Cancer-associated fibroblast-derived stanniocalcin 1 promotes cholangiocarcinoma progression and lymph node metastasis

Abstract

Cholangiocarcinoma (CCA) is a fibroblast-rich tumor. Lymph node metastasis (LNM) is the main prognostic risk factor for CCA. We determined that stanniocalcin 1 (STC1) is highly expressed in cancer-associated fibroblasts (CAFs) with LNM in CCA. However, the roles of CAFs and STC1 in CCA progression and LNM remain unelucidated. Here, we demonstrated that primary CCA tissues with LNM (LN⁺CCA) were enriched with more CAFs and lymphatic vessels than those with LN⁻CCA. LN⁺ CAFs strongly promoted CCA cell migration, invasion, TrEM, and LNM both in vitro and in vivo. LN⁺CAFs exhibited significantly higher STC1 expression than LN⁻CAFs. The in vitro and in vivo results demonstrated that STC1 is mainly responsible for the role of LN⁺CAFs in promoting CCA malignant behavior. Mechanistic studies demonstrated that STC1 activates integrin αVβ3 and its downstream FAK–YAP pathway. Additionally, we found that targeting STC1 and integrin αVβ3 rescued the pro-tumor effects of CAF on CCA in vitro and in vivo. Therefore, CAF-derived STC1 predicts LNM and is a potential therapeutic target in CCA.

Graphical abstract

1. Introduction

Cholangiocarcinoma (CCA) is the second most common primary liver tumor. Other than radical resection and liver transplantation, the effective treatment options are limited. Given the insidious onset and rapid progress of CCA, surgery is often unavailable at diagnosis, leading to poor prognosis. Lymph node metastasis (LNM) is one of the most prominent CCA prognostic risk factors. As an understanding of the mechanism is lacking, there is no targeted treatment for CCA LNM. The main LNM pathways are lymphangiogenesis and cancer cell trans-lymphatic endothelial migration (TrEM) induction, in which cytokines in the tumor microenvironment (TME) are crucial.

In various tumors, the cross-talk between cancer-associated fibroblasts (CAFs) and cancer cells supports tumor growth and progression through the secretion of angiogenic, immune-modulatory, growth-stimulatory, matrix remodeling and pro-invasive factors [1,2]. The CAF-derived secretory proteins include cytokines, chemokines, growth factors, extracellular matrix (ECM) proteins, protease inhibitors, and proteases [3]. Single-cell sequencing recently demonstrated that CAFs are composed of different functional cell subsets with distinct secretion profiles. CAFs are the most abundant stromal cells in the fibrotic TME of CCA. CAF abundance is positively correlated with CCA growth and poor prognosis [4], but differences in the CAF secretion profiles in CCA with or without LNM and the role of the related factors remain unclear. Cadamuro et al. demonstrated the promoting effect of CAFs on LNM, where CCA cell-derived PDGF-D activated the JNK signaling pathway by binding PDGFRβ on CAFs to promote VEGF-C secretion, subsequently recruiting lymphoendothelial cells (LECs) and inducing lymphangiogenesis and promoting cancer cell TrEM, leading to CCA LNM. CCA depletion with navitoclax reduced tumor mass lymphatic vascularization and lymphatic dissemination in animal models of CCA [5]. Although VEGF-C is the most important lymphangiogenesis-promoting factor [6,7], not all lymphatic vessel-rich tumors demonstrate high VEGF-C expression [7,8], indicating the existence of other mechanisms for VEGF-C-independent lymphangiogenesis and LNM in the TME.

Stanniocalcin 1 (STC1) is a mass spectrometry-screened secreted protein with the highest upregulation in LN⁺ CAFs (CAFs from node-positive primary CCA) as compared with that in LN⁻ CAFs. STC1 is a hormone-like protein that exists in almost all tissues and acts as a paracrine/autocrine factor that regulates various cellular biological functions [9]. Tumor cell-derived PDGF-CC coordinates ERα-negative phenotype specification by activating CAFs that induce STC1 secretion in breast cancer [10]. CAF-derived STC1 pretreatment reduced breast cancer cell sensitivity to tamoxifen-induced growth arrest. STC1 expression in CAFs drove colorectal cancer metastasis [11]. However, the potential role of STC1 in CCA progression and LNM remains unclear.

Here, we report the critical role of CAFs in CCA tumor progression and LNM. Functional verification demonstrated the importance of STC1 in the promotion of CAF-mediated CCA. Mechanistic studies demonstrated that STC1 induced lymphangiogenesis and promoted CCA cell proliferation, migration, and TrEM via the integrin αVβ3 pathway. Finally, the feasibility of the STC1/integrin αVβ3 or downstream FAK–YAP pathway as a new CCA therapeutic target in mouse models is discussed.

2. Material and methods

2.1. Patients and clinical samples

Ninety-eight tumor tissue samples were collected from patients diagnosed with CCA who underwent radical surgery at the Second Affiliated Hospital of Sun Yat-sen University between January 2012 and July 2018. Table S1 presents the patients’ clinical information. Blood samples were obtained from another 20 patients with CCA and 20 participants without CCA. Patients in the CCA group were eligible for inclusion if they had pathologically confirmed CCA. All the experiments were approved by the Sun Yat-sen University Human Research Ethics Review Committee (SYSKY-2023-156-01). All participants signed an informed consent form prior to sample collection.

2.2. Immunohistochemistry

Antigens were extracted from 4-mm thick paraffin-embedded sections in 0.01 M trimethylol aminomethane salt liquid acid (pH 9.2) for 15–20 min in a pressure cooker to remove the aldehyde junctions formed during the initial tissue fixation. The samples were incubated at 4 °C overnight with specific antibodies against α-SMA (1:200), podoplanin (PDPN, 1:200), LYVE1 (1:5000), STC1 (1:200), p-YAP (1:200), p-FAK (1:200), CD51/CD61 (1:200), or Ki-67 (1:1000) and immunoassayed the following day using diaminobenzidine (Dako) according to the manufacturer's instructions.

