Potential conflict of interest: Nothing to report.
Chronic inflammation plays a critical role in oncogenesis in various human organs. Epidemiological studies have demonstrated that patients with primary sclerosing cholangitis have a predisposition to develop cholangiocarcinoma (CC). However, the molecular mechanisms that account for the development of bile duct carcinomas are not well defined. We recently provided evidence that activation-induced cytidine deaminase (AID), a member of the DNA/RNA editing enzyme family, is implicated in human tumorigenesis via its mutagenic activity. We found here that ectopic AID production is induced in response to tumor necrosis factor-α (TNF-α) stimulation via the IkappaB kinase-dependent nuclear factor-κB (NF-κB) activation pathway in human cholangiocarcinoma-derived cells. Aberrant expression of AID in biliary cells resulted in the generation of somatic mutations in tumor-related genes, including p53, c-myc, and the promoter region of the INK4A/p16 sequences. In human tissue specimens, real-time reverse transcription polymerase chain reaction (RT-PCR) analyses revealed that AID was increased significantly in 28 of 30 CC tissues (93%), whereas only trace amounts of AID were detected in the normal liver. Immunohistochemistry showed that all of the CC tissue samples examined showed overproduction of endogenous AID protein in cancer cells. Moreover, immunostaining for AID was detectable in 16 of 20 bile epithelia in the tissues underlying primary sclerosing cholangitis. Conclusion: The proinflammatory cytokine-induced aberrant production of AID might link bile duct inflammation to an enhanced genetic susceptibility to mutagenesis, leading to cholangiocarcinogenesis. (HEPATOLOGY 2008;47:888–896.)
Cholangiocarcinoma (CC) is an epithelial neoplasm that originates from the bile duct and can occur at any level of the biliary tree.1, 2 The incidence CC is increasing worldwide; it is the second most common primary hepatobiliary malignancy.2 Although most CC arise in the absence of apparent risk factors, chronic inflammation of the biliary epithelium plays a critical role for their development.2 In fact, primary sclerosing cholangitis (PSC) is the commonest predisposing condition for cholangiocarcinogenesis, and the prevalence of CC in patients with PSC ranges from 9% to 23%, with a cumulative annual risk of 1.5% per year of the disease.1 Other risk factors for cholangiocarcinogenesis are also associated with chronic biliary tract inflammation, including chronic choledocholithiasis, liver fluke infestation, hepatolithiasis, Caroli's disease, and hepatitis C viral infection.1 It has been hypothesized that the increased risk of CC in these conditions occurs because of chronic epithelial inflammation leading to cell proliferation, along with enhanced production of endogenous mutagens in the bile.1 However, the precise molecular mechanism that accounts for the development of CC on the basis of chronic biliary tract inflammation remains unsolved.
Activation-induced cytidine deaminase (AID) was originally identified as an inducer of somatic hypermutation, which diversifies the variable region of immunoglobulin genes in activated B cells in germinal centers.3, 4 However, animal models with constitutive expression of the gene for this enzyme revealed that aberrant AID production resulted in the accumulation of genetic mutations in various tumor-related genes, leading to lymphoid and nonlymphoid malignancies.5 Indeed, most AID transgenic (Tg) mice developed microadenomas of the lung epithelium as well as T cell lymphomas exhibiting frequent point mutations in the T cell receptor and c-myc genes that appeared to be caused by AID activity.5 Strikingly, we recently observed that constitutive expression of AID also caused the development of liver tumors with the morphological characteristics of hepatocellular carcinoma (HCC).6 Although AID gene expression is restricted to the lymphoid organs under physiological conditions, we observed aberrant AID expression in both human hepatocytes and gastric epithelial cells underlying areas of chronic inflammation.7, 8 Consistent with these in vivo findings, we showed that endogenous AID gene expression was induced by proinflammatory cytokine stimulation in human hepatocytes as well as gastric epithelial cells.6, 8 These findings suggest a role for AID in the development of cancers in the setting of chronic inflammation in human epithelial organs.
