Decreased Expression of Inhibitor of Growth 4 Correlated with Poor Prognosis of Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide, with a constantly increasing frequency, especially in China (1, 2). It ranks as the second leading cause of cancer death among males in China (3). With the improvement of the liver surgery, hepatic resection for HCC has evolved into a safe procedure with low operative mortality (4). However, long-term survival is still unsatisfactory mainly because of the recurrence and metastasis of HCC (5-7). Thus, the ability to predict individual risk of recurrence and subsequent prognosis would be critical to guide surgical and chemotherapeutic treatment.

The recurrence and metastasis of HCC is a multistep process that involves many complex biological and pathologic events (8). Angiogenesis is one of the most important steps in HCC invasion and metastasis (9-11). Inhibitor of growth 4 (ING4), a member of the ING family, is a candidate tumor suppressor that plays an important role in tumor growth (12), cell cycle regulation (13), apoptosis (14), and contact inhibition (15, 16), especially in angiogenesis (12, 17). ING4 can regulate brain tumor angiogenesis through transcriptional repression of nuclear factor-κB (NF-κB)-responsive genes by physically interacting with the p65 (RelA) subunit of nuclear factor NF-κB (12, 17). Our previous study has revealed that hypoxia-inducible factor (HIF) played an important role in neovascularization of HCC and its expression correlated with invasion and metastasis of HCC (18). Interestingly, the recent study suggested a critical role of ING4 in regulating HIF stability and activity by directly associating with the HIF prolyl hydroxylase, a Fe(II)-dependent oxygenase that regulates HIF stability as a function of oxygen availability (19, 20).

Although studies with experimental models have suggested important role of ING4 in tumor angiogenesis, clinical evidence of ING4 acting as a tumor suppressor in human HCC and its relevance to human HCC remains lacking. In this study, we first explored the expression level of ING4 in HCC tissues and then explored the relationship of ING4 expression with the clinicopathologic features, microvessel density (MVD), and prognosis of HCC.

In the present study, specimens of HCC tissues were obtained from 136 HCC patients who underwent hepatectomy at the Department of Surgery, Xiangya Hospital of Central South University from November 1998 to December 2005. The 136 patients included 117 males and 19 females with a median age of 47 years (range, 16-73 years). Among these 136 cases of HCC, matched HCC and paracarcinomatous liver tissue (PCLT) specimens from 36 cases were also collected and immediately frozen in liquid nitrogen, which were subsequently stored at −80°C for real-time quantitative reverse transcription-PCR (RT-PCR) and Western blotting analysis. Five cases of normal liver samples were obtained from patients with liver cavernous hemoangioma as controls. All specimens were embedded with paraffin and stained by H&E. The diagnoses of all these patients were confirmed by histopathology that was blinded for examination. The clinicopathologic variables, such as tumor diameter, number of tumor nodules, histopathologic classification, and vein invasion, were recorded. Prior informed consent was obtained from the patients for collection of liver specimens in accordance with the guidelines of Xiangya Hospital, and the study protocols have been approved by Ethics Committee of the Central South University.

Total RNA was isolated by using Trizol reagent (Life Technologies). Total RNA was reverse transcribed with the TaKaRa reverse transcription Kit (TaKaRa) starting with 2 μg total RNA from each sample according to the procedures provided by the supplier. mRNA expressions of ING4 and housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were analyzed by quantitative RT-PCR using a real-time LightCycler rapid thermal cycler (Roche Molecular Biochemicals LightCycler System). Specific ING4 primers [5′-GGGATGTATTTGGAACATTATCTGG-3′ (forward) and 5′-TGCGGGCACTACTCATATACTCA-3′ (reverse)] and internal control GAPDH primers [5′-GCACCGTCAAGGCTGAGAAC-3′ (forward) and 5′-TGGTGAAGACGCCAGTGGAT-3′ (reverse)] were used for mRNA amplifications. To prepare standard curves, ING4 and GAPDH using the above primers were cloned into pMD18-TVector (TaKaRa) and spectrophotometrically quantified. Serial dilutions of cloned ING4 and GAPDH were prepared and used as a standard for PCR amplifications. Reactions were done in a 20 μL volume of the SYBR Premix Taq mix (TaKaRa) with 4 pmol of each primer. Initial denaturation at 95°C for 10 s was followed by 45 cycles of 95°C for 5 s and 60°C for 20 s. To confirm amplification specificity, the PCR products from each primer pair were subjected to a melting curve analysis. The quantification data were analyzed with the LightCycler analysis software as described before (21). The ING4 expression levels in the HCCs and matched PCLTs were determined by standard curve, each of which was normalized for the corresponding GAPDH expression level (HCC: ING4/GAPDH expression ratio in HCC sample; PCLT: ING4/GAPDH expression ratio in matched PCLT).