2.3. Immunofluorescence

The antigens were extracted as described in the previous section. Tissue sections were blocked with TBST containing 5% bovine serum albumin (BSA) for 1 h at room temperature. The samples were then fixed in 4% paraformaldehyde for 15–20 min at room temperature, washed three times with PBS, and permeabilized with 0.2% Triton X-100 in PBS for 15–20 min. Subsequently, the cells were blocked with TBST containing 5% BSA for 1 h at room temperature. After blocking, the samples were incubated overnight at 4 °C with mouse anti-human α-SMA (1:100), rabbit anti-human vimentin (1:200), FAP (1:200), cytokeratin (1:200), fibronectin (1:200), STC1 (1:100), PDPN (1:100), and LYVE1 (1:2000) antibodies, followed by incubation for 1 h at room temperature with an Alexa Fluor conjugated secondary antibody (Invitrogen). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole. Images were captured using a laser scanning confocal microscope (LSM780, Zeiss).

2.4. Popliteal lymphatic metastasis model

BALB/c nude mice (4 weeks old, 18–20 g) were purchased from the Guangdong Province Experimental Animal Center. All experimental procedures were approved by the Sun Yat-sen University Institutional Animal Care and Use Committee (SYSU-IACUC-2023-001286). In vivo imaging was performed 6 weeks after injection (described below) using a bioluminescence imaging system (IVIS Spectrum Imaging System, PerkinElmer). Tumor growth and lymphatic metastasis were examined when the tumors in the control group were of the same size as those in the experimental group. The primary tumor and popliteal lymph nodes (LNs) were removed, paraffin-embedded, and 4-mm serial sections were obtained for hematoxylin–eosin (HE) staining and immunohistochemical analysis. Images were captured using an Eclipse 80I system and NIS-Elements software (Nikon).

2.5. Subcutaneous tumor models

Nude mice were injected subcutaneously with either QBC939 or HuCCT1 cells (2 × 10⁶) for tumor growth. In the co-injection experiment, the nude mice were subcutaneously injected with 2 × 10⁶ QBC939 or HuCCT1 cells and either 2 × 10⁵ LN⁺ CAFs, LN⁻ CAFs, or STC1 knockout CAFs (CAFs^STC1-KO) (10:1 ratio). Tumor growth was measured every 7 days using calipers. Tumor volume was calculated as (length × width²)/2. When the tumor volume was 100 mm³, the mice were treated with DMSO, IgG, anti-STC1 (1 mg/kg, intraperitoneal injection), cyclo(RGD-d-phenylalanine-lysine (cyclo[-RGDfK], 4 mg/kg, intraperitoneal injection), YAP inhibitor (YAPi, verteporfin, HY-B0146, 10 mg/kg, intraperitoneal injection), or FAK inhibitor (FAKi, defactinib, HY-12289, 25 mg/kg, intraperitoneal injection) every 3 days for 3 weeks. At the end of the treatment, the animals were sacrificed and the tumors were excised, fixed in 4% formaldehyde in PBS overnight, and analyzed by HE and immunostaining.

2.6. RNA sequencing data analysis

RNA was sequenced as previously described [12]. Briefly, 1 µg total RNA was isolated from 200 mg/ml STC1-treated CAFs and controls that had been incubated with VAHTS mRNA capture beads (Vazyme), enriched for polyA+ RNA, and then RNA libraries were constructed. The RNA sequencing data were calculated and analyzed using the Ensemble Human Genome Assembly (Genome Reference Consortium GRCH38). The transcript level of each gene was estimated as the number of exon model reads per kilobase per million mapped reads (RPKM). Gene function was annotated using gene set enrichment analysis (GSEA). Hall markers were obtained from the Molecular Signatures database (MSigDB) and GSEA was performed using GSEA v4.0.3. The gene sets revealed that a total of 5,000 permutations to determine the p-values. A false discovery rate of <0.25 and p < 0.01 were considered significant. The genetic sequencing indicators were the signal and noise.

2.7. Proteomics analysis

The LN⁺ and LN⁻ CAFs were cultured using DMEM in 100-mm culture plates at 37 °C in a 5% CO₂ humid atmosphere. The supernatants were harvested after 24 h and the protein content was determined using a Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific). The samples were precipitated with acetone, resuspended in urea buffer (50 mM), and reduced with dithiothreitol (2 mM, 37 °C for 1.5 h). The samples were then alkylated with iodoacetamide (10 mM, 25 °C, 40 min), diluted with 60 mM urea, and digested overnight at 37 °C with trypsin (sequencing-grade modified trypsin, V5111, Promega). The samples were then diluted twice again and trypsin-digested overnight at 37 °C. Tryptic peptide was desalted using an HLB column and evaporated to dryness. In total, 300 µg dried samples from the LN⁺ and LN⁻ CAFs were dissolved in 10 µl 0.1% formic acid in water. Subsequent mass spectrometry was performed as previously described [13].

2.8. Statistical analysis

A t-test and 1-way ANOVA were used for normally distributed data. The Mann-Whitney U test was used for non-normally distributed data. Quantitative data are presented as the mean ± SD of at least three independent experiments. The threshold of significance was set at p < 0.05. The patient samples were divided into two groups based on their mean expression levels before survival analysis. The relationship between protein expression and prognosis was evaluated using Kaplan-Meier survival analysis. The correlation of two proteins was elucidated using Pearson correlation analysis. P-values were calculated using the log-rank test.

3. Results

3.1. LN⁺ CCA specimens demonstrated more CAF aggregation and lymphatic vascularization

To assess CAF distribution and the amount of cancer-associated lymphangiogenesis, we quantified the α-SMA, PDPN⁺, and LYVE1⁺ lymphatic microvessel density (LMVD) in the human CCA tissues. We compared the results of the CCA samples with LNM (LN⁺ CCA) with those of the LN⁻ CCA and normal liver tissues. Compared to the LN⁻ CCA and normal tissues, the LN⁺ CCA tissues exhibited significantly increased CAF aggregation and LMVD (Fig. 1a–d). The results confirmed that lymphatic vasodilatation was a defining feature of the typically desmoplastic LN⁺ CCA in contrast to the relative scarcity of lymphatic vessels in the LN⁻ CCA and normal tissues. Importantly, these results indicated that lymphangiectasia and desmoplasia are associated with LNM status in human CCAs. The spatial relationships between the CCA lymphatic vasculature and CAFs in CCAs investigated using dual immunofluorescence for α-SMA (a CAF marker), PDPN, and LYVE1 (an LEC marker) revealed that CAFs and LECs were spatially close to each other in the tumor-reactive stroma of human CCAs (Fig. 1e).

