Cholangiocarcinoma in human liver continues to be a biomedically challenging cancer because early diag-nosis is rarely achieved and there is presently no effective curative treatment for patients with the advanced disease, nor is there any effective clinical therapy that will prevent the development of cholangiocarcinoma in patients with various predisposing risk conditions. Most of these risk conditions share, as a common feature, a long-standing inflammation and chronic injury to the biliary tract. They include primary sclerosing cholangitis, often associated with chronic inflammatory bowel disease, hepatolithiasis, fibropolycystic diseases of the biliary tract exemplified by choledochal cysts and Caroli's disease, and in the far East, infestation with the liver flukes Opisthorchis viverrini or Clonorchis sinensis.1–3 Intestinal metaplasia in the human biliary tract has also been reported to be an apparent intermediate lesion in the development of intestinal-type cholangiocarcinoma and gallbladder cancers,4–6 and occurs in the human biliary tract in association with specific risk conditions for cholangiocarcinoma, including hepatolithiasis, choledochal cysts, and O. viverrini infection.7–9
Specific mechanisms underlying the cellular and molecular pathogenesis of cholangiocarcinoma and its relationship to putative precursor lesions such as intestinal metaplasia remain unclear. However, there is increasing evidence to suggest that alterations in critical growth factor pathways may be contributing in a significant way to the development of this highly lethal cancer. In addition, aberrant expression of the homeobox intestinal-specific transcription factor CDX1 has recently been related to the histogenesis of intestinal metaplasia and associated intestinal-type cancers formed in the human stomach and esophagus,10 and in rat livers during chemically induced cholangiocarcinogenesis.11
During the past several years, our laboratory has focused on defining potentially critical alterations in specific growth factor and transcriptional regulatory pathways linked to the development of intestinal-type cholangiocarcinoma induced in the liver of rats treated with furan. The pathologic features of the furan rat model of intrahepatic cholangiocarcinogenesis are now well described,1,12–15 with the developed tumors closely resembling the histopathology of human intestinal-type or mucin-producing tubular hepatobiliary adenocarcinoma.
In this review, we will summarize our recently published findings on the overexpression of the growth factor receptor tyrosine kinases c-Neu (the rat homolog of human c-ErbB-2) and c-Met, the receptor for hepatocyte growth factor/scatter factor (HGF/SF) during furan-induced rat cholangiocarcinogenesis. We will also present an overview of our most recent findings concerning the expression of both HGF/SF and cyclo-oxygenase-2 (COX-2) in cancerous epithelium of rat cholangiocarcinoma. The relevance of these experimental findings will be defined in terms of currently available human data in order to underscore common pathways of growth dysregulation for both the rat and human tumors. Moreover, based on the data presented, we will discuss the possibilities of ultimately developing the furan rat model of cholangiocarcinogenesis into a powerful new model system for testing novel preclinical therapeutic strategies that have the potential to be translated into effective approaches for the prevention and/or treatment of the human disease.
c-ErbB-2/c-Neu in human and rat cholangiocarcinogenesis