We also performed a classic quantitative RT-PCR for some of the same samples to confirm our real-time PCR in agarose gel electrophoresis. Reverse transcriptase product (2 μL) was amplified by PCR using the following conditions: 95°C for 5 min and then 32 cycles of 95°C for 40 s, 57°C for 30 s, and 72°C for 30 s and extension 72°C for 5 min. ING4 primers: 5′-CACAAGTCCTGAGTATGGGAT-3′ (forward) and 5′-AGGGGATGTGGAAGAAACTGT-3′ (reverse) and GAPDH primers: 5′-CTGCAGCATCTTCTCCTTCC-3′ (forward) and 5′-CAAAGTTGTCATGGATGACC-3′ (reverse). PCR products (5 μL) were then electrophoresed on 1.5% agarose gel. The sizes of the PCR products were 295 bp for ING4 and 500 bp for GAPDH.

Tissues from HCC and PCLT were lysed in a lysis buffer: 20 mmol/L Tris-HCl (pH 7.4), 10 mmol/L NaCl, 1 mmol/L EDTA (pH 8.0), 1 mmol/L MgCl₂, 1% NP-40, 0.1% SDS, 0.01% phenylmethylsulfonyl fluoride (Sigma), and protease inhibitor (Promega). The concentration of protein was determined by using bicinchoninic acid protein assay kit (Pierce). Total protein (100 μg) was separated by SDS-PAGE and then transferred onto nitrocellulose membrane (Millipore). After blocked with 5% nonfat dry milk in PBS [8 mmol/L Na₂HPO₄, 2 mmol/L KH₂PO₄, 150 mmol/L NaCl (pH 7.4)], the membranes were probed with goat anti-human ING4 polyclonal antibody (Abcam) diluted at 1:1,000. After washing, the membranes were incubated with a 1:3,000 dilution of horseradish peroxidase-conjugated mouse anti-goat antibody (KPL). The blots were developed by enhanced SuperSignal West Pico chemiluminescence (Pierce). β-Actin was also determined by using the specific antibody (Sigma) as a loading control. All experiments were carried out in triplicate and the expression levels of ING4 protein were quantified by densitometry (Stratagene).

A total of 136 HCC specimens, including the 36 cases of HCC fresh specimens (used for real-time quantitative RT-PCR and Western blotting), were evaluated for immunohistochemistry. Sections (4 μm thick) were incubated overnight at 4°C with a goat anti-human polyclonal antibody (Abcam; 1:200 dilution) against ING4. The second antibody (rabbit anti-goat antibody; MaiXin Bio) was applied for 45 min at 37°C. The streptavidin-biotin peroxidase complex tertiary system (Boster) was used according to the manufacturer's instructions. The tissues were visualized by applying 3,3′-diaminobenzidine tetrahydrochloride for 3 min. Sections were counterstained using hematoxylin, dehydrated through gradient alcohols, and mounted for viewing. Negative controls were carried out by omitting the primary antibody, whereas ING4 overexpression confirmed by Western blotting was used as positive controls. The intensity of cytoplasmic staining was scored as 0 to 3+ by comparison with the positive controls. The scoring system has been validated previously (22). Diffuse, moderate to strong cytoplasmic staining characterized ING4 high expression group (scores 2+ and 3+). ING4 low expression group was devoid of any cytoplasmic staining or contained faint, equivocal staining (scores 0 and 1+).

