Piezo1 promoted hepatocellular carcinoma progression and EMT through activating TGF-β signaling by recruiting Rab5c

A total of 150 HCC specimens in training cohort collected from January 2009 to December 2012 were randomly selected from the patients received liver resection at Department of Surgery, Xiangya Hospital of Central South University. Another 130 HCC specimens in validation cohort collected from January 2010 to December 2012 were randomly selected from the Department of Abdominal Surgical Oncology, Affiliated Cancer Hospital of Xiangya School of Medicine, Central South University. The patient demographics and clinicopathological variables of the two cohorts are described in Additional file 3: Table S1. Furthermore, 30 matched fresh HCC tissues and adjacent nontumoral liver tissues (ANLTs) were collected from Xiangya Hospital from September to December 2019. 10 fresh HCC tissues in each clinic subtypes were collected from Xiangya Hospital from June to December 2019. The diagnosis of HCC in all patients was confirmed by two independent histopathologists. The present study was approved by the Ethics Committee of Xiangya Hospital, Central South University. All patients and their families provided written informed consent and agreed to the use of their tissue samples in the study in accordance with the Declaration of Helsinki.

Cells were washed 3 times by PBS and added lysis buffer (250 mM sucrose, 10 mM HEPES, 1 mM EDTA, pH 7.5) supplemented with protease inhibitors, and scraped into centrifuge tube. Extracts were mechanically lysed 50times by homogeniser, and subjected to 90 min of ultracentrifugation at 45,000 rpm (Thermo Fisher Scientific, Waltham, MA) at 4 °C. The resulting supernatant was the cytosolic fraction. The membrane fraction was resuspended in 100 μl immunoprecipitation assay buffer (10 mM Tris–HCl, pH 7.5, 1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS 140 mM NaCl), with protease inhibitors. Cytosol and membrane protein bands were quantified and relativized against their respective fraction markers RhoGDI and transferrin receptor [29].

Immunohistochemical staining on formalin-fixed, paraffin-embedded tissue sections 4 μminthickness wasperformed using the polymer HRP detection system (Zhongshan Goldenbridge Biotechnology, Beijing, China). Immunohistochemical experiments were conducted as previously described [14, 30]. The IHC score of target proteins was independently evaluated by two investigators according to the proportion and intensity of positive cells within five randomly selected fields per slide (magnification, × 400). The intensity was assessed by four grades: 0 for none; 1 for weak; 2 for moderate; 3 for strong. The percentage of positive cells was divided into five degrees: 0, no positive tumor cells; 1 for ≤ 5%; 2 for 6–25%; 3 for 26–75%; 4 for ≥ 76%. Immunoreactive score was calculated by multiplying the staining extent score with the intensity score. As previous reported, high expression was defined as a staining index score > 4, while low expression was defined as a staining index score ≤ 4[31, 32].

Animal xenograft assays were conducted with 6-weekold male BALB/c nude mice (six mice per group). 5 × 10⁶ indicated cells were subcutaneously injected into the right dorsal flank of nude mice. Tumor sizes were measured at the indicated time points and calculated with the following formula: Tumor volume = L × W² × 0.5 (L, length; W, width) [12]. After 4 weeks, the mice were sacrificed, and the tumors were harvested to weigh and undergo further experiments. Orthotopic tumor implantation was performed as described previously [33]. After 8 weeks, the mice were sacrificed, and the livers and lungs were harvested, the tumor size was measured by the vernier caliper as previously described [12], imaged and processed for histopathological examination. All animal experiments were conducted at the Animal Institute of CSU according to the protocols approved by the Medical Experimental Animal Care Commission of CSU. As the proposal of the Animal Institute of CSU, the euthanasia of all the experimental mice were used the pentobarbital sodium, 150 mg/Kg, intraperitoneal injection.

Statistical analysis was performed using SPSS 24.0 software (SPSS Inc., Chicago, IL). The experimental data was presented as the mean ± SD and analyzed using Student’s t test or one-way ANOVA. The Chi-squared test was applied to examine the association between Piezo1 expression and clinicopathological parameters. Survival curves for patients were calculated using the Kaplan–Meier method and analyzed using the log-rank test. Prognostic factors were examined by univariate and multivariate analyses using the Cox proportional hazards model. Spearman’s rank analysis was performed to determine the correlation between different protein levels. Student’s t test was performed to measure the membrane and cytolic protein expression level. All differences were deemed statistically significant at P < 0.05.

Further details of materials and methods are described in the Additional file 1: Materials and methods.

