(TGF), assessed by immunostaining, are elevated in human BTC including GBC [22, 23]. Accumulating evidence suggests that COX-2, an inducible enzyme responsible for conversion of arachidonic acid to prostaglandins, may play a variety of roles in the gastrointestinal tract including pathogenic processes such as neoplasia [24]. A recent study demonstrated a relationship between erbB2 overexpression and COX-2 upregulation in human colorectal cancer cells [25]. Elevated COX-2 expression has been demonstrated in well-differentiated human hepatocellular carcinoma [26, 27] and GBC [28] compared with low or non-detectable COX-2 expression in poorly differentiated tumors. Very recently, Sirica’s group reported a strong positive correlation between the immunostaining intensities of erbB2 and COX-2 in BTC. COX-2 was observed not only in the furan rat cholangiocarcinoma model, but also in human cholangiocarcinomas [29], supporting the possibility that erbB2 plays a key role in regulating COX-2 expression in neoplastic and precancerous biliary tract epithelial cells. Grossman et al. [30] reported a specific COX-2 inhibitor, but not COX-1 inhibitor, decreased mitogenesis, and increased human gallbladder cell apoptosis associated with decreased prostaglandin E2 (PGE2). This suggests that the COX enzymes and the prostanoids may play a role in the development of gallbladder cancer and that COX-2 inhibitors may have a therapeutic role in gallbladder neoplasms [30].
of Growth Factor Signaling Pathways in Biliary Tract Cancer
To date, very few studies have addressed the molecular and cellular mechanisms underlying the development of BTC as described above; however, several lines of evidence suggest a role for the erbB receptor family. Overexpression and activation of erbB2 have been reported in a significant percentage of human BTC [15, 16, 31, 32]. In one study, 30 of 43 cases (69.6 %) and 14 of 43 cases (32.6 %) of GBCs had amplification of erbB2 DNA or overexpression of erbB2 protein, respectively [15]. In another study, 7 of 11 cases (63.6 %) of GBCs showed overexpression of erbB2 protein [16]. Yukawa et al. [17] reported erbB2 protein expression in 9 of 13 cases (69 %) of GBCs considered to be relatively early-stage tumors (all 13 cases were histologically diagnosed as well-differentiated tubular adenocarcinoma), yet erbB2 protein expression was undetectable in tumors that were more advanced. Furthermore, ErbB2 has been shown to be overexpressed in the neoplastic glandular epithelium of furan- and thioacetamide-induced intestinal-type cholangiocarcinomas in rat liver [33, 34]. It has also been reported that erbB2-transformed rat cholangiocytes, which overexpressed activated erbB2, obtained a tumorigenic feature when transplanted into isogenic rats, yielding a 100 % incidence of BTCs [34]. Overexpression and activation of epidermal growth factor receptor (EGFR) have also been reported in 30–60 % of BTC samples [31, 32, 35] and were shown to be correlated with negative clinical and pathologic features, such as distant metastasis and poor dedifferentiation [22, 36–38]. These data suggest that altered expression and activity of erbB2 and EGFR are major mechanisms underlying human BTC carcinogenesis [39].
Several lines of evidence suggest a role for the erbB receptor family as described above. A number of RTKs have been described [40–42]. Among them is the erbB family of RTKs consisting of the epidermal growth factor receptor (EGFR/erbB1), erbB2 (neu), erbB3, and erbB4 [43]. ErbB family RTKs have been shown to be important for normal development as well as in neoplasia [40, 44] (Fig. 1). Although all of the erbB family members share similarities in primary structure, receptor activation mechanism, and signal transduction patterns, they bind to different ligands. EGFR binds to and can be activated by a number of different ligands of the EGF family, including EGF, transforming growth factor- (TGF-), heparin-binding EGF-like growth factor (HB-EGF), amphiregulin (AR), betacellulin [45, 46], epigen (EP), and epiregulin (EREG). The neuregulin subfamily consists of various isoforms referred to as 1–4. These ligands bind to erbB4 and/or erbB3. Betacellulin, HB-EGF, and epiregulin have also been shown to bind to erbB4. Ligand-dependent activation of erbB family receptors can lead to heterodimerization, particularly of EGFR, erbB3 and erbB4 with erbB2. To date, no ligand has been identified for erbB2. ErbB3 cannot generate signals in isolation because the kinase function of this receptor is impaired, thus relying on interaction with erbB2 for signaling.
