A total of 91 patients (48 males and 43 females aged 32-76 years old) with CCA who underwent surgical resection at Southwest Hospital between 2013 and 2016 were included; 91 cancerous samples and 31 paracancerous samples were obtained. The clinical and pathological characteristics of the patients are presented in Tables I and SI. None of the patients received radiotherapy or chemotherapy prior to or following surgery. TNM staging was conducted using the 7th Union for International Cancer Control TNM staging system (40). The present study was approved by the Ethics Committee of Southwest Hospital at Army Medical University. The participants provided written informed consent, and this study was conducted in accordance with the Declaration of Helsinki.

Clinical tissues specimens were fixed in 4% paraformaldehyde for 24 h at room temperature, and then dehydrated and embedded in paraffin. Each TMA containing 91 cancerous and 31 paracancerous paraffin-embedded samples was constructed by Shanghai Outdo Biotech Co., Ltd.; 1.5-mm diameter and 4-µm thick sample tissues were constructed. Immunostaining was performed using a GTVion™ Detection System/Mo&Rb (cat. no. GK500710; Shanghai Genetech Co., Ltd.) according to the manufacturer's protocol. CCA TMAs were incubated at 60°C for 90 min to melt the wax and were deparaffinized twice with xylene for 10 min each, followed by rehydration in a descending graded ethanol series and PBS (5 min per grade).

For antigen retrieval, TMAs were heated for 10 min in sodium citrate buffer (pH 6.0). Endogenous peroxidases were blocked with 3% H₂O₂ for 30 min, and non-specific binding was blocked with 5% BSA (cat. no. ST023; Beyotime) for 30 min at room temperature followed by incubation with primary antibodies overnight at 4°C in a humidified chamber. GATA6 (1:50; cat. no. NBP1-47766; Novus Biologicals LLC), LOXL2 (1:100; cat. no. ab96233), VEGFA (1:200; cat. no. ab46154) and CD34 (1:100; cat. no. ab81289) (all from Abcam) primary antibodies were used. Appropriate anti-mouse/anti-rabbit secondary antibodies from the Detection System kit conjugated to horseradish peroxidase (HRP) were then added for 1 h at 37°C. Expression was detected with 3,3'-diaminobenzidine substrate for 10 sec, and the reaction was stopped with PBS. Hematoxylin was used to stain nuclei for 30 sec at room temperature, and the reaction was stopped by PBS. TMAs were dehydrated in an ascending graded ethanol series (5 min per step), followed by washing with xylene twice for 10 min each. After the slices were dried, a coverslip was mounted onto the slides with neutral balsam (cat. no. G8590; Beijing Solarbio Science & Technology Co., Ltd.). Each sample was observed under a light microscope (magnification, ×40 and ×400; BX41; Olympus Corporation), and three fields per sample were analyzed. The protocol and reagents used for IHC of subcutaneous tumors were as aforementioned.

GATA6 and LOXL2 expression levels were evaluated based on the nuclear score, and VEGFA expression was evaluated based on the cytoplasmic score calculated by two observers without knowledge of the patients' clinical characteristics. The staining intensity was scored as follows: 0 (negative), 1 (weak), 2 (moderate) or 3 (strong staining). The percentage of positively stained cells was calculated by comparing the number of positively stained cells to the number of nuclei (0, no cells stained; 1, 1-25%; 2, 26-50%; and 3, >50%). The final score was obtained by adding the intensity score and the positivity score. A score of 0-1 was defined as negative expression (−), a score of 2-4 was defined as weak positive expression (+), and a score of 5-6 was defined as strong positive expression (++).

An anti-CD34 antibody was used to stain endothelial cells and analyze tumor-associated angiogenesis as aforementioned. The MVD was measured according to the method described by Weidner et al (41), and was assessed by two observers without knowledge of the patient's clinicopathological characteristics. The three highest density vessel fields (hot spots) were identified under a light microscope at low power (magnification, ×100). Any CD34-stained endothelial cells or clusters of endothelial cells were defined as an individual, countable microvessel. The number of microvessels was determined (magnification, ×400), and the MVD was presented as the number of vessels/mm². For each slide, the average MVD of vascular hot spots was calculated based on the vessel count.

