1 Introduction

Colorectal cancer (CRC) is one of the most common carcinomas worldwide, accounting for 8% of cancer incidence and cancer-related deaths in 2018 [[1]]. Distant metastasis accounts for ˜ 90% of cancer-related deaths due to limited treatment options [[2, 3]]. However, the molecular mechanisms of metastasis remain to be fully elucidated.

Adipose tissue is an endocrine organ, and obesity is a national and international public health concern now [[4, 5]]. Obesity has been reported to be a risk factor for several chronic diseases, including colon cancer [[6]]. Adipose-derived stem cells (ADSCs) exist widely in adipose tissue and have the ability to differentiate into numerous cells [[7]]. The impacts of ADSCs on cancer progression are controversial. Although some studies have shown a protective effect of ADSCs mediated by suppressing tumor growth and stimulating apoptosis [[8]], most studies held the opinion that ADSCs promote tumor progression by influencing the tumor microenvironment [[9-12]]. ADSCs can differentiate into carcinoma-associated fibroblasts to promote tumor proliferation [[13, 14]]. In addition, ADSCs secrete multiple cytokines and growth factors, such as TGFβ1, insulin-like growth factor, and VEGF, which contribute to aggressive tumor behavior [[15-17]]. Nevertheless, some novel ADSCs secreted cytokines still remain unexploited.

Cysteine-rich 61 (Cyr61), also known as CCN1, belongs to the CCN protein family. Previous studies found that Cyr61 was upregulated in the serum of certain cancers and associated with poor prognosis, such as breast cancer [[18]], gastric cancer [[19]], hepatocellular carcinoma [[20]], and CRC [[21]]. Cyr61 has diverse biological functions, including promoting cell migration, proliferation, survival, and differentiation, through binding to cell-specific integrin receptors [[22]]. For example, in endothelial cells and vascular smooth muscle cells, Cyr61 stimulates cell migration via binding to αvβ3 and α6β1 [[23]]. In gastric adenocarcinoma cells, Cyr61 activates the NF-κB/cyclooxygenase-2 signaling pathway through binding to integrin αvβ3 [[24]]. In CRC, Cyr61 has been shown to be upregulated and can cooperate with integrin αVβ5 to promote CRC cell migration [[25-27]]. Estrada et al. [[28]] reported that Cyr61 which is secreted by bone marrow-derived mesenchymal stem cells is able to promote angiogenesis. However, the biological functions, whether integrin αVβ5 was the functional receptor, and source of Cyr61 in CRC remain not fully clarified.

In this study, we found that CRC-associated ADSCs (ADSCs-CRC) secreted more Cyr61 than the controls (ADSCs-NC). Elevated serum Cyr61 levels were associated with advanced TNM stages. Mechanistically, we identified that integrin αVβ5 was the functional receptor of Cyr61 and Cyr61 promotes CRC metastasis and vasculogenic mimicry (VM) formation by activing the signaling pathway downstream of integrin αVβ5. Moreover, synergistic effect of anti-VM by integrin αVβ5 inhibitor and anti-VEGF by bevacizumab therapy was found in patient-derived xenograft (PDX) models. Collectively, our findings indicate that Cyr61 derived from ADSCs plays a critical role in promoting CRC progression via integrin αVβ5 and provides a novel antitumor strategy based on targeting the Cyr61/αVβ5.

2 Material and methods 2.1 ADSC isolation and characterization

After obtaining informed consent, adipose tissues were obtained from omentum majus from CRC patients and control donors undergoing surgery for non-neoplastic disease and were split as previously studies [[29]]. First, adipose tissues were washed with PBS to remove debris and red blood cells. Adipose tissues were cut into pieces and treated with 0.25% collagenase type I for 30 min at 37 °C. Then, equal volume Dulbecco's modified Eagle medium (DMEM; Gibco, Thermo Fisher Scientific, St Peters, MO, USA) with 10% FBS (Gibco) was added to neutralize the collagenase activity. Finally, cells were plated on dishes in DMEM with 1 g·L−1 glucose and 10% FBS. After the third passage, we identified the cells by flow cytometric analysis with three stem cell positive markers, CD105, CD90, and CD73, and seven negative markers CD45, CD79a, CD19, CD34, CD14, CD11b, and HLA-DR.

2.2 Patients and samples

Cyr61 levels were analyzed in serum samples from healthy donors (n = 90) and CRC patients (n = 364). Integrin β5 levels were analyzed in 10 paired CRC tissue samples. 293 samples of CRC formalin-fixed, paraffin-embedded tissues were used for integrin β5 expression detection and Kaplan–Meier survival analysis. None of the patients received chemotherapy or radiotherapy before surgery. All the samples were collected from the Sixth Affiliated Hospital of Sun Yat-sen University. All samples were stored at −80 °C refrigerator until further use.