Fig. 1.

LN⁺ CCA tissue demonstrated more CAF aggregation and higher LMVD than LN⁻ CCA and normal liver tissues. (a) CAF density and LMVD were more extensively represented in LN⁺ CCA tissue than in LN⁻ CCA and normal liver tissue, as shown by representative immunohistochemistry staining images for α-SMA, PDPN, and LYVE1. (b–d) The percentages of CCA tissue specimens from 98 patients with high and low α-SMA (b), PDPN (c), LYVE1 (d) levels. (e) LECs were localized close to CAFs in the CCA stroma as shown by double immunofluorescence of PDPN, LYVE1 (green), and α-SMA (red). *p < 0.05; **p < 0.01; ***p < 0.001; 2-tailed t-test. Original magnification: ×200.

3.2. LN⁺ CAFs strongly promoted CCA cell aggressiveness both in vitro and in vivo

Immunofluorescence analysis revealed that the CAFs exhibited high vimentin, fibronectin, α-SMA, and FAP expression and negative cytokeratin expression (Fig. S1). LNM is a complex process; in addition to intratumoral lymphangiogenesis in vivo, tumor metastasis requires enhanced tumor cell invasion and migration [14,15]. To reveal the exact biological roles of LN⁺ and LN⁻ CAFs in CCA, we examined the effect of the CAF supernatant on HuCCT1 and QBC939 cell migration, invasion, TrEM, colony-forming ability, and proliferation. The Transwell assays revealed that the LN⁺ CAF supernatant dramatically increased QBC939 and HuCCT1 cell invasion and migration compared to the LN⁻ CAF and control supernatants (Fig. 2a-d).

Fig. 2.

LN⁺ CAFs strongly promoted CCA cell aggressiveness in vitro and in vivo. (a, b) Representative images of the migration, invasion, and TrEM of QBC939 (a) and HuCCT1 cells (b) treated with LN⁺ or LN⁻ CAF supernatant (magnification: ×100). (c, d) Histogram analysis of the migration, invasion, and TrEM of QBC939 (c) and HuCCT1 cells (d). (e–g) Representative images (e) and histogram analysis of colony formation by QBC939 (f) and HuCCT1 cells (g). (h, i) MTT assays for measuring the proliferation of QBC939 (h) and HuCCT1 cells (i) incubated with LN⁺ or LN⁻ CAF supernatant. (j, m) Representative images of QBC939 (j) and HuCCT1 (m) cells combined with LN⁺ or LN⁻ CAFs on nude mouse xenografts. (k, l) QBC939 tumor volumes (k) and weights (l) (n = 5). (n, o) HuCCT1 tumor volumes (n) and weights (o) (n = 5). Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. (p, q) Representative images and histogram analysis of immunohistochemical staining for Ki67 expression in QBC939 (p) and HuCCT1 (q) cells (n = 5). Scale bars: 100 µm. Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. Error bars represent the SD of three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

We determined whether CCA cell migration across the endothelial barrier was promoted by seeding enhanced green fluorescent protein (EGFP)-expressing QBC939 and HuCCT1 cells on LEC monolayers and adding LN⁺ and LN⁻ CAF-conditioned media. A markedly increased number of CCA cells migrated across the LEC monolayer in the LN⁺ CAF group as compared with the LN⁻ CAFs and control groups (Fig. 2a, b). Colony formation and CCK-8 assays revealed that the LN⁺ CAF supernatant significantly increased the proliferation (Fig. 2e, f) and colony-forming ability (Fig. 2g, i) of both CCA cell lines as compared to the LN⁻ CAF and control groups. In accordance with the in vitro findings, the LN⁺ CAFs greatly promoted tumor growth as compared with the LN⁻ CAFs and control groups. The LN⁻ CAFs and control tumors were light and smaller (Fig. 2j-n) and had lower Ki67 expression levels than the LN⁺ CAFs (Fig. 2o, p). These results indicated that LN⁺ CAFs strongly promoted CCA aggressiveness in vitro and in vivo.

3.3. LN⁺ CAFs greatly promoted CCA lymphangiogenesis and LNM both in vitro and in vivo

Tumor-associated lymphangiogenesis is a key step in LNM and provides a route for tumor cells to enter the lymphatic system [16,17]. We examined whether CAFs promoted lymphangiogenesis in CCA. The LN⁺ CAF supernatant significantly promoted LEC migration and tube formation compared with the LN⁻ CAF and control supernatants (Fig. 3a-c). We used a popliteal LNM (PLNM) nude mouse model that simulated CCA directional drainage and LNM to confirm the critical role of CAFs in guiding tumor lymphangiogenesis and LNM in vivo (Fig. S2a). Lymph node metastasis was evaluated by in vivo imaging, gross anatomy, HE staining, and immunohistochemistry (Fig. S2b, S2c). The plantar LN drainage direction was from the popliteal LNs to the external iliac LNs, which simulated the intraperitoneal LNM process in the CCA. In vivo imaging revealed that LN⁺ CAFs markedly increased the ability of CCA cells to transfer to LNs (as determined by luminescence and the LNM proportion) as compared with the LN⁻ CAFs and controls (Fig. 3d-g). Statistical analysis indicated that the LN⁺ CAF group had a significantly higher LNM ratio than the LN⁻ CAF and control groups, confirming that LN⁺ CAFs significantly promoted the CCA cell metastatic capability (Fig. 3h, i). Immunohistochemistry revealed that the LN⁺ CAF group had significantly higher LMVD in the intratumoral regions than the LN⁻ CAF and control groups (Fig. 3j, k), suggesting that CAFs promoted lymphangiogenesis in vivo. These findings suggested that LN⁺ CAFs strongly promoted intratumoral lymphangiogenesis and LNM in CCA both in vitro and in vivo.