Firstly, we examined the levels of Piezo1 and Piezo2 mRNA in 7 HCC cell lines and 2 normal hepatocytes, the Primary human hepatocytes (PHH) and immortalized hepatocytes (L02) (Fig. 1A). Notably, the expression of Piezo1 mRNA in HCC cells significantly stronger than PHH and L02. But this difference of expression was not observed in Piezo2 (Fig. 1B). Then, we also detected the mRNA expression of Piezo1 and Piezo2 in frozen HCC tissue and the corresponding adjusted nontumor liver tissue (ANLT). Consistently, mRNA expression level of Piezo1 significantly upregulated compared with ANLT, but the expression difference was not detected in Piezo2. Previously, we found a new clinic subtype of HCC. We summarized this subtype as Solitary Large HCC (SLHCC) which tumor major axis > 5 cm, expansive growth, and with intact capsule or pseudocapsule [6]. According to these clinic pathologic features, we classified HCC into 3 clinic subtypes, the SLHCC, nodular HCC (NHCC, node number > 1) and Small HCC (SHCC, tumor diameter ≤ 5 cm). In our previous research, the exhibited a similar long-term overall and disease-free survival with SHCC, but much better than NHCC [4, 7‐9]. We also detected mRNA of Piezo1 in the 3 clinic subtypes of HCC, and the NHCC with high metastatic potentials expressed relatively higher level of Piezo1 than the SLHCC and SHCC with low metastatic potentials, but Piezo2 was not significantly differential expressed. Then, Piezo1 and Piezo2 protein expression in HCC cell lines and PHH, L02 cells was detected by WB (Fig. 1C, D, the quantification result seen in Additional file 2: Fig. S1). Therefore, we supposed that Piezo1 might associate with progression of HCC, but not Piezo2. Expression level of Piezo1 was also analyzed by immunohistochemistry (IHC), which suggested that Piezo1 protein was highly expressed in HCC (Fig. 1E). These data reveal that Piezo1 upregulated in HCC cell lines and tissues, and indicates that it’s very necessary to explore the role of Piezo1 in HCC progression.

After the upregulated expression level of Piezo1 in HCC was determined, we further verified the expression difference through the analyses of mRNA datasets from TCGA (P = 0.0004) and Gene Expression Omnibus (Fig. 2A, GSE76297 P = 0.0364, GSE36376 P < 0.0001, GSE10143 P < 0.0001), which verified our experiments that Piezo1 upregulated in HCC. In order to explore the correlation between Piezo1 expression level and clinic pathologic features of HCC, we performed IHC in 280 patients collected form 2 hospitals, and concluded to Training cohort and Validation cohort (Fig. 2B). Before the research, the comparability of data was verified (Additional file 3: Table S1). Then the correlation test was performed and revealed the Piezo1 expression level correlated with several fatal clinicopathological features. In the training cohort (n = 150), a high expression level of Piezo1 was closely correlated with tumor size, tumor nodular number, capsulation formation, high Edmondson–Steiner grade, micro- and macro-vascular invasion, advanced tumor node metastasis stage (TNM), Barcelona Clinic Liver Cancer (BCLC) stage and China Clinic Liver Cancer (CNLC) stage [2] (all P < 0.05, Table 1), and verified by Validation cohort. Furthermore, uni- and multi-variate analysis revealed that high Piezo1 expression was an independent risk factor for both OS and DFS of HCC patients after liver resection (Table 2), this result was verified in validation cohort (Additional file 3: Table S2). In addition, Kaplan–Meier survival analysis in TCGA showed that HCC patients with low Piezo1 had longer OS and DFS, but Piezo2 showed no significance between low- and high- expression group (Fig. 2C), which consistent with our above research. Notably, in training cohort, survival analysis revealed that the high Piezo1 expression group had worse OS (1-, 3-, and 5-year OS: 91.33%, 67.33%, and 48.67% vs. 79.33%, 41.33%, and 21.33%) and DFS rates (1-, 3-, and 5-year DFS: 82.31%, 61.54%, and 41.54% vs. 70.77%, 31.54%, and 10.77%) than patients in the low-expression group (Fig. 2D), and also verified in Validation cohort. Besides, this result was also observed in integrated cohort combined by training and validation cohort and 3 clinical subtypes of HCC (Fig. 2E). These data fully confirmed that Piezo1 expression level was closely correlated with clinic pathologic features and poor survival, and has the potential to be a novel independent prognosis biomarker for HCC patients after hepatic resection.

