Background
Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related death worldwide [
1]. In most patients, HCC arises in the setting of chronic liver disease of various etiologies, with cirrhosis being present in about 80% of the cases [
2]. Chronic infection with hepatitis B virus (HBV) and hepatitis C virus (HCV) is responsible for over 80% of HCC cases worldwide. HBV was one of the first viruses for which a direct link with the development of HCC was demonstrated [
3],[
4]. Thus, HCC was the first human cancer for which a viral cause was established, and the first to be shown to be preventable by universal vaccination [
5]. Although the availability of an effective vaccine against HBV promises the eventual elimination of HBV-associated HCC, more than 350 million chronic carriers of HBV in the world are still at increased risk of developing cirrhosis and HCC, making HBV, along with tobacco, the most important environmental carcinogen [
6]. Although the causal association between HBV and HCC has been well established, the molecular mechanisms of hepatocarcinogenesis remain elusive.
The advent of post-genomic technologies has provided tools to investigate the pathogenesis of liver cancer, making study of the simultaneous expression of mRNA of thousands of genes in a single array possible [
7]. However, many studies of HCC derived from gene expression profiling have focused mainly on the host and little on the virus. HCC patients are often analyzed as a single group regardless of the etiologic factor involved, and the clinical, virologic and histologic features studied are often limited or missing. Moreover, there is limited information on the gene expression profiles of the surrounding non-tumorous tissue [
8],[
9].
We took advantage of a unique collection of liver specimens from patients who underwent orthotopic liver transplantation (OLT) or partial hepatectomy for HBV-associated HCC to study simultaneously host and viral factors that contribute to hepatocarcinogenesis. To the best of our knowledge, this is the first study that reports the results of an extensive microarray analysis in which up to 17 specimens per patient were analyzed, including samples from the tumor, the neighboring tissue, and the most distant non-tumorous tissues, along with the intrahepatic expression of HBV. Moreover, because the liver contains heterogeneous cell populations, we investigated the gene expression profiles of malignant versus non-malignant hepatocytes isolated by laser capture microdissection (LCM). The combined study of gene expression profiling of whole liver tissue (WLT) with malignant and non-malignant microdissected hepatocytes, along with the analysis of the intrahepatic expression and distribution of HBV, provided new insights into the molecular programs involved in the pathogenesis of HBV-associated HCC, opening new perspectives for the identification of novel tumor markers, which are needed for the early detection of HCC and for the development of novel forms of therapy.
Discussion
Our comparative analysis of multiple liver specimens sampled at various distances from the center of the tumor allowed us to demonstrate that the liver containing HBV-associated HCC is characterized by a sharp change in gene expression at the immediate perilesional area, within millimeters of the tumor margin. Moreover, we documented that this change is highly specific, as all genes down-regulated within the tumor were up-regulated in all non-tumorous liver areas and, conversely, all genes up-regulated in the tumor were down-regulated in all non-tumorous areas.
To the best of our knowledge, this is the first study that integrated gene expression profiling of whole liver tissue with that of microdissected hepatocytes from the same cohort of patients with HBV-associated HCC. Accruing evidence has documented the importance of the LCM technique to study isolated cell populations in tumors that arise in histologically heterogeneous tissues [
24]. The comparison of WLT and LCM data made it possible to distinguish three subsets of genes: genes detected in both WLT and LCM samples, genes detected only in LCM samples, and genes detected only in WLT samples. Interestingly, genes detected in both WLT and LCM samples showed comparable fold changes and the vast majority were down-regulated like most of WLT-unique genes. Conversely, most LCM-unique genes were up-regulated, indicating that malignant hepatocytes and whole tumor tissue diverge not only in the nature of genes differentially expressed but also in the overall direction of the change. Such a discrepancy underlines the importance of LCM to collect specific information on the gene regulation of malignant hepatocytes.
Among genes detected by both WLT and LCM samples, an overwhelming number of genes were involved in the metabolism of lipids and fatty acids, glucose, amino acids and drugs and were down-regulated. In agreement with our observations, previous studies showed down-regulation of metabolism-related genes in HBV-related HCC [
25], a feature that is preferentially linked to HBV- rather than to HCV-associated HCC [
26],[
27]. The reasons for this dramatic down-regulation of metabolism-associated genes are presently unknown, although some studies have suggested that this phenomenon is part of the de-differentiation program of liver tumor cells, which is particularly evident in HCC associated with HBV infection [
28]. Consistent with this hypothesis, Nagata
et al.[
29] found a remarkable down-regulation of cytochrome-associated genes in fetal human liver compared to adult liver, and explained this finding with the absence of hepatocyte-specific function in fetal livers.
Two of the most overexpressed genes in HBV-associated HCC, found both by LCM and WLT, that may be of particular interest are
AKR1B10 and
IGF2BP3. AKR1B10 has recently been associated with several tumors including, but not limited to, pancreatic carcinoma [
30], breast cancer [
31], and papillary renal carcinoma [
32]. However, studies in HCC are limited [
33]. Although its function is still largely unknown,
AKR1B10 was shown to deplete cells of retinoic acid, which controls cell proliferation [
34]. In support of a possible role of retinoic acid depletion, we also detected down-regulation of
RDH16, the key enzyme responsible for retinoic acid synthesis. Regarding
IGF2BP3, whose function is even less well established, its depletion has been correlated with a decrease in cell motility, invasion and transendothelial migration [
35]. In a recent study published by our group [
36],
AKR1B10 and
IGF2BP3 were found to be highly up-regulated in the liver of patients with HBV-associated acute liver failure and evidence of liver regeneration, emphasizing the need to further dissect the relationship between liver regeneration and liver cancer.
