Cathepsin L promotes angiogenesis by regulating the CDP/Cux/VEGF-D pathway in human gastric cancer

The GC cell lines AGS and NCI-N87 used in our study were purchased from the American Type Culture Collection (ATCC) and SUN-1, MGC803, MKN45, MKN28, HGC27, and GES-1 were purchased from Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. All GC cell lines were authenticated by short-tandem repeat analysis and had negative results for mycoplasma. SUN-1, AGS, NCI-N87, MKN45, MKN28, and HGC27 were cultured at 37 °C in a humidified atmosphere of 5% CO₂ with RPMI-1640 medium containing 10% fetal bovine serum with 100 U/ml penicillin and 100 U/ml streptomycin. GES-1 and MGC803 were cultured at 37 °C in a humidified atmosphere of 5% CO₂ with Dulbecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum with 100-U/ml penicillin and 100-U/ml streptomycin.

A total of 174 patients in this study underwent resection of the GC at Shanghai Ruijin Hospital. All samples were obtained with the patients’ informed consent, and the samples were histologically confirmed.

Total RNA was extracted with Trizol reagent (Invitrogen, Carlsbad, CA, USA), which was reversely transcribed into cDNA using a Reverse Transcription system (TOYOBO, Kita-ku, Osaka, Japan). Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to quantify CTSL and VEGF-D mRNA levels with the SYBER Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were used as the endogenous reference and used the \({2}^{{ - \left( {\Delta \Delta C_{{\text{t}}} } \right)}}\) method to calculate the relative abundance of RNA compared with GAPDH expression.

Protein was extracted from cells using radioimmunoprecipitation (RIPA) buffer (Thermo, Rockford, IL, USA) and cell debris were removed by centrifugation. The extracts were quantified using Pierce BCA Protein Assay Kit (Thermo) and then boiled the extracts in sodium dodecyl sulfate (SDS) gel-loading buffer containing 10% β-mercaptoethanol. The proteins were separated in 10% SDS–polyacrylamide gel electrophoresis (SDS–PAGE) gels and transferred them onto immobilon polyvinylidene fluoride (PVDF) membranes (Millipore, Burlington, MA, USA), which were probed with rabbit mAbs against mouse GAPDH (1:5000; 60004-1-Ig; ProteinTech Group, Chicago, IL, USA), CDP/Cux1 (1:1000; sc514008X; Santa Cruz Biotechnology, Dallas, TX, USA), VEGF-D (1:1000; 26915-1-AP; ProteinTech Group, Chicago, IL, USA), and CTSL (1:1000, ab58991; Abcam, Cambridge, MA, USA) and then incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit (SA00001-2) and anti-rabbit immunoglobulin G (IgG, SA00001-1) antibody (1:10,000; 13099-1-AP; ProteinTech Group, Chicago, IL, USA), which was followed by Pierce™ ECL Western Blotting Substrate (Thermo).

Two CTSL small hairpin RNAs (shRNA, sh-CTSL #1 and sh-CTSL #2) and negative control (Ctrl) sequences were designed and constructed into a hU6-MCS-Ubiquitin-EGFP-IRES-puromycin plasmid. HGC27, MGC803, MKN28, and MKN45 cells were transfected with shRNA plasmids using Lipofectamine 3000 reagent (Invitrogen) according to the manufacturer’s protocol. CTSL cDNA was subcloned into the CN550-pLOV-EF1a-PuroR-CMV-eGFP-2A-3FLAG plasmid (Obio Technology Co. Ltd., Shanghai, China). CTSL plasmid and corresponding empty vector were transfected the into GC cells (HGC27, MGC803, AGS, and SUN-1) using Lipofectamine 3000 reagent (Invitrogen) following the manufacturer’s protocol. The stably transfected cell lines were selected by puromycin.

IHC staining was performed on the tissue microarray according to a previously reported standard protocol used in Shanghai Institute of Digestive Surgery [36]. In brief, freshly cut 5-μm-thick tissue microarray was stained with antibodies against CTSL (1:200, ab203028, Abcam, Cambridge, MA, USA), VEGF-D (1:1000; 26915-1-AP; ProteinTech Group, Chicago, IL, USA) and PECAM1 (1:200, ab268100; Abcam, Cambridge, MA, USA). Two independent board-certified pathologists evaluated the protein expression levels in an unbiased fashion. The staining intensity and percentage were used to score the overall tissue sections and the intensity was graded as follows: 0 points (no staining), 1 point (light-brown staining), 2 points (brown staining), and 3 points (dark-brown staining). We divided the percentage of positive cell number into four grades: 1 points (< 5%), 2 point (5–30%), 3 points (31–60%), and 4 points (61–100%). We calculated the staining score as follows: overall staining score = intensity score × percentage score. A final score ≤ 3 was considered to be negative staining, and a score > 3 was considered to be positive staining.

