation of pathways shared by a variety of receptor promoter may explain di€erences in the consequence of tyrosine kinase family members. the presence of the HGF transgene in the liver. The MT The dramatic acceleration of liver tumorigenesis by promoter with adjacent locus control regions induced a DEN in transgenic mice, from over 17 months in 2 ± 5-fold higher levels of serum HGF compared to that untreated mice (Sakata et al., 1996) to as early as 6 achieved by the albumin promoter (Sakata et al., 1996). months in treated mice, suggests that HGF acts as a In addition, the mouse HGF cDNA used in our promoter of DEN-initiated hepatocytes. One mechan- transgenic mice, instead of the human HGF cDNA, ism by which chronically overexpressed HGF may which demonstrates about 10% diversity in terms of collaborate with DEN is through dysregulation of cell amino acid conservation (Liu et al., 1993), might be growth, as documented previously in transgenic required to promote optimal activity of mouse Met. hepatocytes (Sakata et al., 1996), which could facilitate Transgenic mice overexpressing isoforms of HGF with the ®xation of promutagenic lesions induced by DEN. ®ve deleted amino acids (d-HGF) under the regulation of In another mechanism, the anti-apoptotic e€ects of albumin promoter has been documented (Bell et al., HGF on hepatocytes may facilitate hepatocarcinogen- 1999). Although d-HGF is reported to be functionally esis, as it was previously documented that HGF more potent than the full-length HGF, d-HGF trans- blocked massive Fas-mediated liver apoptosis (Kosai genic mice showed no obvious di€erence in liver/body et al., 1998). HGF-Met signaling may encourage the weight ratio compared to control mice. However, d- survival of initiated cells that would otherwise succumb HGF transgenic male mice, generated on a B6D2 to mutation-induced death. In fact, our male and (C57B66DBA2) strain, were three times more likely to female HGF transgenic mice showed more preneoplas- develop HCC beyond 17 months (Bell et al., 1999). tic foci at 16 weeks than FVB. This might explain the Interestingly, the tumor promoting e€ects of d-HGF high incidence of tumors in transgenic female mice in became more apparent in both male and female mice the absence of hormonal manipulation. when initiated DEN (Bell et al., 1999), consistent with Angiogenesis is implicated in cancer development, our data described here. progression, growth, and metastasis (Fidler and Ellis, HGF has been considered for the treatment of 1994; Folkman, 1995). Among various angiogenic patients with liver cirrhosis (Ueki et al., 1999), factors, VEGF is one of the most important factors in fulminant hepatic failure (Kosai et al., 1998) and liver hepatocarcinogenesis (Mise et al., 1996; Suzuki et al., transplantation (Ishiki et al., 1992; Fujiwara et al., 1996). Five di€erent molecular species resulting from 1993) as well as arteriosclerosis (Morishita et al., 2000). alternative splicing of the VEGF gene have been However, we have previously shown that aberrant identi®ed according to their number of amino acids; HGF expression induces developmental anomalies VEGF121, VEGF145, VEGF165, VEGF189 and (Takayama et al., 1996), renal dysfunction (Takayama VEGF206 (Houck et al., 1991; Tischer et al., 1991). All et al., 1997a) and intestinal disease (Takayama et al., isoforms di€er in eciency of secretion but act similarly 2001) in mice, as well as spontaneous neoplastic by stimulating mitogenesis and migration of vascular transformation in older animals (Takayama et al., endothelial cells, and all increase vascular permeability 1997b; Otsuka et al., 1998). Here we present data (Houck et al., 1991). Recent studies have shown that demonstrating that environmental carcinogens can administration of HGF induces upregulation of VEGF cooperate with, and dramatically accelerate, HGF- in various cell types (Van Belle et al., 1998; Moriyama et mediated tumorigenesis in the liver, further supporting al., 1998; Gille et al., 1998). Although Van Belle et al. the oncogenic risk associated with persistent in vivo (1998) reported that administration of recombinant exposure to robust levels of HGF. human HGF induced three principal isoforms (VEGF 121, VEGF 165, VEGF 189) in smooth muscle cells, we found that HGF overexpression in the liver uniquely up- regulated 3.7 kb transcripts, corresponding to VEGF164 Materials and methods (VEGF165 in human). We demonstrated here that VEGF transcription was upregulated in parallel with Animals HGF transgene expression. It is noteworthy that both MT-HGF line MH19 transgenic mice were generated on an macroscopic and microscopic examinations revealed albino FVB genetic background as described previously

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1797 (Takayama et al., 1996). Transgenic and control mice used in osine antibody (Upstate Biotechnology) overnight. Met was this study were typically produced from the mating of HGF visualized by incubation with anti-goat antibody conjugated heterozygote transgenic males with FVB females. Drinking to horseradish peroxidase (Santa Cruz Biotechnology) by water containing zinc was not used in this study. All mouse using enhanced chemiluminescence (Santa Cruz Biotechnol- work was performed in accordance with the guidelines for ogy). Phosphotyrosine was visualized as above except that an animal care and use established by Gunma University School anti-mouse antibody conjugated to horseradish peroxidase of Medicine. (Santa Cruz Biotechnology) was used.

