HOXC6 predicts invasion and poor survival in hepatocellular carcinoma by driving epithelial-mesenchymal transition

Relationship between HOXC6 and clinicopathological features

In order to analyze the importance of HOXC6 in HCC patients, we detected the expression of HOXC6 in 164 HCC samples and matched adjacent non-tumorous liver tissues by immunohistochemistry. Representative images indicated that HOXC6 was mainly localized in cytoplasm of HCC cells (Fig. 1A). A total of 88 samples exhibited high expression, and 76 exhibited low expression. The expression of HOXC6 in tumors is significantly higher than that in adjacent non-tumorous tissues (P < 0.001; Fig. 1B), suggesting that HOXC6 accumulated in HCC samples.

Figure 1. HOXC6 was overexpression in hepatocellular carcinoma samples. (A) Immunohistochemistry assays of HOXC6 expression in HCC tissues and adjacent non-tumorous tissues. The upper left panel represents low HOXC6 expression in adjacent non-tumorous tissues. The upper middle and right panel represents low and high HOXC6 expression in HCC tissues. Lower panels represent magnified pictures of boxed area in the corresponding upper panels. The line scale bar represents 50 μm. (B) HOXC6 expression in HCC was compared with that in adjacent non-tumorous specimens. Statistical analysis was performed by Paired-Samples t-test.

To investigate the biological functions of HOXC6 in HCC patients, we determined the correlation of HOXC6 status with clinicopathological features in 164 HCC tissues. HOXC6 overexpression correlated with high AFP level (P = 0.038), liver cirrhosis (P = 0.044), larger tumor (P = 0.001), vascular invasion (P = 0.046) and Barcelona Clinic Liver Cancer (BCLC) stage (P = 0.005) (Table S1). On the contrary, HOXC6 expression had no correlation with other features, such as gender, age, HBsAg, gamma-glutamyltransferase (GGT), tumor number, satellite nodule, tumor margin, tumor differentiation and tumor-node-metastasis (TNM) stage (all P > 0.05). In addition, HOXC6 expression in tumors gradually increased in TNM and BCLC stage-dependent manner (Fig. S1).

HOXC6 overexpression in tumors indicated poor prognosis

To determine the significance of HOXC6 status on prognosis in HCC patients, we analyzed univariate analysis of clinicopathologic characteristics and HOXC6 expression for Overall survival (OS) and time to recurrence (TTR). Kaplan-Meier analysis showed that the patients with high expression of HOXC6, high AFP level, high gamma-glutamyltransferase (GGT) level, liver cirrhosis, larger tumor or vascular invasion had poorer OS time (all P < 0.05). Furthermore, high HOXC6 expression, high AFP level, high GGT level, liver cirrhosis, larger tumor, tumor number, satellite nodule and vascular invasion were unfavorable predictors for TTR of HCC patients (all P < 0.05, Table 1).

Significant OS and TTR advantages were observed in patients with HOXC6 low-expression (both P < 0.001) (Fig. 2A). Moreover, the median of OS and TTR times in HOXC6 overexpression group (n = 88) were 38.5 months and 17.5 months, while there were 58.0 months and 58.0 months in HOXC6 low expression group (n = 76). In addition, the 5-year OS and TTR rates of the HOXC6 overexpression group were significantly lower than those of the HOXC6 low expression group (OS: 41.5% vs. 72.8%, TTR: 27.4% vs. 70.2%) (Fig. 2A). Moreover, multivariate analysis was used to analyze whether HOXC6 could be an independent prognostic marker for OS and TTR in HCC patients. Our results found that overexpression of HOXC6 remained as an independent predictor in multivariate models and portended worse OS in HCC patients (HR = 2.543, 95% CI = 1.437-4.498, P = 0.001). The patients with HOXC6 overexpression will be more likely to suffer from recurrence than those with low HOXC6 expression (HR = 3.540, 95% CI = 2.047-6.122, P < 0.001), (Table 1).

Figure 2. Overall survival and time to recurrence are shown for patients with HCC. All patients were classified according to tumor size, vascular invasion, TNM stage and BCLC stage. Kaplan-Meier survival estimates and log-rank tests were used to analyze the prognostic value of HOXC6 expression in all patients (A) and each subgroup (B-I).

To discuss prognostic significance of HOXC6 in different subgroups, HCC patients were divided base on tumor size (Fig. 2B, C), vascular invasion (Fig. 2D, E), TNM stage (Fig. 2F, G) and BCLC stage (Fig. 2H, I). Patients with HOXC6 overexpression predicted unfavourable OS and TTR times in all of these subgroups for except OS and TTR in patients who had vascular invasion (P = 0.208, and P = 0.080) or OS in patients with TNM stage II/III (P = 0.111). In general, it reveals that HOXC6 status could act as prognostic predictor in different risk subgroups of HCC patients.