2.3 Cell culture and cell treatment

The CRC cell lines HCT8, HCT116, DLD1, and human embryonic kidney 293T cells were purchased from the American Type Culture Collection (ATCC, Washington, DC, USA). All cells were cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco) at 37 °C under 5% CO2. Recombinant human Cyr61 (rCyr61) protein was purchased from Amyjet Scientific. The neutralizing anti-Cyr61 antibody and the integrin β5 inhibitor (EMD121974; EMD) were purchased from Thermo Fisher Scientific and Selleck (Houston, TX, USA), respectively. Different concentrations of rCyr61, neutralizing antibody, and inhibitor were added prior to experiments.

2.4 ELISA

Cyr61 ELISA kit (RayBiotech, Atlanta, GA, USA) was used to detect the Cyr61 level in supernatants of ADSCs and serum samples according to the manufacturer's protocol. elisacalc software (ELISA Calc, Shanghai, China) was used to generate standard curve.

2.5 RNA extraction and real-time PCR

Total RNA was isolated from cells by TRIzol Reagent (Thermo Fisher Scientific). ReverTra Ace qPCR RT Kit (Toyobo, Kita-ku, Osaka, Japan) was used to perform reverse transcription according to the manufacturer's instructions. The Applied Biosystems 7500 Sequence Detection system was used to carry out quantitative real-time reverse transcription PCR (qRT-PCR) with the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). We generated standard curves and applied the 2−▵▵CT method with normalized to 18S rRNA. All the gene-specific primers were obtained from Invitrogen (Thermo Fisher Scientific), and the oligonucleotide sequences are listed in Table S4.

2.6 Western blot analysis

Cell and tissue samples were lysed with radio-immunoprecipitation assay buffer (RIPA) with protease and phosphatase inhibitor cocktail (Promega, Fitchburg, WI, USA). Proteins were separated by SDS/PAGE and then transferred to polyvinylidene fluoride (PVDF) membranes by the Trans-Blot System (Bio-Rad, Hercules, CA, USA). The membranes were blocked by milk and then incubated with specific primary antibodies against Cyr61 (Abcam, Cambridge, MA, USA, 1 : 1000), integrin β5 (Cell Signaling Technology, Danvers, MA, USA, CST, 1 : 1000), FAK (CST, 1 : 1000), p-FAK (CST, 1 : 1000), P65 (CST, 1 : 1000), GAPDH (Abcam, 1 : 1000), MEK (CST, 1 : 1000), p-MEK (CST, 1 : 1000), ERK (CST, 1 : 1000), p-ERK (CST, 1 : 1000), STAT3 (CST, 1 : 1000), p-STAT3 (CST, 1 : 1000), MMP2 (Abcam, 1 : 2000), and HIF-1α (Abcam, 1 : 500). Finally, membranes were incubated with a specific secondary antibody and visualized by ECL Blotting Detection Reagents. GAPDH served as a control for western blot analysis.

2.7 Cell linages isolation in CRC tissues

Fresh tissue was chopped with a sterile scalpel and then digested for 1 h at 37 °C using collagenase digestion medium (RPMI-medium, collagenase type IV 1 mg·mL−1 and DNAse I 150 U·mL−1) to obtain single cell suspension. Lymphocytes, macrophages, endothelial cells, and CRC cells were purified with CD3+ microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany, 130-050-101), CD14+ microbeads (Miltenyi Biotec, 130-050-201), CD31+ microbeads (Miltenyi Biotec, 130-091-935), and EpCAM+ microbeads (Miltenyi Biotec, 130-061-101), respectively. Finally, cells were plated in adherent conditions in growth medium (DMEM, Pen-Strep 1X, 10% FBS) and passaged regularly to obtain fibroblasts.

2.8 Cell migration assay

Cell migration assays were performed with 24-well plates with 8 μm pore size chamber inserts (Corning, New York, NY, USA). In general, 5 × 104 cells resuspended with 200 μL serum-free DMEM were seeded in the upper chamber well and 800 μL of DMEM with 10% FBS was added into the lower chamber. After 24 h, cells migrating through the membrane were fixed with 4% paraformaldehyde for 15 min and then stained with 0.1% crystal violet for 15 min. The cells were viewed under an inverted microscope (DMI4000B; Leica, Wetzlar, Germany) and quantified using software imagej (Bethesda Softworks LLC, Rockville, MD, USA).

2.9 Wound-healing assay

A total of 2 × 106 cells were seeded in six-well plates and incubated until confluency was reached. A 100 μL pipette tip was used to create a rectilinear scratch. After 24 h, cells were fixed with 4% paraformaldehyde for 15 min and then stained with 0.1% crystal violet for 15 min. An inverted microscope (DMI4000B; Leica) was utilized to image the wound closure.

2.10 Cell counting

A total of 5 × 104 cells were resuspended in DMEM with 10% FBS and seeded in a 12-well plate. The cells were released by trypsinization 3 days later, and cell numbers were counted immediately.