Carcinogen-induced HCC Immunohistochemistry and quantification of blood vessel formation Diethylnitrosamine (DEN) (Sigma Chemical Co., St. Louis, MO, USA) was injected intraperitoneally in 15-day-old mice For immunohistochemical analysis of von-Willebrand Factor at a dose of 5 mg/g of body weight. At 21 days of age, mice (vWF), formalin-®xed and paran-embedded tissues were were separated by sex and their genotype was determined. sectioned at a thickness of 4 mm. An anti-human vWF rabbit Study duration was 48 weeks, with interim sacri®ces at 16, antibody (Dako Corporation, Carpinteria, CA, USA) was 24, 32 and 40 weeks. Mice found moribund or dead were also used at 1 : 200 concentration and antibody localization was examined for tumor development. Tissues were ®xed in either visualized using Vectastain elite avidin-biotin-peroxidase 10% bu€ered formalin or 4% paraformaldehyde, embedded complex kits (Vector Laboratories, Burlingame, CA, USA) in paran, sectioned at 4 mm and stained with hematoxylin according to the instructions provided by the manufacturer. and eosin (H&E) for histopathological analysis. A portion of Staining with vWF antibody, speci®c for endothelial cells, each tissue or tumor was snap frozen in liquid nitrogen and denoted blood vessel formation in liver tumor. The stored at 7808C for molecular analysis. For image analysis vascularized area was expressed as the percentage of blood of hepatocellular foci, four representative sections per mouse vessels to the total area using National Institute of Health were obtained from all liver lobes and the number and area- Image software. Sixteen randomly selected ®elds from eight perimeter of each lesion were determined as described tumors were analysed in each group. previously (Takagi et al., 1993). Data analysis Analysis of RNA transcripts All data were expressed as mean+s.e.m. Di€erences in tumor HGF, c-met, and VEGF transcripts were detected by incidence were examined for statistical signi®cance using the Northern blot hybridization. The mouse HGF cDNA probe w2-test and those in preneoplastic lesion and blood vessel and mouse c-met cDNA probe were synthesized by PCR as formation by one-way ANOVA followed by Fisher-PLSD described previously (Takayama et al., 1996). The human analysis. A P value less than 0.05 denoted the presence of a VEGF cDNA probe was synthesized by PCR using clone statistically signi®cant di€erence. LA737 (kindly provided by Dr William J LaRochelle) as a template and the following set of primers: 5'-GTAAAAC- GACGGCCAGT-3' and 5'-CAGGAAACAGCTATGAC-3'. Abbreviations Total RNA was isolated using Isogen (Wako Pure Chemical HGF, hepatocyte growth factor; DEN, diethylnitrosamine; Industries, Osaka, Japan) and 20 mg were loaded per lane HCC, hepatocellular carcinoma; VEGF, vascular endothe- onto 1% agarose/formaldehyde gels and transferred to nylon lial growth factor; MT, metallothionein; d-HGF, 5 amino membranes after electrophoresis. acid deleted isoform of HGF; vWF, von-Willebrand Factor; TGF-a, transforming growth factor alpha; EGFR, Analysis of Met and Met activation epidermal growth factor receptor; H&E, hematoxylin and eosin; GAPDH, glyceraldehyde-3-phosphate dehydrogen- Quanti®cation of Met and Met tyrosine phosphorylation was ase. performed as described previously (Otsuka et al., 1998). For immunoprecipitation, 500 mg of lysate were incubated with anti-Met antibody (Santa Cruz Biotechnology, Santa Cruz, Acknowledgments CA, USA) for 2 h on ice. After the addition of Gamma-Bind The authors thank Dr William J LaRochelle from G Sepharose (Boehringer Mannheim, Mannheim, Germany) CuraGen Corporation (New Haven, CT, USA), for and washing in RIPA bu€er, the immunoprecipitates were providing clone LA 737. We are indebted to Dr Hiromi fractionated on 10% polyacrylamide gels (Biocraft). After Sakata for critical reading of this manuscript. This work electrophoretic transfer to nitrocellulose membranes (Bio-Rad was supported in part by a Grant-in Aid for Scienti®c Laboratories, Richmond, CA, USA), ®lters were blocked and Research (No.11680817) from the Ministry of Education, then incubated with anti-Met antibody or anti-phosphotyr- Science, Sports and Culture of the Japanese Government.

