CAPZA1 modulates EMT by regulating actin cytoskeleton remodelling in hepatocellular carcinoma

We collected 129 samples from HCC patients who received pathological liver resection at Southwest Hospital between January 2011 and December 2011. Clinicopathological data, including gender, age, tumour size, TNM stage, HCC differentiation, lymph node metastasis, vascular invasion, extrahepatic metastasis and other information, were collected from patient records and pathological examination. We conducted postoperative follow-up with patients until December 31, 2015; postoperative relapse and deaths were also recorded.

HCC tissue specimens were paraffin-embedded and sectioned. After sections were deparaffinized in water, they were placed in sodium citrate solution in a microwave oven at moderate heat for 10 min for antigen retrieval. After naturally cooling, the sections were incubated in 3% H₂O₂ for 10 min and then blocked in 10% goat serum at room temperature for 1 h. Anti-CAPZA1 polyclonal rabbit antibody (1:50 dilution; Proteintech, Rosemont, IL, USA) was added and incubated with the tissues overnight at 4 °C. CAPZA1 immunoreactivity was detected using an anti-mouse/rabbit universal immunohistochemical detection kit (Proteintech). Finally, the sections were dehydrated, after 2 min of haematoxylin staining, and mounted with neutral resin. CAPZA1 staining results were scored by two independent pathologists and grouped according to the percentage of positively stained HCC cells: 0 (0%), 1+ (1–24%), 2+ (25–49%), 3+ (50–74%), 4+ (75–100%) (Fig. 1a).

HepG2 and MHCC97H cell lines (from our laboratory’s cell bank) were cultured in DMEM (Thermo Fisher Scientific, Waltham, MA, USA) containing 10% foetal bovine serum (Biological Industries, Cromwell, CT, USA) and were kept in a humidified incubator containing 5% CO₂ at 37 °C.

A small interfering RNA (siRNA) against CAPZA1 (si-CAPZA1; 5′-GGA ACA AGA UAC UCA GCU A-3′) was purchased from Biomics Biotech (Nantong, China). HepG2 cells were inoculated in 6-well plates 24 h before transfection to achieve 30–50% confluence. Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA, USA) was used to transfect cells with the constructs. Cells were harvested after 24 h, and knockdown efficiency was verified by western blot.

Lentiviruses carrying interference (LV10-CAPZA1) and overexpression (LV8-CAPZA1) constructs for human CAPZA1 were built by (GenePharma, Shanghai, China). Cells were inoculated in 6-well plates 24 h prior to infection to achieve 50–70% confluence. For infection, 1 ml of fresh medium containing 10 μl of lentivirus (1 × 10⁹TU/ml) was added to each well, and after 72 h, the infection rate was observed using a fluorescence microscope, and CAPZA1expression was detected by western blot.

Before invasion assays, 30 μl of Basement Membrane Matrix (diluted 1:6 in PBS; Corning Life Science, Lowell, CA, USA) was applied to the bottom of a Transwell chamber (Millipore, Billerica, MA, USA) and placed in a 37 °C incubator for 5 h. The following steps were the same for the migration and invasion assays: cells (6 × 10⁴) in serum-free media were added into the upper well of the Transwell chamber, and 800 μl of DMEM containing 10% FBS was added to the lower chamber, and then, the chamber was placed in a 37 °C incubator. After18h for migration assays or 36 h for invasion assays, cells were fixed with paraformaldehyde for 30 min, and stained for 3 min with crystal violet. Excess stain was washed away with PBS, and the bottom of the upper chamber was gently wiped with a cotton swab. After drying, the chamber was imaged, and the number of invading and migrating cells was calculated.

Cells (1 × 10³) were inoculated into 96-well plates in triplicate, and the number of cells was detected at 0 h, 24 h, 48 h and 72 h. A Cell Counting Kit-8 (CCK-8; Engreen, Auckland, New Zealand) was used to detect cell numbers. Briefly, 10 μl of CCK-8 was added per well and incubated for 1 h, and then a microplate reader was used to detect absorbance at 450 nm.

Total cellular protein was isolated in RIPA lysis buffer (CWBIO, Beijing, China), and protein concentration was detected using a BCA Protein Assay Kit (CWBIO). Western blotting was performed as previously described [15]. Antibodies against E-cadherin, N-cadherin and Vimentin were purchased from Proteintech. For qPCR assays, an ultrapure RNA Kit (CWBIO) was used to extract total RNA. The primers used to amplifyCAPZA1, E-cadherin, N-cadherin, vimentin and GAPDH areshown in Table 1. SYBR Premix Ex Taq II (Tli RNaseH Plus) (TaKaRa, Shiga, China) was used for reverse transcription and amplification, according to the manufacturer’s protocols.