2.11 Vector construction and generation of stable cell lines

The methods for vector construction and generation of stable cell lines were described in our previous study [[30]]. The oligonucleotides to suppress integrin β5 expression were designed by RiboBio (Guangzhou, China). After we verified their knockdown efficiency, they were cloned into lentiviral expression vector pLKO.1-Pur (Addgene, Cambridge, MA, USA). The plasmids were verified by sequencing. Empty vector pLKO.1-Pur carrying a scrambled shRNA served as a control. 293T cells were incubated with the constructed vectors, pMD2G and psPAX2 (Addgene), according to the manufacturer's protocol. 0.22 μm PVDF filters were used to filter 293T cells supernatant, and then, the supernatant was added into the plate to infect DLD1, HCT116, and HCT8 cells. The oligonucleotide sequences for vector construction are listed in Table S4.

2.12 Mass spectrometry assay

The gel was chopped into small fragments with a razor blade, destained, and subjected to digestion by modified porcine trypsin (50–100 ng per digestion; Promega). After trypsin digestion, peptides were dissolved in 0.1% FA and 2% ACN, directly loaded onto a reversed-phase analytical column (75 μm i.d. × 150 mm, packed with Acclaim PepMap RSLC C18, 2 μm, 100 Å, nanoViper, Thermo Fisher Scientific). The gradient was comprised of an increase from 5% to 50% solvent B (0.1% FA in 80% ACN) over 40 min, climbing to 90% in 5 min, and then holding at 90% for the 5 min. All at a constant flow rate of 300 nL·min−1. The MS analysis was performed on Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific). The peptides were subjected to NSI source followed by tandem mass spectrometry (MS/MS) in Q ExactiveTM (Thermo Fisher Scientific) coupled online to the UPLC. Intact peptides were detected in the Orbitrap at a resolution of 70 000. Peptides were selected for MS/MS using NCE setting as 27; ion fragments were detected in the Orbitrap at a resolution of 17 500. A data-dependent procedure that alternated between one MS scan followed by 20 MS/MS scans was applied for the top 20 precursor ions above a threshold ion count of 1E4 in the MS survey scan with 30.0s dynamic exclusion. The electrospray voltage applied was 2.0 kV. Automatic gain control (AGC) was used to prevent overfilling of the ion trap; 1E5 ions were accumulated for generation of MS/MS spectra. For MS scans, the m/z scan range was 350–1800 m/z. Fixed first mass was set as 100 m/z. Protein identification was performed with MASCOT software by searching Uniprot_Aedis Aegypti.

2.13 Membrane protein extraction and receptor identification

Cell membrane proteins were extracted using the ProteoExtract Native Membrane Protein Extraction Kit (M-PEK Kit; Calbiochem, Thermo Fisher Scientific) according to manufacturer’s recommendation. Before this experiment, cells were treated with 40 ng·mL−1 rCyr61 for 1 h. The IgG group and 0.1% SDS group served as controls. The SDS-Out Precipitation Kit (Pierce, Thermo Fisher Scientific) was used to remove the SDS. Then, the extracts were incubated with 50ul beads coupled with IgG or anti-human monoclonal Cyr61 antibody at 37 °C overnight. Finally, proteins were separated by SDS/PAGE and visualized by Coomassie blue staining. The protein-specific bands were excised for mass spectrometric analysis.

2.14 Immunohistochemistry and IHC scoring

Paraffin-embedded tissues were deparaffinized with dimethylbenzene followed by antigen retrieval. The tissues were blocked with normal goat serum at 37 °C for 30 min. Next, the tissues were incubated overnight at 4 °C with specific primary antibodies against integrin β5 (Bioss, Boston, MA, USA, 1 : 200), p-FAK (CST, 1 : 600), p-ERK (CST, 1 : 250), p-STAT3 (CST, 1 : 200), MMP2 (Abcam, 1 : 200), HIF-1α (Abcam, 1 : 100), and CD31 (CST, 1 : 200). Finally, the tissues were incubated with appropriate secondary antibodies and then incubated with 3, 3′-diaminobenzidine (DAB). We evaluated the marker staining results according to a previous study [[31]]. The staining intensity was graded 4 stages: 0 (none), 1 (weak), 2 (moderate), and 3 (strong). The percentage of expression was graded 5 stages: 0 (< 5% staining), 1 (5–25% staining), 2 (25–50% staining), 3 (50–75% staining), and 4 (> 75% staining). The sum of both scores served as the final score. Two pathologists performed the scoring analyses according to the above criteria.

2.15 Immunofluorescence assays

The cells were incubated overnight at 4 °C with specific primary antibodies Cyr61 (Abcam, 1 : 500), integrin β5 (CST, 1 : 1600), p-FAK (CST, 1 : 100), P65 (CST, 1 : 400), p-MEK (CST, 1 : 500), ERK (CST, 1 : 800), STAT3 (CST, 1 : 100), MMP2 (Abcam, 1 : 250), and HIF-1α (Abcam, 1 : 1000). Then, cells were incubated with Alexa 488- or Alexa 594-conjugated goat antibodies (Thermo Fisher Scientific) against mouse or rabbit IgG. Finally, the samples were counterstained with DAPI and imaged with confocal laser-scanning microscope (Leica TCS-SP8).