The human proto-oncogene c-erbB-2 (also known as HER-2), and its rat gene homolog c-neu encode a 185-kDa transmembrane receptor tyrosine kinase sharing extensive homology with epidermal growth factor receptor (EGF-R). Limited but compelling immunohistochemical data generated from formalin-fixed paraffin-embedded tissue specimens have suggested that an overexpression of c-ErbB-2 receptor may represent a frequent aberration associated with human cholangiocarcinoma. Voravud et al.16 were the first to report increased c-ErbB-2 immunoreactivity in neoplastic epithelium of 46 of 63 (73%) human biliary tract adenocarcinomas from both Western and Eastern patients. They further observed eight of 10 tumors from Thai patients with O. viverrini infection as being positive for c-ErbB-2 immunostaining when reacted with a rabbit polyclonal anti-c-ErbB-2 antibody raised against a synthetic peptide corresponding to a sequence of amino acids mapping to the carboxy-terminus of the c-ErbB-2 protein. Both diffuse and plasma membrane staining were noted, and no correlation was observed between c-ErbB-2 immunoreactivity and tumor differentiation. Furthermore, no difference was reported in the frequency of c-ErbB-2 expression between the Western and Eastern patient cholangiocarcinomas analyzed. In comparison, c-ErbB-2 immunoreactivity was not detected by these investigators in the cells of normal fetal and adult human livers. By using a monoclonal antibody, Motojima et al.17 subsequently reported that 46 of 68 (68%) human biliary cancers of Japanese origin were immunoreactive for c-ErbB-2, and concluded from their results that c-ErbB-2 overexpression might have a prognostic value in bile duct carcinoma. Brunt and Swanson also used a monoclonal antibody to the extracellular domain of c-ErbB-2 to demonstrate the overexpression of c-ErbB-2 receptor protein in the neoplastic epithelial cells of four of six (67%) human cholangiocarcinomas of Caucasian origin.18 Each of the four positive tumors arose in patients with primary sclerosing cholangitis. In contrast, these investigators did not observe c-ErbB-2 immunoreactivity in cases of hepatoblastoma (n = 8), mixed cholangiocarcinoma– hepatocellular carcinoma (n = 1), bile duct adenoma (n = 1), or hepatocellular adenoma (n = 2), and in only two of eight (25%) cases of hepatocellular carcinoma analyzed. They further reported that neither bile ducts nor hepatocytes immunostained for c-ErbB-2 in three analyzed cases of primary sclerosing cholangitis in which no adenocarcinoma was present, that only weak immunostaining was seen in the larger bile ducts in two of seven cases of submassive or massive necrosis and in two examples of post-necrotic necrosis secondary to either hepatitis B or C virus infection, and that normal human fetal hepatocytes and bile ducts were non-reactive for c-ErbB-2. By using a monoclonal anti-ErbB-2 antibody comparable to that used by Brunt and Swanson, Chow et al. detected an overexpression of c-ErbB-2 in the neoplastic epithelium of five of 18 (28%) analyzed cases of intrahepatic cholangiocarcinoma diagnosed in Taiwan, but did not immunohistochemically detect enhanced expression of this proto-oncogene product in areas of biliary cell hyperplasias in livers of patients with hepatolithiasis, nor in atypical bile duct hyperplasia.19 In contrast to Motojima et al.,17 Chow et al. found no correlation between c-ErbB-2 overexpression in the biliary tract cancers analyzed and conventional prognostic indicators (e.g. such as tumor grade and stage). More recently, Terada et al. used an antigen-retrieval method combined with a commercially available monoclonal anti-c-ErbB-2 antibody to investigate c-ErbB-2 expression in cholangiocarcinoma and in hepatolithiasis compared with normal adult and fetal livers of Japanese origin.20 Similar to the initial findings of Voravud et al., these investigators found that 33 of 47 (70%) cases of cholangiocarcinoma overexpressed c-ErbB-2 receptor protein at neoplastic cell plasma membranes compared with intrahepatic biliary epithelia of normal adult and fetal human livers, which showed no detectable immunoreactivity. Also in agreement with the earlier reported immunohistochemical findings of Voravud et al.16 and Chow et al.,19 Terada et al. found that membranous expression of c-ErbB-2 in the cholangiocarcinomas did not correlate with tumor grade. However, in contrast to the negative findings found by Chow et al. for c-ErbB-2 expression in hepatolithiasis, Terada et al. detected positive immunohistochemical plasma membrane staining for c-ErbB-2 protein in hyperplastic intrahepatic bile ducts and proliferated peribiliary glands in 15 of 20 (75%) analyzed cases of hepatolithiasis, without evidence of intrahepatic biliary malignancy.