The human CCA cell line QBC939 was established from an extrahepatic CCA lesion and maintained at the Hepatobiliary Surgery Institute, Southwest Hospital, Army Medical University. RBE cells were purchased from the Japanese Collection of Research Bioresources Cell Bank (39). The two human CCA cell lines were cultured in RPMI 1640 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Zeta Life). Human umbilical vein endothelial cells (HUVECs) were purchased from the American Type Culture Collection and cultured in DMEM (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS.

QBC939 cells were infected with lentiviral vectors (LV1-shGATA6 and LV10-shLOXL2) encoding short hairpin RNAs (shRNAs) targeting human GATA6 and/or LOXL2 (shGATA6, shLOXL2 and shGATA6/LOXL2) or scrambled shRNA (LV1-shControl and LV10-shControl), which served as a control (shControl; all from Shanghai GenePharma Co., Ltd.). shRNAs were inserted into LV1 or LV10 plasmids (Shanghai GenePharma Co., Ltd.). RBE cells were infected with lentiviral vectors (LV5-GATA6 and LV8-LOXL2) containing LV5 or LV8 plasmids (Shanghai GenePharma Co., Ltd.) overexpressing human GATA6 and/or LOXL2 (exGATA6, exLOXL2 or exGATA6/LOXL2), or containing empty vector (LV5 or LV8), which served as a control (exControl; all from Shanghai GenePharma Co., Ltd.). In total, 1x10⁵ CCA cells were seeded in 6-well plates overnight. After the cells were rinsed with sterile PBS, 100 µl (10⁷ units) lentivirus particles and 2 µl (5 µg/µl) polybrene (Shanghai GenePharma Co., Ltd.) were added to 2 ml RPMI-1640 medium for transfection. Fresh medium was added after 24 h. Cells were cultured for another 48 h, and the transfection efficiency was assessed. These cells were used in subsequent experiments. The shRNA sequences are listed in Table SII.

Plasmids (pcDNA3.1 backbone) containing the full-length (Flag-LOXL2-full) and different deletion mutants of LOXL2 (Flag-LOXL2-Δ1: deleted aa 548-774; Flag-LOXL2-Δ2: deleted aa 1-547) were constructed by Shanghai Genechem Co., Ltd. QBC939 or RBE cells at 70-80% confluency in 10-cm diameter dishes were transfected with 20 µg plasmids using the Lipofectamine™ 3000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Cells were cultured for 72 h after transfection. The collected cells were used for co-IP according to the subsequent protocol.

QBC939 cells at 70-80% confluency in 96-well plates were transfected with 0.1 µg shRNA plasmids [shControl, shGATA6, shLOXL2 or shGATA6/LOXL2 in pGPU6 vectors (Shanghai GenePharma Co., Ltd.)], 0.1 µg 2-kb pGL3-VEGFA promoter firefly luciferase plasmid (Shanghai GenePharma Co., Ltd.) and 0.01 µg pGL4.74-TK Renilla luciferase plasmid (Promega Corporation). Plasmids were transfected using the Lipofectamine™ 3000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.). After 48 h, luciferase activity was detected with a Dual-Glo^® Luciferase Assay system (cat. no. E2920; Promega Corporation) using Renilla luciferase for normalization.

RBE cells were transfected with GATA6, LOXL2, or GATA6/LOXL2 overexpression plasmids, or an empty plasmid as a control (pcDNA3.1(+) vector; Shanghai GenePharma Co., Ltd.). The luciferase activity of the VEGFA promoter was measured as described above. For LOXL2 mutant plasmids, RBE cells in 96-well plate were transfected with 0.1 µg mutant plasmids, 0.1 µg GATA6 plasmid, 0.1 µg pGL3-VEGFA promoter firefly luciferase plasmid and 0.01 µg pGL4.74-TK Renilla luciferase plasmid. The luciferase activity of the VEGFA promoter was measured as described above. The experiments were performed in triplicate.

