In order to detect differences in HNRNPC protein levels between tumor tissues and para-tumor tissues, we collected 12 pairs of fresh tissues (tumor and matched para-tumor tissue) from patients diagnosed with HCC by a pathologist after undergoing hepatectomy between Dec 2018 and Jan 2019 in the Department of Hepatobiliary Surgery of the Second Affiliated Hospital of Chongqing Medical University. For analysis of prognosis and clinical information, our study enrolled 147 available cases from Jul 2014 to Nov 2016 from the same department who had not received chemotherapy, radiofrequency ablation, or molecular targeted therapy before liver resection. These 147 cases included 128 male and 19 female patients, with an average age of 53.67 years. Specific patient information is given in Table 1 . The selection criteria were as follows. Inclusion criteria: HCC confirmed by biopsy after surgery; exclusion criteria: patients with other cancers, distant metastasis, or grade C Child–Pugh liver function. A total of 147 tumor tissues and 82 available corresponding para-tumor tissues were used to construct tissue microarrays (TMAs). This research was approved by the Ethics Committee at the Second Affiliated Hospital of Chongqing Medical University ((2017)36), and informed consent was obtained from all patients.

Follow-up was carried out by phone, and alpha fetoprotein (AFP), liver function, and chest radiography were obtained at least every 2 months in the first 6 months after surgery, every 3 months for 6 months to 3 years following surgery, and every 6 months for 3–5 years after surgery. If necessary, magnetic resonance imaging or computed tomography was also used. Overall survival was defined as the time from liver resection to death or last follow-up; disease-free survival was defined as the time from liver resection to first recurrence or last follow-up.

Fresh tissues were immobilized with 4% formalin for 48 h at 37°C and subsequently washed with cool phosphate-buffered saline (PBS) three times for 5 min each time. Tissue samples were dehydrated using ethanol, embedded in paraffin, and then cut them into 4-μm-thick sections of 2 mm diameter to produce TMAs. The paraffin sections were dewaxed in an incubator at 65°C overnight, before being deparaffinized in xylene and rehydrated using a decreasing ethanol gradient. Sections were placed in ethylene diamine tetraacetic acid (cat.no. 0085; Beyotime) and heated in a microwave for 10 min for antigen repair. A 3% hydrogen peroxide was used to eliminate endogenous peroxidase at 37°C for 30 min. Sections were incubated with anti-HNRNPC (dilution 1:50, cat. no. ET1611-2; HUABIO), anti-ki67 (dilution 1:10000, cat. no. 27309; Proteintech), anti-CD34 (dilution 1:50, cat. no. ET1606-11; HUABIO), anti-Vimentin (dilution 1:100, cat. no. ET1610-39; HUABIO), and anti-E-Cadherin (dilution 1:50, cat. no. ET1607-75; HUABIO) antibodies at room temperature overnight, and then with secondary antibody (dilution: 1:1, cat.no. K5007; Dako) for 1 h at room temperature. Finally, sections were stained with an IHC kit (cat. no. K5007; Dako) according to the manufacturer’s protocol.

The proportion of positive cells and staining strength were used to evaluate the expression of HNRNPC. Color intensity was ranked using four grades: no staining (score = 0), weak staining (score = 1), moderate staining (score = 2), and intense staining (score = 3); and the proportions of positive cells were divided into five classifications: < 5% (score = 0), 5–25% (score =1), 26–50% (score = 2), 51–75% (score = 3), and 76–100% (score = 4). These two values were multiplied to acquire the final score. All patients were divided into high and low HNRNPC expression groups according to the median score of 7. The IHC results were analyzed by two pathologists; uncertain results were judged by a third pathologist.