Fig. 3.

LN⁺ CAFs promoted marked lymphangiogenesis and LNM in CCA. (a–c) Representative images (a) and quantification of Transwell migration (b) and tube formation (c) by HLECs treated with supernatant from LN⁺ or LN⁻ CAFs. (d) Representative gross anatomy images of the PLNM and non-LNM nude mouse model of QBC939. (e, h) Representative images of bioluminescence analysis (PLNs) of footpad tumor and PLNM in the QBC939 (e) and HuCCT1 (h) (n = 6 per group). (f, i) Enucleated LNs and immunohistochemical staining with anti-luciferase antibody in QBC939 (f) and HuCCT1 (i) (n = 6 per group). Scale bars: red, 50 µm. (g, j) Statistical significance in the QBC939 (g) and HuCCT1 (j) groups was assessed using 1-way ANOVA followed by Dunnett's tests and 2-tailed t-tests. Representative images of intratumoral microlymphatic vessels stained with anti-LYVE1 (left, black arrows) and histogram quantification of LMVD (right) in the QBC939 (k) and HuCCT1 (Ll) groups. Scale bars: black, 100 µm Error bars represent the SD of the mean. *p < 0.05; **p < 0.01; ***p < 0.001.

3.4. CAF-derived STC1 was associated with LNM and poor CCA prognosis

The LN⁺ CAFs strongly promoted CCA cell aggressiveness and CCA lymphangiogenesis and LNM in vitro and in vivo. Accordingly, we performed proteomics analysis of the CCA CAFs, which indicated that LN⁺ CAFs had markedly increased STC1 expression as compared with LN⁻ CAFs (Fig. 4a). Western blotting and ELISA confirmed that LN⁺ CAFs had upregulated STC1 expression as compared with LN⁻ CAFs (Fig. 4b, c). Gene Expression Profiling Interactive Analysis (GEPIA2) database analysis demonstrated that STC1 was significantly overexpressed in CCA (Fig. 4d), which was confirmed by immunohistochemistry and ELISA (Fig. 4e, f).

Fig. 4.

CAF-derived STC1 was correlated with LNM in CCA. (a) Proteomics analysis volcano plot showing that STC1 was significantly upregulated in LN⁺ CAF supernatant compared with LN⁻ CAF supernatant. X-axis: the log2-fold change for each identified secretory protein from the supernatant. Y-axis: the −log10 of the p-values. Red dots indicate significantly upregulated proteins; blue dots indicate significantly downregulated proteins (p < 0.05, fold change > 2 [log2 = 1]). Horizontal dotted line: original p = 0.05. Dots with fold change < 2 (log2 = 1) and/or p > 0.05 are in gray. Statistical tests were 2-sided. (b) Secreted STC1 concentration in LN⁺ and LN⁻ CAF supernatants. (c) Western blotting results of STC1 expression levels in LN⁺ and LN⁻ CAFs. (d) GEPIA database analysis of the relative STC1 mRNA expression in matched CCA tissues (n = 36) and adjacent non-CCA tissues (n = 9). (e) Representative images of STC1 immunohistochemical staining in human CCA and non-cancer tissues. (f) Secreted STC1 concentration in the serum of 20 patients with CCA and 20 non-cancer controls (Health). (g) Representative images of STC1 immunohistochemical and double-labeled immunofluorescent staining in LN⁺ and LN⁻ CCA tissues. (h) Representative images of STC1 immunohistochemical and double-labeled immunofluorescent staining in CCA⁺ and CCA⁻ LN tissues. All in vitro experiments were performed with at least three biological replicates. *p < 0.05; **p < 0.01; ***p < 0.001; using 2-tailed t-test. Original magnification: ×200.

We assessed basal STC1 secretion by QBC939, HuCCT1, Huh28, HCCC-9810, ZJU-1125, and HiBEpiC cells and patients’ normal liver fibroblasts (NFs) and CAFs by qRT-PCR, western blotting, and ELISA. The CAFs secreted significantly higher STC1 levels than the NFs and the HiBEpiC and CCA cells (Fig. S3a-c). Double immunofluorescence experiments demonstrated that STC1 was mainly derived from the CAFs in human CCA specimens (Fig. S3d-e). Numerous studies have demonstrated that LNM is an independent risk factor for CCA prognosis [18], which was confirmed by our data (Fig. S3f). Immunohistochemistry and immunofluorescence assay comparison of the STC1 expression in LN⁺ and LN⁻ CCA revealed that the LN⁺ CCA tissue had markedly increased STC1 expression compared to that in the LN⁻ CCA tissue (Fig. 4g).

Immunohistochemistry and immunofluorescence comparison of the CCA-positive LN tissues (CCA⁺ LN) and CCA⁻ LN revealed that the CCA⁺ LN demonstrated markedly increased STC1 expression as compared to the CCA⁻ LN (Fig. 4h). STC1 expression staining intensity was evaluated and the results were correlated with the CCA clinicopathological parameters. STC1 expression was positively correlated with LNM status (Fig. S3h). High stromal STC1 expression was associated with prognosis (Kaplan-Meier survival analysis) and was a risk factor for poor prognosis (univariate Cox regression analysis) (Fig. S3g, Table S2). These findings confirmed that CAF-derived STC1 correlated with poor CCA prognosis and LNM.