ChIP was performed according to the EZ-Magna ChIP™ A/G kit protocol (cat. no. 17-10086; Merck KGaA). CCA cells at 80-90% confluency in a 15-cm culture dish containing 20 ml medium were incubated for 10 min at room temperature with 37% formaldehyde (550 µl) for cross-linking. The reaction was quenched by the addition of 2 ml 2X glycine. Then, the cells were washed with cold PBS and were collected with a cell scraper. The collected cells were lysed with cell lysis buffer (included in the ChIP A/G kit) and chromatin was sheared to fragments of 200-1,000 bp by sonication for 14 cycles at 200 W on ice (cycles of 9 sec on followed by 9 sec off). Then, 3 µg of rabbit anti-GATA6 (cat. no. 5851; Cell Signaling Technology, Inc.) or rabbit IgG control antibody (cat. no. 2729; Cell Signaling Technology, Inc.) were adsorbed onto protein G magnetic beads and incubated with chromatin extract at 4°C overnight. Cross-linking of DNA fragments was reversed with ChIP elution buffer, and subsequent incubation at 62°C for 2 h and 95°C for 15 min. Then, DNA was purified in spin columns by sequentially using bind reagent A, wash reagent B and elution buffer C from the kit, and the spin columns were centrifuged at 12,000 x g for 30 sec at room temperature.

Purified DNA was subjected to RNase treatment and analyzed via quantitative (q)PCR. the qPCR reaction was conducted in 25 µl of reaction buffer containing 12.5 µl SYBR Premix Ex Taq™ II (cat. no. RR820A; Takara Bio, Inc.), 2 µl of 10 µM forward and reverse primers, 2 µl purified DNA and 8.5 µl ddH₂O. The PCR conditions were: 95°C for 30 sec, followed by 40 cycles of amplification at 95°C for 5 sec and 60°C for 30 sec. Then, the samples were electrophoresed on 1.5% (w/v) agarose gels and stained with Gel-Red (cat. no. D0139; Beyotime Institute of Biotechnology) in TAE buffer. Primers targeting the VEGFA promoter are presented in Table SIII.

QBC939 or RBE cells at 80-90% confluency in two 10-cm diameter dishes were scraped on ice into cold phosphate-buffered saline (PBS), pooled and lysed with 1 ml of Pierce™ IP lysis buffer (Thermo Fisher Scientific, Inc.) supplemented with protease inhibitors (Roche Diagnostics). Protein G magnetic beads (Bio-Rad Laboratories, Inc.) were washed three times with PBS-0.1% Tween-20 and incubated with 3 µg of antibody in a final volume of 200 µl for 30 min at room temperature. The beads were washed three times and incubated overnight at 4°C. The beads were washed again, and 30 µl of 1X loading buffer was added. The slurry was incubated at 100°C for 10 min, and the beads were discarded. The supernatant was collected for western blotting conducted as described below. The following antibodies were used: GATA6 (cat. no. 5851; Cell Signaling Technology, Inc.); LOXL2 (cat. no. MAB2639; R&D Systems, Inc.); Flag (cat. no. 14793; Cell Signaling Technology, Inc.); rabbit IgG (cat. no. bs-0295P; BIOSS), and mouse IgG (cat. no. bs-0296P; BIOSS). HRP-conjugated VeriBlot for IP (1:1,000; cat. no. ab131366; Abcam) was used as the secondary antibody, and the membrane was incubated for 2 h at room temperature.

Total RNA from cells was extracted using an Eastep Super Total RNA Extraction kit (Promega Corporation) according to the manufacturer's protocols. RT was performed according to the manufacturer's protocols (PrimeScript™ RT reagent kit; cat. no. RR037A; Takara Bio, Inc.). qPCR was performed with SYBR Premix Ex Taq™ II (cat. no. RR820A; Takara Bio, Inc.) using a CFX96 Real-Time system (Bio-Rad Laboratories, Inc.). The PCR conditions for all assays were: 95°C for 30 sec, followed by 40 cycles of amplification at 95°C for 5 sec and 60°C for 30 sec. GAPDH mRNA was used to normalize the RNA input. The relative mRNA expression of indicated genes was using the 2^−ΔΔCq method (42). The primer sequences are presented in Table SIII.

