bHLH/PAS proteins play important roles in tumor progression. Lost or reduced expression of single-minded homolog 2 (SIM) as well as aryl hydrocarbon receptor repressor (AHRR) has been observed in cancerous human tissues. Here, we investigated the role of aryl hydrocarbon receptor nuclear translocator (ARNT), another bHLH/PAS protein, in hepatocellular carcinoma (HCC). Using tissue microarray and immunohistochemistry, we found that intratumoral ARNT was inversely correlated with time to recurrence and overall survival of HCC patients after resection. Knockdown of ARNT in HepG2, HCCLM3 and HCCLM6 cells significantly shortened cell doubling time, increased S-phase cell populations and accelerated in vivo HCCLM6 growth and metastasis. After ARNT expression was rescued, prolonged cell doubling time and decreased S-phase cell populations were observed in HepG2, HCCLM3 and HCCLM6 cells. And, HCCLM6 growth and metastasis in vivo were remarkably inhibited. Screening by quantitative reverse-transcription PCR and PCR arrays revealed that cyclin E1, CDK2, Fos and Jun were negatively regulated by ARNT, whereas CDKN1C, CNKN2A, CDKN2B, MAPK11 and MAPK14 were positively regulated in HCC. According to the results of immunoprecipitation assay, both ARNT/ARNT and ARNT/AHRR complexes were clearly formed in HCCLM6 xenograft with increased ARNT expression. In summary, ARNT is an important regulator of HCC growth and metastasis and could be a promising prognostic candidate in HCC patients.
Hepatocellular carcinoma (HCC) is the third leading cause of cancer death in the world and the second in China.1, 2 The long-term prognosis of HCC patients after hepatectomy still remains a challenge, largely due to its high recurrence rate and metastasis.3 The molecular mechanisms of HCC progression need be further investigated for new insights and interventions against metastatic recurrence.
Aryl hydrocarbon receptor nuclear translocator (ARNT), also known as hypoxia-induced factor-1β (HIF-1β), is one of the most important nuclear transcription factors within the basic helix-loop-helix/Per-ARNT-SIM (bHLH/PAS) superfamily. It is widely expressed in human cells, including hepatocytes. When responding to different extracellular stimuli, ARNT can form a heterodimeric complex with aryl hydrocarbon receptor (AHR), hypoxia-inducible factor-1α (HIF-1α) and its homologous factors (HIF-2α, HIF-3α), or with single-minded homolog 2 (SIM) to mediate various biological actions, such as hypoxia reaction, xenobiotic metabolism, teratogenesis, immunosuppression4 and embryonic development.5 However, its onco-biological function is still unclear.
Several prototypes of bHLH/PAS factors have been identified as being involved in tumor progression. HIF-1α promotes tumor progression and metastasis via its regulation of cancerous glycolysis,6 proliferation, apoptosis7 and angiogenesis.8 AHR appears to be a regulator of cell proliferation, albeit growth-inhibitory and promoting roles in MCF-7 and HepG2 cells, respectively, are observed.9 SIM is frequently lost or reduced in primary breast tumors, promoting malignant transformation of cells and tumor invasion.10 More recently, aryl hydrocarbon receptor repressor (AHRR) has been identified as a tumor suppressor in multiple human cancers.11 These findings collectively suggest that nuclear transcription factors of bHLH/PAS play critical roles in tumor progression. Although ARNT has been widely studied in tumor angiogenesis and tumorigenesis in the past 2 decades, its role in tumor progression, especially in HCC, has not been explored. In our study, the expression levels of ARNT in HCC surgical specimens were investigated and their prognostic significance was then analyzed. Furthermore, the effects of ARNT on HCC growth and metastasis were experimentally tested to confirm clinical observations.
One hundred five patients were randomly enrolled in this study. All patients underwent curative liver resection between January 1999 and March 2006 and had histological confirmation of HCC.12 None of them received any preoperative anticancer treatment. Of the enrolled patients, 90 had a history of hepatitis B. Preoperative liver functions of all patients were classified as Child-Pugh A. Tumor stages were determined by TNM classification according to the 2002 International Union Against Cancer guidelines.13 Tumor differentiation was graded by the Edmondson grading system. The Scheuer system was applied in 100 patients for grading (necroinflammatory activity) and staging (fibrosis and cirrhosis) of the nontumor liver tissue (Supporting Information Table 1; the surrounding liver tissue was not adequate for scoring in five patients)14, 15 The study was approved by the Zhongshan Hospital Research Ethics Committee. Informed consent was obtained from each patient according to the committee's regulations.