3.5. STC1 knockdown in CAFs reduced their promoting effect on CCA cells both in vitro and in vivo

To determine the exact biological function of CAF-derived STC1 in CCA, we examined the effect of recombinant human STC1 on CCA cell proliferation, colony-forming ability, and migration by adding STC1 to the cell culture medium. STC1 significantly promoted HuCCT1 and QBC939 cell proliferation, migration, invasion, TrEM, and colony-forming ability in a concentration-dependent manner (Fig. S4a-m). To clarify whether STC1 participates in epithelial–mesenchymal transition (EMT), we measured vimentin, E-cadherin, and N-cadherin expression in CCA cells after STC1 treatment. Immunofluorescence and western blotting demonstrated that E-cadherin expression was decreased, whereas vimentin and N-cadherin expression was elevated after the STC1 treatment, suggesting its EMT promotion effect (Fig. S5a-d). To evaluate the role of STC1 in CAF supernatant, STC1 in CAFs was knocked out using CRISPR/Cas9 technology and verified by western blotting and ELISA (Fig. 5a–c) and termed as CAF^STC1-KO, while the parental is termed as CAF. Knocking out STC1 significantly inhibited the inducing effect of the CAF supernatant on HuCCT1 and QBC939 cell proliferation (Fig. 5d, 5e). In accordance with this change, CAF^STC1-KO markedly reduced HuCCT1 and QBC939 cell abilities of migration, invasion, TrEM, and colony formation (Fig. 5f-l). Consistent with the in vitro results, CAF^STC1-KO in vivo markedly suppressed the tumor growth-promoting effect. The CAFs^STC1-KO and control tumors were smaller and lighter (Fig. 5m-r) and had lower Ki67 expression than the CAFs^wt tumors (Fig. 5s-t). These results indicated that CAF-derived STC1 promoted CCA cell aggressiveness in vitro and in vivo.

Fig. 5.

STC1 knockdown in CAFs reduced the promoting effect on CCA cells in vitro and in vivo. (a–c) Stable knockdown of STC1 expression in CAFs was confirmed by western blotting (a, b) and ELISA (c). (d, e) MTT assays for measuring the proliferation of QBC939 (d) and HuCCT1 cells (e) incubated with CAF or CAF^STC1-KO supernatant. (f, g) Representative images of migration, invasion, and TrEM assays of QBC939 (f) and HuCCT1 cells (g) treated with supernatant from CAFs or CAFs^STC1-KO (magnification: ×100). (h, i) Histogram analysis of migration, invasion, and TrEM of QBC939 (h) and HuCCT1 cells (i). (j–l) Representative images (j) and histogram analysis of colony formation by QBC939 (k) and HuCCT1 cells (l). (m, p) Representative images of QBC939 (m) and HuCCT1 (p) cells combined with CAFs or CAFs^STC1-KO on nude mouse xenografts. (n, o) QBC939 tumor volumes (n) and weights (o) (n = 5). HuCCT1 tumor volumes (q) and weights (r) (n = 5). Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. (s, t) Representative images and histogram analysis of immunohistochemical staining of Ki67 expression in QBC939 (s) and HuCCT1 cells (t) (n = 5). Scale bars: 100 µm. Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. Error bars represent the SD of three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

3.6. STC1 knockdown in CAFs reduced their promoting effect on CCA lymphangiogenesis and LNM

Investigation of the effect of recombinant human STC1 on LEC migration and tubule formation revealed that STC1 markedly promoted LEC migration and tube formation ability in a concentration-dependent manner (Fig. S6a-c). The CAF-derived STC1 significantly promoted LEC migration and tube formation compared to the CAFs^STC1-KO and control (Fig. 6a-c). CAF STC1 knockdown reduced CCA cell growth rates (colony formation assay and CCK-8 assay) and significantly affected CCA cell invasiveness and migration cells, demonstrating that CAF-derived STC1 is important in LNM.

Fig. 6.

STC1 knockdown in CAFs reduced the promoting effect on lymphangiogenesis and LNM in CCA. (a–c) Representative images (a) and quantifications of Transwell migration (b) and tube formation (c) by HLECs treated with CAF and CAF^STC1-KO supernatant. (d) Representative gross anatomy images of the PLNM and non-LNM nude mouse model. (e, h) Representative images of bioluminescence analysis of footpad tumor and PLNM in QBC939 (e) and HuCCT1 groups (h) (n = 6 per group). (f, i) Enucleated LNs and immunohistochemical staining with anti-luciferase antibody in QBC939 (f) and HuCCT1 (i) (n = 6 per group). Scale bars: red, 50 µm. (g, j) Statistical significance in the QBC939 (g) and HuCCT1 (j) groups was assessed using 1-way ANOVA followed by Dunnett's tests and 2-tailed t-tests. Representative images of intratumoral microlymphatic vessels stained with anti-LYVE1 (left, black arrows) and histogram quantification of LMVD (right) in the QBC939 (k) and HuCCT1 (l) groups. Scale bars: black, 100 µm. Error bars represent the SD of the mean. *p < 0.05; **p < 0.01; ***p < 0.001.

Furthermore, we established a PLNM model to examine the effect of STC1 on LNM in vivo. In vivo imaging (Fig. 6d, e), gross anatomy (Fig. 6f), HE staining, and immunohistochemistry (Fig. 6g) demonstrated that CAF-derived STC1 significantly enhanced the ability of CCA cells to metastasize to the LNs as compared with the control and CAFs^STC1-KO. Statistical analysis indicated that the CAF groups had a significantly higher metastatic LN ratio than the CAF^STC1-KO and control groups (Fig. 6h, i). Immunohistochemistry revealed that the intratumoral regions of the CAFs group had significantly higher LMVD than those of the control and CAFs^STC1-KO groups (Fig. 6k-l), suggesting that CAF-derived STC1 promoted lymphangiogenesis in vivo. These findings suggested that CAF-derived STC1 is crucial for CCA lymphangiogenesis and LNM metastasis both in vitro and in vivo.

3.7. STC1-induced CCA aggressiveness required integrin αVβ3

RNA sequencing was performed to identify the downstream target genes of STC1, and revealed that integrin αVβ3 was significantly upregulated in tumor cells after STC1 treatment as compared with the control (Fig. S7a). Western blotting confirmed that the recombinant human STC1 and CAF supernatants induced integrin expression in CCA cells in a time- and concentration-dependent manner (Fig. S7b-g). The addition of the STC1-neutralizing antibody to the medium containing recombinant human STC1 and CAF supernatant did not significantly increase integrin αVβ3 expression (Fig. S7h, i). These results indicated that CAF-derived STC1 induced increased integrin αVβ3 expression in tumor cells.