References

Bell A, Chen Q, DeFrances MC, Michalopoulos GK and Brinkmann V, Foroutan H, Sachs M, Weidner KM and Zarnegar R. (1999). Oncogene, 18, 887 ± 895. Birchmeier W. (1995). J. Cell Biol., 131, 1573 ± 1586. Bellusci S, Moens G, Gaudino G, Comoglio PM, Nakamura Craig JR, Peters RL and Edmondson HA. (1989). Atlas of T, Thiery JP and Jouanneau J. (1994). Oncogene, 9, 1091 ± Tumor pathology, Washington, DC: Armed Forces 1099. Institute of Pathology. Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE, Vande Woude GF and Aaronson SA. (1991). Science, 251, 802 ± 804.

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1798 Di Renzo MF, Olivero M, Giacomini A, Porte H, Chastre E, ParkWS,DongSM,KimSY,NaEY,ShinMS,PiJH,Kim Mirossay L, Nordlinger B, Bretti S, Bottardi S, Giordano BJ,BaeJH,HongYK,LeeKS,YooNJ,JangJJ,PackS, S, Plebani M, Gespach C and Comoglio PM. (1995). Clin. Zhuang Z, Schmidt L, Zbar B and Lee JY. (1999). Cancer Cancer Res., 1, 147 ± 154. Res., 59, 307 ± 310. D'Errico A, Fiorentino M, Ponzetto A, Daikuhara Y, Prat M, Narsimhan RP, Crepaldi T, Nicotra MR, Natali PG Tsubouchi H, Brechot C, Scoazec JY and Grigioni WF. and Comoglio PM. (1991). Int. J. Cancer, 49, 323 ± 328. (1996). Hepatology, 24, 60 ± 64. Pressumann R. (1988). Mechanism of chemical carcinogen- Farber E and Sarma DS. (1987). Lab. Invest., 56, 4±22. esis. In: Biochemistry of Chemical Carcinogenesis, RC Fidler IJ and Ellis LM. (1994). Cell, 79, 185 ± 188. Garner and J Hradec (eds), New York: Plenum Publishing Folkman J. (1995). Nat. Med., 1, 27 ± 31. Corp. pp. 25 ± 35. Fujiwara K, Nagoshi S, Ohno A, Hirata K, Ohta Y, Rahimi N, Tremblay E, McAdam L, Park M, Schwall R and Mochida S, Tomiya T, Higashino K and Kurokawa K. Elliot B. (1996). Cell Growth Di€er., 7, 263 ± 270. (1993). Hepatology, 18, 1443 ± 1449. Rong S, Je€ers M, Resau JH, Tsarfaty I, Oskarsson M and Gille J, Khalik M, Konig V and Kaufmann RJ. (1998). J. Vande Woude GF. (1993). Cancer Res., 53, 5355 ± 5360. Invest. Dermatol., 111, 1160 ± 1165. Rong S, Segal S, Anver M, Resau JH and Vande Woude GF. Gohda E, Tsubouchi H, Nakayama H, Hirono S, Sakiyama (1994). Proc. Natl. Acad. Sci. USA, 91, 4731 ± 4735. O, Takahashi K, Miyazaki H, Hashimoto S and Rong S, Bodescot M, Blair D, Dunn J, Nakamura T, Mizuno Daikuhara Y. (1988). J. Clin. Invest., 81, 414 ± 419. K, Park M, Chan A, Aaronson SA and Vande Woude GF. Houck KA, Ferrara N, Winer J, Cachianes G, Li B and (1992). Mol. Cell. Biol., 12, 5152 ± 5158. Leung DW. (1991). Mol. Endocrinol., 5, 1806 ± 1814. Rosen EM, Lamsuzus K, Laterra J, Polverini PJ, Rubin JS Hsia CC, Axiotis CA, Di Bisceglie A and Tabor E. (1991). and Goldberg ID. (1997). Ciba Found. Symp., 212, 215 ± Cancer, 70, 1049 ± 1056. 