Patients were followed until March 2008, with a median follow-up time of 31.2 months. Briefly, all patients were evaluated every other month during the first year and thereafter at least every 3–4 months. At each visit, alpha fetoprotein measurement and liver ultrasonography were performed. A computed tomography scan was performed on the abdomen every 6 months. A bone scan or magnetic resonance imaging was conducted if localized bone pain was reported.12, 16 Treatment modalities following a relapse were administered based on uniform guidelines as previously described.12, 17, 18 Overall survival (OS) was defined as the interval between surgery and death, and time to recurrence (TTR) as the interval between surgery and recurrence. Patients in whom recurrence was not detected were censored on the date of death or the last follow-up.
Tissue microarray, immunohistochemistry and evaluation
Immunohistochemical staining for molecules of interest was performed on tissue microarrays made of formalin-fixed, paraffin-embedded tumor resection specimens. Two cores were taken from the margin of the tumor and from normal liver tissue within 10 mm of the tumor (Shanghai Biochip Company Ltd, Shanghai, China).16 Rabbit primary antibodies against human ARNT, HIF-1α, CD31 and CD34 (1:200; Santa Cruz Biotechnology, Santa Cruz, CA) were used with the components of the Envision-plus detection system (EnVision + HRP/Mo; Dako, Carpinteria, CA). The reaction products were visualized via incubation with 3,3′-diaminobenzidine. The negative controls were treated identically but without primary antibody. The expression of ARNT and HIF-1α was analyzed comprehensively by the extent of staining and the range according to the Fromowitz standard.19, 20 In brief, five visual fields were randomly observed and 100 cells in each field were counted. The range scores were graded as 0, 1, 2, 3 and 4 when 0–5%, 6–25%, 26–50%, 51–75% and >75% of the examined cells were positively stained, respectively. The extent scores were graded as 0 or 1, 2, or 3 when the examined cells were unstained or stained light yellow, brown or dark brown, respectively. The expression status of the target protein was judged jointly by its range and extent score. If the total scores were ≤3, the protein was regarded as having low expression; otherwise, it was considered as highly expressed. CD31 as well as CD34 levels in cancerous and liver tissues were detected as previously described.21
Human hepatoma HepG2, HCCLM3 and HCCLM6 cell lines and their derivates were used in the study; HCCLM3 and HCCLM6 are hepatitis B virus (HBV) positive. All cell lines were cultured in Dulbecco's modified Eagle's medium (Gibco BRL, Grand Island, NY) supplemented with 10% fetal bovine serum and maintained at 37°C in a humidified incubator containing 5% CO2.
Construction and infection of lentiviral vectors for modulating ARNT expression
ARNT v1 (NM_001668) and ARNT v2 (NM_178426), two transcripts of approximately 2.37 kb and 1.0 kb, respectively, were constitutively expressed in liver tissues. Four sequence-specific shRNAs against human ARNT v1 and ARNT v2 were designed and evaluated in 293T cells by Western blot and immunofluorescence assays. The stem-loop DNA oligonucleotides with the highest knockdown efficiency were (1) sense: 5′-CTC AGA TGA AAT TGA GTA C-3′ and antisense: 5′-GTA CTC AAT TTC ATC TGA G-3′ against ARNT v1; (2) sense: 5′-ATG ACC CAG CCT GAG GTC T-3′ and antisense: 5′-AGA CCT CAG GCT GGG TCA T-3′ against ARNT v2 (Supporting Information Fig. 1A). A mock oligonucleotide (sense: 5′-TTC TCC GAA CGT GTC ACG T-3′ and antisense: 5′-AAG AGG CTT GCA CAG TGC A-3′) was used against a scrambled human gene. One mutant of ARNT v1, starting at oligonucleotide position 1,348 and ending at position 1,368 was successfully changed from TAC TCA GAT GAA ATT GAG TAC to TAT TCT GAC GAG ATA GAA TAT. The mutant together with a FLAG tag encoding sequence was cloned into a pLVTHM vector (Shanghai Genechem Co., China; Supporting Information Fig. 2 and Supporting Information Table 2) and was able to be translated into the same amino acids as the wild-type gene. All lentiviral particles were prepared as previously described.22 Wild-type (WT) HepG2, HCCLM3 and HCCLM6 were infected with either Lenti-ARNTi-v1 virus, Lenti-ARNTi-v2 virus or Lenti-Mock virus. In addition, HCCLM6 cells were co-infected with Lenti-ARNTi-v1 and Lenti-ARNT-v1 viruses to rescue ARNT expression. All vectors except Lenti-ARNT-v1 expressed the green fluorescent protein (GFP) signal.