HGC27-Vector and HGC27-p110Cux1 cells were seeded in 24-well plates and transfected the cells with pGL3-VEGF-D, pGL3-VEGF-D-del 1, pGL3-VEGF-D-del 2, pGL3-VEGF-D-del 3, and pRL-TK plasmid using a Lipofectamine™ 3000 Transfection Reagent (Invitrogen) according to the manufacturer’s instructions. Three days after transfection, dual-luciferase reporter assay was performed using Dual-Luciferase^® Reporter Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions.

Human umbilical vein endothelial cells (HUVECs) were cultured in tumor supernatants from each cell line in 96‐well plates at a density of 1 × 10⁴ cells per well for 6 h. Plates were pre-coated with 50-μL Matrigel (R&D Systems, 3533-010-02) at 37 °C for 1 h. After a 6-h incubation, tubule images were acquired and analyzed by Image‐Pro Plus software, which we quantified by counting the number of cell junctions and tubes in 10 randomly chosen fields of view. Data were obtained from three independent experiments.

HUVECs were suspended in serum-free medium (1 × 10⁵ cells/insert) and added to the upper chamber of the 24-well insert (membrane pore size, 8 μm; Corning Life Science, Tewksbury, MA, USA). Medium containing 10% serum was added to the lower chamber. After incubation for 12 h, the cells that had migrated to the bottom of the membranes were fixed and stained them with 0.1% crystal violet for 30 min.

A chorioallantoic membrane (CAM) was performed as previously described [8]. After cutting a round window into an egg, 30 μL of cell culture supernatant was dropped onto a filter paper disk and sealed with transparent tape for 3 days. On day 10, the eggs were photographed with a MacroPATH dissecting microscope (Milestone, Sorisole, Italy) and the number of blood vessels around the filter paper disc was counted.

A Human Angiogenesis Antibody Array C1000 (RayBiotech, Norcross, GA, USA) was used to analyze the conditioned medium (CM), following the manufacturer's protocol. The results of the antibody array were quantified with a chemiluminescence system (Bio-Rad, Hercules, CA, USA) similar to the Western blotting procedure.

Four-week-old male BALB/c nude mice were purchased from the Institute of Zoology Chinese Academy of Sciences and housed at a specific pathogen-free environment in the Animal Laboratory Unit, Ruijin Hospital, China. HGC27 cells transduced by sh-CTSL or sh-Ctrl (2 × 10⁷) were harvested and resuspended in 500 μL of PBS. Cells (4 × 10⁶/100 μL PBS) were subcutaneously or peritoneally injected into the mice. For tumorigenesis in vivo, all mice were killed after 28 days, and subcutaneous lumps were removed, imaged, and placed in 4% formaldehyde for paraffin embedding. For peritoneal dissemination model, all mice were killed after 30 days and peritoneal metastasis nodules were counted.

All experiments were repeated at least three times, and the results were summarized as means ± SD. Student t test and one-way analysis of variance (ANOVA) were used to analyze the data and Chi-square tests were used to analyze categorical variables. All statistical tests were performed with SPSS 25.0 (SPSS, Chicago, IL, USA) and p values < 0.05 were regarded as indicating statistically significant differences (*p < 0.05, **p < 0.01, ***p < 0.001).