2.16 In vitro vasculogenic mimicry assay

Briefly, 30 μL of Matrigel (BD, BD Biosciences, Franklin Lakes, NJ, USA) was plated in 96-well plates and incubated at 37 °C for 30 min to allow polymerization. Next, 1 × 105 cells per well were added to the Matrigel layer and grown for 12 h. Randomized fields were captured using an inverted microscope, and tubes were quantified from each image.

2.17 Flow cytometric analysis

Adipose-derived stem cells were incubated with fluorochrome-conjugated specific antibodies and matched control IgG at room temperature for 30 min. Then, flow cytometry (BD Biosciences) was used to analyze the cells. The data were analyzed by software flowjo software (v10.0.7, Tree Star, San Carlos, CA, USA).

2.18 Luciferase assays

Cyr61 promoter sequences (from −2000 to +100 relative to the transcription start site) and sequential deletion were cloned into pGL3-Basic vector. STAT3 cDNA was cloned in pcDNA3.1. These plasmids were transfected into ADSCs by Lipofectamie 3000 (Invitrogen) according to the manufacturer's instructions. A Dual-luciferase Reporter Assay System (Promega) was used according to the manufacturer's instructions.

2.19 Exosome extraction and identification

Exosomes were extracted and identified according to our previous study [[30]]. In briefly, supernatants were harvested and centrifuged sequentially at 300 g for 15 min, 2000 g for 15 min, 10 000 g for 30 min, and 120 000 g for 70 min (Beckman Coulter, Brea, CA, USA) twice. Then, Particle Metrix (PMX), transmission electron microscopy (TEM), and western blot analysis were used to identify the exosomes.

2.20 Animal experiments

All animal experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University, Guangzhou, China. Four- to five-week-old male BALB/c nude mice were purchased from the Animal Experiment Center of Sun Yat-Sen University and were fed in SPF environment. A spleen injection model was used for liver colonization assays (n = 5 per group). We analyzed liver metastasis by autopsy and hematoxylin and eosin staining (H&E). After CRC cells were treated with Cyr61 (20 ng·mL−1) for one week, 1 × 106 cells in 100 μL PBS were intrasplenically injected. An immediate splenectomy was performed following intrasplenic injection. Subcutaneous tumor growth assays were performed as previous study [[32]] (n = 5 per group). Subcutaneous tumors were evaluated by H&E and immunohistochemistry (IHC). GFP-positive circulating tumor cells (CTCs) in PBMCs of mice were analyzed by flow cytometry as the previous study [[33]].

2.21 Patient-derived xenograft models and intratumoral injection assay

Fresh tumor tissues were obtained from two CRC patients and implanted into NCG mice. When the tumor size reached 1.5 cm3, the tumors were divided into equal volume ˜ 2 mm3 and were subcutaneously implanted into 4- to 5-week-old male BALB/c nude mice. When the tumor size reached about 100 mm3, all mice were randomized into four groups (n = 4 per group): control group, cilengitide (EMD121974, 10 mg·kg−1, once a week) group, bevacizumab (5 mg·kg−1, once a week), and EMD121974 plus bevacizumab group. Intratumor injection of rCyr61 (0.2 μg·kg−1) was performed 24h later after inhibitor intratumor injection. All reagents were injected precisely into the center of the tumors. All mice were sacrificed 4 weeks later, and subcutaneous tumors were subjected to H&E and IHC.

2.22 Statistical analysis

graphpad prism Software (GraphPad Software, La Jolla, CA, USA) was used to perform statistical analysis. Two-tailed Student's test, one-way ANOVA, and Pearson's correlation analysis were performed for statistical comparisons. All statistics analysis data are expressed as mean ± standard error of the mean. A P value < 0.05 was considered statistically significant.

2.23 Study approval

All samples from human tissues were collected with written informed consent from donors, and all procedures were performed with the approval of the Institutional Review Board of The Sixth Affiliated Hospital of Sun Yat-sen University. Animal experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University and conformed to the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (National Academies Press, 2011) in China.

3 Results 3.1 Cyr61 expression is upregulated in ADSCs derived from CRC patients

We first identified the characteristics of the ADSCs after isolation. ADSCs exhibited a typical spindle-shaped fibroblast-like appearance (Fig. S1A). Furthermore, ADSCs possessed the capacity for multilineage differentiation into adipocytes (Fig. S1B) and osteocytes (Fig. S1C). Flow cytometric analysis of the cell surface markers showed that the isolated ADSCs were positive for CD105, CD90, and CD73 and were negative for CD45, CD79a, CD19, CD34, CD14, CD11b, and HLA-DR (Fig. S1D). Thus, these data confirmed that ADSCs were successfully isolated.