Cyclo(-RGDfK) is a specific integrin αVβ3 inhibitor [19]. We examined whether blocking STC1–integrin αVβ3 signaling using an STC1-neutralizing antibody and cyclo(-RGDfK) could inhibit STC1-induced CCA aggressiveness. Consistent with our hypothesis, the STC1-neutralizing antibody and cyclo(-RGDfK) significantly inhibited HuCCT1 and QBC939 cell proliferation, migration, invasion, TrEM, and colony-forming ability (Fig. 7a-f). Consistent with the in vitro results, the STC1-neutralizing antibody and cyclo(-RGDfK) markedly suppressed the promoting effect of CAFs on tumor growth in vivo. Tumors in the STC1-neutralizing antibody and cyclo(-RGDfK) groups were smaller and lighter (Fig. 7g-j) and had lower Ki67 expression than those in the DMSO and IgG groups (Fig. 7k-m). The results provided evidence that CAF-derived STC1 promoted CCA aggressiveness through integrin αVβ3.

Fig. 7.

CAF-derived STC1 promoted CCA aggressiveness in vitro and in vivo through integrin αVβ3. (a, b) Representative images and histogram analysis of the migration, invasion, and TrEM assays for QBC939 (a) and HuCCT1 cells (b) incubated with CAF supernatant (CAF-CM) treated with DMSO, IgG, anti-STC1, or cyclo(-RGDfK). (c, d) MTT assays for measuring the proliferation of QBC939 (c) and HuCCT1 cells (d) incubated with CAF supernatant treated with DMSO, IgG, anti-STC1, or cyclo(-RGDfK). (e, f) Representative images (e) and histogram analysis (f) of QBC939 and HuCCT1 colony formation. (g, i) Representative images of QBC939 (g) and HuCCT1 cells (i) combined with CAFs on nude mouse xenografts treated with DMSO, IgG, anti-STC1, or cyclo(-RGDfK) (n = 5). (h) QBC939 tumor volumes and weights (n = 5). (j) HuCCT1 tumor volumes and weights (n = 5). Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. (k–m) Representative images (k, l) and histogram analysis (m) of immunohistochemical staining for Ki67 expression (n = 5). Scale bars: 100 µm. Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. Error bars represent the SD of three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001. G: Group.

3.8. CAF-derived STC1 activated the FAK and YAP signaling pathways in CCA via integrin αVβ3

Integrin αVβ3 mediates downstream signal transduction by activating FAK and YAP phosphorylation [20,21]. Here, STC1 stimulation increased the p-FAK and p-YAP levels in the CCA cells. The activation of these intracellular signaling components was time-dependent. The addition of cyclo(-RGDfK) to the STC1-supplemented medium inhibited the time-dependent phosphorylation of FAK and YAP in the CCA cells (Fig. 8a, b). These results indicated that STC1 activated the FAK and YAP signaling pathways through integrin αVβ3.

Fig. 8.

FAK and YAP signaling pathways in CCA were activated by CAF-derived STC1 through integrin αVβ3. (a, b) Representative western blotting results of FAK and YAP signaling pathway proteins in QBC939 (a) and HuCCT1 (b) cell lines treated with STC1 (200 ng/ml) or STC1 plus cyclo(-RGDfK) (50 µmol/ml). (c, d) Representative western blotting results of FAK and YAP signaling pathway proteins in QBC939 (c) and HuCCT1 (d) cell lines treated with CAF supernatant treated with DMSO, IgG, anti-STC1 (50 µg/ml), or cyclo(-RGDfK). (e, f) Immunohistochemical staining of p-FAK and p-YAP in QBC939 (e) and HuCCT1 (f) tumors. Scale bar, 100 µm. n = 5 animals per group.

We examined whether blocking the STC1–integrin αVβ3 pathway in CAFs using an STC1-neutralizing antibody and cyclo(-RGDfK) would block FAK and YAP phosphorylation in CCA. Consistent with our hypothesis, the STC1-neutralizing antibody and cyclo(-RGDfK) significantly inhibited FAK and YAP phosphorylation in HuCCT1 and QBC939 cells. The CAFs^STC1-KO, anti-STC1, and cyclo(-RGDfK) groups exhibited significantly lower FAK and YAP phosphorylation levels than the DMSO and IgG groups. STC1 knockdown in the CAFs led to reduced p-FAK and p-YAP levels in the HuCCT1 and QBC939 cells (Fig. 8c, d). Immunohistochemical staining of the animal models revealed that the DMSO and IgG groups had significantly increased p-FAP and p-YAP expression levels compared with the anti-STC1, cyclo(-RGDfK), CAF^STC1-KO, and control groups (Fig. 8e). These findings indicated that CAF-derived STC1 activated the FAK and YAP signaling pathways in CCA through integrin αVβ3 both in vitro and in vivo.

3.9. Blocking the FAK and YAP pathways reduced their promoting effect on CCA both in vitro and in vivo

Blocking the FAK and YAP pathways by adding FAKi and YAPi to the medium containing the CAF supernatant ameliorated the promoting effect of CAFs on HuCCT1 and QBC939 cell migration, proliferation, invasion, TrEM, and colony-forming ability (Fig. 9a-i). Consistent with the in vitro results, FAKi and YAPi markedly inhibited the tumor growth-promoting effect of CAFs in vivo. The FAKi, YAPi, and control group tumors were smaller and lighter, respectively (Fig. 9j-o) and had lower Ki67 expression than the DMSO group (Fig. 9p-s). The results indicated that blocking the FAK and YAP signaling pathways of CAFs reduced their promoting effect on CCA cells in vitro and in vivo. We tested the inhibitory effects of FAKi and YAPi on the YAP and FAK signaling pathways in animals, and p-FAK and p-YAP levels were determined via immunohistochemical staining. As expected, the FAKi and YAPi groups demonstrated significantly decreased p-FAP and p-YAP expression compared to the DMSO group. Interestingly, the FAK and YAP signaling pathways had no regulatory relationship with each other (Fig. S8a, b). The results indicated that blocking the FAK and YAP signaling pathways in CAFs reduced their promoting effect on CCA cells in vitro and in vivo.