Blood vessels were highlighted by staining endothelial cells with anti-CD34 antibody (Santa Cruz Biotechnology) as described previously (20). Any red-staining endothelial cell cluster that was clearly separated from adjacent microvessels, tumor cells, and other connective tissue elements was considered a single, countable microvessel. Vessel lumens were not necessary for a structure to be defined as a microvessel. We selected three vascular hotspots both intratumorally and peritumorally for counting CD34⁺ microvessels.

Follow-up data were obtained after hepatic resection for all 136 patients. The follow-up period was defined as the interval between the date of the operation and that of patient's death or the last follow-up. Deaths from other causes were treated as censored cases; the follow-up time ranged from 30 to 2,800 days, with a median follow-up time of 630 days. Recurrence and metastasis were diagnosed by clinical examination, α-fetoprotein (AFP) mensuration, B-ultrasound, and computed tomography. To determine factors influencing survival after operation, 11 conventional variables together with ING4 expression were tested in all 136 patients: age (<60 versus ≥60 years), gender, liver cirrhosis (presence versus absence), Child-Pugh classification (A versus B and C), preoperative AFP level (≤20 versus >20 ng/mL), hepatitis history (hepatitis B or C, yes versus no), Edmondson-Steiner grade (1 and 2 versus 3 and 4), size of the tumor (≤5 versus >5 cm), number of tumor nodes (solitary versus multiple), venous invasion (macroscopic or microscopic; absence versus presence), and ING4 protein expression level (0 and 1+ versus 2+ and 3+).

The Fisher's exact test was done to assess differences in immunostaining positive rate between HCC tissues and PCLTs. Quantitative values were presented as mean ± SD or median (range). Independent Student's t test was used to compare ING4 mRNA and protein expression in HCC and PCLT samples. A similar comparison was made for MVD. The correlations between ING4 immunohistochemical staining and clinicopathologic variables of 136 HCCs cases were analyzed by Mann-Whitney U test. Survival curves were constructed using the Kaplan-Meier method and the differences in survival were evaluated using the log-rank test. The Cox proportional hazards model was employed to identify factors that were independently associated with survival. In this model, a stepwise selection was used for variable selection with entry and removal limits of P < 0.05 and P > 0.10, respectively. All statistical evaluations were done using the SPSS statistical software (version 11.5). All tests were two tailed and P < 0.05 was considered statistically significant.

Real-time quantitative RT-PCR analysis of 36 paired HCC and PCLT specimens using the ING4-specific primers indicated that the ING4 mRNA levels in HCC tissues were significantly lower than that in PCLTs (0.44 ± 0.12 versus 0.86 ± 0.21; P = 0.039; Fig. 1). A classic quantitative RT-PCR analysis in parallel confirmed the real-time PCR results (Fig. 1E). To determine whether the difference in the mRNA levels can be reflected at the protein level, we analyzed the same 36 paired specimens by Western blot. Consistent with the mRNA expression, the ING4 protein levels in HCC tissues were significantly lower than those in the corresponding PCLTs (1.09 ± 0.32 versus 1.97 ± 0.45; P = 0.012; Fig. 1E and F). Furthermore, the percentage of positive ING4 expression in HCC tissues was also significantly lower than that in PCLTs [52.8% (19 of 36) versus 91.7% (33 of 36); P < 0.001]. However, both mRNA and protein levels have no difference between PCLTs and normal liver tissues (data not shown; Fig. 1E and F).

The immunohistochemistry data from 136 HCC specimens were analyzed for the correlation of ING4 levels with clinicopathologic features. Among 136 HCC samples, there were 13 (9.6%) cases scored 3+, 16 (11.8%) cases scores 2+, 36 (26.5%) cases scores 1+, and 71 (52.2%) cases scores 0. The representative positive immunostaining of HCC specimens was shown in Fig. 2. We used the Mann-Whitney U test to assess the correlations between the staining intensity of ING4 protein and clinicopathologic variables of HCC. Interestingly, the ING4 protein staining intensity in HCC with Edmondson-Steiner grades 3 and 4 were significantly lower than those with Edmondson-Steiner grades 1 and 2 (P = 0.013). Furthermore, a difference in expression of the ING4 protein between the HCCs with vein invasion and those without vein invasion was also evident (P = 0.016). However, the ING4 protein staining intensity showed no significant relationship with gender, age, preoperative AFP, hepatitis history, Child-Pugh score, liver cirrhosis, tumor size, and tumor nodule number (P > 0.05; Table 1).