HOXC6 overexpression predicts unfavorable prognosis independent of tumor invasiveness

The correlation of HOXC6 and matrix metallopeptidase 9 (MMP-9) expression was evaluated by immunohistochemistry (IHC) assays in serial sections of HCC samples (Fig. 3A). Our results revealed that HOXC6 expression had positively correlation with MMP-9 expression in HCC tissues (r = 0.476, P < 0.001, Fig. 3B) and HCC cells (Fig. 3D). We further investigated the effect of HOXC6 expression on prognosis independent of tumor invasiveness by using MMP-9 marker as an indicator for invasive potential of tumor cells. The HCC patients were stratified into low invasiveness group (low MMP-9 expression; n = 79) and high invasiveness group (high MMP-9 expression; n = 85). In patients with low invasiveness subgroup, HOXC6 overexpression was associated with worse TTR (P = 0.005), while there was no significance in OS time (Fig. 4A). In the high tumor invasiveness group (Fig. 4B), patients with high HOXC6 expression were likely to death (P = 0.001) and recurrence (P < 0.001) compared with those with low HOXC6 expression. Therefore, HOXC6 status may be a prognostic predictor for HCC patients independent of tumor invasiveness.

Figure 3. HOXC6 expression level correlated with the expression of epithelial-mesenchymal transition (EMT) markers. (A) Serial sections of human HCC tissue were subjected to IHC staining with antibodies against HOXC6, E-cadherin, Vimentin and MMP-9. In patient #1, high expression of HOXC6 in HCC tissues was accompanied by the absence of E-cadherin and elevated Vimentin, MMP-9. In patient #2, low expression of HOXC6 was accompanied by elevated E-cadherin and the absence of Vimentin, MMP-9. The scale bar represents 50 μm. (B) HOXC6 expression was negatively correlated with E-cadherin expression and positively associated with vimentin and MMP-9 expression. (C) Decreased expression of Vimentin, α-SMA and N-cadherin with increased expression of E-cadherin and ZO-1 in HOXC6-silenced cells compared with the control cells. An opposite expression pattern of these genes was observed in HOXC6-transfected cells. (D) The mRNA of E-cadherin was up-regulated, while the vimentin and MMP-9 was down-regulated, when HOXC6 was silenced. And opposite expression was observed in HOXC6-transfected cells.

Figure 4. Overall survival and time to recurrence are shown for patients with low MMP-9 expression (A) and high MMP-9 expression (B). Kaplan-Meier survival estimates and log-rank tests were used to assess the association between HOXC6 expression and overall survival or time to recurrence in patients with low MMP-9 expression (n = 79) or high MMP-9 (n = 85).

Combined influence of HOXC6 and serum AFP on risk of HCC prognosis

Previous study revealed that high serum AFP level was associated with poor prognosis for HCC patients [27]. Prognostic analysis showed patients with preoperative serum AFP level above 400 ng/mL had worse OS (P < 0.001) and shorter TTR (P = 0.001), and could be an independent predictor in HCC patients (Table 1). Furthermore, we combined the prognostic significance of HOXC6 expression with serum AFP levels in HCC patients. According to the expression of HOXC6 and serum AFP levels, the prognosis of patients were classified into four different risk subgroups: patients with HOXC6 (-) and AFP ≤ 400 ng/ml (HOXC6-/AFP-) had better OS and TTR time; patients with HOXC6 (-) and AFP > 400 ng/ml (HOXC6-/AFP+), or patients with HOXC6 (+) and AFP ≤ 400 ng/ml (HOXC6+/AFP-), had intermediate prognosis and intermediate-risk of relapse; patients with HOXC6 (+) and AFP > 400 ng/ml (HOXC6+/AFP+) were prone to death and relapse (Fig. 5). Multivariate analysis showed that co-index of HOXC6/AFP could be used as an independent prognostic marker for OS (HR = 1.650, 95% CI = 1.307-2.083, P < 0.001) and TTR (HR = 1.841, 95% CI = 1.478-2.292, P < 0.001) (Table 2).

Figure 5. Combined value of HOXC6 and serum AFP identify different risks of HCC death and recurrence. The associations of HOXC6/AFP co-expression with Overall survival (log-rank P < 0.001) (A) and Time to recurrence (log-rank P < 0.001) (B) in 164 HCC patients.

