Adjacent-cancer hepatocytes promote malignant progression of ICC
This study analyzed a total of 280 cholangiocarcinoma patient records with complete follow-up and clinical data from Sun Yat-sen Memorial Hospital of Sun Yat-sen University between 2009 and 2018 (Fig. 1). The data demonstrated that patients with ICC had a worse prognosis compared to those with ECC (Fig. 1a). In addition, compared to the ECC patients, patients with ICC had a higher chance of distant metastasis (ICC vs ECC: 39.2% vs 24.2%) and liver metastases (ICC vs ECC: 28.4% vs 12.9%) (Fig. 1b). Analysis of the SEER database showed the same trend in survival or prognosis of ICC and ECC patients. The SEER database analysis demonstrated that the median survival of the ICC patients was significantly shorter than that of the patients with ECC (Fig. 1a). Besides, ICC patients were significantly more susceptible to distant metastasis (ICC vs ECC: 35.5% vs 26.1%) (Fig. 1c). These findings showed that ICC is more prone to metastasis and has a worse prognosis than ECC. Thus, we hypothesized that these differences in disease development might be due to differential TME. Analysis of clinical data showed that hepatocytes, an important component of the liver microenvironment, played a crucial role in ICC metastasis. As shown in Fig. 1d, our data demonstrated that the ICC cells grew infiltratively along the intrahepatic bile duct, and then into adjacent liver parenchyma, which was in contact with hepatocytes.
To dissect the functional relationship between hepatocytes and ICC development and progression, we co-cultured the ICC cell lines RBE and HuCCT1 with hepatocytes (THLE3). The growth of the RBE and HuCCT1 cells after co-culture with THLE3 was measured using cell counting kit-8 (CCK-8) assays. The data showed that the co-culture with THLE3 significantly promoted the proliferation of RBE and HuCCT1 cells (Fig. 1e). Moreover, the transwell assay revealed that co-culture with THLE3 robustly enhanced the migratory and invasive ability of the ICC cells (Fig. 1f). In addition, we conducted cell cycle progression and cell apoptosis analysis by flow cytometry. As indicated in Fig. 1g, the co-culture with THLE3 resulted in a significant increase in S-phase cells and a reduction in cells at G0/1 phases. The apoptosis assays showed that the percentage of apoptotic cells was significantly higher in the co-culture group (Fig. 1h). Overall, these data indicated that hepatocytes promote the metastasis of ICC.
Hepatocyte-secreted CCL3 promotes migration and invasion of the ICC cells
To assess the key chemotactic molecule in hepatocytes that affect the metastasis of ICC cells, we employed the cytokine antibody assay to examine the level of cytokines secreted by hepatocytes after co-culture with the ICC cells. As shown in Fig. 2a and Additional file 3: Fig. S1, our results showed a high secretion of cytokine CCL3 and CCL23 in hepatocytes after co-culture with RBE, an ICC cell. We then used qRT-PCR and ELISA to verify the prominent hypersecretion of cytokine CCL3 in supernatants collected from the THLE3 cell line after co-culture with the ICC cells (Fig. 2b and d), while a slightly elevated level of CCL23 secretion was observed in Fig. 2c. Next, to evaluate the role of different cytokines in the regulation of migration and invasion ability of the ICC cells, we conducted transwell migration assays and matrigel invasion assays of the RBE cells and HuCCT1 cells stimulated with CCL3 and CCL23. Our findings showed that CCL3 significantly enhanced the migration and invasion ability of HuCCT1 and RBE cells compared with the control group (Fig. 2e–f). On the contrary, there was no significant difference between the control and the treatment group with CCL23, which indicated that CCL3 was a critical factor in ICC infiltration and metastasis.
The expression of CCL3 in the hepatocytes was significantly upregulated after co-culture with the ICC cells, while there was no significant correlation in the CCL3 expression in ICC cells after co-culture with THLE3 (Fig. 2g, h). Moreover, immunohistochemical staining was performed to profile the localization of CCL3 protein in ICC tumor and adjacent liver tissues. The data showed that the CCL3 was mainly expressed in the cytoplasm of hepatocytes and intercellular matrix adjacent to the tumor tissue, with cancer cells and distant hepatocytes staining negative (Fig. 2i). Functional enrichment analysis showed that genes that interacted with CCL3 were enriched for functions related to RNA metabolism (Fig. 2j). These results suggested that CCL3 is a key chemokine in the interplay between hepatocytes and ICC cells, and could influence RNA metabolic processes such as RNA modification, RNA synthesis, RNA cleavage, and RNA degradation in ICC.