Detection of ARNT-regulated genes by quantitative reverse-transcription PCR (qRT-PCR) and qRT-PCR array
Total RNA was extracted by the RNeasy® Mini kit (Qiagen, Valencia, CA). The quality of RNA from HCCLM6 and its derivates (Supporting Information Fig. 1B) was assessed via A260/280 absorbance, and 1.0 μg of total RNA was used to synthesize the first-strand cDNA with MuLV reverse transcriptase (Applied Biosystems, Foster City, CA) at 42°C for 60 min and then at 95°C for 5 min. The forward and reverse primers of qRT-PCR were 5′-AGT GGC GTT TAA GTC CCC TGA-3′ and 5′-GGG ATA CTG CGG CAG TAG CA-3′ for cyclin E and 5′-CAA GCC AGT ACC CCA TCT TCG-3′ and 5′-CAA ATA GCC CAA GGC CAA GC-3′ for CDK2, respectively. The reactions were performed on a DNA Engine Opticon system (MJ Research, Reno, NV) using SYBR® Green PCR Master Mix (Applied Biosystems). Following each cycle, SYBR green fluorescence was monitored and the melting curve was analyzed to ensure that a single PCR product was obtained. Afterwards, the size and specificity of amplicons were confirmed by 2.5% agarose gel electrophoresis. All reactions were repeated in three separate runs and evaluated with the Opticon Monitor software (Version 1.02). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was employed to normalize the samples. RNase-free water (Qiagen) was included as a negative control in RNA extraction and in each run. More cell cycle-control genes were screened by RT Profiler PCR Arrays (PAHS-061A, SABiosciences, Frederick, MD) and performed by Kangchen Bio-tech Co. (Shanghai).
Detection of ARNT protein by Western blotting
ARNT protein in HCCLM6 cells and their derivates was evaluated by Western blotting. About 20 μg of total protein was extracted and separated by 10% SDS-PAGE, transferred onto polyvinylidene fluoride membranes, and then reacted with primary antibodies against ARNT (1:200) and GAPDH. After being extensively washed with phosphate-buffered saline (PBS) containing 0.1% Triton X-100, the membranes were incubated with alkaline phosphatase-conjugated goat anti-rabbit antibody for 30 min at room temperature. The bands were visualized using 1-step™ NBT/BCIP reagents (Thermo Fisher Scientific, Rockford, IL) and detected by an Alpha Imager (Alpha Innotech, San Leandro, CA).
Cell proliferation and cell doubling time
HCCLM6 cells and their derivates in exponential growth phase were trypsinized to yield a single-cell suspension. A total of 2 × 104 cells in 1 ml of medium was added to each well of 24-well plates and incubated at 37°C in 5% CO2. Cell numbers in three replicate wells were counted by Coulter Counter after 72 h incubation. Cell doubling time was calculated by the following formula: TD = T(log 2)/log(N/N0) (TD, doubling time; T: time interval; N0, initial cell number; N, endpoint cell number).23 Viable cells were also determined by Cell Count Kit-8 (CCK-8) assay (Dojin Laboratories, Kumamoto, Japan).24 Briefly, WT HepG2, HCCLM3 HCCLM6 and their derivates with ARNT knockdown or rescued expression were seeded into 96-well plates at an initial density of 2 × 103 cells/well. After 24-, 48- or 72-h incubation, 10 μl of the kit reagent was added to each well and the plates were incubated for another 3 h. All plates were then scanned by a microplate reader at 450 nm. Cell proliferation was calculated on the basis of absorbency.