A SureBeads reagent (BIO-RAD, California, USA) was thoroughly resuspended in its solution and 100 μl of SureBeads was transfered to 1.5 ml tubes. The beads were magnetized with magnetic separation rack and the supernatant was discarded. After washed with 1,000 μl PBST three times, the beads were resuspended with 200 μl final volume solution containing 3 μg of CAPZA1 antibody or actin antibody (Proteintech), and the total solution was rotated at least 30 min at room temperature. And then, the beads were magnetized and the supernatant was discarded. After the beads were washed with PBST three times, the beads were resuspended with 500 μl of the antigen-containing lysate, and the tubes were rotated for the night at 4 °C. The next morning, the beads were magnetized and the supernatant was discarded. After washed with PBST three times, the beads were resuspended with 30 μl SDS-PAGE loading buffer, and then, all tubes were boiled for 10 min. The residual buffer was aspirated from the tubes to the new one after the beads were magnetized. The rest of operation referenced the western blotting protocols.

HCC cells (sh-CAPZA1-expressing and sh-control-expressing HepG2 cells, CAPZA1-overexpressing and control MHCC97H cells) were harvested with trypsin and resuspended in DMEM. Nude mice were anesthetized with 1% sodium pentobarbital (100 ml/kg; Sigma-Aldrich, St. Louis, MO, USA), and 1 × 10⁵ HCC cells were injected under the liver capsule. After 6 weeks, a 7.0 T small animal MRI (Bruker Biospec, Ettlingen, Germany) was used to scan the chest and abdomen of the mice to observe metastases of the orthotopic HCC cells. The time of death of the mice was observed and recorded. After death, the liver and lungs were removed, and the surfaces were observed; then, the liver was sectioned, paraffinized and stained with haematoxylin and eosin, and tumour lesions were observed.

Tissue samples from 129 HCC patients were collected for this study; the patient cohort had an average age of 47.8 ± 10.7 years and contained 115 men (89.1%). Average tumour size was 7.4 ± 3.2 cm (range: 1.0–18.0 cm). By TNM staging the patients were divided into stage I: 18.8% (n = 26), stage II: 9.3% (n = 13), stage III: 47.7% (n = 61) and stage IV: 24.2% (n = 29). By HCC differentiation the patients were divided into well differentiated: 10.4% (n = 10), moderately differentiated: 73.3% (n = 94), and poorly differentiated: 16.3% (n = 25). 9.3% (n = 12) of patients had lymph node metastasis, 44.1% (n = 53) of patients had vascular invasion, 28.7% (n = 37) of patients had extrahepatic metastasis before or after surgery, 70.5% (n = 91) of patients had recurrence within 5 years of surgery, 62.8% (n = 81) of patients died from cancer-related deaths within 5 years of surgery, and 3.1% (n = 4) of patients died from non-cancer-related deaths (Table 2).

CAPZA1 immunohistochemistry was conducted on the 129 HCC tissue samples and scored according to staining intensity. Based on these metrics, samples were divided into the following scoring groups: 0 accounted for 10.1% (n = 13) of patients, 1+ accounted for 35.7% (n = 46), 2+ accounted for 22.5% (n = 29), 3+ accounted for 20.2% (n = 26), and 4+ accounted for 11.6% (n = 15). Patients who scored 0 and 1+ were defined as the CAPZA1 underexpression group, and patients who scored 2+ and above were defined as the CAPZA1 overexpression group (Fig. 1a). We then analysed differences in TNM staging and HCC differentiation and compared the incidence of lymph node and vascular invasion, as well as patient prognosis, in the two groups. By TNM staging, the percentage of patients with stage I or II disease was significantly lower in the CAPZA1 underexpression group (3.4%) than in the CAPZA1 overexpression group (52.9%); conversely, the percentage of patients with stage III or IV disease was significantly higher in the CAPZA1 underexpression group (96.6%) than in the CAPZA1 overexpression group (47.2%; P < 0.001). By HCC differentiation class, the differentiation in the CAPZA1 underexpression group was poorer than in the CAPZA1 overexpression group (P < 0.001). The CAPZA1 underexpression group also had a higher rate of lymphoid tissue invasion (15.3% vs 4.3%; P = 0.033) (Fig. 1b), vascular invasion (55.9% vs 28.6%; P = 0.002) (Fig. 1c) and extrahepatic metastasis (40.7% vs 18.6%; P = 0.006) (Fig. 1d) compared with the overexpression group. Additionally, postoperative follow-up showed that the recurrence rate within 5 years of surgery in the underexpression group was significantly higher than in the CAPZA1 overexpression group (P < 0.001) (Table 3). Finally, Kaplan–Meier analysis showed that the mean postoperative survival time in the CAPZA1 underexpression group (14.6 ± 2.0 months) was significantly lower than in the CAPZA1 overexpression group (40.4 ± 2.9 months) (Fig. 1e).