CCL3 promotes ICC tumor migration and invasion by regulating m6A modification via VIRMA
Our dot blot assays demonstrated that m6A was among the genes that were shown to interact with CCL3. Global m6A abundance was detected in the mRNA of RBE cells co-cultured with THLE3 and stimulated with CCL3. As shown in Fig. 3a, b, the results indicated that the m6A in total RNA was upregulated in ICC cells after co-culture with THLE3 and stimulation with CCL3. In addition, the mRNA and protein expression profile of VIRMA, one of the major components of the m6A methyltransferase complex, was upregulated in all m6A-related writers and readers (Fig. 3c).
Further analysis revealed that VIRMA expression was significantly upregulated in ICC cells treated with 100 ng/ml CCL3 for 6 h (Fig. 3d, e). It was shown that CCL3 could not significantly increase VIRMA expression in the ICC cells treated with CCR5 receptor antagonist (Fig. 3f). These results suggested that the binding of CCL3 to the CCR5 receptor on the membrane of cancer cells promotes ICC migration and invasion by regulating VIRMA, an m6A methyltransferase. We then conducted transwell migration and invasion assays of the RBE and HuCCT1 cells with different treatments to assess the role of CCL3/VIRMA pathway on the migratory and invasive ability of the cells. As demonstrated in Fig. 3g, human recombinant cytokine CCL3 effectively promoted migration and invasion of ICC cells, a phenomenon that was inhibited by knockdown of VIRMA, thus confirming the pro-metastatic effects of CCL3/VIRMA pathway in ICC.
VIRMA promotes ICC development and progression in vitro
To examine the roles of VIRMA in the CCA cells, we profiled the VIRMA expression in CCA cell lines using qRT-PCR and Western blot assays. The results illustrated significant upregulation of the VIRMA mRNA and protein levels in ICC cell lines (RBE and HuCCT1) compared with human intrahepatic biliary epithelial cells (HIBEpiC) (Fig. 4a). Thereafter, we performed VIRMA overexpression and knockdown studies in RBE and HuCCT1 cells. The ICC cells were transfected with control siRNA (siNC) or siRNA targeting VIRMA (siVIRMA-1 and siVIRMA-2). On the other hand, the overexpression cell lines were constructed as oeVIRMA while the matched control cell lines were named as vector. The VIRMA expression in the cells was confirmed with qRT-PCR and Western blot (Additional file 3: Fig. S2).
The growth of the RBE and HuCCT1 cells after knockdown or overexpression of the VIRMA was measured using cell counting kit-8 (CCK-8) and colony formation assays. As shown in Fig. 4b–d, there was a declining trend in cell proliferation in RBE and HuCCT1 cells with VIRMA knockdown while overexpression resulted in significantly increased cell proliferation. Moreover, the transwell migration assay demonstrated that downregulation of VIRMA impeded the migratory ability of the ICC cells, while upregulation of VIRMA significantly elevated the migration of the ICC cells. Similarly, the transwell invasion assay showed that the invasive ability of HuCCT1 and RBE cells was significantly suppressed in response to the downregulation of VIRMA, while it was enhanced by VIRMA upregulation (Fig. 4e, f). We then conducted wound healing assays to display the slow migration of cells with VIRMA knockdown, which was in contrast to the increase observed in VIRMA overexpression cells (Fig. 4g, h). In the meantime, we detected the expression of the Epithelial-Mesenchymal Transition (EMT) marker, angiogenic marker vascular endothelial growth factor (VEGF), and proliferation marker Cyclin D1 in ICC cells with VIRMA knockdown or overexpression. As displayed in Fig. 4i, j, the mRNA and protein levels of N-cadherin (mesenchymal marker), Vimentin (mesenchymal marker), VEGF, and Cyclin D1 were significantly positively correlated with VIRMA expression, while there was a prominent inverse correlation between the E-cadherin (epithelial marker) expression and VIRMA expression. These results implied that VIRMA promotes ICC proliferation, invasion, and metastasis in vitro.
Furthermore, we performed cell cycle progression and cell apoptosis analyses. As demonstrated in Fig. 5a, b, VIRMA-deficient cells had a partial cell cycle arrest at the G0/1 transition but a significant increase in the G0/1 phase and a reduction in cells at S and G2-M phases. The apoptosis examination detected by flow cytometry (Fig. 5c, d) and TUNEL assay (Fig. 5e, f) indicated that the percentage of apoptotic cells was significantly elevated in cells transfected with siVIRMA. There appeared an obvious decrease in Bcl-2 expression, the antiapoptotic protein, in ICC cells with down-expression of VIRMA (Fig. 5g, h). From this, we can deduce that downregulation of VIRMA induces ICC cell cycle arrest and promotes cell apoptosis.