An important finding, which further highlights the value of LCM, is the up-regulation of four CTA genes (MAGEA3, NUF2, CEP55, and TTK) that we detected in microdissected malignant hepatocytes, and confirmed by real-time PCR. MAGEA3, one of the first CTAs associated with HCC, is a member of the MAGE gene family and a candidate for specific HCC immunotherapy [
37]. It has been proposed that MAGEA3 may enhance the ubiquitin ligase activity of TRIM28 and stimulate p53/TP53 ubiquitination by TRIM28 [
38]. The other three CTAs are directly involved, at various levels, in the mitotic machinery. CEP55 has previously been associated with HCC [
39], whereas, to our knowledge, NUF2 and TTK are novel HCC-associated CTAs. Members of the CTA family have been suggested as potential targets for cancer immunotherapy because, unlike most auto-antigens, they are highly immunogenic, even in autologous cancer-bearing patients.
One of the major goals of our study was to investigate the relationship between gene expression and viral biomarkers in the liver. The sharp change in gene expression that we documented between the perilesional area and the periphery of the tumor was mirrored by a significant decrease in HBsAg expression. Conversely, the levels of intrahepatic HBV replication were uniformly low in all the areas within and outside the tumor. Accordingly, HBcAg was not detectable in any liver specimens with the exception of a single non-tumorous area from a single patient. Although the low levels of HBV replication may be explained by the fact that these patients were all anti-HBe positive and under antiviral therapy with nucleos(t)ide analogues prior to surgery, the dramatic and significant decrease in HBsAg expression within the tumor cannot be explained by antiviral therapy, because nucleos(t)ide analogues have no direct effect on transcription and translation of HBsAg, which was not suppressed in the tumor-surrounding tissue. The reasons for the dramatic decrease of HBsAg within the tumor remain to be fully elucidated. Previous reports have found a lower expression of HBsAg in HCC as compared with matched non-tumorous tissues [
40],[
41]. One of the reasons proposed to explain this observation is an increased rate of integration of HBV DNA into the host genome inside the tumor [
40],[
42],[
43]. However, integration of HBV has also been reported in non-malignant hepatocytes [
40],[
42],[
43]. In a recent extensive genome-wide study, HBV integration was reported in 86% of HCC and in 31% of the surrounding non-tumorous tissue [
44].
Although a role for HBV integration in our cohort of patients cannot be excluded, one of the most interesting findings in our study was the down-regulation within the tumor of canonical pathways that are part of the nuclear receptor (NR)-associated network. This network is critically involved not only in the metabolic functions of the liver but also in the life cycle of HBV, acting as essential transcription factors for viral gene expression [
45],[
46]. Most notably, we observed down-regulation of PPARγ coactivator-1 alpha (PGC-1α), a key transcriptional co-activator that acts as a master switcher for a large number of nuclear receptors [
46]. PGC-1α serves critical functions in the control of cellular energy metabolic pathways [
47]. In the liver, it is a key regulator of gluconeogenesis and fatty acid oxidative metabolism, and coordinates adaptation to metabolic alterations. Shlomai et al. [
48], demonstrated in a mouse model that starvation, by turning on the gluconeogenic program, robustly induces HBV gene expression through the induction of PGC-1α, which serves as a co-activator of HBV transcription. The importance of PGC-1α in the HBV life cycle is further highlighted by the strong inhibition of HBV expression observed upon degradation of PGC-1α using the natural phenolic compound curcumin [
49]. In addition to NRs, an important non-NR partner of PGC-1α, FOXO1, was found to be down-regulated in tumor tissue. FOXO1 is a central mediator of glucose metabolism in the liver that was shown to bind HBV and activate its transcription [
50]. Thus, down-regulation of these genes may contribute to the sudden and significant decrease of HBsAg expression that we documented in the tumor. Through the exploitation of several liver-specific transcription factors and coactivators that regulate vital metabolic functions, HBV has acquired the ability to specifically replicate in liver tissue, reducing the risk of development of host cell resistance and leading to the definition of HBV as a “metabolovirus” [
51]. However, our data documented down-regulation of essential hepatic metabolism genes, including PGC-1α and FOXO1, suggesting that the malignant hepatocyte has developed an autonomous metabolic regulation. Given the fact that PGC-1α and FOXO1 are essential coactivators of HBV transcription, we hypothesize that down-regulation of these genes may contribute to the sudden and significant decrease of HBsAg expression that we documented in the tumor. Another possibility that we cannot rule out is whether the reduced expression of HBsAg may be related to the increased proliferation rate of tumor cells.
In conclusion, microarray analysis of multiple areas of livers with HBV-associated HCC provided evidence for a remarkable change of gene expression in the perilesional area, within millimeters of the tumor margin. This sudden change is paralleled by a sharp decrease in HBsAg expression. The combined application of WLT and LCM techniques validated most of the gene expression changes associated with the tumor. Moreover, LCM allowed us to identify a series of genes not found in whole liver tissue (WLT) that may play a role in HCC, adding a new tool for dissecting pathogenesis and discovering new cancer markers.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests
Authors’ contributions
Study concept and design: PF, FZ; Analysis and interpretation of data: MM, GD, DEK, FZ, JK, JL, GM, AT, REE, SB, CRB, JCH, JR-C, ME-B, SG, PF; Drafting of the manuscript: MM, PF, GD; Critical revision of the manuscript for important intellectual content: PF, GD, MK, DEK; Statistical analysis: GD; Study supervision: PF. All authors read and approved the final manuscript.