Total protein was extracted from cells using RIPA lysis buffer (Sigma-Aldrich; Merck KGaA) containing protease inhibitors (Roche Diagnostics); the protein concentration was measured using a BCA reagent kit (Beyotime Institute of Biotechnology). Equal amounts of protein (30 µg/lane) were separated via 5-10% SDS-PAGE and transferred to PVDF membranes (Merck KGaA). Membranes were blocked with TBST with 0.1% Tween-20 containing 5% nonfat milk for 2 h at room temperature, and incubated overnight at 4°C with the following primary antibodies: Anti-GATA6 (1:2,000; cat. no. 5851; Cell Signaling Technology, Inc.); anti-GAPDH (1:10,000; cat. no. 10494-1-AP; ProteinTech Group, Inc.); anti-LOXL2 (1:2,000; cat. no. ab96233; Abcam); anti-VEGFA (1:2,000; ab46154; Abcam); and anti-Flag (1:1,000; cat. no. 14793, Cell Signaling Technology, Inc.). Membranes were washed with TBST for 30 min and incubated with an HRP-conjugated anti-rabbit secondary antibody (1:4,000; cat. no. ab6721; Abcam) for 2 h at room temperature. The immunocomplexes were visualized using the Immobilon™ Western Chemiluminescent HRP substrate (cat. no. WBKLS0100; Merck KGaA). ImageJ v1.51 software (National Institutes of Health) was used to quantify protein expression level.

CCA cells (1×10⁴) were seeded on chamber slides overnight, fixed with 4% formaldehyde for 15 min at room temperature, permeabilized with 0.3% Triton X-100 for 10 min and blocked with 1% BSA (cat. no. ST023; Beyotime Institute of Biotechnology) for 30 min at room temperature. Cells were incubated with the following primary antibodies overnight at 4°C: Anti-GATA6 (1:100; cat. no. AF1700; R&D Systems, Inc.) and anti-LOXL2 (1:200; cat. no. ab179810; Abcam). Cells were washed three times and incubated with Cyanine3-conjugated donkey anti-goat (1:50; cat. no. SA00009-3) and Alexa Fluor^® 488-conjugated donkey anti-rabbit (1:50; cat. no. SA00006-6) secondary antibodies (ProteinTech Group, Inc.) for 1 h at 37°C in dark, and nuclei were stained with DAPI (Beyotime Institute of Biotechnology) for 5 min at room temperature. A coverslip was mounted onto the slides, which were stored at 4°C in the dark. Cells were imaged using a confocal laser scanning microscopy (Leica Microsystem GmbH). Three fields per section were photographed (magnification, ×200).

QBC939 cells transduced with lentiviruses carrying shRNAs were cultured in 6-well plates as described above. The medium was removed 48 h later, and the cells were washed three times with PBS and cultured in serum-free RPMI-1640 medium for 24 h. This serum-free medium was then centrifuged at 12,000 x g for 10 min at 4°C to obtain the CM, and stored at −80°C until use.

The levels of VEGFA in the CM were detected using an ELISA kit (cat. no. AE11718HU; Wuhan Abebio Science Co., Ltd.) according to the manufacturer's protocols. Standard curves were created using purified VEGFA.

HUVECs were cultured in 24-well plates coated with Matrigel (Corning Inc.) at a density of 1x10⁵ cells/well with CM prepared as described above for 10 h at 37°C. Tubes were imaged using an inverted microscope (Nikon Corporation). A total of ten random fields per sample were photographed (magnification, ×4). The number of connected tubes was determined and then compared between groups.