To understand the function of Piezo1 in HCC cells, we manipulated Piezo1 expression in cells by short hairpin RNA (shRNA). Due to the molecular weight of Piezo1 is too large to construct overexpression vector, we interfered Piezo1 in two highest expression cell lines HCCLM3 and Hep3B cells by three shRNAs separately, as named HCCLM3^shPiezo1 and Hep3B^shPiezo1. The corresponding control lentivirus transfected HCCLM3 and Hep3B, as named HCCLM3^shCtr and Hep3B^shCtr. The expression level of Piezo1 was identified by real-time PCR and western blotting. shRNA2 was the most effective one and was chosen for further study (Additional file 2: Fig. S2A, B). The wound-healing and transwell assays were used to investigate migration and invasion capacity. The results showed that HCCLM3^shCtr cells had a faster wound closure rate and more invasion than HCCLM3^shPiezo1 cells, and Hep3B^shPiezo1 cells had also markedly reduced migratory and invasive capacity (Fig. 3A, B). Compared to HCCLM3^shPiezo1, cell clone formation assay (Fig. 3C) and MTT assay (Additional file 2: Fig. S3) indicated HCCLM3^shCtr had a higher proliferation rate. Consistently, Hep3B^shCtr cells also formed more colonies in colony formation assay. In IF stained F-actin (Fig. 3D), we also noticed the morphological change when Piezo1 was interfered, in which the spindle-like mesenchymal morphology of HCCLM3^shCtr and Hep3B^shCtr changed into cobblestone-like epithelial morphology of HCCLM3^shPiezo1 cells and Hep3B^shPiezo1 cells. To verify the above findings in vivo, we established subcutaneous (SC) xenograft tumor and orthotopic xenograft tumor models, as previously described [34]. After 6 weeks, HCCLM3^shCtr and Hep3B^shCtr cell-derived tumors at the SC implantation sites were larger and grew more rapidly than HCCLM3^shPiezo1 and Hep3B^shPiezo1 derived tumors (Fig. 3E, and the growth curve was shown in Additional file 2: Fig. S4 and tumor weight in Additional file 2: Fig. S5). Consistently, liver orthotopic xenograft tumor also showed that HCCLM3^shPiezo1 and Hep3B^shPiezo1 derived tumors significantly smaller than HCCLM3^NC and Hep3B^NC-derived tumors. Ki67 for the subcutaneous xenograft tumors and orthotopic xenograft tumors was detected by IHC (Additional file 2: Fig. S6), which indicates that Ki67 negative in Piezo1 interfered HCCLM3 and Hep3B cells and positive in the control cells. All these results indicate that Piezo1 knockdown inhibited tumor growth in vivo (Fig. 3F). Taking these results together, our results shows that Piezo1 promotes HCC growth, progression in vitro and in vivo.

In the present study, the morphological changes in IF stained F-actin attracted our attention when we silenced Piezo1 expression in HCCLM3 and Hep3B cells. In addition, the hallmarks of EMT enriched in high Piezo1 group in the gene set enrichment analysis (GSEA) of TCGA database (NES = 1.678, NOM p-val = 0.004, Fig. 4A). We speculated that Piezo1 promoted HCC progression through EMT. To confirm our hypothesis, cytological and histological experiments were conducted to verify the expression relationship between Piezo1 and EMT markers. At the protein level, Piezo1 knockdown in HCCLM3 and Hep3B cells resulted in the expression of E-cadherin increased and vimentin decreased, whereas high Piezo1 expressed HCCLM3^shCtr and Hep3B^shCtr cells resulted in opposing results (Fig. 4B). IF analysis showed that the expression of the mesenchymal marker vimentin was significantly reduced in Piezo1 interfered HCCLM3^shPiezo1 and Hep3B^shPiezo1 cells and increased the expression of the epithelial marker E-cadherin, while relatively high expression level of Piezo1 induced inverse results in HCCLM3^shCtr and Hep3B^shCtr (Fig. 4C). IHC staining of consecutive HCC sections also showed that vimentin expression levels were up-regulated in high Piezo1-expressing cell-derived tumors, whereas E-cadherin expression was reduced (Fig. 4D). Moreover, IHC revealed that Piezo1 expression was negatively correlated with E-cadherin expression and positively correlated with vimentin expression (Fig. 4E). These results indicated that Piezo1 promotes EMT in HCC cell lines.