HOXC6 expression promoted HCC cell invasion via EMT pathway

We further determined the significance of HOXC6 in tumor cell migration and invasion, which has been implicated from clinical data. SK-Hep1 cells migrated slower and had less ability to invade through Matrigel when HOXC6 was knocked down (Fig. 6A and B). In contrast, HuH7 cells migrated faster and had more invasive ability when HOXC6 was upregulated (Fig. 6C and D). We further evaluated the effect of HOXC6 on tumor metastasis in SK-Hep1 and HuH7 xenografts in vivo. After 8 weeks, the mice were sacrificed, and the metastatic nodules at the lung surfaces were counted. There were less numbers of metastatic nodules at the surface of the lungs of mice when HOXC6 was silenced. In contrast, with the HuH7 cells, tumors developed from HOXC6-transfected cells were significantly larger than tumors from vector cells. Hematoxyliin and eosin (H&E) staining confirmed that the nodules on the surfaces of mice lungs were metastatic tumors (Fig. 6E). These data suggest that enhanced HOXC6 may promote the motile and invasive abilities of HCC cells.

Figure 6. HOXC6 promotes cell migration and invasion in HCC cell lines. (A, B) HOXC6 silenced in SK-Hep1 cells and inhibited cell migration and invasion. (C, D) HOXC6 up-regulated in HuH7 cells and promoted cell migration and invasion. (E) Lung H&E staining of nude mice inoculated SK-Hep1 and HuH7 transfected cells via tail vein. The number of lung metastatic foci in each group were also calculated, * P < 0.05.

As EMT is one of the key events in tumor invasion and metastasis, the effect of HOXC6 on EMT markers was analyzed. The expression of HOXC6 was negatively correlated with E-cadherin expression and positively associated with Vimentin and MMP-9 expression in HCC samples (Fig. 3A and B). Interestingly, our western blot results found that decreased expression of Vimentin, α-SMA and N-cadherin with increased expression of E-cadherin and ZO-1 in HOXC6-silenced cells compared with the control cells. An opposite expression pattern of these genes was observed in HOXC6-transfected cells (Fig. 3C). Similar to the results of western blotting, our RT-PCR results showed that up-regulation of E-cadherin and down-regulation of Vimentin and MMP-9 when HOXC6 was silenced, while opposite expression was observed in HOXC6-transfected cells (Fig. 3D). Together, our results suggest that HOXC6 may promote HCC cells invasion by driving EMT pathway.

Patients and specimens

The research was approved by the Institutional Review Board and Human Ethics Committee of Cancer Center, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology. Written informed consent was obtained from all the patients enrolled in this study.

All HCC samples were obtained from 164 consecutive patients who underwent hepatectomy from February 2009 to December 2010 at the Cancer Center, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (Wuhan, China). The inclusion criteria of this study was as follows: Firstly, all the cases were tested for pathological proof. Secondly, none of the patients had distant metastasis or received previous therapy before liver resection. Finally, there were no serious complications in perioperative period or no other malignant tumor. All the patients were chosen consistently according to availability of resection samples and follow-up data. The clinicopathological characteristics of HCC patients were showed in (Table S2). Tumor stage was determined based on the 7th Edition tumor-node-metastasis (TNM) classification of the American Joint Committee on Cancer Staging and the Barcelona Clinic Liver Cancer (BCLC) staging system. Overall survival was defined from date of operation to date of death or last follow-up. Time to recurrence was computed as from date of operation to date of recurrence or last follow-up.

Cell culture

Two human HCC cell lines were used in this study: SK-Hep1 and HuH7. The two cell lines were obtained from Cell Bank of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (Shanghai, China). SK-Hep1 and HuH7 were maintained in DMEM supplemented with 10% fetal bovine serum and 100 units/ml penicillinstreptomycin (Invitrogen, USA) at 37°C in 5% CO₂.

IHC staining

All 164 HCC tissues and their para-carcinoma samples were detected by IHC method. Slides were baked at 55 °C for 1.5 h, deparaffinized with xylene and rehydrated using an alcohol gradient (100% alcohol, 95% alcohol, 80% alcohol, and 70% alcohol). The tissue slides were then treated with 3% hydrogen peroxide in methanol for 30 min to quench endogenous peroxidase activity, and the antigens were retrieved in 0.01 M sodium citrate buffer (pH 6.0) using a microwave oven. After antigen retrieval, the sections were incubated with 10% normal goat serum to block any nonspecific reaction, the sections were incubated using a primary antibody against HOXC6 (Anti-HOXC6 antibody produced in mouse, sc-376330, Santa Cruz Biotechnology, USA) or E-cadherin (Anti-E-cadherin antibody produced in mouse, Abcam, UK) or Vimentin (Anti-Vimentin antibody produced in rabbit, Abcam, UK) or MMP-9 (Anti-MMP-9 antibody produced in rabbit, Abcam, UK), at 4 °C overnight. The tissue slides were treated with a non-biotin horseradish-peroxidase detection system according to the manufacturer’s instructions (Gene Tech). Immunohistochemical signals were scored by two independent investigators blinded to the patients’ identity and clinical status. In discrepant cases, a pathologist reviewed the cases, and a consensus was reached.