Although the origin of CC is not well understood, it has been proposed that both HCC and CC could develop from a common origin, such as hepatic stem or progenitor cells.9 The histogenesis of intestinal-type CC and combined hepatocellular and cholangiocellular carcinoma observed in experimental rodent models of liver carcinogenesis and in humans is consistent with the concept that at least some subtypes of CC derived from pluripotent liver stem cells.10, 11 Thus, as aberrant expression of AID in the liver can be genotoxic, leading to hepatocarcinogenesis, we were prompted to speculate that AID might be involved in cholangiocarcinogenesis. Therefore, in this study, we investigated the production and regulation of endogenous AID in human biliary epithelial cells in association with proinflammatory cytokine stimulation. We also examined whether there was aberrant AID production in human liver tissue specimens of PSC and bile duct cancers.
The study group consisted of 30 patients who had undergone potentially curative resection for a primary intrahepatic cholangiocarcinoma (ICC) at Kyoto University Hospital from 1995 to 2006. Selection of patients enrolled in this study was based on the availability of a sufficient amount of tissue for analysis. The patients included 15 men and 15 women, with a mean age at the time of surgery of 61.7 ± 12.3 years [mean ± standard deviation; range, 29–78 years; Table 1]. Moreover, the liver tissue specimens of 20 patients with PSC who received liver transplantation from 1999 to 2006 were examined for AID expression. As a control, 6 samples of normal liver tissues from patients with metastatic liver cancer were also examined. Biopsy specimens of tumor tissues at the proximal edge of freshly resected specimens were obtained and frozen immediately in liquid nitrogen. Written informed consent for the use of their resected tissues was obtained from all patients in accordance with the Declaration of Helsinki, and the Kyoto University Graduate School and Faculty of Medicine Ethics Committee approved the study.
Table 1. Clinicopathological Features of Patients with ICC
Total RNA was extracted from tissue specimens using the guanidinium–phenol–chloroform method (Sepasol; Nacalai Tesque, Kyoto, Japan).12 Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) for human AID messenger RNA (mRNA) amplification was carried out using the 7300 Real-Time PCR System (PE Applied Biosystems, Foster City, CA). The 6-carboxyfluorescein–labeled probe used for human AID mRNA was 5′-TCGGCGTGAGACCTACCTGTGCTAC-3′.6 Standard curves were generated for every target using a 10-fold serial dilution series of 5 independent transcripts derived from BL2-lymphoma cells that contained a high endogenous level of AID.8 Target complementary DNAs were normalized to the endogenous mRNA levels of the housekeeping reference gene 18S ribosomal RNA (18S rRNA).7 For simplicity, the ratios are given as relative values compared with the levels in a lysate from the nontreated cholangiocarcinoma-derived cells (RBE). Reproducibility was examined by comparing the results obtained from replicate samples during the same reaction run and those from independent runs on different days.8 The PCR procedures were performed at least 3 times for each sample, and results are expressed as the mean ± standard deviation in Figs. 1B through D and 2B and as the mean ± SEM (standard error measurement) in Fig. 3A, B.
The human CC cell lines HuCCT-1 and TFK-1 were obtained from the Cell Resource Center for Biochemical Research, Tohoku University; OZ was from the Japan Health Science Foundation (Tokyo); SSP-25 and RBE cells were from the RIKEN Bioresource Center (Tsukuba). These were cultured at 37°C in Dulbecco's modified Eagle's medium (Gibco-BRL, Tokyo, Japan) supplemented with 10% fetal bovine serum.
Plasmids and Reagents.
The expression plasmids pcDNA3-IκB kinase (IKK) pcDNA3-α, pcDNA3-IKK-β, and pcDNA3-RelA [nuclear factor-κB (NF-κB)] were as described.13 The expression plasmids pcDNA3-IκB-α-ΔN, pcDNA3-IKK-α (K44A), and pcDNA3-IKK-β (K44A), encoding the super-repressor form of IκB-α, and dominant negative mutants of IKK-α and IKK-β, respectively, were also as described.6 Small interference RNA (siRNA) duplexes composed of 21-nucleotide sense and antisense strands used for targeting IKK-γ/NEMO and AID were obtained from Dharmacon Research (Lafayette, CO). Recombinant human tumor necrosis factor-α (TNF-α) was purchased from Peprotech EC Ltd. (London, UK).