To test the effects of CAPZA1 expression on the proliferation, migration and invasion of HCC cells, we knocked down CAPZA1 with siRNA in HepG2 cells, which have weak metastatic potential (Fig. 2a). A CCK-8 assay was used to detect the proliferation of HepG2 cells transfected with control siRNA or CAPZ1 siRNA. The data showed that CAPZA1 down regulation had no effect on HepG2 proliferation (Fig. 2b). However, the results from Transwell migration and invasion assays showed that HepG2 migration and invasion were enhanced by CAPZA1 siRNA (Fig. 2d–e).

To further verify these effects, we constructed a CAPZA1 overexpression lentivirus and infected MHCC97H cells, which have strong invasion capabilities (Fig. 2a). Consistent with the results from the siRNA experiments, CAPZA1 overexpression had no effect on the proliferation of MHCC97H cells (Fig. 2c), but significantly reduced migration and invasion compared with control MHCC97H cells (Fig. 2g–i).

In vitro experiments verified that CAPZA1 inhibits the migration and invasion of HCC cells. To verify these results in vivo, we constructed stable sh-CAPZA1- expressing HepG2 cells, CAPZA1-overexpressing MHCC97H cells and corresponding control cells in vitro and grew them in the liver capsule of nude mice. After six weeks, a small animal MRI was used to scan the chest and abdomen of the mice, and the results showed that extensive intrahepatic metastases occurred in 3 of 5 (60%) mice in the sh-CAPZA1-expressing HepG2 cells group, while a few intrahepatic metastases were found in each of the 5 mice in the sh-control-expressing HepG2 cells group; no pulmonary metastases occurred in either HepG2 groups (Fig. 3a). Similarly, in the MHCC97H cell-treated nude mice, a few intrahepatic metastases were found in the CAPZA1-overexpressing group of 5 nude mice, and extensive intrahepatic metastases were observed in 4 of 5 mice (80%) in the control MHCC97H group (Fig. 3a). The liver surface of mice in the sh-CAPZA1-expressing HepG2 group and MHCC97H control group were covered with micro-hepatoma lesions (Fig. 3b). The haematoxylin and eosin-stained nude mice hepatic sections revealed that the number of liver lesions in the sh-CAPZA1-expressing HepG2 group and MHCC97H control group were significantly increased compared with the sh-control-expressing HepG2 group and CAPZA1-overexpressing MHCC97H group (Fig. 3c-e). Finally, the survival time of mice in the sh-control-expressing HepG2 group was significantly longer than that of mice in the sh-CAPZA1-expressing HepG2 group (Fig. 3f). Meanwhile, the survival time of CAPZA1-overexpressing MHCC97H group mice was significantly longer than that of mice in the MHCC97H control group (Fig. 3g).

Actin filaments are an important component of the cytoskeleton that play a critical role in maintaining cell structure and regulating cell movement [16]. CAPZA1 is an actin-binding protein that is intimately involved in actin filament assembly [17]. To investigate whether CAPZA1 regulates EMT in HCC cells, we conducted siRNA and overexpression experiments and measured changes in the expression of the EMT markers E-cadherin, N-cadherin and Vimentin. When CAPZA1 expression was down-regulated, E-cadherin expression was decreased, while the mesenchymal markers N-cadherin and Vimentin were up-regulated (Fig. 4a–c); in contrast, when CAPZA1 was overexpressed, E-cadherin expression was up-regulated, while N-cadherin and Vimentin were down-regulated (Fig. 4d–f). Thus, CAPZA1 significantly inhibited EMT in HCC cells. To explore which transcription factor is involved in the EMT process, we detected EMT transcription factors of Snail family, Twist family and ZEB family. As the results, we found that the expression of Snail1 and ZEB1 were up-regulated when CAPZA1 expression was down-regulated. Conversely, the expression of Snail1 and ZEB1 were down-regulated when CAPZA1 was overexpressed (Fig. 4g). We also detected the interaction between CAPZA1 and F-actin, and the immunoprecipitation assay showed that CAPZA1 and actin can pull each other down (Fig. 4h). Thus, CAPZA1 inhibits EMT by regulating actin cytoskeleton remodeling via Snail1/ZEB1.