Both the extent and intensity of immunohistochemical staining were taken into consideration when analyzing the data. The intensity of staining was scored from 0 to 3, and the extent of staining was scored from 0% to 100%. The final quantitation of each staining was obtained by multiplying the two scores. HOXC6 expression was classified as high expression if the score was higher than the median score of 1.1, if the score was 1.1 or less, the case was classified as low expression.

Plasmid constructs and transfection

cDNA containing open reading frames of HOXC6 was amplified by PCR and cloned into pcDNA3.1 vector (Invitrogen, Carlsbad, CA), and then transfected into HuH7 cell using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Cell transfected with empty vector was used as control. Stable HOXC6-expressing clones were selected by Geneticin (Rache Diagnostics, Indianapolis, IN) at the concentration of 500ug/ml.

Establishment of HOXC6 knockdown cells

Short hairpin RNAs (shRNA) targeting the human HOXC6 gene was purchased from GeneRay Biotechnology (GeneRay Biotechnology Co., Ltd.) and transfected into SK-Hep1 cell using Lipofectamine 2000 (Invitrogen) base on the manufacturer’s guidelines. Cell transfected with empty vector was used as control. Two days after transfection, cell culture media was collected and concentrated. Real-time PCR and Western blot were used to validate gene silencing efficiency of shRNA plasmids 48 hours after transfection. The recombinant lentivirus was stored at -80°C.

Total RNA extraction and qRT-PCR

Total cellular RNA was extracted from HCC cell lines using TRIzol reagent (Life Technologies, USA) base on the manufacturer’s instructions. RNA (2 µg) was reverse transcribed using a PrimeScript RT Kit (Takara, Japan). Real-time PCR was performed using SYBR master mix (Takara) on the Bio-Rad Connect Real-Time PCR platform. The primer sequences used for human HOXC6 were 5’-CACCGCCTATGATCCAGTGAGGCA-3’ (forward) and 5’-GCTGGAACTGAACACGACATTCTC-3’ (reverse). Relative mRNA quantities were determined using comparative cycle threshold methods and normalized against GAPDH.

Western blotting

Western blot analyses were performed according to the standard protocol. The primary antibodies used for western blot were described in IHC method.

Transwell invasion assays

Invasion assay was performed with BD BioCoat Matrigel Invasion Chambers (Becton Dickinson Labware, Franklin Lakes, NJ) following the manufacturer’s instructions. The matrigel membrane was stained with crystal violet, and migrated cells were counted under a microscope.

Experimental metastasis assay

All animal experiments were performed in accordance with NIH guidelines for the use of experimental animals. Male athymic BALB/c nude mice were purchased from the Shanghai Experimental Animal Center of Chinese Academic of Sciences (Shanghai, China). Two groups of 8 mice each were given intravenous tail vein injections of 1×10⁶ shControl-SK-Hep1 cells and shHOXC6-SK-Hep1 cells, or 2×10⁶ vector-HuH7 cells and HOXC6-HuH7 cells, respectively. After 8 weeks, the mice were sacrificed, and the tumor nodules formed on the lung surfaces were counted. Lungs were excised and embedded in paraffin for further study.

Follow-up

The last follow-up date was on 30 March 2017. In all the HCC patients (21 females and 143 males), the median follow-up period was 52.5 months (ranging 3.0 - 73.0 months). All patients were conducted once in every 3 months in the first 2 year, once in every 6 months from 2 to 5 year. The follow-up protocol included physical examination, serum AFP level and abdominal ultrasonography. Computed tomography and/or magnetic resonance imaging and/or positron emission tomography were performed when intrahepatic relapse or distant metastasis was suspected. During the follow-up period, there were 83 (50.6%) patients who suffered relapse and 70 patients (42.7%) died of cancer-related causes. Ninety-four patients were still alive at the time of the last follow-up report.

Statistical analysis

Statistical analysis was performed using SPSS 16.0 statistical software (SPSS Inc, Chicago, IL). The relationship between the expression of HOXC6 and clinicopathologic characteristics was tested by χ2 test. The Student’s t-test was used for comparisons. Pearson χ2 test was applied to analyze the correlation of HOXC6 with EMT markers. Multivariate analysis was performed using the Cox proportional hazards model. Kaplan-Meier analysis and log-rank tests were also performed to compare prognostic differences between two groups. P values < 0.05 were considered statistically significant.