The CCN protein family comprises six members, including Cyr61 (CCN1), connective tissue growth factor (CTGF, CCN2), nephroblastoma overexpressed protein (Nov, CCN3), wnt-1-induced secreted protein 1 (WISP-1, CCN4), WISP-2 (CCN5), and WISP-3 (CCN6) [[34]]. These proteins are involved in multiple biological processes, such as cell proliferation, migration, survival, and angiogenesis [[35, 36]], but the functions of these proteins in ADSCs and CRC remain to be fully clarified. We performed qRT-PCR to analyze the mRNA levels of CCN protein family members in ADSCs isolated from nine healthy donors and 11 CRC patients. Among all the CCN protein family members tested, only Cyr61 was upregulated in ADSC-CRC (Fig. S1E, Fig. 1A). Furthermore, ELISAs indicated that ADSCs-CRC secreted abundant amounts of Cyr61 protein into their culture supernatants (Fig. 1B).

Serum Cyr61 is a diagnostic marker for CRC. (A) qRT-PCR analysis of Cyr61 mRNA levels in ADSCs-NC (n = 9) and ADSCs-CRC (n = 11). (B) ELISA of Cyr61 levels in the medium of ADSCs-NC (n = 9) and ADSCs-CRC (n = 11). (C) ELISA of serum Cyr61 levels in healthy donors (n = 90) and CRC patients (n = 364). (D) ELISA of serum Cyr61 levels in different CRC tumor stages. (E) ROC curves for diagnosis of CRC via serum Cyr61, CA125, CEA, or CA199 levels. (F) Decision curve for diagnosis of CRC via serum Cyr61, CA125, CEA, or CA199 levels. Values are represented as mean ± SD. ***P < 0.001, by two-tailed Student's t-test (A–C) and one-way ANOVA (D).

Next, ELISAs were performed to analyze Cyr61 protein levels in serum from 90 healthy donors and 364 CRC patients. Serum Cyr61 protein was upregulated in CRC patients and was positively correlated with tumor TNM stages (Fig. 1C,D). By assessing the relationship between serum Cyr61 and the clinicopathologic characteristics of CRC patients, we found that Cyr61 was significantly elevated in patients with more advanced TNM stages (P < 0.001; Table S1). We then conducted receiver operating characteristic (ROC) curve analysis to compare the diagnostic power of Cyr61 with the traditional biomarkers of CRC. The results showed that the area under the ROC curve (AUC) for Cyr61 was 0.933, better than the traditional diagnostic biomarkers (Fig. 1E and Table S2). Furthermore, decision curve analysis showed that Cyr61 levels provided greater net diagnostic power than three traditional biomarkers, regardless of the threshold used (Fig. 1F). Compared with the Cyr61 levels in ADSC culture supernatants, Cyr61 protein secreted by CRC cell lines was negligible (Fig. S1F). Moreover, western blot analysis of tumor tissues and major tumor infiltration cells, such as lymphocytes, macrophages, endothelial cells, and fibroblasts, also showed negligible expression of Cyr61 protein tissues in CRC tissues (Fig. S1G). These data demonstrated that ADSCs-CRC appeared to be the main source of Cyr61 protein and serum Cyr61 may be a potential diagnostic biomarker of CRC.

3.2 ADSC-derived Cyr61 promotes CRC cell invasion and migration in vitro

To determine whether ADSCs contribute to cancer cell invasion and migration, we employed cell migration and wound-healing assays. For cell migration assays, CRC cells were seeded in the upper wells with ADSCs in the lower wells. For wound-healing assays, ADSCs were seeded in the upper wells with CRC cells in the lower wells. Compared with the ADSCs-NC, ADSCs-CRC significantly promoted cell invasion and migration (Fig. S2A,B). A polyclonal anti-Cyr61 antibody with neutralized function was used to determine the ability of Cyr61 to contribute to cancer cell invasion and migration. Compared with 10 μg·mL−1 IgG, the addition of 5 μg·mL−1 anti-Cyr61 antibody to HCT8 and DLD1 cells reduced the number of invasive cancer cells by 36% and 27%, respectively, and the addition of 10 μg·mL−1 anti-Cyr61 antibody reduced the cell number by 69% and 54%, respectively (Fig. 2A). CRC cells migrated slowly to close the scratched wounds while adding neutralized anti-Cyr61 antibody (Fig. 2B). Furthermore, treatment of HCT8, DLD1, and HCT116 cells with recombinant Cyr61 (rCyr61, 0–40 ng·mL−1) enhanced the invasion and migration of cells in a dose-dependent manner (Fig. 2C,D). Cell counting and MTS assays revealed that incubation of CRC cells with ADSCs-CRC significantly promoted cell proliferation compared to ADSCs-CRC (Fig. S2C,D). However, adding neutralized antibody to the coculture system did not influence CRC cell proliferation (Fig. S2E,F). Collectively, these data suggested that ADSCs-CRC promoted CRC cell invasion and migration via Cyr61.