Fig. 9.

Blocking CAF FAK and YAP signaling pathways reduced the promoting effect on CCA cells in vitro and in vivo. (a–d) Representative images (a, b) and histogram analysis (c, d) of the migration, invasion, and TrEM assays of QBC939 and HuCCT1 cells incubated with CAF supernatant treated with DMSO, FAKi, or YAPi. (e, f) MTT assays for measuring the proliferation of QBC939 (e) and HuCCT1 cells (f) incubated with CAF supernatant treated with DMSO, FAKi, or YAPi. (g–i) Representative images (g) and histogram analysis (h, i) of QBC939 and HuCCT1 colony formation. (j, m) Representative images of QBC939 (j) and HuCCT1 (m) cells combined with CAFs on nude mouse xenografts treated with FAKi and YAPi (n = 5). (k, l) QBC939 tumor volumes (k) and weights (l) (n = 5). (n, o) HuCCT1 tumor volumes (n) and weights (o) (n = 5). Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. (p–s) Representative images (p, r) and histogram analysis (q, s) of immunohistochemical staining for Ki67 expression (n = 5). Scale bars: 100 µm. Statistical significance was assessed using 1-way ANOVA followed by Dunnett's tests. Error bars represent the SD of three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

3.10. Clinical correlations among STC1, integrin αVβ3, p-FAK, and p-YAP in CCA samples

We conducted a study on 50 randomly selected patients out of the initial pool of 98 diagnosed with CCA to assess the clinical significance of our findings. Specifically, the correlation between STC1, integrin αVβ3, p-FAK, and p-YAP within these 50 CCA tissues was examined. The GEPIA2 database analysis demonstrated that STC1, integrin αVβ3, FAK, and YAP were significantly overexpressed in CCA (Fig. 10a). Immunohistochemical analysis indicated a significant positive correlation between these four markers (Fig. 10b), which was validated using Pearson analysis (Fig. 10c). Kaplan-Meier analysis revealed that the high integrin αVβ3, p-FAK, and p-YAP levels were significantly associated with poor overall survival (OS) (Fig. 10d). These findings suggested that targeting the STC1–integrin αVβ3–p-FAK/p-YAP pathway might be beneficial for treating human CCA.

Fig. 10.

Clinical correlations between STC1, integrin αVβ3, p-FAK, and p-YAP in CCA tissues. (a) GEPIA database analysis of the relative mRNA expression of STC1, integrin, FAK, and YAP in matched CCA tissues (n = 36) and adjacent non-cancer tissues (n = 9). (b) STC1, integrin αVβ3, p-YAP, and p-FAK expression in human CCA specimens. Bars: 100 µm. (c) Pearson correlation coefficient analysis of STC1, integrin αVβ3, p-YAP, and p-FAK expression in patients with CCA (n = 50) by immunohistochemistry. (d) Kaplan-Meier analysis of the OS data of patients with CCA stratified by STC1 level with integrin αVβ3, p-YAP, and p-FAK.

4. Discussion

TME lymphangiogenesis is key to CCA progression. Early metastasis through lymphangiogenesis occurs in 60–70% of patients with CCA, often leading to poor outcomes [22,23]. However, the detailed mechanisms that regulate CCA lymphangiogenesis are unclear; therefore, tumor lymphangiogenesis remains a major cancer research focus. CAFs are abundant in the CCA microenvironment, and CCA cells interact with CAFs to promote tumor progression [24,25]. In the present study, we explored the potential roles of CAFs in CCA LNM. Proteomics screening and functional verification demonstrated the importance of STC1 in CAF-mediated LNM. Mechanistic studies revealed that STC1 induced lymphangiogenesis and promoted CCA cell proliferation, migration, and TrEM via the integrin αVβ3 pathway. Our findings highlighted the mechanism of STC1-mediated LNM and revealed that the STC1–integrin αVβ3–FAK–YAP signaling axis is a potential new target for CCA anti-lymphangiogenic treatment.

Here, we observed that CAFs and LECs are in close proximity to the tumor reactive stroma of human CCAs and the lymphatic vascular bed was significantly dilated in LN⁺ CCA compared with LN⁻ CCA and the control. This spatial proximity of tumor cells, CAFs, and LECs suggested a close cross-talk among the three cell types. Other cancers demonstrate similar expression patterns, such as that in ovarian [26] and colon cancers [27]. These findings supported the idea that CCA cells escape their primary growth site into the lymphatic system via lymphatic vessels. Here, we demonstrated that LN⁺ CAFs were more potent than LN⁻ CAFs in promoting CCA aggressiveness and lymphangiogenesis both in vitro and in vivo. Furthermore, the proteomic analysis confirmed the highly upregulated expression of STC1 in the LN⁺ CAFs. Other studies supported the premise that STC1 promotes tumor progression and metastasis [28,29]. Elevated circulating STC1 mRNA levels were correlated with poorer lung cancer outcomes [30]. STC1 expression in tumor-associated fibroblasts promoted colorectal cancer metastasis. The correlation between STC1 expression and tumor size and LNM was analyzed in that study, but there was no evidence that STC1 is directly involved in tumor LNM [11].