All animal experiments were approved by the Institutional Animal Care and Use Committee of Southwest Hospital. A total of 20 female nude (BALB/c-nude) mice (age, 4-6 weeks; weight, 20-25 g) were randomly divided into 4 groups and were housed under specific pathogen-free conditions (temperature, 24-26°C; humidity, 40-60%; ventilation, 15 times/h; 12:12-h light/dark cycle; free access to food and water). To evaluate whether GATA6/LOXL2 affects tumor angiogenesis, ~1×10⁶ lentivirus-transfected CCA cells were subcutaneously injected into the right groin region of nude mice. Each group (shControl, shGATA6, shLOXL2 and shGATA6/LOXL2 group) consisted of five mice. All mice were sacrificed via cervical dislocation after 4 weeks. Tumor size was measured using calipers, and tumor volume was calculated according to a formula (length × width² × ½). The maximum tumor volume reached 0.82 cm³ and no mice developed multiple tumors. All mice maintained good body condition throughout the experiment. Tumor tissues were soaked in 4% paraformaldehyde for 24 h at room temperature, and then dehydrated and embedded in paraffin. The samples were sliced into 4 µm thickness and stained as described above.

Tissue sections of subcutaneous tumors were stained with H&E according to the manufacturer's protocols (cat. no. BL700B; Biosharp Inc.). Deparaffinization and rehydration of sections were performed as previously described. Then, the sections were stained with hematoxylin for 30 sec at room temperature, and followed by eosin staining for 30 sec at room temperature. The sections were observed under a light microscope (Olympus BX41, Japan) at ×200 magnification.

Data were analyzed using SPSS 22.0 software (IBM Corp.). Survival curves were analyzed with the Kaplan-Meier method, and the log rank test was used to evaluate overall survival and disease-free survival. Multivariate analysis was performed using the Cox proportional hazard regression model. Student's two-tailed t-test was used to compare the mean value of two groups. Comparisons among multiple groups were made using ANOVA followed by least significant difference post hoc test. For categorical data, a χ² analysis or Fisher's exact test was used. The data are presented as the mean ± SEM. P<0.05 was considered to indicate a statistically significant difference.

IHC staining was performed to analyze GATA6 and LOXL2 expression levels in two TMAs. GATA6 was exclusively located in the nucleus; however, LOXL2 was detected in both the nucleus and cytoplasm. The scores for the two proteins were calculated in the nucleus only (Fig. 1A). All paracancerous samples were negative for GATA6, LOXL2 and VEGFA staining in bile duct epithelium (Fig. 1B). Among these CCA TMA samples, 38 samples that were weakly or strongly positive for GATA6 and LOXL2 were defined as double-positive, which was designated as GATA6/LOXL2-positive; 28 were defined as double-negative (GATA6/LOXL2-negative). The remaining 25 cases included 12 GATA6-negative/LOXL2-positive and 13 GATA6-positive/LOXL2-negative samples (Table SI).

Our previous studies reported that aberrant expression of GATA6 and LOXL2 were associated with poor prognosis in patients with CCA (38,39). In the present study, the coexpression of GATA6 and LOXL2 in CCA was analyzed using TMAs. The 2-year survival rate of GATA6/LOXL2-positive patients was significantly lower than that of GATA6/LOXL2-negative patients (16 vs. 39%; P=0.03; Table I). According to the results of the Kaplan-Meier analysis, patients with CCA with GATA6/LOXL2-positive staining exhibited worse overall survival (P=0.01; Fig. 1C) and shorter disease-free survival (P=0.02; Fig. 1D) than patients with CCA with GATA6/LOXL2-negative staining. Using Cox regression analysis to assess hazard ratios (HRs) for age, sex, tumor location, TNM stage and GATA6/LOXL2 expression, GATA6/LOXL2 expression was an independent predictor of poor overall survival and poor disease-free survival (HR=1.871, 95% CI=1.070-3.271, P=0.028, and HR=1.781, 95% CI=1.013-3.129, P=0.045, respectively; Table II). Additionally, advanced TNM stage (III-IV) was another independent predictor of overall survival and disease-free survival (HR=2.333, 95% CI=1.311-4.150, P=0.004, and HR=1.861, 95% CI=1.053-3.291, P=0.033, respectively; Table II). Collectively, these findings indicated that GATA6/LOXL2 expression was a risk factor for poor prognosis in patients with CCA.

