Human colorectal cancer-derived carcinoma associated fibroblasts promote CD44-mediated adhesion of colorectal cancer cells to endothelial cells by secretion of HGF

Colorectal cancer tissues were obtained from patients who had undergone surgery at the Tongji Hospital, Huazhong University of Science and Technology. Written informed consent was obtained from all participants, and all procedures were authorized by the Ethical Committee of Tongji Hospital.

Fresh colorectal adenocarcinoma tissues were obtained from eight patients who had undergone surgery, colorectal cancer-derived cancer associated fibroblasts (CC-CAFs) and primary colorectal normal fibroblasts (NFs) were isolated from colorectal cancer tissues and adjacent normal tissues respectively as our previous work [11]. Briefly, the specimens were soaked into 75% ethanol, then washed in phosphate buffered saline (PBS) with 1% penicillin/streptomycin. After mincing and digesting with collagenase IV (Invitrogen, CA, USA) for 3 h at 37 °C, PBS was used to wash the tissues twice, which were then passed through a 70 µm cell strainer (BD Falcon, CA, USA). The cells were centrifuged and cultivated in red blood cell lysis buffer to eliminate red blood cells, and then washed the remaining cells three times with PBS. Finally, the cells were disseminated on culture dishes. After the fourth passage, the human CC-CAFs were ready for use in subsequent experiments. For the co-culture experiment, CC-CAFs were grown on the 0.4 μm pore size transwell insert (Corning, USA) and the CRC cells were grown in the bottom well of the transwell chamber.

Cell lines (LOVO, HUVEC, CT26 and SW48) were obtained from the American Type Culture Collection. The SW48, LOVO, and HUVEC were incubated in Dulbecco’s modified Eagle’s medium (DMEM) with high glucose content (Invitrogen, CA, USA) containing 10% fetal bovine serum (FBS; gibco, USA) and 1% penicillin/streptomycin, CT26 was maintained in RPMI1640 (Invitrogen, CA, USA) supplemented with 10% FBS and 1% penicillin/streptomycin. All cells were cultured in humidified atmosphere of 5% CO2 in air at 37 °C. SU11274 (Medchem express, USA), a specific chemical inhibitor of Met; Anti-HGF antibody (Sinobiological, China), a neutralizing antibody for HGF; recombinant human HGF (Peprotech, USA). GFP lentiviral particles were purchased from Genechem Co (china, shanghai). The transfection of lentiviral was conducted following instruction of manufacturer. For lentiviral transduction, 5000 cells/well were seeded on 96 well tissue culture plates and infected the following day with lentiviral particles (Santa-Cruz Biotechnology, Dallas, TX) at a MOI of 10 in the presence of 10 mg/ml polybrene, purchased from). GFP-expressing cells were selected with 2 μg/ml puromycin (Medchem express, USA) and enriched by three cycles of fluorescence-activated cell sorting (FACS).

Cells were prepared as previously mentioned [11]. The labeled cells were analyzed on BD FACSverse (BD, Bioscience), and the data were processed by FlowJo software (Treestar).

The condition medium was collected as our previously mentioned [11]. 2 × 106 CC-CAFs were seeded onto a 10-cm plate and cultivated with the complete medium. After 36 h, the medium was replaced with 5 ml of fresh F12/DMEM without serum and the cells were cultured for a further 24 h. The CC-CAFs culture medium (CC-CAFs CM) was collected and subjected to filter, stored it at − 80 °C.

The colorectal cancer primary tumor and CRC liver metastases were acquired from 10 patients who had undergone surgery. The isolated tissues were then fixed in 4% paraformaldehyde for 24 h, dehydrated in 25% saccharobiose solution overnight at 4 °C. After embedded in optimal cutting temperature (OCT) compound (Sakura Finetech, Tokyo, Japan), serial cryosections of 10 μm were cut and mounted on glass slides.

Immunofluorescence assay was carried out according to the standard protocols. The specific biomarkers of CAFs, α-SMA, FSP, vimentin and CD44 induced by the CAFs in SW48 and LOVO cells were determined by immunofluorescence assay. Cells planted on the glass slide and frozen sections were washed with PBS, following by fixing with 4% paraformaldehyde for 30 min and then permeabilized in 0.1% Triton X-100 for 25 min. After being blocked with goat sera at room temperature for 1 h, cells and sections were incubated with antibody against α-SMA (1:100; proteintech, China), vimentin (1:100; proteintech, China), FSP (1:100; proteintech, China) or CD44 (1:100; proteintech, China) overnight at 4 °C, rinsed with PBS, incubated with suitable Dylight 594 red-conjugated or Alexa Flour 488 green-conjugated secondary antibodies (1:500; Multi-Sciences Biotech Co. Ltd, Hangzhou, China) for 1 h at room temperature, washed with PBS thrice and costained with 10 μg/ml 40,60-diamidino-2-phenylindole (DAPI) (Sigma, USA) for 10 min and finally observed and imaged with an Olympus Fluoview 500 IX 71 confocal microscope (Tokyo, Japan). Images were digitally recorded at the same magnification and time of exposure.