Fig. 1
ErbB family signaling system. Cross talk between erbB2/EGFR and other erbB receptor tyrosine kinase members and downstream signaling which leads to cell proliferation, survival, and migration (see text in detail)
Post-receptor signaling by activated erbB family members includes signaling through Ras/MEK/MAPK/Erk (extracellular signal-regulated kinase), phospholipase , signal transducer and activation of transcription (STATs), and phosphatidylinositol 3-kinase (PI3K) pathways that are common to nearly all RTKs (Fig. 1). Although the membrane-anchored peptide can be biologically active through juxtacrine signaling, in most cases, the extracellular domain is proteolytically cleaved by a metalloprotease activity present in the cell membrane. This process is known as “ectodomain shedding” and leads to the release of the soluble growth factor, which may act in an endocrine, paracrine, or autocrine fashion [47].
To allow paracrine or autocrine interaction of the EGFR ligands with the receptor, the membrane-tethered ligand precursors need to be released by a proteolytic reaction. This important step is mediated mainly by membrane-anchored metalloproteases of the ADAM (a disintegrin and metalloprotease) family [48]. ADAM17, which is also known as tumor necrosis factor- (TNF-)-converting enzyme, or TACE, together with ADAM10, is thought to play a central role. ADAM17 can cleave the AR, EREG, TGF-, and HB-EGF membrane-anchored precursors, while ADAM 10 is a key sheddase for EGF and BTC, and can also cleave the HB-HGF transmembrane precursor [45, 46, 49]. Transactivation of the EGFR by ligands of G-protein-coupled receptors (GPCRs) is perhaps the best characterized example of EGFR activation by heterologous ligands [48]. These include angiotensin II (ANG II), lysophosphatidic acid (LPA), endothelin-I, thrombin, IL-8, and prostaglandins such as PGE2 [48]. Different mechanisms have been proposed to mediate ADAM activation by GPCRs. Elevation of the intracellular levels of Ca2 or reactive oxygen species (ROS) is likely to be involved as well as phosphorylation reactions involving protein kinase C (PKC), ERK, or c-Src [48]. As previously indicated, transactivation of the EGFR is not exclusive of GPCR-triggered signaling. Studies carried out in keratinocytes have established that the expression and release of EGFR ligands can be elicited by the cytokines TNF- and interferon- (INF-) [50]. This has been recently observed also for the proapoptotic factor Fas ligand (FasL). Interestingly, it was shown that transactivation of the EGFR through the secretion of ligands such as AR contributed to mediate part of the inflammatory responses to FasL in human epidermis [51] (Fig. 2).
Fig. 2
ErbB2/EGFR transactivation through TLRs, GPCR, and TACE and cross talk between src which leads to the development of biliary tract cancer (see text in detail)
As mentioned, very few studies have attempted to decipher the molecular and cellular mechanism(s) involved in the development of BTC; thus, very little is known regarding the sequence of events that lead to this disease. A limiting factor has been the lack of relevant animal models for the study of early events in BTC. Presently available animal models are based on exposure to chemical carcinogens, and in most of these models, the latency between the treatment and tumor development is long and the tumor incidence is relatively low. However, the furan rat model described by Sirica et al. gives rise to a very high incidence of BTC, intrahepatic cholangiocarcinoma [52, 53]. In this model, treatment of rats with furan rapidly induced intestinal metaplasia and associated cholangiofibrosis in the right/caudate liver of rats [54]. Long-term treatment with furan (daily dose of 30 mg/kg of body weight, five times weekly by gavage for 9–13 weeks) resulted in the preferential development of cholangiocarcinoma [53]. The incidence of cholangiocarcinoma was 70–90 % in rats treated with furan by 16 months. The furan-induced cholangiocarcinoma in this rat model characteristically overexpressed erbB2, COX-2, and c-Met [54]. In addition to this model, combined treatment of Syrian golden hamster with dihydroxy-di-n-propyl nitrosamine and liver fluke infestation was shown to be associated with the enhancement of cholangiocarcinomas and preneoplastic lesions in the gallbladder [55].