Cell cycle analysis
About 1 × 106 HCCLM6 cells and their derivates were harvested, washed with cold PBS twice and fixed with 70% ethanol solution. The pellets were then resuspended in PBS and incubated with RNase A solution at a final concentration of 100 μg/ml at 37°C for 30 min. After staining with 1 mg/ml of propidium iodide, the cell cycle was measured by flow cytometry (BD Biosciences, San Jose, CA).
Tumor growth assays in vivo
Ten nude mice (Institute of Materia Medica, CAS, Shanghai, China) were divided into groups of two mice apiece. Both mice in each group were injected in the upper right flank region with 1 × 107/0.2 ml of HCCLM6-WT, HCCLM6-Mock, HCCLM6-ARNTi-v1, HCCLM6-ARNTi-v2 or HCCLM6-ARNT-v1, respectively, to establish subcutaneous xenograft models. Four weeks later, subcutaneous tumors 1 cm in diameter were removed, cut into 1-mm3 pieces and implanted into the livers of other 30 mice to establish orthotopic xenograft models.25In vivo green fluorescence imaging of HCCLM6-Mock, HCCLM6-ARNTi-v1, HCCLM6-ARNTi-v2 and HCCLM6-ARNT-v1 xenografts was done once a week (stereomicroscope: Leica MZ6; illumination: Leica L5 FL; C-mount: 0.63/1.25; CCD: DFC 300FX) and quantified with Image Pro Plus software. On day 42, all mice were sacrificed and spontaneous metastasis to the lung was observed by the presence of GFP signals. All procedures were approved by the Animal Care and Use Committee of Shanghai, China.
Approximately 1 mg of tissue from HCCLM6-WT, HCCLM6-Mock, HCCLM6-ARNTi-v1 and HCCLM6-ARNT-v1 xenografts was lysed with 3 ml of RIPA buffer at 4°C for 30 min. One ml of lysate was incubated with 0.25 μg of rabbit IgG together with 20 μl of protein A-agarose at 4°C for 30 min and centrifuged at 10,000 rpm for 10 min. The supernatant was incubated with 2 μg of rabbit anti-human ARNT (Santa Cruz Biotechnology) or anti-FLAG antibody (Sigma-Aldrich, St. Louis, MO) at 4°C for 2 h, and then with 20 μl of protein A-agarose at 4°C overnight. Pellets were washed three times with RIPA buffer before analysis for HIF-1α, p300, AHR, ARNT (1:200; Santa Cruz Biotechnology, Santa Cruz, CA), AHRR (1:1,000; Abcam, Cambridge, MA) and FLAG levels (1:5,000; Sigma-Aldrich, St. Louis, MO) by the method described in Detection of ARNT protein by Western blotting.
Statistical analysis was performed with SPSS 16.0 for Windows (SPSS, Chicago, IL). Pearson chi-square test or Fisher exact test was applied for the comparison of qualitative variables, and quantitative variables were analyzed by ANOVA and Pearson's correlation test. Kaplan-Meier analysis was performed to determine survival probability; Log-rank test, to compare patients' survival probability between subgroups; and Cox regression model, to conduct a multivariate analysis.