Previous studies have demonstrated that CTSL is involved in a wide range of human malignancies, including ovarian, breast, prostate, lung, pancreatic, and colon cancers [30]. In accordance with previous studies, bioinformatic analysis of CTSL expression profiles through Gene Expression Profiling Interactive Analyses (GEPIA; https://​gepia.​cancer-pku.​cn) showed that the CTSL mRNA expression levels were significantly increased in majority of tumor types (Supplementary Fig. 1). Notably, CTSL was obviously upregulated in GC tissues (Fig. 1a). Moreover, bioinformatic analyses showed that expression of CTSL was positively related to TNM stage (Fig. 1b). We then examined CTSL protein expression in 134 pairs of GC tissues (GC cohort 1) and examined corresponding nontumor tissues by IHC. Typical immunostaining of CTSL in GC tissues and corresponding non-tumor tissues is shown in Fig. 1c. We primarily identified positive staining of CTSL protein in the cytoplasm and nuclei of GC cells. CTSL has been demonstrated to promote the development of GC [20, 22]; however, the exact role of nuclear CTSL in GC progression remains undefined. To further evaluate the function of nuclear CTSL in GC, we next analyzed the expression level of the nuclear fractions of CTSL in GC, and the results showed that the average expression score for the nuclear fractions of CTSL was significantly higher in GC tissues than that in adjacent nontumor tissues (Fig. 1d). Furthermore, we analyzed the correlation of nuclear CTSL with the clinical parameters of GC, as shown in Table 1. A high nuclear CTSL staining score was associated with more poor differentiation, local invasion, positive lymphatic metastasis status, and higher TNM stage. In addition, we assessed the correlation between CTSL expression and the survival of GC patients. Kaplan–Meier Plotter analysis (https://​kmplot.​com/​analysis) showed that CTSL expression was inversely proportional to survival (Fig. 1e, f). We further identified that prognosis of cases with high nuclear CTSL expression was worse (Fig. 1g) and multivariate analyses of different prognostic parameters revealed that nuclear CTSL expression was an independent factor for overall survival (p = 0.026) (Fig. 1h). It has been reported that cysteine cathepsins play an important role in angiogenesis [25]. To further define the clinic role of CTSL in angiogenesis of GC, we detected the VD of GC tissues by IHC staining of PECAM1 (also called CD31, a marker of angiogenesis). The result showed that VD in tumor tissue was significantly higher than in non-tumor tissues (Fig. 1i, j), and further analysis showed that the nuclear CTSL expression was positively related to the VD of GC tissues (Fig. 1k). Similarly, bioinformatic analyses of the correlation between CTSL and PECAM1 through GEPIA (Fig. 1l) confirmed these findings, which indicated that nuclear CTSL may play a vital role in the angiogenesis of GC. Overall, these data suggest that upregulation of nuclear CTSL is positively correlated with VD in human GC and predicts poor prognosis of GC patients.

Given a potential link between CTSL overexpression and angiogenesis in GC, we determined whether knockdown or ectopic overexpression of CTSL in GC cells could affect the tubular formation and migration of endothelial cells. We identified the expression levels of CTSL in seven GC cell lines and one human gastric epithelial cell line (GES-1). Most GC cell lines expressed relatively high levels of CTSL (Supplementary Fig. 2A and Supplementary Fig. 2B). To determine whether the expression level of CTSL changed in GC cells and contributed to endothelial cells tubule formation and migration, we established an in vitro co-culture system (Supplementary Fig. 2C), in which HUVECs were indirectly cocultured with GC cells and separated from GC cells by a semipermeable membrane (pore size of 0.6 μm). Interestingly, we found that CTSL-knockdown GC cells elicited significantly less tubule formation by HUVECs compared with their respective control group (Fig. 2a, b); whereas, CTSL-overexpressing GC cells stimulated more tubule formation by HUVECs in comparison with their respective control group (Fig. 2c, d). Tumor cells may regulate HUVEC tubule formation by altering endothelial cell migration. We then performed Transwell assays, and the result revealed that overexpression of CTSL in GC cells promoted the migration of HUVECs (Fig. 2e, f); whereas, knockdown of CTSL expression in GC cells led to the reduced migratory capacity of HUVECs (Fig. 2g, h). Furthermore, a wound-healing assay confirmed that knockdown of CTSL in GC cells markedly increased HUVECs migration (Fig. 2i–m, Supplementary Fig. 3A); whereas, overexpression had the opposite effect (Supplementary Fig. 3B and Fig. 2n). These results indicate that CTSL could promote the tubular formation and migration of HUVEC cells in vitro.

To determine the effect of CTSL on angiogenesis in vivo, we performed a chick embryo CAM assay, which has emerged as a standard in in vivo angiogenesis assay [32]. Treatment with CM from the CTSL-knockdown GC cells significantly reduced the vessel number compared with their respective control group (Fig. 3a, b); whereas, CM from CTSL-overexpressing GC cells induced the formation of more vessels than that from their respective control group (Fig. 3c, d). In addition, we further identified the tumor-promoting effect of CTSL in vivo. We subcutaneously transplanted HGC27/sh-CTSL and HGC27/sh-Ctrl cells into nude mice. As shown in Fig. 3e–h, the tumors generated by HGC27/sh-CTSL cells exhibited smaller volume and less weight than those generated by HGC27/sh-Ctrl cells. Moreover, we further detected VD in subcutaneous tumors derived from HGC27/sh-CTSL and HGC27/sh-Ctrl cells. Anti-CD31 immunostaining revealed that vascellum in tumors from HGC27/sh-Ctrl cells was more developed; however, less vascellum formed in tumors from HGC27/sh-CTSL cells (Fig. 3i). Consistently, the assessment of averaged VD per field in each group indicated a threefold higher level in tumors from HGC27/sh-Ctrl cells when compared with tumors from HGC27/sh-CTSL cells (Fig. 3j). Thus, these findings suggest that CTSL in GC cells enhances tumor progression by promoting the angiogenesis of GC.