Static adhesion assay was performed as previously described with modifications [1]. HUVECs were seeded onto 6-well plates at a density of 5 × 104 cells/well. Then LOVO or sw48 cells which stably transfected with GFP (pretreated with or without HGF or CC-CAFs CM) were plated (5 × 104 cells/well) and incubated for 30 min at 37 °C. The unattached cells were gently washed twice with 10% FBS-containing DMEM, the count of tumor cells adhered to HUVECs was examined under fluorescence microscopy. Then cells were digested with 0.25% trypsin, the SW48-GFP/LOVO-GFP cells and HUVECs were quantified by flow cytometry (BD, FACSVerse). The ratio of adhered sw48-GFP/LOVO-GFP cells to HUVECs were calculated as the ratio of sw48-GFP/LOVO-GFP cells to 5 × 104 cells.

Conditioned medium from the CC-CAFs was collected and HGF ELISA kit (eBioscience, CA, USA) was used to determine the concentration of HGF in CC-CAFs according to instruction of manufacturer. All tests were performed in duplicate. The plate was read at a wavelength of 450 nm. The concentration of HGF (pg/ml) was defined by making a standard curve with recombinant HGF. F12 without supplements served as the control. Expression of SDF-1 was investigated in a similar way using a human SDF-1 ELISA kit (eBioscience, CA, USA).

Western blotting was carried out according to the standard protocols described previously [11]. We used primary antibodies raised against GAPDH (Santa Cruz Biotechnology, Biotechnology, CA, USA), Met, AKT, phospho-Met, phospho-AKT (Cell Signaling Technology, MA, USA), α-SMA, FSP and CD44. Goat anti-mouse and anti-rabbit antibodies conjugated with horseradish peroxidase were used as secondary antibodies (1:2000, Jackson ImmunoResearch, PA, USA). Protein bands were detected using enhanced chemiluminescence (ECL) reagents (Dura, Pierce, NJ, USA).

We performed the migration of SW48 and LOVO cells using Transwell chambers (Corning, MA, USA) in a 24-well plate containing 8-µm pores as our previously mentioned [11]. Tumor cells in serum-free F12/DMEM were placed in the upper chambers and F12/DMEM containing 20% FBS was placed in the lower chamber. After 24 h, the cells that had migrated towards lower chamber were fixed with 4% paraformaldehyde for 10 min, then stained with crystal violet, and finally counted by bright-field microscopy.

Total RNA was extracted using TRIzol Reagent (Invitrogen, CA, USA) following the manufacturer’s protocol and was reverse-transcribed into complementary DNA (cDNA) using a Superscript Reverse Transcriptase Kit (Transgene, France). Super SYBR Green Kit (Transgen, France) was used to carry out real-time PCR in ABI7300 real time PCR system (Applied Biosystems). Following specific primer was used: GAPDH (5′-GAGAGACCCTCACTGCTG-3′ and 5′-GATGGTACATGACAAGGTGC-3′); vimentin (5′-AGTCCACTGAGTACCGGAGAC-3′ and 5′-CATTTCACGCATCTGGCGTTC-3′); fibroblast activation protein (FAP) (5′-ATGAGCTTCCTCGTCCAATTCA-3′ and 5′-AGACCACCAGAGAGCATATTTTG-3′); ACTA2 (5′-AAAAGACAGCTACGTGGGTGA-3′ and 5′-GCCATGTTCTATCGGGTACTTC-3′). PCR was performed according to instruction: 95 °C for 10 min; and 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The relative gene expression levels for vimentin, FAP and ACTA2 were normalized against that of GAPDH in each sample, and each sample was run in triplicate and averaged.

For HGF, treatment cells were seeded at a concentration of 6 × 105 cells/well in a six well plate. After 24 h, the cell media was changed to serum-free media for an additional 24 h. Recombinant human HGF (PeproTech, USA) was then added to the wells at different concentration for 12 h. For shRNA transduction, CD44 shRNA and control shRNA expression lentiviral particles were purchased from Genechem Co (China, shanghai), lentiviral transduction was performed as mentioned previously.