Expression levels of ARNT and HIF-1α on HCC prognosis
To understand ARNT expression on HCC prognosis, we first investigated the protein level in resection specimens from 105 HCC patients using a tissue microarray and immunohistochemical staining (Supporting Information Fig. 1C). Our results showed that ARNT protein was primarily found in the nucleus and occasionally in the cytoplasm of hepatocytes and tumor cells and strongly stained to be a brown or a dark brown mass. The expression level of ARNT was significantly higher in normal liver tissues compared with HCC tissues (p = 0.007). Specifically, 75 of 100 peritumoral tissues (75.0%) and 56 of 105 intratumoral tissues (53.3%) had a high level of ARNT (Figs. 1a–1d). Overall survival of patients with a high intratumoral ARNT level was significantly longer than survival of those with a low ARNT level, and recurrence incidence was lower in patients with a high intratumoral ARNT level than in those with a low ARNT level (p < 0.001 and p = 0.016, respectively; Figs. 1e and 1f). Median survival time of patients with high and low ARNT levels was 84.0 and 26.3 months, respectively (p < 0.001). Thirty-five patients with a low intratumoral ARNT level experienced tumor recurrence within 2 years, whereas 11 patients with a high ARNT level experienced tumor recurrence during the same period of time. The median survival and recurrence time for patients with low and high ARNT expression were significantly different (p < 0.001 for both). Intratumoral ARNT level was a prognostic factor both for OS and TTR in univariate analysis, but only for OS in multivariate analysis (Supporting Information Table 3). HIF-1α was mainly found in the cytoplasm of hepatocytes and tumor cells appeared as weakly stained light yellow or yellow granules. CD31+ and CD34+ blood vessels were detected both in tumor tissues and in normal liver tissues. Staining hotspots were observed at the interface of tumor and liver tissues. Neither HIF-1α, CD31 nor CD34 levels had any relationship with OS or TTR in postoperative patients (Supporting Information Fig. 3).
Proliferation inhibition of ARNT in HCC cells in vitro
To confirm clinical observations, we successfully constructed four lentiviral vectors to modulate ARNT expression in HCC cells. Among them, Lenti-ARNTi-v1 and Lenti-ARNTi-v2 were used to suppress ARNT expression (Supporting Information Figs. 1A and 1B), Lenti-ARNT-v1 for rescued expression and mock vector for the control (Supporting Information Figs. 1B and 2, Supporting Information Table 2). After the vectors were introduced into HCCLM6 cells, the relative ARNT levels were significantly down-regulated in both HCCLM6-ARNTi-v1 and HCCLM6-ARNTi-v2 cells, especially in the former, and up-regulated in HCCLM6-ARNT-v1 cells (p < 0.001, Fig. 2a). There was no significant difference in ARNT levels between HCCLM6-Mock and HCCLM6-WT cells.
Using these stable infected cell lines, we next evaluated the roles of ARNT on HCC proliferation and cell cycle progression. HCCLM6-ARNTi-v1 cells grew the fastest, while HCCLM6-ARNT-v1 cells were the slowest. The doubling times of HCCLM6-ARNTi-v1, HCCLM6-ARNTi-v2, HCCLM6-Mock, HCCLM6-WT and HCCLM6-ARNT-v1 were 29.54 ± 0.98 h, 40.81 ± 1.05 h, 48.06 ± 0.71 h, 47.74 ± 0.98 h and 72.93 ± 3.12 h, respectively (Fig. 2b). To further examine the effect of ARNT on cell cycle progression, HCCLM6 cells and their derivates were analyzed on the third day after subculture. The S-phase cell populations of HCCLM6-ARNTi-v1 and HCCLM6-ARNTi-v2 cells were conspicuously increased, especially for the former (p < 0.01), and the S-phase cell populations of HCCLM6-ARNT-v1 cells were dramatically decreased (p < 0.01, Figs. 2c and 2d).
To explore the universal significance of ARNT, cell proliferations in ARNT-modulated HepG2, HCCLM3 and HCCLM6 cells were tested again by a cell counting kit. As expected, cell proliferation was significantly accelerated in ARNT-decreased HepG2, HCCLM3 and HCCLM6 cells and slowed in ARNT-increased cells (Fig. 2e).
Anti-tumor effects of ARNT on HCCLM6 xenograft in vivo
The average tumor size of HCC patients with high intratumoral ARNT expression was statistically smaller than that of patients with low expression (Fig. 3a, p = 0.001). To elucidate the effects of ARNT on HCC progression, we compared tumor metastatic foci in lung in parallel by observing GFP signals. On day 42 after orthotopic implantation, the numbers of lung metastatic foci of HCCLM6-ARNTi-v1 and HCCLM6-ARNTi-v2 xenografts, especially of the former, were much more numerous than those of HCCLM6-Mock xenografts, while the numbers of lung metastatic foci of HCCLM6-ARNT-v1 were also fewer than those of HCCLM6-Mock (Figs. 3b and 3c). To dynamically measure in vivo tumor growth, the fluorescence area of HCCLM6-Mock, HCCLM6-ARNTi-v1, HCCLM6-ARNTi-v2 and HCCLM6-ARNT-v1 xenografts were monitored in parallel once a week for six successive weeks. Again, HCCLM6-ARNTi-v1 xenografts grew the fastest, whereas HCCLM6-ARNT-v1 xenografts were the slowest (Figs. 3d and 3e). Our in vivo experiments indicated once more that ARNT played an inhibitory role on HCCLM6 growth and metastasis.