Anti-VEGFA and anti-CD34 antibodies were used to examine the associations between GATA6/LOXL2 and VEGFA expression, and between GATA6/LOXL2 expression and the MVD in CCA TMAs. VEGFA staining was observed in the cytoplasm (Fig. 1A). The weak and strong positive VEGFA classifications were defined as VEGFA-positive. CD34 was expressed in vascular endothelial cells and was used to determine the MVD (Fig. 1E). The mean value of MVD, defined as the cut-off value, was 23.69/mm² in 38 GATA6/LOXL2-positive and 28 GATA6/LOXL2-negative samples. Samples above the mean value were defined as exhibiting high MVD (40/66), and those below than the mean value as exhibiting low MVD (26/66; Fig. 1E). In the same TMA (Fig. 2A and B), 38 GATA6/LOXL2-positive tissues, 25 (65.8%) VEGFA-positive tissues, and 27 (71.1%) tissues with a high MVD were observed. Of note, GATA6/LOXL2 expression positively associated with VEGFA expression (P=0.02; Table III) and MVD (P=0.04; Table IV). Based on these findings, the coexpression of GATA6 and LOXL2 was significantly associated with VEGFA expression and MVD in human CCA samples.

Using the clinical data of patients, an association between GATA6/LOXL2 and VEGFA expression was identified in patients with CCA. GATA6 and LOXL2 protein levels were evaluated in two CCA cell lines, QBC939 and RBE cells, to investigate this association in vitro. Increased levels of GATA6 and LOXL2 were observed in QBC939 cells compared with in RBE cells (Fig. S1A and B). These results were consistent with findings from our previous studies (38,39). Therefore, GATA6 and LOXL2 expression was knocked down in QBC939 cells by transfecting lentiviruses carrying GATA6-shRNA or LOXL2-shRNA, and GATA6 and LOXL2 were overexpressed in RBE cells transfected with GATA6 and LOXL2-encoding lentiviruses. The effects of GATA6 and LOXL2 on VEGFA expression were then assessed. VEGFA protein levels were significantly increased in RBE cells overexpressing GATA6 and LOXL2 (Fig. 3B and D) but were reduced in QBC939 cells with GATA6 and LOXL2 knockdown (Fig. 3C and E). Similar changes in mRNA expression were observed in CCA cells (Fig. 3B-E); however, no regulatory association between GATA6 and LOXL2 was detected (Fig. S2A and B). Thus, GATA6 and LOXL2 regulate VEGFA expression in human CCA cells.

The localization of the two proteins was determined via confocal laser scanning microscopy to confirm that GATA6 interacted with LOXL2. In CCA cells, GATA6 was located in the nucleus, and LOXL2 was located in the nucleus and cytoplasm. These results are consistent with the IHC staining of CCA tissues (Fig. 3A). A co-IP experiment was then performed using protein extracts from QBC939 and RBE cells. IP with an anti-GATA6 antibody resulted in the successful co-IP of LOXL2 in both cell lines (Fig. 4A and C). Similarly, an anti-LOXL2 antibody also successfully immunoprecipitated GATA6 in both cell lines (Fig. 4B and D). Interactions between GATA6 and LOXL2 in QBC939 and RBE cells were observed in co-IP experiments, revealing that GATA6 and LOXL2 physically interact in CCA cell lines.

A plasmid in which the VEGFA promoter was inserted upstream of the luciferase gene was generated, and dual-luciferase reporter assays were performed to examine the regulatory activities of GATA6 and LOXL2 at the VEGFA promoter, and to investigate the underlying mechanisms via which these protein regulate VEGFA expression at the transcriptional level. Knockdown of GATA6 or LOXL2 significantly decreased VEGFA promoter activity compared with the control (P<0.05 and P<0.05, respectively), and knockdown of both GATA6 and LOXL2 further significantly decreased promoter activity compared with GATA6, LOXL2 or shControl alone (P<0.01, P<0.01 and P<0.001, respectively; Fig. 4E). In addition, whether GATA6 and LOXL2 activate the VEGFA promoter was evaluated. VEGFA promoter activity was increased in cells overexpressing

GATA6 or LOXL2 (P<0.05 and P<0.05, respectively), and the simultaneous overexpression of GATA6 and LOXL2 significantly increased VEGFA promoter activity compared with the overexpression of GATA6 (P<0.01), LOXL2 (P<0.01), or exControl alone (P<0.001; Fig. 4F).