While the immunohistochemical studies described above suggest that c-ErbB-2 overexpression may represent a frequent change associated with human cholangiocarcinogenesis, it is also evident that in some of these studies, the number of cases of human cholangiocarcinoma analyzed were relatively few in number.18,19 Also, it is not clear from a number of the studies cited above to what extent strong plasma membrane immunostaining, considered to be a hallmark of c-ErbB-2 overexpression in malignant cells,21 was used as a criterion for identifying positive cases. In this context, a study by Collier et al.22 indicated an absence of membrane immunoreactivity for c-ErbB-2 in a total of 10 analyzed cases of cholangiocarcinoma of European Caucasian origin, when immunostaining of archival tissue sections was performed by using a primary monoclonal antibody raised against a synthetic peptide corresponding to the carboxy-terminus of the internal domain of the c-ErbB-2 protein. In contrast, it seems apparent that at least some of the contradictory immunostaining results obtained for c-ErbB-2 in the different human cholangiocarcinoma studies described above may be explained by differences in both sensitivity and specificity of the antibodies and immunohistochemical procedures used, as well as to variations in tissue sample preparation and condition of the archival human tissue specimens analyzed by the different research groups. As also noted, disagreement exists between different investigators17,19 as to whether c-ErbB-2 overexpression has prognostic value for human cholangiocarcinoma, although the overall immunohistochemical findings to date seem to indicate that c-ErbB-2 participates in human cholangiocarcinogenesis. However, a more comprehensive and better controlled molecular approach involving the analysis of a large number of cases of human cholangiocarcinomas of different etiologies, geographic and ethnic origins, and of different histiotypes, tumor grades, and stages, compared with precancerous and non-neoplastic intrahepatic biliary proliferative conditions is needed before the significance of c-ErbB-2 overexpression in human cholangiocarcinogenesis can be fully appreciated.
In order to begin to clarify the role played by c-ErbB-2 in cholangiocarcinogenesis, we used immunohistochemistry, western blot analysis, in situ hybridization, and northern and Southern blotting, to investigate the expression of c-Neu, the rat homolog of c-ErbB-2, in primary and transplantable cholangiocarcinomas that originated in the liver of furan-treated rats, and in a novel rat cholangiocarcinoma cell line, designated C611B, recently established by us from a furan-derived transplantable tumor.23 Briefly, our results demonstrated a marked overexpression of c-Neu receptor protein and mRNA transcripts, as representing a prominent feature of the neoplastic epithelium of furan-induced primary and transplantable rat cholangiocarcinomas when compared with rat hyperplastic bile ductular epithelium induced in rat liver by bile duct ligation or observed within putative precancerous cholangiofibrotic tissue formed in the liver of furan-treated rats.23–25 Moreover, we did not detect either c-Neu protein or mRNA in normal adult rat liver. In contrast, c-Neu protein and mRNA transcripts were also observed by us to be significantly increased in intestinal metaplastic glands within putative precancerous cholangiofibrotic tissue formed in rat liver after 6 weeks of furan treatment when compared with both normal and hyperplastic intrahepatic biliary epithelia.24,25 Here it is noteworthy that these earlier appearing intestinal metaplastic epithelial glands overexpressing c-Neu, as well as the neoplastic glandular epithelial cells of subsequently developed primary intestinal-type cholangiocarcinomas induced by furan, each exhibited increased proliferating cell nuclear antigen (PCNA) labeling indices that were three- to-fourfold higher than that of hyperplastic bile ductular epithelial structures formed within either cholangiofibrotic or bile duct-ligated livers.25
c-Neu protein overexpressed in transplantable rat cholangiocarcinomas derived from the furan model, as well as by the C611B rat cholangiocarcinoma cell line, showed in each case strong immunoreactivity for phosphotyrosine residues, indicating an active state (Fig. 1).25 However, we did not detect evidence of a common activating mutation at nucleotide T2012 of the transmembrane domain of c-Neu overexpressed in different in vivo passages of transplantable rat cholangiocarcinoma by using PCR followed by direct DNA sequencing analysis.25 Our preliminary Southern blot analysis also failed to detect evidence of c-neu gene amplification in these tumors,25 although this possibility needs to be investigated further by using more sensitive approaches, such as those using fluorescence in situ hybridization26,27 or laser capture tissue microdissection combined with PCR.28,29
Figure 1. Western blot of p185Neu immunoprecipitated (Imppt) with primary anti-Neu antibody from cultured C611B rat cholangiocarcinoma cells and from neu-transformed WB-F344 rat liver epithelial stem-like cells (WBneu) compared with cultured untransformed WB-F344 cells and RIPA buffer (negative non-cellular control). Each cell culture was maintained in basal medium supplemented with 10% fetal bovine serum without the addition of exogenous growth factors. (a) Membrane reacted with anti-Neu antibody; (b) after stripping, same membrane reacted with antiphosphotyrosine (anti-PTyr) antibody. Experimental conditions for western blotting and immunochemical demonstrations of phosphotyrosine immunoreactivity were similar to those described.25 These data support constitutive activation by tyrosine phosphorylation of p185Neu overexpressed in the respective tumorigenic cell lines C611B and WBneu relative to untransformed WB-F344 cells.