We collected total protein from cells or tissues using RIPA buffer with protease and phosphatase inhibitors. After separation by sodium dodecyl polyacrylamide gel electrophoresis, proteins were transferred onto polyvinylidene fluoride membranes. Subsequently, the membranes were blocked with 5% skim milk at room temperature for 1 h, then incubated with primary antibody at 4°C overnight. On the second day, the membranes were incubated with secondary antibody for 1 h at 37°C. The antibodies used were as follows: HNRNPC (dilution 1: 2000, cat. no. ET1611-2; HUABIO), Ras (dilution 1:1000, cat. no. 3339T; CST), c-Raf (1:1000, cat. no. 9422T; CST), P-c-Raf (dilution 1:1000, cat. no. 9427T; CST), MEK1/2 (dilution 1:1000, cat. no. 8727T; CST), P-MEK1/2 (Ser217/221) (dilution 1:1000, cat. no. 9154T; CST), Erk1/2 (dilution 1:1000, cat. no. 4695T; CST), P-Erk1/2 (Thr202/Tyr204) (dilution 1:1000, cat. no. 4377T; CST), c-Myc (dilution 1:1000, cat. no. 5605T; CST), CDK4 (dilution 1:2000, cat. no. ET1612-1; HUABIO), cyclin E1 (dilution 1:1000, cat. no. ET1612-16; HUABIO), Vimentin (dilution 1:2000, cat. no. ET1610-39; HUABIO), E-Cadherin (dilution 1:1000, cat. no. ET1607-75; HUABIO), β-actin, (dilution 1:2000, cat. no. BM0627; Boster), goat anti-mouse (dilution 1:5000, cat. no. ZB-2305; ZSGB-BIO), and goat-anti-rabbit (dilution 1:5000, cat. no. ZB-2301; ZSGB-BIO). Finally, the membranes were imaged using a chemiluminescent horseradish peroxidase substrate (cat. no. WBKL S0100; Millipore). β-actin was used as the internal reference, and the western blot analysis was performed using Image Lab 6.0. The tests mentioned above were operated three times respectively.

Total RNA was extracted from cells using an Animal Total RNA Isolation Kit (cat. no. RE-03011; Foregene) according to the manufacturer’s instructions. Total RNA was reverse transcribed to cDNA with an iScript cDNA synthesis kit (cat. no. 1708891; Bio-Rad), following the manufacturer’s reaction protocol: 5 min at 25°C 20 min at 46°C, 1 min at 95°C, and holding at 4°C. Quantitation of cDNA was performed using a SYBR Green PCR kit (cat. no. 208054; QIAGEN). Relative HNRNPC expression was calculated by the 2^-△△Cq method, with GAPDH as an internal reference. The primers were designed as follows: GAPDH forward, 5′-GGAGTCCACTGGCGTCTTCA-3′, reverse, 5′-GTCATGAGTCCTTC CACGATACC-3′; HNRNPC forward, 5′-GCAGAGCCAAAAGTGAACCG-3′, reverse, 5′-ACGTTTCGAGGGCACTACAG-3′. The reaction protocol was given by the manufacturer as follows: 2 min at 95°C, 5 sec at 95°C, and 10 sec at 60°C, repeated for 35 cycles. Each test was repeated dividedly in triplicate.

Huh-7 and Hep 3B HCC cell lines were purchased from the Chinese Academy of Sciences Stem Cell Bank (Shanghai, China) and identified with STR Profile. Both cell lines were cultured in DMEM (cat.no. SH30022.01; HyClone) or MEM (cat.no. SH30024.01; Hyclone) supplied with 10% fetal bovine serum (cat.no. 04-001-1ACS; BI) in an incubator containing 5% CO₂ at 37°C.

Huh-7 and Hep 3B cells were cultured in six-well plates. When they reached 20–30% density, they were stably infected with HNRNPC lentivirus (Lv-HNRNPC) or a negative control lentivirus (Lv-NC) for 14 h (viral volume=MOI × cell number/viral titer; GeneChem, Shanghai, China). The lentiviral interference system was produced using a GV248 carrier loaded consecutively with the hU6 promoter, a multiple cloning site, the ubiquitin promoter, the EGFP gene, an IRES site, and the puromycin gene. A short hairpin RNA was designed to target the HNRNPC sequence (sh-HNRNPC): 5′-CTTCGTTCAGTATGTTAAT-3′; the negative sequence (sh-NC) was 5′-TTCTCCGAACGTGTCACGT-3′. After culture for 48 h, cells were filtrated with 2 µg/ml puromycin (cat. no. REVG1001; GeneChem, Shanghai, China) for 48 h; this was repeated twice. Infection and knockdown efficiency were detected by inverted fluorescence microscopy, and by western blotting and q-RT-PCR, respectively.