ADSC-derived Cyr61 promotes CRC cell invasion and migration in vitro. (A) Representative images of transwell migration assays for HCT8 and DLD1 cells cocultured with culture medium alone (Med) or ADSCs in the presence or absence of 5 or 10 μg·mL−1 anti-Cyr61 antibody, or isotype-matched IgG control (IgG). Scale bar = 100 μm, n = 3. (B) Representative images of wound-healing assays for HCT116 and DLD1 cells cocultured with Med or ADSCs in the presence or absence of an anti-Cyr61 antibody at 5 or 10 μg·mL−1, or an IgG. n = 3 (C) Representative images of transwell migration assays for HCT8 and DLD1 cells with rCyr61 at different concentration. Scale bar = 100 μm, n = 3. (D) Representative images of wound-healing assays for HCT116 and DLD1 cells with rCyr61 at different concentration. n = 3. Values are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, by one-way ANOVA.

3.3 Integrin αVβ5 is the functional receptor of Cyr61

To investigate the existence of a receptor of Cyr61 on the CRC cell membrane, HCT8 cells were treated with rCyr61. Immunofluorescence (IF) assays showed that rCyr61 was localized to the cell membrane, implying the presence of a Cyr61-specific receptor on CRC cell membrane (Fig. 3A). To identify the unknown receptor, we extracted the membranous proteins from HCT8 cells treated with rCyr61. Specific protein bands were detected after immunoprecipitation with Cyr61-specific antibody (Fig. 3B) and were subjected to mass spectrometry. Nine candidate membranous proteins were identified (Table S3). Among these proteins, integrin αVβ5 got the highest score in the mass spectrometry results and only integrin αVβ5 belonged to the integrin family (Fig. S3A,B). Subsequent western blot analysis with specific antibodies for the detection of integrin αV and integrin β5 confirmed the combination of Cyr61 with integrin αVβ5 in HCT8 and DLD1 cells (Fig. 3C). Besides, purified integrin αVβ5 was used to confirm the direct interaction with Cyr61 (Fig. 3D). Furthermore, CRC cells were incubated with rCyr61 and the IF assays results demonstrated that Cyr61 colocalized with integrin αVβ5 on the plasma membrane (Fig. S3D). Following short hairpin RNA (shRNA)-mediated knockdown of integrin β5 expression, Cyr61 localization on the cell surface was decreased (Fig. S3C,D). Moreover, integrin β5 was overexpressed in the majority of CRC cell lines (HCT8, DLD1, and HT29) compared to the normal colonic cell lines HIEC-6 and NCM460 (Fig. 3E).

Cyr61 binds to integrin αVβ5 on CRC cells. (A) Confocal microscopy for HCT8 cells in the presence or absence of rCyr61 at 40 ng·mL−1. Scale bar = 10 μm. (B) Immunoprecipitation of the membrane extracts of rCyr61-treated HCT8 cells with anti-Cyr61 antibody. (C) Western blot validation of mass spectrometric results with anti-integrin αV, anti-integrin β5, and anti-Cyr61 antibody in immunoprecipitation products of the membrane extracts from rCyr61-treated HCT8 and DLD1 cells with anti-Cyr61 antibody. (D) Western blot analysis validation of the direct interaction between Cyr61 and integrin αVβ5. (E) Western blot analysis of integrin β5 expression in normal colonic cell lines and CRC cell lines. (F) Western blot analysis of integrin β5 expression in 10 CRC tissues and paired normal adjacent tissues. (G) IHC analysis of integrin β5 expression in the paraffin-embedded CRC tissues and paired normal adjacent tissues. White scale bar = 200 μm. Black scale bar = 50 μm. (H, I) Kaplan–Meier analysis for OS and DFS of CRC patients with low or high expression of integrin β5.

Next, integrin β5 protein levels were analyzed in 10 CRC patient tissues and matched normal tissues. Western blot analysis indicated that integrin β5 was notably increased in 9 out of 10 CRC tissues (Fig. 3F). To investigate the clinical significance of integrin β5, we detected integrin β5 expression levels in a large cohort of CRC patients. Immunohistochemical (IHC) analysis also suggested that integrin β5 was upregulated in CRC tissues (Fig. 3G). Moreover, survival analysis showed that high integrin β5 expression was associated with poor overall survival (OS; P = 0.002, Fig. 3H) and disease-free survival (DFS) in CRC patients (P < 0.001; Fig. 3I). Taken together, integrin αVβ5 was the functional receptor of Cyr61 and played an important role in CRC progression.

3.4 Cyr61 promotes CRC cell migration and invasion via αVβ5/FAK/NF-κB signaling pathway

To confirm the ability of Cyr61 to promote CRC migration and invasion via integrin αVβ5, we employed the integrin αVβ5 inhibitor EMD 121974 (EMD) and integrin β5-specific shRNA (shβ5). Compared with the control group, the number of invasive cancer cells was reduced and CRC cells migrated slowly to close the scratched wounds in the presence of EMD or shRNA (Fig. 4A,B). Aberrant nuclear factor-κB (NF-κB) activation promotes cancer invasion and metastasis in many cancers, including CRC [[37, 38]]. As FAK activates a number of downstream molecules, including NF-κB [[39]], we hypothesized that Cyr61 binds to integrin αVβ5 to activate the FAK-NF-κB signaling pathway to promote CRC cell migration and invasion. As expected, western blot analysis showed that the protein expression of p-FAK and p-P65 increased following incubation of HCT8 and DLD1 cells with rCyr61 as determined (Fig. 4C and Fig. S4A). Treatment with shβ5 or EMD abrogated the Cyr61-induced αVβ5/FAK/NF-κB signaling (Fig. 4C and Fig. S4A). Further analysis of the protein levels of p-FAK and subcellular localization of P65 by IF revealed upregulated p-FAK expression and intense nuclear staining of P65 following Cyr61 stimulation. Thus, our results indicated that αVβ5/FAK/NF-κB signaling pathway is inhibited by shβ5 or EMD (Fig. 4D and Fig. S4B).