Download figure to PowerPoint
As is the case with the furan model, only limited studies to date have addressed possible mechanisms for c-ErbB-2 overexpression in human cholangiocarcinogenesis. It is of interest, however, that Shiraishi et al.30 recently demonstrated a gain of chromosome arm 17q, which harbors the c-erbB-2 gene, in six of 18 (33%) human biliary tract cancers. Shiraishi et al. also showed that this gain in 17q was not accompanied by c-erbB-2 amplification. In contrast, no information is presently available concerning the patterns of expression and tyrosine phosphorylation of other members of the EGF-R family (EGF-R itself, c-ErbB-3, c-ErbB-4) relative to c-ErbB-2/c-Neu in either human or rat cholangiocarcinogenesis. Defining these relationships is important because unlike EGF-R, the c-ErbB-2/c-Neu receptor has no homodimerizing ligand, but can be transactivated in the presence of epidermal growth factor by forming EGF-R/c-ErbB-2 heterodimers.31 In addition, transactivation of c-ErbB-2 has been demonstrated to occur in response to heregulin, a ligand for cErbB-3 and c-ErbB-4, inducing the formation of c-ErbB-2/c-ErbB-3 or c-ErbB-2/c-ErbB-4 heterodimers, and resulting in transphosphorylation of c-ErbR-2 receptor kinase tyrosine residues.32,33
c-Met in human and rat cholangiocarcinogenesis
There have only been a limited number of studies aimed at investigating c-Met expression in human cholangiocarcinogenesis.34,35 By using immunohistochemistry, our laboratory previously detected enhanced c-Met immunoreactivity in the cancerous epithelial cells of a significant percentage (58%) of human cholangiocarcinomas of Japanese origin when compared with bile ducts in age- and sex-matched normal adult control livers, as well as with intrahepatic biliary epithelia in normal fetal and postnatal preadult livers.34 c-Met immunoreactivity appeared to directly correlate with the degree of tumor differentiation. Most notably, well-differentiated tubular adenocarcinomas resembling in their morphology that of furan-induced rat cholangiocarcinomas exhibited the strongest immunoreactivity for c-Met, which was localized to both the plasma membrane and cytoplasm. Interestingly, and similar to the immunohistochemical findings of Terada et al.20 for c-ErbB-2, strong immunoreactivity for c-Met was also observed in hyperplastic and dysplastic biliary epithelia in a high percentage of analyzed cases of human liver with hepatholithiasis.34 However, as is the case for c-ErbB-2, no data are currently available relating c-Met mRNA levels to receptor protein overexpression in human primary cholangiocarcinomas compared with associated hyperplastic or dysplastic biliary cell types. It is also not known to what extent gene amplification might account for overexpression of c-Met in cancerous epithelial cells of human cholangiocarcinomas relative to non-neoplastic or precancerous hepatobiliary proliferative conditions.
More recently, we demonstrated c-Met protein and mRNA transcripts to be coordinately overexpressed with those of c-Neu in the neoplastic epithelium of primary and transplantable rat cholangiocarcinomas originated in the liver of furan-treated rats compared with adult rat hyperplastic bile ductules and normal intrahepatic biliary ducts.25 Like c-Neu, c-Met was also observed to be prominently overexpressed in intestinal metaplastic glandular epithelium within earlier-appearing putative precancerous hepatic cholangiofibrotic tissue, suggesting that an upregulation of c-Met, together with the upregulation of c-Neu represents a relatively early change in the cholangiocarcinogenesis process induced in rat liver by furan. Cultured tumorigenic C611B rat cholangiocarcinoma cells were also found by using western blotting to overexpress both c-Met and c-Neu when compared with rat hyperplastic bile ductular epithelial cells in primary culture (Fig. 2).23 We further reported on the constitutive tyrosine phosphorylation of c-Met overexpressed in transplantable cholangiocarcinomas derived from the furan model, but in a preliminary Southern blot analysis, we did not detect evidence of c-Met gene amplification in these tumors.25 Further studies, however, are needed to confirm and extend these latter preliminary findings. Nevertheless, in close agreement with our in vitro and in vivo findings for c-Met in rat cholangiocarcinoma cells, Yokomuro et al. recently demonstrated an enhanced expression of tyrosine phosphorylated c-Met protein and mRNA in the human cholangiocarcinoma cell line SG231 in vitro compared with primary cultures of normal human intrahepatic biliary epithelial cells.35 Immunohistochemical staining for c-Met was also reported by Yokomuro et al. to be stronger in the tumor cells within cholangiocarcinomas developed in the liver of severe combined immunodeficient mice (SCID) mice following intrahepatic inoculation of SG231 cells than that seen in reactive biliary epithelial cells in some areas of liver.