226. Ishiki Y, Ohnishi H, Muto Y, Matsumoto K and Nakamura Rosen EM, Nigam SK and Goldberg ID. (1994). J. Cell T. (1992). Hepatology, 16, 1227 ± 1235. Biol., 127, 1783 ± 1787. Je€ers M, Rong S, Anver M and Vande Woude GF. (1996). Rubin JS, Chan AM, Bottaro DP, Burgess WH, Taylor WG, Oncogene, 13, 853 ± 856. CechAC,Hirsch®eldDW,WongJ,MikiT,FinchPWand Jhappan C, Stahle C, Harkins RN, Fausto N, Smith GH and Aaronson SA. (1991). Proc.Natl.Acad.Sci.USA,88, Merlino G. (1990). Cell, 61, 1137 ± 1146. 415 ± 419. Kosai K, Matsumoto K, Nagata S, Tsujimoto Y and Sakata H, Takayama H, Sharp R, Rubin JS, Merlino G and Nakamura T. (1998). Biochem. Biophys. Res. Commun., LaRochelle WJ. (1996). Cell Growth Di€er., 7, 1513 ± 244, 683 ± 690. 1523. Kuniyasu H, Yasui W, Kitadai Y, Yokozaki H, Ito H and Santoni-Rugiu E, Preiseggar KH, Kiss A, Audolfsson T, Tahara E. (1992). Biochem. Biophys. Res. Commun., 189, Shiota G, Schmidt EV and Thorgeirsson SS. (1996). Proc. 227 ± 232. Natl. Acad. Sci. USA, 93, 9577 ± 9582. Liu ML, Mars WM and Michalopoulos GK. (1995). SchmidtL,DuhFM,ChenF,KishidaT,GlennG,Choyke Carcinogenesis, 16, 841 ± 843. P, Scherer SW, Zhuang Z, Lubensky I, Dean M, Allikmets Liu Y, Michalopoulos GK and Zarnegar R. (1993). Biochim. R, Chidambaram A, Bergerheim UR, Feltis JT, Casade- Biophys. Acta, 1216, 299 ± 303. vallC,ZamarronA,BernuesM,RichardS,LipsCJ, Matsumoto K and Nakamura T. (1996). J. Biochem., 119, Walther MM, Tsui LC, Geli L, Orcutt ML, Stackhouse T, 591 ± 600. LipanJ,SlifeL,BrauchH,DeckerJ,NiehansG,Hughson Mise M, Arii S, Higashitsuji H, Furutani M, Niwano M, MD, Moch H, Storkel S, Lerman MI, Linehan WM and Harada T, Ishigami S, Toda Y, Nakayama H, Fukumoto Zbar B. (1997). Nat. Genet., 16, 68 ± 73. M, Fujita J and Imamura M. (1996). Hepatology, 23, 455 ± Shiota G, Kawasaki H, Nakamura T and Schmidt EV. 464. (1995). Res. Commun. Mol. Pathol. Pharmacol., 90, 17 ± Miyazaki M, Gohda E, Tsuboi S, Tsubouchi H, Daikuhara 24. Y, Namba M and Yamamoto I. (1992). Cell Biol. Int. Shiota G, Rhoads DB, Wang TC, Nakamura T and Schmidt Rep., 16, 145 ± 154. EV. (1992). Proc. Natl. Acad. Sci. USA, 89, 373 ± 377. Montesano R, Matsumoto K, Nakamura T and Orci L. Shiota G, Wang TC, Nakamura T and Schmidt EV. (1994). (1991). Cell, 67, 901 ± 908. Hepatology, 19, 962 ± 972. MorishitaR,AokiM,NakamuraS,HigakiJ,KanedaYand SuzukiK,HayashiN,MiyamotoY,YamamotoM,Ohkawa Ogihara T. (2000). Ann. N. Y. Acad. Sci., 902, 369 ± 376. K, Ito Y, Sasaki Y, Yamaguchi Y, Nakase H, Noda K, Moriyama T, Kataoka H, Hamasuna R, Yokogami K, Enomoto N, Arai K, Yamada Y, Yoshihara H, Tujimura Uehara H, Kawano H, Goya T, Tsubouchi H, Koono M T, Kawano K, Yoshikawa K and Kamada T. (1996). and Wakisaka S. (1998). Biochem. Biophys. Res. Commun., Cancer Res., 56, 3004 ± 3009. 