Recombinant Retrovirus Production and Infection of Biliary Cells.
The retroviral system for measuring the expression of the AID gene in cultured biliary cells was as described.14 A full-length complementary DNA for AID was subcloned into the EcoRI and XhoI restriction sites of the pFB vector (Stratagene, La Jolla, CA). The plasmids and packaging plasmids, pCL-Ampho (Imgenex, San Diego, CA), were transfected into 293T cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Preactivated cells were suspended in the medium containing retrovirus supplemented with 16 g/mL Polybrene (Sigma-Aldrich, St. Louis, MO), centrifuged for 40 minutes at 32°C, and incubated for 48 hours.
Subcloning and Sequencing of Tumor-Related Genes.
The oligonucleotide primers for human p53, c-myc, INK4A/p16, and k-ras are shown in Supplementary Table 1. Amplification of these genes was carried out using high-fidelity Phusion polymerase (Finnzymes, Espoo, Finland), and the products were subcloned by insertion into the pcDNA3 vector (Invitrogen).15 The resulting plasmids were subjected to sequence analysis using a DYEnamic ET terminator kit with AmpliTaq DNA polymerase (Amersham Pharmacia Biotech, Piscataway, NJ) on an automated sequencer (Applied Biosystems).
Immunoblotting and Immunohistochemistry.
Protein samples were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on 12% (wt/vol) polyacrylamide gels and subjected to immunoblotting analyses. The polyclonal antibodies against human AID were used in this study as described.6 Anti-α-tubulin antibodies were obtained from Calbiochem (San Diego, CA). Immunohistochemistry was carried out as described.16 To semi-quantitate the immunostaining results, the slides were scored independently by two evaluators (J.K. and H.H.) for AID staining. Visual assessment based on degree and intensity of immunoreactivity was classified as no staining (−), weak positive staining (+), moderate positive staining (++), and strong positive staining (+++).
Statistical differences in AID gene expression levels were analyzed using the Mann-Whitney nonparametric U test for real-time PCR results and the chi-squared test for immunohistochemical results. P < 0.05 was considered statistically significant.
Aberrant AID Expression Is Induced by Proinflammatory Cytokine Stimulation in Human Cholangiocarcinoma-Derived Cells.
To gain preliminary insights into the expression of the AID gene in human bile duct epithelium, expression of AID mRNA transcripts was first analyzed by RT-PCR in several cholangiocarcinoma-derived cells in the absence or presence of TNF-α, a proinflammatory cytokine that plays a central role in the pathogenesis of human sclerosing cholangitis.17–19 We found that endogenous AID mRNA expression was enhanced by TNF-α stimulation in RBE and SSP-25 cells, whereas only small amounts were detectable in the quiescent cells (Fig. 1A). Quantitative RT-PCR analyses revealed that TNF-α stimulation induced up-regulation of AID gene expression in all the biliary cells examined, including HuCCT-1, OZ, and TFK-1 (Fig. 1B). Notably, the expression of AID mRNA in RBE cells was increased more than 15-fold after stimulation with TNF-α, a comparable level of AID expression to that in the B cell lymphoma-derived cells, Tree92. Real-time RT-PCR analysis with 6-carboxyfluorescein–labeled probes also revealed that TNF-α treatment induced a dose-dependent increase in AID mRNA expression in RBE cells (Fig. 1C). Moreover, TNF-α induced a time-dependent transcriptional upregulation of AID mRNA in RBE cells, peaking at 8 hours, whereas the expression of 18S rRNA transcripts was unchanged (Fig. 1D). To confirm the TNF-α–mediated induction of AID mRNA expression in human bile duct–derived cells, we carried out immunoblotting analysis for the detection of endogenous AID protein. Only trace amounts were detectable in RBE cells without any stimulation. However, marked up-regulation of AID protein production was observed after treatment with TNF-α (Fig. 1E).