VIRMA promotes ICC proliferation and metastasis in vivo and is correlated with poor prognosis of ICC patients
We demonstrated that the downregulation of VIRMA inhibits the proliferation, invasion, and metastasis of ICC cells in vitro. Here, we analyzed the effect of VIRMA in the development of ICC in mice. Stable cells with modified VIRMA expression were subcutaneously injected into the BABL/c nude mice. First, we investigated the effect of VIRMA on the proliferation of ICC. As shown in Fig. 6a–d, VIRMA knockdown (HuCCT1-shVIRMA) led to significantly low tumor growth rate and tumor weight compared with the other groups (HuCCT1: 0.75 ± 0.05 g, HuCCT1-shNC: 0.77 ± 0.02 g, HuCCT1-shVIRMA: 0.28 ± 0.01 g). The H&E staining demonstrated that VIRMA knockdown significantly reduced the tumor density and differentiation degree of the ICC. On the other hand, immunohistochemical staining showed that the expression of tumor progression markers such as N-cadherin, Vimentin, VEGF, Ki67, and Cyclin D1 was decreased along with the downregulation of VIRMA, except E-cadherin exhibited a contrary tendency (Fig. 6j).
In parallel, we analyzed the expression of VIRMA in ICC patients using the TCGA dataset and showed that the VIRMA expression was remarkably increased in CCA tissues (Fig. 6g). Moreover, immunohistochemical staining in 110 human ICC patients and normal adjacent tissues revealed that VIRMA expression was significantly higher in ICC cancer compared to the adjacent counterparts (Fig. 6e, f). In addition, we conducted a Kaplan–Meier survival analysis to investigate the correlation between VIRMA expression and ICC patient prognosis (Fig. 6h, i). The results demonstrated that ICC patients with high VIRMA expression had poorer overall survival (OS) and disease-free survival (DFS) (*P < 0.05). Taken together, these findings suggested that the higher expression of VIRMA was associated with adverse prognosis in ICC patients.
SIRT1 was selected as a downstream target of VIRMA-mediated m6A modification
To identify the target genes of VIRMA, MeRIP-seq and RNA-seq were performed in stable VIRMA downregulation ICC cells. According to the results of the MeRIP-seq, it was found that more than 80% of m6A binding sites were located in the protein-coding region (CDS region) and were enriched in the 5′ untranslated region (5′ UTR) as well as 3′ untranslated region (3′ UTR). Further analysis showed that the proportion of specific m6A binding sites in the 3′UTR was reduced in the siVIRMA group as compared with that in the siNC group and this was consistent with findings that VIRMA was mainly involved in regulating m6A modification in the 3′UTR region of mRNA (Fig. 7a).
The results of the Motif for the m6A binding site indicated that the most significant m6A Motif was GATG (Additional file 3: Fig. S3c), which provided a basis for subsequent screening of the downstream targets. Network enrichment analysis was then conducted according to MeRIP-seq results and it was identified that the cell cycle pathway, as well as the Foxo signaling pathway, were significantly associated with the oncogenic activities of VIRMA (Fig. 7b). Kyoto Encyclopedia of Genes and Genomes (KEGG) and gene ontology (GO) enrichment analysis were carried out using gene set enrichment analysis (GSEA) (Additional file 3: Fig. S3). Results of the RNA-seq showed that 1188 transcripts were markedly downregulated on VIRMA knockdown. Further, MeRIP-seq presented that m6A peaks of 1011 transcripts exhibited decreased abundance. In addition, it was found that 2072 genes were significantly correlated with VIRMA in TCGA dataset (Fig. 7c).
Intriguingly, results of the transcriptome-sequencing, MeRIP-seq in the genes selected in the TCGA dataset revealed that 16 transcripts were overlapped (Fig. 7d). Further, it was noted that SIRT1 showed the most pronounced difference and was reported to be close to tumor progression among the candidate genes. The m6A usually happens in RRACH (R = G or A, H = A, C or U) consensus sequence, results of MeRIP-seq in the present study showed that an m6A peak was detected around the stop codon of SIRT1 mRNA in non-target control siRNA cells and were all decreased upon VIRMA knockdown (Fig. 7g). The SIRT1 mRNA was also significantly downregulated in VIRMA-knockdown ICC cells both at mRNA and protein levels (Fig. 7e, f). Furthermore, the RNA pull-down and RIP assays were also conducted to verify the positive association between SIRT1 and VIRMA (Fig. 7h, i). The correlation analysis between SIRT1 and VIRMA based on the TCGA ICC dataset in Fig. 7j reached a similar conclusion. Therefore, a preliminary conclusion was drawn that SIRT1 was selected for further studies as a candidate target of VIRMA-mediated m6A modification.