249, 73 ± 77. Tajima H, Matsumoto K and Nakamura T. (1991). FEBS Nakamura T, Nishizawa T, Hagiya M, Seki T, Shimonishi Lett., 291, 229 ± 232. M, Sugimura A, Tashiro K and Shimizu S. (1989). Nature, Takagi H, Scharp R, Hammermeister C, Goodrow T, 342, 440 ± 443. Bradley M, Fausto N and Merlino G. (1992). Cancer Ogasawara H, Hiramoto J, Takahashi M, Shirahama K, Res., 52, 5171 ± 5177. Furusaka A, Hiyane S, Nakada T, Nagayama K and Takagi H, Sharp R, Takayama H, Anver M, Ward JM and Tanaka T. (1998). Gastroenterology, 114, 775 ± 781. Merlino G. (1993). Cancer Res., 53, 4329 ± 4336. Otsuka T, Takayama H, Sharp R, Celli G, LaRochelle WJ, Takayama H, LaRochelle WJ, Anver M, Bockman DE and BottaroDP,EllmoreN,VieiraW,OwensJW,AnverM Merlino G. (1996). Proc. Natl. Acad. Sci. USA, 93, 5866 ± and Merlino G. (1998). Cancer Res., 58, 5157 ± 5167. 5871. Park M, Dean M, Cooper CS, Schmidt M, O'Brien SJ, Blair Takayama H, LaRochelle WJ, Sabnis SG, Otsuka T and DG and Vande Woude GF. (1986). Cell, 45, 895 ± 904. Merlino G. (1997a). Lab. Invest., 77, 131 ± 138.

Oncogene HGF overexpression promotes hepatocarcinogenesis N Horiguchi et al 1799 Takayama H, LaRochelle WJ, Sharp R, Otsuka T, Kriebel P, Van Belle E, Witzenbichler B, Chen D, Silver M, Chang L, Anver M, Aaronson SA and Merlino G. (1997b) Proc. Schwall R and Isner JM. (1998). Circulation, 97, 381 ± 390. Natl. Acad. Sci. USA, 94, 701 ± 706. Wu JC, Merlino G, Cveklova K, Mosinger B and Fausto N. Takayama H, Takagi H, LaRochelle WJ, Kapur RP and (1994a). Cancer Res., 54, 5964 ± 5973. Merlino G. (2001). Lab. Invest., 81, 297 ± 305. Wu JC, Merlino G and Fausto N. (1994b). Proc. Natl. Acad. TischerE,MitchellR,HartmanT,SilvaM,Gospodarowicz Sci. USA, 91, 674 ± 678. D, Fiddes JC and Abraham JA. (1991). J. Biol. Chem., Yaono M, Hasegawa R, Mizoguchi Y, Futakuchi M, 266, 11947 ± 11954. Nakamura T, Ito N and Shirai T. (1995). Jpn. J. Cancer Tuck AB, Park M, Sterns EE, Boag A and Elliott BE. (1996). Res., 86, 718 ± 723. Am.J.Pathol.,148, 225 ± 232. Yeh YC, Tsai JF, Chuang LY, Yeh HW, Tsai JH, Florine Ueki T, Kaneda Y, Tsutsui H, Nakanishi K, Sawa Y, DL and Tam JP. (1987). Cancer Res., 47, 896 ± 901. Morishita R, Matsumoto K, Nakamura T, Takahashi H, Zarnegar R and Michalopoulos GK. (1995). J. Cell Biol., Okamoto E and Fujimoto J. (1999). Nat. Med., 5, 226 ± 230. 129, 1177 ± 1180.

Oncogene