Because endogenous AID protein production was induced in response to TNF-α stimulation in biliary cells, we tested whether the AID gene would be regulated transcriptionally by the NF-κB. Activation of the NF-κB pathway converges on the IKK signals, a protein complex composed of 2 kinase subunits (IKK-α and IKK-β) and a noncatalytic subunit IκB kinase-γ (IKK-γ/NEMO). NF-κB is rendered inactive in unstimulated cells through binding of a specific NF-κB inhibitor, IκB-α protein. First, we examined whether synthesis of the positive regulators of NF-κB signaling affected AID expression and found that the expression of AID protein was substantially up-regulated by coproduction of the wild-type IKK-α, IKK-β or NF-κB itself (Fig. 2A). In contrast, the TNF-α–mediated AID mRNA expression was substantially reduced in biliary cells by coproduction of negative regulators of NF-κB, the dominant negative forms of IκB kinases or super-repressor form of IκB-α (Fig. 2B). Moreover, knockdown of endogenous IKK-γ by small interfering RNA resulted in the substantial reduction in the TNF-α–mediated AID expression (Fig. 2C). Taken together, these findings suggest that the proinflammatory cytokine TNF-α induces endogenous AID mRNA expression via NF-κB signaling in human bile duct–derived cells.
AID Activation Achieved Accumulation of Nucleotide Alterations in Tumor-Related Genes of the Human Cholangiocarcinoma-Derived Cells.
We demonstrated previously that aberrant AID gene expression is capable of triggering the accumulation of genomic mutations in human hepatocytes.6 To clarify whether proinflammatory cytokine-induced aberrant AID gene expression is genotoxic in biliary cells, we investigated whether AID caused somatic mutations in several tumor-related genes. For this purpose, the mutagenic effects of AID were determined using a retroviral vector-mediated AID gene expression system in HuCCT-1 cells. We investigated the overall somatic mutation frequencies in the p53 gene and in the promoter region of INK4A/p16, both of which have been reported to contain nucleotide alterations in human CC tissues.20, 21 In addition, we also investigated mutations in c-myc, which is thought to be the common target for abnormal gene editing in lymphoma cells of AID Tg mice.5 Accordingly, over 70 clones were randomly picked from the cells seven days after AID expression and subjected to sequence analyses. We first confirmed that no nucleotide alterations were detected in all of those tumor-related genes subcloned from the control cells (mutation frequencies less than 0.20 per 104 nucleotides). In contrast, nucleotide alterations appeared in both c-myc and the promoter region of the INK4A/p16 gene of the cells with AID gene expression (mutation frequency 0.51 and 0.72 per 104 nucleotides; Table 2). Interestingly, the nucleotide alterations induced by AID gene activation were clustered in exons 5 to 6 of the p53 gene, whereas exons 1 and 2 through 4 of the p53 sequences had no mutation among the clones isolated from the same cells with AID gene expression. In contrast to those 3 genes, no somatic mutations emerged in k-ras sequences after AID activation. These findings suggest that aberrant AID gene expression plays a role as a DNA mutator for some of the tumor-related genes in human biliary epithelium cells.
Table 2. Genomic Mutations in HuCCT-1 Cells with AID Expression
Number of mutated clones/number of clones examined is shown.
Values in parenthesis are number of mutated bases/number of total bases examined.
Endogenous AID Expression Is Up-regulated in the Human Bile Duct Epithelium Underlying Sclerosing Cholangitis and Cholangiocarcinomas.
The in vitro findings that endogenous AID gene expression was induced by proinflammatory cytokine stimulation prompted us to test whether aberrant AID expression is involved in human cholangiocarcinogenesis via bile duct inflammation. To examine AID gene expression in human bile ducts under physiological or pathological conditions, we first quantified the AID transcripts in normal liver and ICC tissues. Quantitative real-time PCR analyses revealed that 28 of the 30 ICC tissue samples (93%) showed up-regulation of AID gene expression, whereas it was transcribed only in trace amounts in the normal liver tissues (Fig. 3A). The mean AID/18S rRNA ratio in tumorous tissues (42.7 ± 15.6) was significantly higher than in normal liver (0.2 ± 0.1, P < 0.01). Next, we focused on the expression levels of endogenous AID mRNA in noncancerous liver tissues of the patients with ICC. Eight of 30 such samples (27%) exhibited the histological features of chronic cholangitis or hepatitis. In contrast, the remaining 22 showed no evidence of inflammatory changes in the liver tissue around the tumors. We found that the mean AID expression level of noncancerous tissues underlying chronic inflammation was 35.2 ± 23.9, significantly higher than those of the noncancerous tissue lacking inflammatory features (P < 0.01) (Fig. 3B).