Figure 2. Western blot demonstrating overexpression of Neu (p185Neu) and Met (p170Met and p140Met) in C611B rat cholangiocarcinoma cells cultured for 4 days on rat type I collagen-coated plastic tissue culture wells compared with rat hyperplastic bile ductular epithelial cells in 4 day primary culture. β-Actin (structural protein) was used for normalization of the blot. (Reproduced with permission from Carcinogenesis 1999; 20: 2335–39). (a) Neu, (b) Met and (c) Actin.
Download figure to PowerPoint
Hepatocyte growth factor/scatter factor expression in rat and human cholangiocarcinoma epithelial cells
Hepatocyte growth factor/scatter factor has been previously shown to be a mitogen for both rat and human intrahepatic biliary epithelial cells,36–38 and to induce the upregulation of its receptor, c-Met.39,40 In a normal adult rat liver, HGF/SF is not produced by either the hepatocytes or intrahepatic biliary epithelial cells, but principally by hepatic stellate cells,41 and functions via paracrine signaling as a hepatotrophic factor in liver regeneration.42 Hepatocyte growth factor/scatter factor has also been observed not to be expressed in either rat43 or human35 hyperplastic (reactive) intrahepatic biliary epithelia, but is expressed in periductular fibroblastic cells associated with hyperplastic bile ductules induced in the liver of bile duct-ligated rats.43 In striking contrast, we recently demonstrated HGF/SF protein and mRNA to be aberrantly expressed in cholangiocarcinoma epithelium, as well as in putative precancerous intestinal metaplastic epithelium induced in the liver of furan-treated rats.43 Through the use of both in situ hybridization (Fig. 3) and RT-PCR, we further showed specific expression of HGF/SF mRNA in cultured C611B rat cholangiocarcinoma cells. We also detected prominent in situ hybridization signals for HGF/SF mRNA transcripts in the cytoplasm of all of the neoplastic epithelial glands observed in tumors developed in recipient rats after in vivo cell transplantation of C611B cells (Fig. 3).43 Comparable results on aberrant HGF/SF expression were also recently reported by Yokomuro et al.35 for the SG231 human cholangiocarcinoma cell line. Specifically, these investigators demonstrated that SG231 cells, like rat C611B cells, uniquely expressed HGF/SF mRNA and protein in culture, which was associated with the upregulation and autophosphorylation of the c-Met receptor. They also detected focal positivity for HGF/SF mRNA transcripts in the neoplastic epithelium of tumors formed following inoculation of the human SG231 cell line into SCID mice by using in situ hybridization.
Figure 3. Localization by in situ hybridization of hepatocyte growth factor/scatter factor (HGF/SF) mRNA transcripts in the cytoplasm of C611B rat cholangiocarcinoma cells cultured for 4 days in type I collagen gel culture and in resulting tumor glands after in vivo cell transplantation of cultured C611B cells into syngeneic Fischer 344 male rat recipients. In situ hybridization was performed by using a digoxigenin-labeled antisense rat HGF/SF riboprobe, as previously described.43 (a) Histological section of cultured C611B rat cholangiocarcinoma cells in type I collagen gel exhibiting strong cytoplasmic in situ hybridization signals (red granular staining) for HGF/SF mRNA. Arrows highlight the positive granular cytoplasmic staining reaction (× 132). (b) Representative histological section of tumor developed in the right inguinal fat pad of a recipient rat after cell transplantation of C611B cells showing positive in situ hybridization signals for HGF/SF mRNA transcripts in the cytoplasm of the neoplastic glandular epithelium (× 132). (c) Sense negative control for the cultured C611B cells in (a) (× 132). (d) Sense negative control for C611B cell-derived tumor in (b). (Reproduced with permission from Hepatology 2000; 31: 1257–65.)