Recently, we developed BK5.erbB2 transgenic mice, where expression of the rat erbB2 cDNA is targeted to the basal layer of multiple epithelial tissues, including the biliary tract epithelium [7, 56] (Fig. 3a). Adenocarcinoma of the gallbladder develops in 90 % of the homozygous BK5.erbB2 transgenic mice by 2–3 months of age [7]. The BK5.erbB2 transgenic mouse line represents the first genetically engineered mouse model for investigating the mechanism(s) underlying the development of GBCs and other BTCs. The remainder of this section will be devoted to a summary of this model and its initial utilization for preclinical therapeutic studies.
Fig. 3
a The DNA construct used to generate BK5.erbB2 mice. b Gross appearance and histological evaluations of BTC in BK5.erbB2 mice. A Gallbladder of wild-type mouse, B BK5.erbB2 mouse at 3 months of age, C anomalous fasciculus form of gallbladder in the early stage of gallbladder development (2 weeks of age) in BK5.erbB2 mouse, D H and E staining of gallbladder in wild-type mouse and E BK5.erbB2 mouse, F BrdU staining of gallbladder in wild-type mouse and G BK5.erbB2 mouse, H H and E staining of the ampulla of Vater, I intrahepatic cholangiocarcinoma, and J the junction of the pancreaticobiliary duct (JPBD) in a 3-month-old BK5.erbB2 mouse. c Two different pathways of development of GBC in BK5.erbB2 mice. (Upper figures) Carcinoma arising from hyperplasia in situ shown in an adenoma/hyperplasia/carcinoma sequence. (Lower figures) Carcinoma arising from hyperplasia shown in a de novo sequence. (Figure on left) Normal gallbladder from wild-type control mouse. Some of figures are adopted from Kiguchi K et al. [7]
Necropsy of adult BK5.erbB2 mice revealed that the gallbladder was dramatically enlarged and had a white, opaque appearance (Fig. 3bB). Enlarged gallbladders were often associated with a significantly dilated common bile duct (Fig. 3bB). This enlarged hepatic duct from the liver and the cystic duct from the gallbladder unite to form the enlarged common bile duct, which extends posteriorly through the pancreas and intestinal wall, where it opens to the mucosal surface of the duodenum as the ampulla of Vater (Fig. 3B and H). Most of the gallbladders in young BK5.erbB2 mice (<3 weeks) possess an anomalous fasciculus structure (Fig 3bC). The majority of the GBCs completely filled the lumen (Fig. 3E), although some showed focal lesions.
Analysis of the mucosa adjacent to the GBC observed in the mice allowed segregation into two categories based on etiology: carcinoma arising from hyperplasia in situ (HIS, 14 cases out of 34 GBCs from BK5.erbB2 mice, 41 %) or hyperplasia in whole mucosa (HIW, 59 %) (Fig. 3c). GBC tumors arising from HIW were more likely to be invasive (70 %, p < 0.01) compared to those arising from HIS (14 %). Tumors were characterized by branching structures with finger-like projections covered with high columnar epithelium and hyperchromatic nuclei. Most of the tumors were diagnosed as well-differentiated adenocarcinomas. Carcinoma cells frequently invaded into the surrounding connective tissues. In addition, hypervascularization was a characteristic feature of these tumors. Staining with CD31, a marker for endothelial cells, revealed extensive vascularization in the adenocarcinomas from BK5.erbB2 mice [7]. Adenocarcinomas from BK5.erbB2 mice exhibited a significantly elevated labeling index (a marker of proliferation) compared to normal gallbladder epithelium as determined by staining with antibromodeoxyuridine (BrdU) antibody (Fig. 3bG). Tumor cells of the common bile duct often invaded into the pancreatic duct (Fig. 3bJ). The ampulla of Vater was dilated, and hyperplasia of the epithelium was observed in transgenic mice. Pronounced congestion of bile, inflammation, necrosis, hyperplasia of biliary duct cells, and/or tumor development was also frequently observed in intrahepatic biliary ducts of transgenic mice (Fig. 3bI).
Persistent expression of the erbB2 transgene was observed in the epithelia of both gallbladder and intrahepatic biliary duct as well as in gallbladder adenocarcinoma (Fig. 4a) and cholangiocarcinomas [7