In order to investigate how the Piezo1 promotes EMT and progression of HCC, we turn to see potential signaling pathways manipulated by Piezo1 in HCC. Firstly, the GSEA of the TCGA database showed that TGF-β signaling was the most significantly enriched pathway in Piezo1 high group (NES = 1.775, NOM p-val = 0.002, Fig. 5A. The result of GSEA shown in Additional file 3: Tables S3, S4). Then, we conducted the Cignal Finder Cancer 10-Pathway Reporter Array (Qiagen, Germany) in Piezo1 expression interfered HCC cells to further screen and confirm the signaling pathway regulated by Piezo1. The results showed that TGF-β signaling was the most significantly altered pathway (Fig. 5A) between Piezo1 expression interfered HCC cells and the corresponding control cells. So, we further detected the expression of key members of the TGF-β signaling pathway in Piezo1 manipulated HCC cells, the result showed that in shPiezo1 cells, the protein levels of AKT, p-AKT, ERK, p-ERK and smad2/3 had no significant difference, only phosphorylated-smad2/3 was down-regulated in Piezo1 knockdown cells (Fig. 5B), which indicated the canonical TGF-β signaling. Meanwhile, several genes related to EMT were analyzed. The data showed that E-cadherin was up-regulated and vimentin was down-regulated in Piezo1 knockdown cells. Moreover, IHC staining in 40 HCC samples randomly selected from the training and validation cohorts showed that p-Smad2/3 protein was highly expressed in HCC tissues and positively correlated with Piezo1 expression (P < 0.001, r = 0.578, Fig. 5C). Then, the four cell lines were treated with LY2109761 (10 μM, 2 h), a TGF-β/smad2/3 inhibitor, and we found that LY2109761 did not affected Piezo1 expression. Compared with the control cells, HCCLM3^shPiezo1 and Hep3B^shPiezo1cells showed a decreasing trend in phosphorylation level of Smad2/3, which was similar to LY2109761 treated HCCLM3^shCtr and Hep3B^shCtr cells. It proves that Piezo1 activates TGF-β signaling in HCC cells (Fig. 5D). To further study the role of Piezo1 and TGF-β signaling in progression, Wound-healing (Fig. 5E) and transwell (Fig. 5F) assays showed that LY2109761 treatment obviously decreased the migration and invasion capacity of Piezo1 high expressed cells, but had none or less effect for cells in HCCLM3^shPiezo1 and Hep3B^shPiezo1cells which Piezo1 expression and TGF-β signaling activity was low. Taken together, the public database analysis and our results showed that Piezo1 could activate the canonical TGF-β/ Smad2/3 signaling induced EMT in HCC and, promoting HCC progression.

The above results revealed that Piezo1 promote HCC progression via activating the TGF-β signaling, to further explore the molecular mechanism by which Piezo1 exerted the functions in HCC. In consideration of that Piezo1 is a stretch-activated ion channel and the activation of Piezo1 channels increased the intracellular Ca²⁺ concentration, we treated HCCLM3 and Hep3B cells with Yoda1 (the activator of Piezo1, 20 mM) to increase the intracellular Ca²⁺ concentration and GsMTx4 (the inhibitor of Piezo1, 2.5 μM) to decrease Ca²⁺ concentration [19]19, then examined p-Smad2/3 and marker of EMT, found that Ca²⁺ influx was not the dominant factor in Piezo1 activated TGF-β signaling (Additional file 2: Fig. S7A, B). Then we reviewed literatures and searched BioGrid 4.4 database (Additional file 2: Fig. S7C), found that Piezo1 might interact with Rab5c, an isoform of Rab5, which belongs to the small GTPases of Rab family. Double IF revealed the colocalization of Piezo1 and Rab5c in HCCLM3^NC and Hep3B^NC cells. More than that, we also observed that the Rab5c has a tendency to enriched in membrane when Piezo1 highly expressed (Fig. 6A). Then, co-IP results revealed Piezo1 and Rab5c could interact with each other in HCC cells (Fig. 6B, C). Next, we investigated whether the interaction between Piezo1 and Rab5c could affect rab5c expression in HCC cells. Interestingly, we found that silenced Piezo1 in HCCLM3 and Hep3B cells did not changed the protein expression level of Rab5c, but the phosphorylation level of Smad2/3 and expression level of EMT markers have changed correspondingly (Fig. 6D). Combined with the location of Rab5c in the IF result, we further analyzed membrane enrichment of Rab5c by western blotting, and confirmed that Rab5c was significantly enriched in membrane fractions in cells highly expressed Piezo1 (Fig. 6E). Summarily, these results indicate that Piezo1 activates TGF-β/Smad2/3 signaling by recruiting Rab5c without change the total expression level of Rab5c.

It has proved that Piezo1 activates TGF-β signaling by recruiting Rab5c in the level of protein, but the function of Rab5c in this progress still unknown. Next, we need to test whether Rab5c is indispensable for Piezo1-mediated activation of TGF-β signaling and promotion of HCC progression and EMT. we transfected Rab5c in HCCLM3^shPiezo1 and Hep3B^shPiezo1 cells. MTT, wound healing and transwell assay results revealed overexpression of Rab5c reversed the down-regulation of the invasion, migration and proliferation potential caused by Piezo1 silencing in HCCLM3^shPiezo1 and Hep3B^shPiezo1 cells (Fig. 7A–C). The morphology and expression of EMT marker also reversed along with the overexpression of Rab5c in Piezo1 interfered HCC cells (Fig. 7D, E). In summary, the results indicate that Piezo1 promote HCC progression by activating TGF-β signaling via recruiting Rab5c (Fig. 7F).