GATA6 is a transcription factor, whereas LOXL2 is not. Thus, ChIP analysis was performed using the GATA6 antibody; it was observed that GATA6 bound to VEGFA promoter regions in QBC939 and RBE cells (Fig. 4G). Thus, GATA6 and LOXL2 may physically interact with each other and cooperatively target the VEGFA promoter in CCA cells.

The SRCR domain of LOXL2 interacts with other proteins (43). Two LOXL2 deletion mutants lacking or retaining the SRCR domain were constructed to further explore the interaction between LOXL2 and GATA6 (Fig. 5A). Based on the results from the co-IP experiment, GATA6 bound to the SRCR domain of LOXL2 independent of the catalytic domain in CCA cells (Fig. 5B and C). Additionally, Flag-LOXL2-Δ1 and Flag-LOXL2-full significantly increased VEGFA promoter activity compared with the control (P<0.05 and P<0.05, respectively); however, Flag-LOXL2-Δ2 did not increase VEGFA promoter activity (Fig. 5D). In addition, knockdown of GATA6 or LOXL2 in QBC939 cells decreased the interaction of LOXL2 with GATA6 (Fig. 5E and F). Based on the data, the SRCR domain of LOXL2 interacts with GATA6 to increase VEGFA promoter activity.

VEGFA secreted by tumor cells promotes the proliferation and migration of vascular endothelial cells and tumor angiogenesis in the tumor microenvironment. CM was collected from QBC939 cells following transduction with shRNA lentivirus, and ELISAs were performed to evaluate the VEGFA concentration and determine whether GATA6/LOXL2 affected VEGFA secretion and angiogenesis. Significantly reduced levels of VEGFA were secreted from cells in the shGATA6/LOXL2 group compared with those in the shGATA6 (P<0.05), shLOXL2 (P<0.05) and shControl (P<0.001) groups (Fig. 6A).

Matrigel-based capillary tube formation assays were performed in HUVECs using CM from CCA cell lines to investigate whether GATA6/LOXL2 promoted angiogenesis in vitro. Different CM samples were added to Matrigel-coated wells, and HUVECs were added to the wells and allowed to form capillary tubes. Tube formation was significantly reduced in the shGATA6/LOXL2 group compared with in the shGATA6 (P<0.01), shLOXL2 (P<0.01) and shControl (P<0.001) groups (Fig. 6B).

Stable lentivirus-transduced CCA cell lines were established and subcutaneously injected into nude mice. After 4 weeks, the nude mice were sacrificed; the subcutaneous tumors were surgically removed, and measurements were obtained. The tumors were also sectioned and subjected to H&E staining and IHC (Fig. 7A). IHC revealed GATA6 expression in the nucleus and LOXL2 expression in the nucleus and cytoplasm of tumor tissues (Fig. 7A), consistent with the IHC results for human CCA tissues and IF staining of human CCA cell lines. Knockdown of GATA6 or LOXL2 reduced VEGFA expression compared with the control, and knockdown of GATA6/LOXL2 markedly reduced VEGFA expression compared with the other groups (Fig. 7A); however, IHC did not reveal an association between GATA6 and LOXL2 expression (Fig. 7A).

The MVD in the shGATA6/LOXL2 group was significantly reduced compared with that in the shGATA6 (P<0.01), shLOXL2 (P<0.01) and shControl (P<0.001) groups (Fig. 7B). Furthermore, the tumors in the shGATA6/LOXL2 group were significantly smaller than those in the shGATA6 (P<0.001), shLOXL2 (P<0.001) and shControl (P<0.001) groups (Fig. 7C). Thus, the findings indicated that GATA6/LOXL2 induced VEGFA expression, tumor growth and angiogenesis in subcutaneous tumors in nude mice.