Approximately 500 cells were seeded in each well of the six-well plates and cultured in an incubator with 5% CO₂ at 37°C for 2 weeks. After washing with PBS, the cells were immobilized with 4% paraformaldehyde for 15 min at room temperature and then stained with 0.1% crystal violet solution (cat. no. G1062; Solarbio) for 30 min. Finally, the colony numbers were counted under a microscope. All experiments were operated independently three times.

Approximately 3000 (Huh-7) or 8000 (Hep 3B) cells were seeded in each well of a 96-well plate and cultured for 72 h. After being seeded in a 96-well plate, cell viability was measured using a CCK-8 kit (cat. no. ZP328-3; ZOMANBIO) according to the manufacturer’s instructions at 6, 24, 48 and 72h. Briefly, CCK-8 (10 µl) was added to each well, and the cells were incubated at 5% CO₂ and 37°C. The absorbance (optical density) was detected by enzyme-linked immunosorbent array at a wavelength of 450 nm. Each test was performed dividedly in triplicate.

When the cells had grown to a density of 90% in the six-well plates, we used a sterile 200-µl pipette tip to scratch a wound at the middle line of each well bottom. Subsequently, the suspended cells were removed by PBS, and the culture medium was replaced with serum-free medium. Cells were sequentially cultured for 48 h in an incubator containing 5% CO₂ at 37°C. The wound width was recorded under a light microscope at 0, 24 and 48h. The experiments mentioned above were operated three times respectively.

To analyze the invasion ability of cells, a Corning BioCoat Matrigel Invasion Chamber (cat. no. 354480; Corning) was used according to the manufacturer’s instructions. Specifically, cells (5×10⁴/well) were cultured with serum-free medium in the upper chambers, while the lower chambers were loaded with DMEM or MEM supplied with 10% fetal bovine serum. After culture for 48 h, the cells were washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature. Invading cells were captured with 0.1% crystal violet solution under a light microscope, and non-invading cells were eliminated with cotton swabs. Specifically, five fields in each chamber were randomly recorded, and average values were calculated. Each test was repeated independently in triplicate.

Approximately 25×10⁴/well cells were seeded in six-well plates. Subsequently, the cells were collected in 1.5-ml EP tubes and then fixed with 70% ethanol at 4°C overnight. The next day, the cells were stained with propidium iodide with RNase A at 37°C for 30 min in the dark. Finally, the results were analyzed by flow cytometry. A total of 2×10⁴ cells were recorded for each sample. All experiments were performed three times respectively.

To verify the upstream and downstream relationships between HNRNPC and the MAPK signaling pathway, we used an agonist for Ras (ML-098) (cat. no. HY-19800; MCE). The specific treatments were as follows. To measure the efficiency of the activation of the MAPK signaling pathway by ML-098, the Lv-NC group was treated with dimethyl sulfoxide (DMSO) for 48 h, and the Lv-HNRNPC group was treated with ML-098 (20 nmol/l) and DMSO, for 48 h. CCK-8 tests, cell cycle analysis, scratch wound assay, and cell invasion tests were performed using the method described above. To detect the rescue efficiency of ML-098 for colony formation, we used ML-098 (20 nmol/l) to treat the Lv-HNRNPC group twice sequentially for 5 days each time. The Lv-NC group was treated with the same amount of DMSO as a control.

A total of 12 5-week-old male BALB/c nude mice were purchased from Beijing Huafukang Biotechnology and randomly divided into two groups (six per group). Lv-HNRNPC- and Lv-NC-group Huh-7 cells were suspended in cool PBS and subsequently injected into the backs of mice subcutaneously (2.5×10⁶/mice). The mice were fed at 23–25°C and 60% humidity with a 12-h light/dark cycle. Tumor size was measured with a caliper every 3 days; the formula for the size calculation was as follows: volume = (length × width²)/2 cm³. Twenty-seven days after implantation, the mice were killed by cervical dislocation and the tumors were removed. Finally, the tumor samples were fixed with 4% formalin for IHC analysis. All animal experiments were approved by the Ethics Committee of Chongqing Medical University.

The expression of HNRNPC in HCC tissues was analyzed by UALCAN (ualcan.path.uab.edu), using RNA sequencing data downloaded from The Cancer Genome Atlas for 50 normal liver tissues and 371 primary HCC tissues. HNRNPC-correlated genes were also acquired from UALCAN; we selected 3517 significantly correlated genes (Pearson correlation coefficient ≥ 0.5) for Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis by WebGestalt 2013 (www.webgestalt.org/2013).