To gain a better understanding of the seemly counterintuitive correlation that ING4 contributes to HCC angiogenesis, the MVD of HCC was determined by CD34 immunohistochemical staining to test whether it correlated with the ING4 protein level (Fig. 2E). The results indicated that the ING4 protein high expression group had a lower MVD than that in the low expression group (95.7 ± 28.0 versus 142.4 ± 31.7; P = 0.005; Fig. 2F).

Based on the immunohistochemistry data, 136 cases of HCCs were divided into low expression group (score 0 or 1+; n = 107) and high expression group (score 2+ or 3+; n = 29). The Kaplan-Meier method was used to analyze the correlation of ING4 expression level with the prognosis of HCC patients. The results indicated that the ING4 high expression group correlated with a longer survival time (1,860 versus 872 days, median survival time) than the low expression group and the overall survival rate for the patients with low and high ING4 levels was significantly different (P = 0.0001, log-rank test; Fig. 3A). Furthermore, the overall metastasis rate for the patients with low and high ING4 levels was also significantly different (P = 0.0065, log-rank test; Fig. 3B). The ING4 high expression group had a longer tumor-free survival time than the ING4 low expression group (1,221 versus 690 days, median disease-free survival time).

To test whether the ING4 expression level was an independent prognostic factor for survival of HCC patients, univariable and multivariable Cox regression analyses were used to identify factors that may predict survival after hepatic resection. Univariable Cox regression analysis results indicated that the ING4 expression level [hazard risk (HR), 1.48; P = 0.018], tumor size (HR, 2.95; P = 0.002), tumor nodule number (HR, 2.03; P = 0.032), Edmondson-Steiner grade (HR, 9.69; P < 0.001), and vein invasion (HR, 14.41; P < 0.001) were all significantly associated with survival. However, gender, age, hepatitis B virus infection, serum AFP level, liver cirrhosis, type of hepatectomy, Child-Pugh classification, and margin status were not significantly correlated with survival (Table 2). By using the multivariable Cox regression analysis, the ING4 expression level (HR, 9.63; P = 0.001), Edmondson-Steiner grade (HR, 9.90; P < 0.001), and vein invasion (HR, 5.87; P < 0.001) were found to be independent prognostic factors for survival (Table 2). The other clinicopathologic variables did not add any independent prognostic information.

ING4, a member of the ING family encoding potential tumor suppressors, has been implicated in a variety of cellular processes including cell growth and proliferation, cell cycle arrest, apoptosis, and contact inhibition (13, 15, 23). More importantly, recent studies have confirmed that ING4 emerged as a candidate tumor suppressor protein affecting angiogenesis (12, 17). As everyone knows, HCC is a special solid tumor with abundant blood vessel. Angiogenesis has a critical role in the pathophysiology of HCC (11). However, the role of ING4 in human HCC still remains unknown. Thus, we carried out this study to investigate the role of ING4 in HCC.

We firstly measured ING4 mRNA and protein expression levels in 36 paired HCC and PCLT specimens and 5 normal liver tissue specimens. Our results revealed that expression levels of ING4 mRNA and protein were significantly lower in HCC tissues than that in the corresponding PCLTs. Furthermore, the positive rate of ING4 protein expression in HCC tissues was significantly lower than that in PCLTs. These results were consistent with the previous study in gliomas and melanoma, which showed that ING4 mRNA and protein expression levels were dramatically reduced in tumor tissues (12, 21). ING4 is localized at chromosome 12p12-13 region, which has known to be deleted in various cancers (21). Gunduz et al. found that 66% head and neck squamous cell carcinoma patients had loss of heterozygosity at 12p12-13 region and that 76% tumor tissues had decreased ING4 mRNA expression compared with matched normal samples (21). Kim et al. also found that 10% to 20% of human breast cancer cell lines had ING4 locus deletion (15). These study suggested that deletion of ING4 gene may be a common event in tumorigenesis, which can possibly explain the reduced expression level of ING4 in HCC tissues. However, whether ING4 is mutated or deleted in human HCC remains to be determined.