Download figure to PowerPoint
Our previous findings23,25,43 and those of Yokomuro et al.35 strongly support the establishment of a HGF/SF–Met autocrine loop as being a potentially significant common feature of both rat and human cholangiocarcinoma epithelial cells. The detection of HGF/SF protein and mRNA in the cytoplasm of putative precancerous intestinal metaplastic glandular epithelium in the furan model, along with evidence of increased expression of c-Met in these cells, further suggests that acquisition of an HGF/SF–Met autocrine loop might represent a critical early event in cholangiocarcinogenesis, contributing to increased growth and promoting tumor development. Further studies are now needed to establish this possibility, as well as to determine if aberrant expression of HGF/SF in precancerous and cancerous biliary epithelial cells might have significance for improved early diagnosis and treatment of cholangiocarcinoma. In this regard, we are now pursuing an analysis of a large archival sampling of human cholangiocarcinomas and precancerous lesions associated with defined high-risk conditions, compared with various non-neoplastic hepatobiliary proliferative disorders, in order to evaluate if aberrant HGF/SF expression might serve as a unique marker for cholangiocarcinogenesis in the human liver.
Cyclo-oxygenase-2 expression in human and rat cholangiocarcinoma
Cyclo-oxygenase functions as the rate-limiting enzyme in the conversion of arachidonic acid to prostaglandins.44 Cyclo-oxygenase (COX)-1 and COX-2 represent two separate isoforms of cyclo-oxygenase. Although both of these isoforms catalyze the same reaction, COX-1 is constitutively expressed in most tissues to generate prostaglandins for normal physiologic functions, whereas COX-2 is not expressed in most tissues under normal physiologic conditions, but can be induced by various mitogens, cytokines, and tumor promoters.44–46 Cyclo-oxygenase-2 has also been demonstrated to be overexpressed in a variety of human gastrointestinal tract cancers, including colorectal, gastric and esophageal, and pancreatic adenocarcinomas,47–50 as well as in rodent colon adenocarcinomas.51 Furthermore, there is increasing evidence to strongly suggest that COX-2 may be playing an important role in colorectal carcinogenesis.44,52,53
In cholangiocarcinogenesis, COX-2 has only just recently become a focus of study. In a recent preliminary report, Jin et al.54 demonstrated increased immunohistochemical staining for COX-2 in the cancerous epithelium of 18 of 23 cases (78%) of human intrahepatic cholangiocarcinomas analyzed. These investigators further reported increased COX-2 immunoreactivity in dysplastic biliary epithelium around a tumor in a significant number of cases, whereas histologically normal bile ducts or bile ducts with reactive epithelial change were found to be only weakly positive for COX-2, irrespective of the presence of inciting agents, such as hepatoliths or C. sinensis. These findings prompted us to investigate COX-2 expression in cholangiocarcinomas induced in the liver of furan-treated rats, as well as in the C611B rat cholangiocarcinoma cell line.
Consistent with the human data of Jin et al.,54 we recently observed positive COX-2 immunoreactivity in the cytoplasm of neoplastic epithelium in five separate rat cholangiocarcinomas analyzed to date (Fig. 4), but did not detect COX-2 immunostaining in normal adult rat livers. We were further able to demonstrate by western blotting (Fig. 5) and RT-PCR (data not shown) selective prominent upregulation of COX-2 protein and mRNA in cultured C611B cholangiocarcinoma cells overexpressing c-Neu, as well as in a neu-transformed WB-F344 rat liver epithelial stem-like cell line, designated WBneu, which was recently established in our laboratory. Interestingly, the upregulation of the COX-2 protein and mRNA in both C611B cells and in WBneu cells paralleled markedly enhanced expression of tyrosine phosphorylated Neu receptor protein by both of these tumorigenic cell lines when compared with untransformed WB-F344 cells in culture. In addition, like C611B cells, WBneu cells spontaneously expressed HGF/SF mRNA, which was not detected in untransformed WB-F344 cells (data not shown). This latter finding complements the recently published findings of Presnell et al.,55 demonstrating acquired HGF/SF expression in nine of 16 spontaneously transformed WB-F344 rat tumor cell lines, together with the establishment of a functional autocrine HGF/SF-Met loop in a subset of spontaneous transformants concomitantly expressing HGF/SF and c-Met.