IBM SPSS Statistics 23 was used for statistical analysis. Categorical data were analyzed by chi-square test, while continuous data between two groups were analyzed by t-test. Comparisons between multiple groups were performed by one-way analysis of variance (ANOVA), and the results were analyzed by LSD test and presented as mean ± standard deviation (SD). Univariate and multivariate Cox proportional hazards regression models were used to assess the risk factors associated with HCC. In addition, overall and disease-free survivals were computed by the Kaplan–Meier method and examined by log-rank test. Finally, a P-value less than 0.05 was considered to indicate a statistically significant difference.

Using the UALCAN database, we found that HNRNPC mRNA levels were higher in 371 primary HCC tissues than in 50 normal liver tissues (P < 0.001, Figure 1A ); in addition, the expression of HNRNPC mRNA was elevated in stage 2 (P = 0.033, Figure 1B ) and stage 3 (P < 0.001, Figure 1B ) HCC patients compared with stage 1 HCC patients. To compare HNRNPC protein expression in tumor and para-tumor tissues, western blotting and IHC were performed. HNRNPC protein levels were obviously higher in tumor tissues than in para-tumor tissues, based on western blot analysis of 12 HCC tumor tissues and their matched para-tumor tissues (P< 0.001, Figures 1C, D ). The IHC results showed high expression of HNRNPC protein, mainly located in nuclei, in 80 of 147 tumor tissues (54.40%) and in 13 (15.90%) of 82 paired para-tumor tissues (P<0.001, Figures 1E, F ). Correlations of HNRNPC protein levels with clinicopathological features in HCC were analyzed by chi-square test; this demonstrated that HNRNPC protein levels were significantly associated with tumor size (P = 0.007) and tumor TNM stage (P=0.028) ( Table 1 ). To explore the effects of HNRNPC protein levels on survival of HCC patients, Kaplan–Meier survival curves were constructed, showing that patients with high HNRNPC protein expression had poorer overall survival (P = 0.001, Figure 1G ) and disease-free survival (P < 0.001, Figure 1H ) than those with low HNRNPC protein expression. Univariate and multivariate analysis were used to evaluate the risk factors associated with HCC patients’ survival. According to our results, tumor size (P = 0.005, HR = 1.957), tumor TNM stage (P = 0.014, HR = 1.786), and high HNRNPC expression (P = 0.039, HR = 1.637) could be regarded as independent risk factors for overall survival of HCC patients ( Table 2 ). Moreover, tumor TNM stage (P < 0.001, HR = 2.188) and high HNRNPC level (P = 0.002, HR = 1.883) could independently predict disease-free survival in HCC patients ( Table 2 ).

To observe the impact of HNRNPC on HCC proliferation, invasion, and migration, Lv-NC and Lv-HNRNPC lentiviruses were used to infect the Huh-7 and Hep 3B cell lines. After selection by puromycin, the infection efficiency of Huh-7 and Hep 3B cells was >90% ( Supplementary Figure 1A ), and lower HNRNPC expression was detected in Lv-HNRNPC Huh-7 and Hep 3B cells via western blotting and q-RT-PCR ( Supplementary Figures 1B, C ). CCK-8 and colony formation assays were used to analyze cell proliferation in vitro. The colony formation assay revealed that HNRNPC knockdown significantly inhibited proliferation of Huh-7 and Hep 3B cells ( Figure 2A ). In the CCK-8 test, the viability of Huh-7 and Hep 3B cells was also significantly suppressed by silencing of HNRNPC ( Figure 2B ). A transwell assay was performed to observe invasion; the results showed that inhibition of HNRNPC impaired the invasion activity of Huh-7 and Hep 3B cells ( Figure 2C ). The scratch wound assay showed decreased migration ability of Huh-7 and Hep 3B cells in the Lv-HNRNPC-group ( Figure 2D ).