Here, we reported STC1 expression in CCA for the first time. Immunohistochemical experiments demonstrated that STC1 protein was expressed in both tumor cells and stromal cells, but ELISA, western blotting, PCR, and immunofluorescence experiments confirmed that STC1 was mainly derived from CAFs. Therefore, these uncertain analyses must be extended. A possible reason for this contradiction is that tumor cells may take up CAF-derived STC1 after it is secreted. Univariate and multivariate analyses revealed that STC1 expression in epithelial cells did not affect the prognosis. However, the univariate analysis indicated that increased stromal STC1 expression was associated with poor prognosis. Moreover, the current assay relied on STC1 protein expression analysis by immunohistochemistry, which may not reflect gene-level expression. Here, we used a PLNM nude mouse model, which has also been used in bladder cancer [7] and fibrosarcoma [31] studies. The direction of plantar LN drainage was PLNM, which was used to simulate CCA intraperitoneal LNM. The advantage of this model is that the definite lymphatic drainage produced by footpad injection renders in vitro lymphatic measurements more sensitive and quantitative. Here, the LNM evaluation included in vivo imaging of gross anatomy, HE staining, and luciferase immunohistochemical staining of LN tissue to ensure the accuracy of the experimental results.

However, this animal model has several limitations. First, the local footpad microenvironment is markedly different from that in the liver. Moreover, the intratumoral plantar interstitial fluid pressure is higher than that of the liver tissue, affecting the lymphatic return and filtration rates and leading to differing LNM results [32]. We isolated fibroblasts from CCA tumor tissues and classified them into LN⁺ and LN⁻ CAFs, and detected functional differences between the CAF types from different tissue sources in vitro and in vivo. We screened the differential protein STC1 between the LN⁺ and LN⁻ CAFs using proteomics analysis. The in vitro and animal model results strongly supported the stimulatory effect of fibroblast-derived STC1 on tumor growth and LNM. Some studies reported tumor CAF heterogeneity in CCA [33], pancreatic cancer [34], and breast cancer [35], where different CAF subsets in the TME have different functions.

Integrin is a transmembrane receptor and cell adhesion molecule that mediates cell–cell and cell–ECM interactions and is involved in intracellular signaling activation. Currently, eighteen α and eight β subunits have been identified [36]. Integrins can promote tumor survival, proliferation, differentiation, angiogenesis, and EMT by binding to ECM ligands [37]. Integrins are found in both tumor cells and stromal cells in the TME [38]. Here, RNA sequencing revealed that integrin αVβ3 was significantly upregulated after STC1 treatment of tumor cells. Western blotting confirmed that recombinant human STC1 and CAF supernatants induced integrin expression in CCA cells. The integrin increasing effect disappeared when anti-STC1 neutralizing antibodies were added to the CAF supernatant. Recent studies demonstrated that STC1 binds directly to integrin β6 to activate the PI3K signaling pathway. Moreover, targeting the STC1–FOXC2–ITGB6 signal transduction axis was associated with cisplatin resistance in ovarian cancer [39]. Here, blocking STC1–integrin αVβ3 signaling using an STC1-neutralizing antibody and integrin αVβ3 inhibitor inhibited the STC1-induced CCA aggressiveness both in vitro and in vivo. Our data provided evidence that CAF-derived STC1 promoted CCA aggressiveness via integrin αVβ3. The RNA sequencing results demonstrated that STC1 induced significantly increased integrin expression in CCA cells, which was subsequently verified by western blotting. We did not obtain relevant positive results regarding whether STC1 could also directly bind integrin αVβ3 to alter the downstream signaling pathways.

Accumulating evidence suggests that YAP signaling is a downstream target of integrins in breast [20], liver [40], lung [41], prostate [42], and ovarian cancers [43,44]. Some reports indicated that FAK signaling is also a downstream target of integrins [45]. The integrin αVβ3–FAK–MAPK axis may be key to angiogenesis in CCA tissues [21]. In the present study, p-FAK and p-YAP levels were increased in CCA cells stimulated with STC1 and CAF supernatants. FAK and YAP phosphorylation was inhibited in CCA cells when the STC1-neutralizing antibody and cyclo(-RGDfK) were added to the medium containing the CAF supernatant, which was consistent with our hypothesis. The in vivo experiments yielded consistent results.

We confirmed that CAF-derived STC1 activated the FAK and YAP signaling pathways in CCA through integrin αVβ3 in vitro and in vivo. Recently, many studies demonstrated that the YAP and FAK signaling pathways are critical for CCA development. FAK was upregulated and activated in human and mouse intrahepatic CCA samples. FAK activation is involved in CCA initiation and progression by inducing the YAP proto-oncogene [46]. In the present study, FAKi and YAPi reduced the CAF-promoting effect on CCA cell aggressiveness in vitro and in vivo. Immunohistochemistry used in vivo to detect the phosphorylation level of the corresponding pathways blocked by FAKi and YAPi revealed that there was no cross-talk between the FAK and YAP pathways. The results indicated that the FAK and YAP pathways were independent pathways downstream of integrin αVβ3. Finally, we validated the STC1–integrin αVβ3–FAK/YAP pathway in human CCA tissue samples, which was consistent with the in vitro experimental results.

5. Conclusion

In conclusion, the secretion of CAF-derived STC1 induced CCA lymphangiogenesis and promoted LNM and CCA progression through STC1–integrin–FAK/YAP signaling. STC1 promoted CCA cell aggressiveness by upregulating integrin expression in tumor cells and activating the downstream FAK and YAP pathways. Our findings provided new strategies for selecting molecular targets for CCA therapy.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Jian Yan received his master degree in 2013 and his doctor's degree in 2023 from Sun Yat-Sen University. His supervisor was Professor Liu Chao, a famous biliary tract surgeon in China. Currently, he is an attending physician in the Department of Hepatobiliary Surgery of the Third Affiliated Hospital of Southern Medical University. His research interests focus on clinical and basic research of lymphatic metastasis of cholangiocarcinoma.

Rui Zhang(BRID: 05100.00.51809) is currently a deputy chief physician of Biliary-Pancreatic Surgery Department of Sun Yat-Sen Memorial Hospital. He received his master's degree from Sun Yat-sen University in 2011 and his doctor's degree from the Medical School of the University of Duisburg-Essen (West German Oncology Center) in 2017. He has been engaged in clinical work for more than 10 years, and is good at the diagnosis and minimally invasive surgical treatment of benign and malignant tumors of liver and pancreas. His current research interests focus on clinical and basic research on cholangiocarcinoma and hepatocellular carcinoma.