Heterodimeric and homodimeric complexes formed by ARNT in HCCLM6 xenografts
Previous studies have shown that bHLH-PAS proteins usually exert their biological functions as heterodimers or homodimers. To elucidate this issue, the total level of ARNT in ARNT-modulated HCCLM6 xenografts was first analyzed by immunoprecipitation assays. Similar to the results in Western blotting, a relatively low level of the ARNT complex was pulled down from the HCCLM6-ARNTi-v1 xenograft and a relatively high level was from HCCLM6-ARNT-v1 xenograft, as compared with the HCCLM6-WT and HCCLM6-Mock xenografts. Although a tiny amount of HIF-1 α/ARNT and AHR/ARNT heterodimer was identified, its level as well as the cofactor P300 were not markedly changed. However, a significant decease of AHRR level in the HCCLM6-ARNTi-v1 xenograft and an increase in the HCCLM6-ARNT-v1 xenograft were found in ARNT pull-down complexes, as compared with the HCCLM6-WT and HCCLM6-Mock xenografts. In addition, the homodimer of ARNT in the HCCLM6-ARNT-v1 xenograft was clearly detected by use of the antibody against the FLAG tag (Fig. 4a).
Cell cycle-control genes regulated by ARNT
To address the underlying mechanisms, mRNA levels of cell cycle-control genes were then screened by qRT-PCR. Both cyclin E1 and CDK2 were found to be up-regulated in HCCLM6-ARNTi-v1 and HCCLM6-ARNTi-v2 cells, especially in the former, and to be down-regulated in HCCLM6-ARNT-v1 cells (Fig. 4b, Supporting Information Fig. 1D). To gain a comprehensive understanding, more genes involved in cell cycle progression were then screened by using qRT-PCR assays. Our results showed that the expression of CDKN1C, CNKN2A, CDKN2B, MAPK11 and MAPK14 was positively regulated by ARNT, while the expression of CDKN2C, Fos and Jun was negatively regulated (Fig. 4c).
Several members of the bHLH/PAS superfamily have been shown to be involved in tumor progression; however, little is known about the role of ARNT protein per se on clinical HCC progression. The preliminary clinical results from our immunohistochemical study suggest for the first time that ARNT might be gradually lost during HCC progression and could serve as a prognostic factor for postoperative HCC patients.
To confirm clinical observation, loss- and gain-of-function studies were performed by using HCC cell lines and xenograft models in nude mice. We found attenuated ARNT expression in HepG2, HCCLM3 and HCCLM6 cells with specific siRNA, especially with Lenti-ARNTi-v1 vector, greatly shortened in vitro cell doubling time, increased S-phase cell populations and promoted in vivo tumor growth and lung metastasis in a HCCLM6 xenograft. After ARNT expression was rescued, prolonged cell doubling time and decreased S-phase cell populations were observed in HepG2, HCCLM3 and HCCLM6 cells. And HCCLM6 growth and metastasis in vivo were remarkably inhibited. Therefore, we believe that ARNT is a robust and negative regulator of HCC progression.