To determine whether the increased AID expression in inflammatory liver tissues was derived from the biliary epithelial cells, hepatocytes, or infiltrating lymphocytes, we carried out immunostaining in various noncancerous liver specimens using antibodies specific for human AID. Specificity was confirmed by control blotting performed on AID-expressing lymphoid tissues (Fig. 3C–a). Immunoreactivity for endogenous AID was absent in the normal bile duct epithelium and hepatocytes in patients lacking hepatic inflammation (Fig. 3C–b). In contrast, AID protein expression was observed in both bile duct epithelium and inflammatory cells in the liver exhibiting chronic biliary inflammation (Fig. 3C–c). Immunoreactivity for AID was mainly detectable in hepatocytes as well as lymphocytes in the liver with underlying chronic hepatitis (Fig. 3C–d). Conversely, no immunoreactivity was detected when we used nonimmunized serum on those specimens (Fig. 3C–e, C–f). Taken together, these findings indicate that there was aberrant AID expression in bile duct epithelium in the liver with chronic inflammation and human CC tissues.
To further study the specific expression and precise localization of the AID protein in bile duct epithelium underlying chronic inflammation, we expanded the analyses regarding AID immunohistochemistry on liver tissue specimens of patients with PSC. We found that immunoreactivity for AID was detectable in the bile duct epithelium as well as in some of the lymphocytes infiltrating around the portal area in 16 of 20 (80%) liver specimens underlying sclerosing cholangitis (Table 3; Fig. 4). In contrast, no immunostaining for AID was observed in any of the normal liver tissues. Thus, the frequency of expression of moderate to strong positive immunostaining for AID protein in bile duct epithelium was significantly higher in the PSC livers than in the normal liver tissues (P < 0.05; Table 3). In CC tissues, all 20 tumor specimens examined showed positive AID immunostaining, and AID protein was observed in neoplastic cells mainly in the cytoplasm (Table 3; Fig. 5). We also confirmed that no immunostaining was obtained when nonimmunized serum or phosphate-buffered saline were used instead of the antibodies against AID in any of the tissue specimens (Fig. 5D,H,L). Taken together, these findings revealed that the AID protein is produced aberrantly in a substantial proportion of human bile epithelial cells with chronic inflammation and cholangiocarcinoma cells.
Table 3. Semiquantitation of AID Immunoreactivity in Normal Liver, Primary Sclerosing Cholangitis (PSC) and Intrahepatic Cholangiocarcinoma (ICC)
Various molecular alterations in relation to dysregulation of cell growth and survival pathways, invasion and metastasis, and tumor microenvironment have been reported to occur during the development of CC.10 In fact, many mutations in oncogenes and tumor suppressor genes have been identified in human CC tissues, suggesting that biliary neoplastic cells may arise from cellular and consequent DNA injury.22 However, how somatic mutations accumulate through the process of human cholangiocarcinogenesis is unknown. In the current study, we demonstrated that a recently identified DNA editing enzyme, AID, is induced by proinflammatory cytokine stimulation in biliary epithelial cells. Moreover, AID production caused multiple somatic mutations, which accumulated in some genes possibly involved in oncogenic pathways of the biliary cells. These findings suggest the involvement of aberrant AID gene expression in biliary epithelial cells in causing a high susceptibility to somatic mutations, which may lead to the development of bile duct neoplasms.