Figure 4. Immunohistochemical demonstration of cyclo-oxygenase-2 (COX-2) in the neoplastic glandular epithelium of a furan-induced rat cholangiocarcinoma. (a) Negative immunohistochemical control in which a COX-2 sequence-specific blocking peptide was used to completely neutralize COX-2 immunostaining in the tissue section (× 66). (b) Tissue section from the same tumor as in (a) (without blocking peptide pretreatment of primary COX-2 antibody) showing uniformly positive cytoplasmic COX-2 immunoreactivity in the neoplastic glandular epithelium of the cholangiocarcinoma (× 66). (c) Higher magnification photomicrograph demonstrating the positive immunoperoxidase staining reaction for COX-2 in the cytoplasm of the cholangiocarcinoma epithelium (× 132).
Download figure to PowerPoint
Figure 5. Western blot demonstrating p72/74 cyclo-oxygenase-2 (COX-2) to be distinctly upregulated in cultured neoplastically transformed WBneu rat liver epithelial stem-like cells and in C611B rat cholangiocarcinoma cells compared with cultured untransformed WB-F344 rat cells and radioimmunoprecipitation assay buffer (negative non-cellular control). The cultured cell types were maintained in basal medium that did not contain exogenous growth factors. Cyclo-oxygenase-2 immunoprecipitation and subsequent western blot analysis were carried out by using a goat polyclonal IgG directed against an epitope mapping at the carboxy terminus of rat COX-2, purchased from Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA.
Download figure to PowerPoint
c-ErbB-2/c-Neu and hepatocyte growth factor/scatter factor-Met regulation of cyclo-oxygenase-2
Data from a limited number of recent studies have suggested that activated c-ErbB-2/c-Neu receptor and HGF/SF-Met each may be contributing in significant ways to the regulation of the COX-2 pathway in cells. With respect to c-ErbB-2/c-Neu, Vadlamudi et al.56 have recently shown that the inhibition of c-ErbB-2/c-ErbB-3 signaling in cultured human colorectal cancer cells downregulates COX-2 expression in these cells. In contrast, activation of the c-ErbB-2/c-ErbB-3 signaling pathway by neu-differentiation factor beta 1 (NDF) resulted in induced activation of the COX-2 promoter, expression of COX-2 mRNA and protein, and accumulation of prostaglandin E2 in the culture medium. These findings suggest a possible role of the COX-2 pathway in the mitogenic action of NDF in colorectal cancer cells that appears to be regulated through activation of c-ErbB-2/c-ErbB-3 heterodimers. Cyclo-oxygenase-2 gene activation in a cultured normal rat gastric epithelial cell line (RGM1) was also recently demonstrated by Jones et al.57 to be triggered by HGF/SF through tyrosine phosphorylation of c-Met receptor and activation of the extracellular-regulated kinase 2 (ERK2) signaling pathway. Cyclo-oxygenase-2 has further been implicated an as inducer of interleukin-6 (IL-6).53,58 This latter finding is of potential significance for cholangiocarcinogenesis, as IL-6 has been demonstrated to function as an autocrine growth factor in different human cholangiocarcinoma cell lines.35,59,60 Park et al. have further reported that selective inhibition of either the p44/p42 mitogen-activated protein kinase (MAPK) pathway by PD98059 or of the p38 MAPK pathway by SB203580 blocked the proliferation response of cultured KMCH-1 human cholangiocarcinoma cells to IL-6.60 Interestingly, Jones et al. also observed that pretreatment of RGM1 non-malignant rat gastric cells with PD98059 abolished both HGF/SF-mediated induction of COX-2 and increased ERK2 activity in these cultured cells.57 However, in contrast to the inhibitory effect produced by SB203580 on the IL-6-mediated proliferative response of KMCH-1 cells, pretreatment of RGM1 cells with SB203580 had no effect on either the HGF/SF-induced increase in COX-2 protein levels or on the HGF/SF-induced increase in ERK2 activity.