Cyr61 promotes CRC cell migration and invasion via αVβ5/FAK/NF-κB signaling pathway. (A) Representative images of transwell migration assays for HCT8 and DLD1 cells with integrin β5 knockdown or treated with integrin αVβ5 inhibitor EMD 121974 (EMD) before treated with rCyr61 (20 ng·mL−1). Scale bar = 100 μm, n = 3. (B) Representative images of wound-healing assays for HCT116 and DLD1 cells with integrin β5 knockdown or treated with integrin αVβ5 inhibitor EMD before treated with rCyr61 (20 ng·mL−1). n = 3. (C) Western blot analysis the expression of p-FAK, FAK, p-Ikb-α, t-Ikb-αP65, and P65 in HCT8 cells with integrin β5 knockdown or with integrin αVβ5 inhibitor EMD. GAPDH, β-actin, and Lamin A were used as the controls. (D) Confocal microscopy assays were performed to detect the expression of p-FAK and subcellular localization of P65 in HCT8 cells with integrin β5 knockdown or treated with integrin αVβ5 inhibitor EMD. Scale bar = 50 μm. (E) Autopsy and H&E staining of the livers in the nude mice by in vivo liver metastasis assays (n = 5 per group). The arrows indicated the liver metastasis. Black scale bar = 1000 μm. White scale bar = 100 μm. (F) Incidence of liver metastasis in the nude mice in vivo liver metastasis assays. (G) Number of tumor foci on liver surface in the nude mice in vivo liver metastasis assays. (H) Kaplan–Meier survival analysis of the nude mice. NC, negative control. Values are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, by two-tailed Student's t-test and one-way ANOVA (A, B, G).

Next, we performed in vivo metastasis assays by intrasplenic injection of nude mice with HCT8 cells pretreated with rCyr61, shβ5, or EMD (n = 5 per group). Incubation of HCT8 with rCyr6 increased the number of metastatic nodules in the liver and decreased the survival time compared to the control group. However, shRNA-mediated knockdown of integrin β5- or EMD-induced inhibition of the receptor decreased the number of metastatic nodules and increased the survival time compared to the Cyr61 group (Fig. 4E–H). These findings suggested that Cyr61 promotes CRC cell migration and invasion via the integrin αVβ5/FAK/NF-κB signaling pathway in vitro and in vivo.

3.5 Cyr61 promotes VM formation to promote CRC growth and metastasis

Our previous study indicated that Cyr61 did not influence CRC cell proliferation in vitro (Fig. S2E,F). We performed subcutaneous xenograft assays to further confirm this effect in vivo (n = 5 per group). Intriguingly, the results suggested that Cyr61 increased the volume and weight of tumors. Treatment with shβ5 or EMD abrogated the effect of Cyr61 on tumor growth (Fig. 5A–C). A previous study indicated that Epstein–Barr virus-infected cancer cells promoted VM formation by upregulating the expression of some genes, including Cyr61 [[40]]. Therefore, we hypothesized that Cyr61 promotes CRC growth by promoting VM formation. The criteria of VM formation were positive for periodic acid-Schiff (PAS) but negative for CD31 (PAS+/CD31−) and the existence of erythrocytes in the vascular-like channels [[41]]. The results showed that tumors derived from cells treated with rCyr61 exhibited more VM structures. As expected, treatment with shβ5 or EMD decreased the number of VM structures (Fig. 5D,E).

Cyr61 promotes VM formation to promote CRC progression via integrin αVβ5. (A) Tumors in nude mice subcutaneously injected with HCT8 cells pretreated with rCyr61, EMD, or knockdown integrin β5 expression (n = 5 per group). (B) Growth curves of the xenograft tumors. (C) Tumor weight of the HCT8 xenograft tumors. (D) H&E, PAS, and CD31 staining in xenograft tumors. Red arrows indicate the presence of red blood cells; brown arrows indicate typical blood vessels with CD31+ staining; pink arrows indicate PAS+/CD31− VM channels. Black scale bars = 200 μm, white scale bars = 20 μm. (E) The numbers of PAS+/CD31− VM channels in xenograft tumors. (F) IHC analysis of p-FAK, HIF-1α, p-STAT3, and MMP2 expression in xenograft tumors. Scale bars = 100 μm. (G) Statistics of the flow cytometric analysis of the GFP labeled CTCs from whole blood in the nude mice. (H, I) Representative images and statistics of tube formation in HCT8 and HCT116 cells pretreated with rCyr61, EMD, or knockdown integrinβ5 expression. Scale bars = 100 μm, n = 3. (J) Western blot analysis of αVβ5/FAK/HIF-1α/STAT3/MMP2 signaling cascade in HCT8 cells pretreated with rCyr61, EMD, or knockdown integrin β5 expression. NC, negative control. Values are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, by 2-tailed Student’s t-test and one-way ANOVA (C, E, G, I).