To study the potential molecular mechanisms by which HNRNPC suppressed tumor growth, migration and invasion, we selected 3517 genes that were significantly correlated with HNRNPC (Pearson correlation coefficient ≥ 0.5) from the UALCAN database; these genes were subjected to KEGG analysis using WebGestalt (2013). Among the enriched KEGG pathways, “Cell cycle” and “Pathways in cancer” were highly related to tumor growth ( Supplementary Figure 1D ), and “Pathways in cancer” mainly involved the MAPK and AKT signaling pathways ( Supplementary Figure 1E ). Subsequently, the activity of the MAPK signaling pathway was detected by western blotting. The results showed that HNRNPC knockdown statistically decreased the expression of Ras and the relative levels of p-Raf, p-MEK1/2, and p-Erk1/2 in Huh-7 and Hep 3B cells ( Figures 3A, B ). As the MAPK pathway activates a series of downstream genes to regulate the cell cycle, thereby regulating cell proliferation, we used flow cytometry for cell cycle analysis. The results showed that HNRNPC downregulation blocked more cells at G0/G1 phase and reduced the proportion of S-phase cells among both Huh-7 and Hep 3B cells ( Figures 3C, D ). CDK4, cyclin E1, and c-myc are well known as important molecules that positively regulate G1-S transition. In the present study, silencing of HNRNPC markedly downregulated CDK4, cyclin E1, and c-myc protein levels, as confirmed by western blotting ( Figures 3E, F ). Ras/MAPK signaling pathway not only positively regulate G1-S transition to promote tumor proliferation, but also accelerate EMT process to facilitate tumor migration and invasion. We also detected the marker of EMT by western blotting, our results indicated that HNRNPC silencing decreased Vimentin level, and upregulated E-cadherin expression statistically ( Figures 3G, H ).

To explore the impact of HNRNPC on tumor growth in vivo, Lv-NC and Lv-HNRNPC Huh-7 cells were implanted in nude mice subcutaneously. The tumors of the Lv-HNRNPC group showed obvious reductions in both size and weight compared with those of the Lv-NC group ( Figures 4A–C ). Subsequently, hematoxylin eosin (H&E) and IHC staining were performed to confirm the tumor tissues and HNRNPC knockdown efficiency, respectively ( Figure 4D ). In addition, Ki-67 was used to observe the proliferation of tumor cells via IHC; the Ki-67 staining intensity was stronger in the Lv-NC group than in the Lv-HNRNPC group ( Figure 4D ). Tumor migration and invasion are highly correlated with angiogenesis and EMT process, so Vimentin, E-cadherin, and CD34 were used to evaluate the ability of tumor invasion and migration by IHC. The results showed that Vimentin and CD34 staining intensity were weaker, and E-cadherin staining intensity was stronger in the Lv-HNRNPC group than in Lv-NC group ( Figure 4E ).

CDK4, cyclin E1, and c-myc are classical downstream genes of the MAPK signaling pathway. To confirm the upstream and downstream relationships between the MAPK signaling pathway and these cell cycle-related molecules, a Ras agonist (ML-098) was used to treat cells in the Lv-HNRNPC group. Details of this process are given in the Materials and Methods section. First, we performed western blotting to detect the recovery of MAPK pathway activity. After treatment with ML-098 for 48 h, the activity of the MAPK pathway was reversed completely in the Lv-HNRNPC group ( Figures 5A, B ). Moreover, colony formation and CCK-8 assays revealed that the proliferation ability of cells in the Lv-HNRNPC group was partly rescued by ML-098 ( Figures 5C–E ). Furthermore, the cell cycle assay showed that ML-098 treatment could partially reverse the G0/G1 arrest caused by HNRNPC knockdown in both Huh-7 and Hep 3B cells ( Figures 6A, B ). Finally, using western blotting, we showed that the effects of HNRNPC inhibition on CDK4, cyclin E1, and c-myc levels in HCC cells were reversed by ML-098 to some extent ( Figures 6C, D ). On the other hand, Ras/MAPK signaling pathway take part in EMT process. In present study, cell invasion and scratch wound assays showed that the invasion ( Figures 7A, B ) and migration ( Figures 7C, D ) abilities of the Lv-HNRNPC group cells were recovered by ML-098 partly. Furthermore, the western blotting results demonstrated that after treatment with ML-098, the EMT process inhibition of cells in the Lv-HNRNPC group was reversed in part too ( Figures 7E, F ). Taken together, these results indicate that HNRNPC inhibition not only arrests HCC cells in G0/G1 phase to inhibit tumor proliferation, but also suppresses EMT process to block invasion and migration of HCC cells in part via the Ras/Raf/MEk/Erk signaling pathway.