In conclusion, the evidence presented in this review strongly suggest that prominent overexpression and activation of the growth factor receptor tyrosine kinases, c-ErbB-2 and c-Met, together with the upregulation of COX-2 may represent potentially critical alterations associated with the molecular pathogenesis of both human and rat cholangiocarcinoma induced in the furan model. In addition, recent demonstrations of HGF/SF mRNA as being aberrantly expressed in human and furan-induced rat cholangiocarcinoma cells, and in a subset of putative precancerous intestinal metaplastic epithelium appearing in the liver of furan-treated rats subsequent to the development of frank malignancy, indicate a novel change not seen in normal or reactive non-neoplastic intrahepatic biliary epithelial cells of either species, but one that may possibly herald an expression of the neoplastic phenotype. Clearly, further studies are needed to validate this possibility, as well as to determine if relevant functional relationships exist between c-ErbB-2/cNeu, HGF/ SF-Met, and COX-2 in human and rat cholangiocarcinogenesis. It also remains to be determined if MAP(ERK) kinase signaling may be playing a significant role in the upregulation of COX-2 in cholangiocarcinoma cells overexpressing c-ErbB-2/c-Neu and/or HGF/SF-Met.
The findings reviewed herein also justify future research aimed at the likelihood of selectively exploiting c-ErbB-2, HGF/SF-Met, and COX-2 in cholangiocarcinogenesis for the development of novel diagnostic, chemoprevention, and/or therapeutic strategies. A number of distinct approaches have already been developed to inhibit the expression or function of c-ErbB-2/c-Neu in malignant cell types in vivo and in vitro,61 and the humanized monoclonal antibody Trastuzumab (Herceptin, Genentech Inc., South San Francisco, CA, USA) against c-ErbB-2 receptor protein has yielded promising results alone and particularly in combination with chemotherapy in the treatment of a subset of breast cancer patients.62,63 Recent studies have also demonstrated the clinically approved selective COX-2 inhibitor, celecoxib (Celebrex, Pharmacia, Peapack, NJ, USA), to exert chemopreventative and antitumor activity in experimental models of colon cancer,64–66 and to produce a significant reduction in the number of colorectal polyps in patients with familial adenomatous polyposis.67 In a preliminary study, we have also found that the COX-2 inhibitor, NS-398, produced dose-dependent growth inhibition in vitro of C611B rat cholangiocarcinoma cells (Fig. 6), as well as suppressed anchorage-independent growth of these cells in soft agar (Fig. 7). It has been further shown that the four-kringle antagonist HGF/NK4 for HGF/SF significantly suppressed tumor growth in vivo in nude mice subcutaneously transplanted with GB-d1 human gallbladder carcinoma cells.68 Based on these promising results, it is reasonable for us to now propose the development of a novel strategy that combines selective targeting of c-ErbB-2/c-Neu with that of COX-2 and/or autocrine c-Met-HGF/SF as having the potential of being a useful new approach in the treatment and/or chemoprevention of cholangiocarcinoma. In this regard, it is anticipated that the furan model of cholangiocarcinogenesis together with the C611B rat cholangiocarcinoma cell line will prove to be powerful preclinical models for testing the merits of this strategy and hopefully provide support for possible future clinical trials.
Figure 6. Dose-dependent inhibition of cell growth induced by the cyclo-oxygenase (COX)-2 inhibitor NS-398 (Cayman Chemical, Ann Arbor, MI, USA) in C611B rat cholangiocarcinoma cell cultures. Experimental cultures were exposed for 4 days to NS-398 at the indicated μmol/L levels administered in dimethyl sulfoxide (DMSO) at a final concentration of 0.1% per culture. Solvent control cultures contained only 0.1% DMSO. After 4 days of treatment, the amount of cell growth per individual cultures was determined by using the Cell Titre 96® Aqueous non-Radioactive Proliferation Assay Kit from Promega, Madison, WI, USA. For each dose point, n = 15. (●) C611B ChC cells.
Download figure to PowerPoint
Figure 7. (a) Phase contrast photomicrograph of a dimethyl sulfoxide (DMSO) solvent control culture demonstrating prominent anchorage-independent growth of C611B rat cholangiocarcinoma cells in soft agar following 3 weeks of daily exposure to 0.1% DMSO (× 33). (b) Phase contrast photomicrograph of NS-398-treated culture demonstrating marked inhibition in both size and number of C611B rat cholangiocarcinoma cell colonies (arrows) formed in soft agar following 3 weeks of daily treatment with the cyclo-oxygenase-2 inhibitor NS-398 (200 μmol/L) added in 0.1% DMSO (× 33).
Download figure to PowerPoint