Downregulation of SIRT1 repressed tumor progression of ICC cells in vitro
As a kind of histone deacetylase, SIRT1 may regulate the chromatin acetylation process. Therefore, we observed that the H3K4 acetylation (H3K4ac) protein level was decreased after SIRT1 knockdown in both RBE and HuCCT1 cells (Fig. 8a). Cell viability was determined by CCK8 assays, from which we observed that downregulation of SIRT1 notably inhibited tumor cell growth in ICC (Fig. 8b). Moreover, we conducted cell cycle progression analysis and cell apoptosis assay of RBE and HuCCT1 cells after SIRT1 knockdown. As illustrated in Fig. 8c, d, depletion of SIRT1 induced tumor cell arrest in the G0/1 phase. And it also markedly increased cell apoptosis in ICC (Fig. 8e, f). Transwell assays (Fig. 8g, h) and wound healing assays (Fig. 8i, j) were performed through SIRT1 knockdown to evaluate the role of SIRT1 in the regulation of migration and invasion ability in the ICC cells. Results of the current study showed that the downregulation of SIRT1 significantly decreased the migration and invasion ability of HuCCT1 and RBE cells compared with the control group.
CCL3/VIRMA/SIRT1 pathway accelerates the ICC malignant process
It was evident that the CCL3 can abolish the effect of VIRMA knockdown in SIRT1 expression (Fig. 9a, b and Additional file 3: Fig. S4) during the rescue experiments in vitro. Therefore, SIRT1 is a downstream target of CCL3/VIRMA in ICC which served as an oncogene to promote migration and invasion of ICC cells.
To explore the effect of the CCL3/VIRMA/SIRT1 pathway on the ICC malignant process, we constructed subcutaneous and orthotopic tumor models in nude mice.
We first injected different cells (HuCCT1 + PBS, HuCCT1 + CCL3, HuCCT1-shNC + CCL3, and HuCCT1-shVIRMA + CCL3) subcutaneously into nude mice. As shown in Fig. 9c–f, tumor growth of HuCCT1 ICC cells was significantly faster than that in the PBS group after CCL3 treatment, while HuCCT1 cells with VIRMA knockdown grew slower even after receiving CCL3 stimulation. Our data showed that the tumor weight in the CCL3 treatment group was significantly higher than that in the PBS group, while the VIRMA knockdown group was not affected by CCL3 treatment, and the tumor weight was significantly lower than that in the control group (HuCCT1 + PBS: 0.81 ± 0.02 g, HuCCT1 + CCL3: 1.45 ± 0.08 g, HuCCT1-shNC + CCL3: 1.53 ± 0.04 g, and HuCCT1-shVIRMA + CCL3: 0.27 ± 0.02 g). The immunohistochemistry analyses confirmed that the expression of VIRMA, SIRT1, and Ki67 was upregulated in the tumor tissues with CCL3 stimulation, and it was also reversed in the tumor tissues with VIRMA down-expression (Fig. 9k). Together, these data demonstrated that CCL3 promotes the growth of ICC cells by regulating VIRMA/SIRT1 in ICC subcutaneous tumor mice model.
We further showed that CCL3 promoted the metastasis of intrahepatic bile duct cancer cells by regulating VIRMA/SIRT1 pathway in the orthotopic ICC tumor model. The liver metastasis was significantly enhanced with CCL3 stimulation, whereas the progression was significantly inhibited by VIRMA knockdown (Fig. 9g–j). Furthermore, IHC staining of the nude mice tissues (Fig. 9l) also verified that SIRT1 was positively correlated with VIRMA expression. With CCL3 stimulation, the Ki67 was up-regulated in the tumor tissues and it was also reversed in the tumor tissues with VIRMA down-expression. Consistent with the above observations, it was evident that the CCL3 had positive effects on tumor growth and liver metastasis of ICC and the progression could be abolished through VIRMA knockdown.
In conclusion, the results of the current study confirmed that CCL3/VIRMA/SIRT1 pathway promotes ICC proliferation and tumor metastasis in vivo.