Numerous studies have demonstrated that the HIF-1 signal pathway is one of the most important pathways in solid tumor growth and metastasis.26–28 HBx protein was reported to cross-talk with HIF-1α and promote its nuclear translocation, phosphorylation, stability and transactivation function in HBx-inducible Chang liver cells as well as in nonhepatic cells. This kind of cross-talk may lead to transcriptional activation of HIF-1α target genes [e.g., an increased expression of vascular endothelial growth factor (VEGF) in HBx+ cells] and thus stimulate endothelial cell proliferation and tumor angiogenesis.29, 30 Because about 90% of the patients enrolled in this study had a history of HBV infection, we first evaluated the expression level of HIF-1α. Given that CD31+ and CD34+ endothelial cells are two major target cells in VEGF response and their tissue densities can be used to evaluate the biological functions of VEGF and HIF-1 signaling pathway, we observed the prognostic roles of CD31 and CD34 molecules in the same patient cohort. However, neither was found to have any prognostic values when evaluated by TTR and OS. Our findings were consistent with observations of epithelial ovarian tumors,31 but different from those in HCV-related HCC,32 oropharyngeal cancer33 and esophageal cancer.34 Such differences in the expression level of HIF-1α might arise from the types of tumor studied, the tumor tissue regions, or the specific pretreatments received. To avoid as far as possible tissue necrosis and a secondary hypoxia response caused by radiotherapy and chemotherapy, all tissues analyzed in this study were taken from the margins of HCC tumors that had not received any pretreatment. In such a circumstance, HIF-1α was predominantly localized in the cytoplasm, but not in the nucleus as reported by Wada et al.,32 suggesting that the periphery of HCC in our patients may not have been hypoxic. Although a low basal level expression of VEGF was found in HBx-bearing HCCLM3 and HCCLM6 cells in normoxic culture conditions, no significant changes of this factor were detected after ARNT was down-regulated or up-regulated (data not shown). In addition, little protein interaction between HIF-1α and ARNT was found in vivo by using HCCLM6 xenograft tissues. Therefore, we concluded that the regulation of ARNT on HCC growth and metastasis was not primarily mediated by hypoxia- or HBx-induced HIF-1 signal pathway.
Recently, endogenous AHR was identified as a regulator of cancerous cell proliferation in a ligand-independent pattern.9 Further, AHRR was reported to repress transcription activity of AHR by competing with this transcription factor for heterodimer formation with ARNT.11 It is very likely that the AHRR/ARNT heterodimer plays an antagonistic role to the AHR/ARNT heterodimer as well as the ARNT homodimer in HCC growth and metastasis. To test this hypothesis, we analyzed AHR and AHRR levels in ARNT pulled-down complexes. Although no obvious changes of AHR levels were found, a marked increase in AHRR levels was detected from HCCLM6-ARNT-v1 xenografts as compared with those from HCCLM6-Mock and HCCLM6-ARNTi-v1 xenografts. Thus, we supposed that the transcriptional activities of ARNT homodimer-driven gene expression with the E box core sequence could be reversed, at least partly, in the presence of the AHRR/ARNT heterodimer.35, 36 The relative balance between the ARNT homodimer and the ARNT/AHRR heterodimer will be important in determining HCC progression.
Cell cycle is strictly controlled by sequential activation of cyclin/Cdk complexes, which are negatively regulated by Cdk-inhibitors (CKI).37–39 Two families of CKI, Cip/Kip and INK4, have been found in human cells. CDKN1C (p57kip2), a member of the Cip/Kip family, can strongly inhibit the activities of several G1 cyclin/Cdk complexes. CDKN2A (p16INK4a) and CDKN2B (p15INK4b), two members of the INK4 family, can function as inhibitors of CDK4 kinase. Because cell cycle arrest is an obligate step in cell differentiation,40 the activity of the CyclinE/cdk2 complex is usually inhibited by the members of the Cip/Kip family at this juncture.41, 42 Therefore, up-regulation of CDKN1C, CNKN2A and CDKN2B together with down-regulation of cyclin E1 (CCNE1) and Cdk2 by enhanced ARNT expression may cause cell cycle arrest from G1 to S phase, resulting in HCC stagnancy and deceasing HCC metastasis. In addition, up-regulations of MAPK11 and MAPK14 together with down-regulations of Fos and Jun by enhanced ARNT expression may inhibit cell proliferation and promote cell differentiation in a mechanism as reported by Johnstone et al.43 and Hui et al.44
In conclusion, ARNT is an important regulator of the bHLH/PAS superfamily in HCC progression and a useful biomarker for predicting HCC prognosis after curative resection.