AID is a member of the DNA/RNA-editing cytidine deaminase, apolipoprotein B mRNA-editing enzyme catalytic-polypeptide (APOBEC) family that includes APOBEC-1, APOBEC-2, APOBEC-3A, APOBEC-3B, APOBEC-3C, APOBEC-3DE, APOBEC-3F, APOBEC-3G, APOBEC-3H, and APOBEC-4.23 The inappropriate expression of APOBEC family molecules could act as a DNA/RNA mutator and thus contribute to tumorigenesis.24 The first evidence for the oncogenic potential of the APOBEC family was shown using animal models with constitutive expression of the gene for APOBEC-1. APOBEC-1 Tg animals developed HCC via APOBEC-1–induced mutagenesis of inappropriate target genes including NAT-1.25 However, more remarkable phenotypical changes were observed in mice producing AID. Interestingly, AID Tg mice developed various forms of neoplasia, including T cell lymphomas, lung cancers, and HCC,5, 6 suggesting that AID acts as a genome mutator in various tissues including the liver and that aberrant AID gene expression might play a role in producing neoplastic cells in these organs.
One of the intriguing findings in the current study is that AID production was significantly up-regulated in human biliary epithelium cells in the setting of PSC as well as in CC tumor cells. PSC is characterized by chronic inflammatory damage of the biliary tree, and patients with PSC have a predisposition to develop CC.19, 26 How biliary epithelia underlying chronic inflammation develop cancers remains unclear. One hypothesis is that chronic biliary inflammation leads to the generation of cytokines and reactive oxygen species, causing irreversible DNA damage.2 Our current data showing that AID mRNA expression is mediated by proinflammatory cytokine stimulation via NF-κB in biliary epithelium could provide a link between chronic biliary inflammation and the development of CC. In fact, proinflammatory cytokine levels including TNF-α are up-regulated in patients with PSC.27, 28 It has been shown that proinflammatory cytokine-mediated NF-κB signaling pathways play a critical role in tumorigenesis. The mechanism by which IKK-β–dependent NF-κB activation drives tumor promotion is thought to be due to the transcriptional upregulation of anti-apoptotic target genes or cyclin D1 and other growth factors such as interleukin-6.29–31 In this study, we identified AID as a target gene of the IKK-β–dependent NF-κB activation pathway in bile epithelial cells. Thus, 1 possible mechanism for increased susceptibility to CC development under chronic inflammation is due to the aberrant expression of DNA mutator, AID, in the bile tract via NF-κB activation.
Various molecular alterations have been described during the development of CC.10 Among them, p53 is the most commonly mutated tumor suppressor gene and was shown to be implicated in CC developing in patients with PSC.20, 32 For example, mutated p53 protein was detectable in 31% of the CC tumors in patients with PSC, as opposed to negative findings in the control bile duct specimens.33 Another study revealed the accumulation of p53 protein in 79% of patients with CC, most of whom had underlying PSC.32 Our findings that aberrant expression of AID in biliary cells resulted in the emergence of nucleotide alterations in the p53 gene suggest that AID production might lead to the generation of a mutated p53 gene that plays a critical role of tumorigenesis. Alterations in the INK4A/p16 signaling pathway by homozygous deletions, exon mutations, promoter mutations, and methylation were also shown in CC, and more importantly in PSC.21AID gene expression in biliary cells induced the nucleotide alterations in the promoter region of the INK4A/p16 gene preferentially, whereas the k-ras gene was not mutated at all. It is unclear why the p53 and INK4A/p16 genes were more sensitive to AID activation compared with the k-ras gene in cholangiocarcinoma-derived cells. However, our current findings may be consistent with a previous observation that target gene selection for AID-mediated somatic hypermutation is variable among target cells.6
In conclusion, we demonstrated that proinflammatory cytokine stimulation is responsible for the aberrant AID gene expression in human biliary epithelial cells, providing a possible link between chronic biliary inflammation and the development of CC. Further analyses will be necessary to determine the significance of AID production on leading precancerous cells to acquire a critical number of genetic changes.
We thank Dr. T. Honjo and Dr. I.M. Okazaki for their useful suggestions and critical reading of our manuscript, and Dr. K. Tashiro for the establishment of retrovirus vector systems.