In this study, the ING4 expression data obtained from immunohistochemistry staining were analyzed for the correlation with clinicopathologic features. We found the expression of ING4 significantly correlated with vein invasion and Edmondson-Steiner grade of HCCs. Our clinical data also showed a highly significant association between the ING4 expression level and MVD of HCC. ING4 protein low expression group had higher MVD than that in the high expression group. The decreased expression of ING4 and its correlation with MVD suggested that ING4 might be a potent tumor suppressor in HCC and ING4 could regulate angiogenesis of HCC. Abundant evidence has indicated that the Edmondson-Steiner grade, vein invasion, and angiogenesis are highly correlated with invasion and metastasis as well as the prognosis of HCC (24-26). Consistent with these results, we found a highly significantly association between the ING4 expression level and prognosis/metastasis of HCC. Multivariable Cox regression analyses indicated that the low ING4 expression was an independent risk factor for the prognosis of HCC patients as well as Edmondson-Steiner grade and venous invasion. These results suggested that ING4 might play an important role in HCC carcinogenesis and development. Recent study also showed that reduced ING4 staining is associated with poor overall and disease-specific 5-year survival in primary melanoma (27).

Many tumor-suppressive functions of ING4 contribute to the inverse correlation between ING4 expression and survival of HCC patients. First, ING4 can regulate cell migration and invasion. Shen et al. found that ING4 can regulate cell migration and invasion by interacting with liprin α1, which colocalized at lamellipodia with ING4 (16). Li et al. also found that overexpression of ING4 can lead to suppression of melanoma cell invasion by inhibiting the gelatinolytic activities of both matrix metalloproteinase-2 and -9 (27). ING4 can inhibit melanoma cell migration by down-regulating RhoA activity through NF-κB pathway (27). Second, ING4 can regulate angiogenesis as an angiogenic repressor. ING4 can regulate brain tumor angiogenesis through transcriptional repression of NF-κB-responsive genes by physically interacting with the p65 (RelA) subunit of nuclear factor NF-κB (12). ING4 was recently shown to be a critical regulator of NF-κB-mediated interleukin-8 transcription and subsequent angiogenesis in gliomas (17). HIF played an important role in neovascularization of HCC and its expression correlated with invasion and metastasis of HCC (18). Recent study showed a critical role of ING4 in regulating HIF pathway. ING4 can regulate HIF stability and activity by directly associating with the HIF prolyl hydroxylase, a Fe(II)-dependent oxygenase that regulates HIF stability as a function of oxygen availability (19, 20). ING4, which can be recruited to HIF under hypoxic conditions by HIF prolyl hydroxylases, mediates the ability of HIF to activate transcription of its downstream target genes rather than affecting hydroxylase activity or HIF stability (20). Moreover, ING4 exerts an inhibitory effect on HIF-1α activity and its target gene NIP-3 expression in hypoxic conditions and consequently inhibits angiogenesis (28). Other functions of ING4 on tumor growth (12), cell cycle regulation (13), apoptosis (14), and contact inhibition (15, 16) may also contribute to the survival of HCC patients.

As ING4 has potential effect on tumor inhibition via multiple pathways and plays an important role in HCC carcinogenesis and development, it may serve as a powerful therapeutic target. Xie et al. constructed a recombinant adenoviral vector Ad-ING4 and transfected A549 human lung carcinoma cells to explore its therapeutic effect on human lung carcinoma (29). They found that ING4 can induce cell apoptosis, altered cell cycle with S-phase reduction and G₂-M-phase arrest, suppressed cell invasiveness, and down-regulated interleukin-6 and -8 and matrix metalloproteinase-2 and -9 expression of tumor cells. In athymic mice bearing A549 lung tumors, intratumoral injections of Ad-ING4 suppressed the tumor growth and reduced the tumor microvessel formation. However, its therapeutic effect on human HCC remains to be determined.

In summary, we found that ING4 is down-regulated in human HCC tissues and the ING4 expression level is significantly correlated with the prognosis and metastatic potential of HCC, which suggests that ING4 may serve as a powerful prognostic marker and therapeutic target for HCC.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.