The molecular mechanism of VM formation was complicated, in which HIF-1α and MMP2 play critical roles [[40, 42]]. Furthermore, p-STAT3 upregulated MMP2 expression, which promoted the formation of VM [[42]]. In previous experiments, we identified that Cyr61 activates the FAK signaling pathway (Fig. 4C,D). Since FAK activation can stimulate the FAK/MEK/ERK signaling pathway to activate transcription factors [[43]], we hypothesized that Cyr61 promotes VM formation via the αVβ5/FAK signaling cascade to activate HIF-1α and MMP2. We then verified this hypothesis in subcutaneous tumors, and the results suggested that these two transcription factors were active after rCyr61 treatment. Consistently, shβ5 or EMD treatment suppressed this signaling cascade (Fig. 5F and Fig. S5A). Emerging evidence indicated that VM was related to clinical stage [[44]] and tumor metastasis [[45]], with a relationship between VM and the existence of CTCs [[46]]. Flow cytometric analysis of CTCs in whole-blood samples from the mice showed that treatment with rCyr61 increased the number of CTCs compared to the control group. Conversely, treatment with shβ5 or EMD decreased the number of CTCs (Fig. 5G and Fig. S5B).

Next, we applied a 3D culture system to identify VM formation in vitro. HCT8 and HCT116 cells treated with rCyr61 showed vessel-like structures, while CRC cells treated with shβ5 or EMD spread evenly on the matrix surface in a pattern that was similar to that observed in the control group (Fig. 5H,I). Moreover, western blot analysis was performed to confirm the ability of Cyr61 to activate HIF-1α and MMP2 in vitro. As expected, the protein level of HIF-1α and MMP2 were increased via the FAK/MEK/ERK signaling activation pathway (Fig. 5J and Fig. S5C). We further analyzed the protein levels of integrin αVβ5/FAK/HIF-1α/STAT3/MMP2 signaling cascade via IF assays. Compared with the controls, IF assays revealed upregulated expression of p-FAK, p-MEK, and MMP2 and intense nuclear staining of ERK, HIF-1α, and STAT3 in CRC cells treated with rCyr61 (Fig. S5D). Taken together, these results suggested that Cyr61 enhances VM formation to promote CRC growth and metastasis in vivo and in vitro via the integrin αVβ5/FAK/HIF-1α/STAT3/MMP2 signaling cascade.

3.6 Cyr61 is related to VM formation in CRC tissues

We further investigated VM formation and the expression levels of proteins related to the FAK/HIF-1α/STAT3/MMP2 signaling cascade in CRC tissues with or without metastasis. Forty cases of CRC tissues, half of which were with metastasis, were selected from 364 CRC patients whose serum Cyr61 protein levels had been analyzed before. Compared with the cases without metastasis, we found that the number of PAS+/CD31− vascular-like channels increased in CRC tissues with metastasis (Fig. 6A,B). IHC analysis showed that the protein levels of p-FAK, HIF-1α, p-STAT3, and MMP2 were also upregulated in CRC tissues with metastasis (Fig. 6A). Moreover, the numbers of PAS+/CD31− vascular-like channels correlated positively with serum Cyr61 protein levels and the levels of p-FAK, HIF-1α, p-STAT3, and MMP2 protein in tissues (Fig. 6C–G). Thus, these data indicated the existence of VM in CRC tissues and further confirmed the association between VM formation, metastasis, and activation of the FAK/HIF-1α/STAT3/MMP2 signaling cascade.

The relationship between serum Cyr61 levels and VM in CRC tissues. (A) H&E staining, VM formation, p-FAK, HIF-1α, p-STAT3, and MMP2 expression in CRC patients with or without metastasis. n = 20 per group. Black scale bars = 200 μm; white scale bars = 20 μm. (B) Statistics of PAS+/CD31− VM channels in CRC patients with or without metastasis. (C–G) Pearson correlation of PAS+/CD31− VM channels with serum Cyr61 levels, p-FAK, HIF-1α, p-STAT3, and MMP2 IHC scores in CRC tissues. Values are represented as mean ± SD. ***P < 0.001, by two-tailed Student's t-test (B).

3.7 Synergistic effect of anti-VM by integrin αVβ5 inhibitor EMD and anti-VEGF by bevacizumab therapy in CRC PDX models

Since angiogenesis plays a vital role in CRC growth and metastasis, anti-VEGF therapy is a useful treatment strategy [[47]]. PDX models retain the characteristics of the original cancer and are used for curative effect analysis and preclinical drug evaluation [[