The angiogenesis of a tumor is regulated by the coordination of the angiogenic and anti-angiogenic factors secreted by tumor cells or stromal cells in tumor microenvironment. To identify the factors from GC cells involved in the angiogenesis of tumors, we performed a human angiogenesis antibody array on CM from HGC27/sh-CTSL and HGC27/sh-Ctrl cells. The result showed a remarkable decrease in VEGF-D in CM from HGC27/sh-CTSL cells (Supplementary Fig. 4A and 4B). We, therefore, quantified the expression level of VEGF-D in HGC27/sh-CTSL cells and HGC27/sh-Ctrl cells by qRT-PCR, and we found a fivefold reduction of VEGF-D mRNA in HGC27/sh-CTSL cells compared with HGC27/sh-Ctrl cells (Fig. 4a). In addition, we collected the CM from HGC27/sh-CTSL cells and HGC27/sh-Ctrl cells and detected the protein expression of VEGF-D in CM by enzyme-linked immunosorbent assay (ELISA). The result also showed that silencing CTSL led to a significant decrease in VEGF-D secretion by GC cells (Fig. 4b).

CCAAT-displacement protein/cut homeobox (CDP/Cux) belongs to a family of transcription factors and has four evolutionarily conserved DNA binding domains: the Cut homeodomain (HD) and the three Cut repeats (CR1, CR2, and CR3) [11]. The full-length protein, named p200 Cux1, is abundant but only transiently binds DNA [23]. P110 Cux1, proteolytic processing of p200 Cux1 by CTSL, stably interacts with DNA, and functions as a transcriptional repressor or activator depending on promoter context [26, 29]. To identify whether upregulating CTSL expression in GC cells could increase CDP/Cux degradation to p110 Cux1, we performed Western blot to detect a change in the protein level of CDP/Cux. We found that overexpression of CTSL in GC cells (MGC803 and HGC27) promoted the cleavage of CDP/Cux and significantly increased the expression of p110 Cux1 isoform (Supplementary Fig. 4C and 4D). To further determine whether p110 Cux1 could directly bind the VEGF-D promoter and transcriptionally activate its expression in GC cells, we next analyzed the promoter region of VEGF-D, which ranged from − 2000 to + 100 bp with respect to the transcription start site (TSS), using the JASPAR database (Fig. 4c). We identified three putative p110 Cux1 binding sites (score > 5) at site A (− 1799 to − 1790), site B (− 903 to − 894), and site C (− 406 to − 397). Dual-luciferase reporter assay showed that p110 Cux1 significantly increased the relative luciferase activity compared with the control group by cotransfecting p110 Cux1 and the full-length promoter of VEGF-D, the promoter 1 vector (deletion at site B: − 903 bp to − 894 bp), or the promoter 2# vector (deletion site C: − 406 to − 397). Deletion at site A (− 1799 bp to − 1790 bp), however, significantly decreased p110 Cux1-induced VEGF-D promoter activity (Fig. 4d), which demonstrated that p110 Cux1 could bind to the promoter of VEGF-D at − 1799 to − 1790 bp and then could activate VEGF-D transcription.

To further evaluate whether CTSL promotes the angiogenic ability of HUVECs through the CDP/Cux/VEGF-D pathway, we first established an in vitro co-culture system. We indirectly co-cultured HUVECs with GC cells (HGC27 and MGC803) after transfection with sh-CTSL, p110 Cux1-expressing vector, or Si-VEGF-D. Next, we performed tubule formation and Transwell assays, and the results revealed that upregulation of p110 Cux1 in GC cells promoted the tubule formation and migratory ability of HUVECs, which could be reversed by co-transfection with Si-VEGF-D (Fig. 4e–h). To ascertain whether CTSL promotes the transcription of VEGF-D by increasing the expression of p110 Cux1, we next performed Western blot and the results showed that the overexpression of p110 Cux1 can rescue the sh-CTSL-mediated suppression of VEGF-D in HGC27 and MGC803 GC cells (Fig. 4i–l). In general, we found that CTSL enhances the cleavage of CDP/Cux to p110 Cux1, which promotes the transcriptional activation of VEGF-D to subsequently induce angiogenesis of GC.

We further examined the correlation between the expression level of CTSL and VEGF-D in human GC. We performed IHC analysis to check VEGF-D expression in 157 GC patients (GC cohort 2), and the results revealed that VEGF-D expression was significantly upregulated in GC tissues compared with adjacent non-tumor tissues (Fig. 5a, b). Moreover, we found that the expression level of VEGF-D in GC was positively correlated with VD and the expression of nuclear CTSL (Fig. 5c, d). Similarly, bioinformatic analysis of the correlation of VEGF-D and VD revealed that VEGF-D was positively correlated with VD in GC (Fig. 5e). In addition, we analyzed the correlation between CTSL and VEGF-D in GC using TCGA data. As shown in Fig. 5f, CTSL expression was positively correlated with the expression of VEGF-D.