Animal assay were performed according to Wuhan Medical Experimental Animal Care Guidelines and were carried out in 6–10-week-old BALB/c mice bred in specific pathogen-free condition of the Animal Research Center of Tongji Medical College of Huazhong University of Science and Technology (Wuhan, China) approved by The Tongji Hospital of Huazhong University of Science and Technology. The liver metastasis model was based on orthotopic CRC model established previously with modification. Briefly, an abdominal midline incision was performed, the cecum exposed, and 5 × 105 CT26-GFP cells (in 50 μl Matrigel) were injected into the cecal wall. Then, the cecum was rinsed with distilled water to kill leaked tumor cells and repositioned into the abdomen. The abdomen and skin were closed using running suture. After 3 weeks, the mice were sacrificed, and the metastatic sites of liver were isolated. Then, Metastatic tumor fragments were minced into 1-mm cubes and digested in collagenase IV (Invitrogen, CA, USA) for 3 h at 37 °C, digested cells were washed twice with complete cell culture media and transferred into RPMI1640 (Invitrogen, CA, USA) containing 10% FBS and 1% penicillin/streptomycin and 1 μg/ml puromycin to eliminate cells except CT26-GFP cells derived from liver metastatic site, after selection process, the cells were analyzed by flow cytometry and western blotting.

We bred BALB/c (nu/nu) mice under specific pathogen-free (SPF) conditions of Animal Research Center of Tongji Medical College of Huazhong University of Science and Technology (Wuhan, China) approved by The Tongji Hospital of Huazhong University of Science and Technology. Mice were used for experiments at 6 weeks old. The animal experiment was carried out as described previously [11]. For in vivo lung metastasis model, sw48 cells with or without CC-CAFs were injected at a density of 1 × 106 in 100 μl PBS via the tail veins of the randomized mice (n = 7 per group). For intrasplenic injection of CRC cells, BALB/c (nu/nu) mice were anesthetized with pentobarbital sodium. A 1 cm cutaneous incision was made in the left flank and carried down through the peritoneal wall. The spleen was carefully exposed, and SW48 cells with or without CC-CAFs (5 × 106/100 μl) were injected under the spleen capsule. For short-term metastasis assay, 1 × 106 SW48-GFP cells were injected via tail vein, and 48 h later, pulmonary circulation perfusion was performed with PBS to eliminate circulating tumour cells, lungs were then taken out and fixed in 4% paraformaldehyde for frozen sections. The fluorescence cells per fields of lung sections were counted.

SPSS 21.0 software (IBM Corp., Armonk, New York, USA) was applied for data analysis. All experiments were repeated 3 times in each group. The mean value of the measurement data was expressed as the mean and SEM. Comparisons among groups were by one-way analysis of variance (ANOVA), and multiple comparisons between the average number of samples were performed by LSD analysis. P < 0.05 indicated that the difference was statistically significant.

The CC-CAFs and NFs were isolated from colorectal cancer tissue and adjacent normal tissue, respectively. Fibroblast specific protein (FSP), smooth muscle actin (α-SMA) were used to identify CC-CAFs. As shown in Fig. 1a, CC-CAFs were positive for the expression of FSP, α-SMA, Western blotting and real-time PCR confirmed that the expression of those markers in CC-CAFs were up-regulated compared with adjacent NFs (Fig. 1b, c). These results showed that we have isolated primary CC-CAFs and NFs successfully. The CC-CAFs was subsequently infected by hTERT expression lentiviral vector to produce immortalization, as shown in Fig. 1d, real-time PCR confirmed that mRNA expression of hTERT was increased after infection,primary CC-CAFs undergo aging after 8 generations, the CC-CAFs-hTERT can continue dividing even after 20 generations.

As shown in the Fig. 1e, a higher level of CD44 expression in CRC liver metastases compared with primary tumor was observed by immunofluorescence. The CD44 expression of liver metastasis site and primary tumor was further confirmed by western-blot, as shown in Fig. 1f, 3 pair of samples were evaluated which proved that up-regulation was not clinical sample-specific. To confirm correlation between liver metastatic propensity and expression of CD44, orthotopic liver metastatic model was established. In this study, mouse colon cancer cell line CT26 which engineered to express GFP and puro-resistance gene, the cell line was named CT26-GFP. CT26-GFP was used to injected into cecal wall, after 3 weeks, liver metastases were harvested, the cells were isolated and designated as CT26-LM (liver metastasis), and the cells was cultured in 5 μg/ml puromycin for 1 week to eliminate stroma cells. Western blotting showed the expression of CD44 in CT26-LM was up-regulated significantly compared with CT26 (Fig. 1g), the result was further conformed by flow cytometry analysis (Fig. 1i). to exclude possibility that in vivo environment might effect on expression of CD44, the expression of CD44 of primary tumor site in orthotopic liver metastatic model was evaluated by western-blot, as shown in Fig. 1h, similar CD44 expression can be observed compared primary tumor site with in vitro cultured CT26-GFP, taken together, those results indicated that CD44 expression was associated with metastatic propensity in CRC.

To evaluate effect of CC-CAFs on adhesion of CRC cells, adhesion assay was performed. As shown in the Fig. 2a, after co-cultured with CC-CAFs, adhesion capacity of SW48-GFP cells or LOVO-GFP cells increased significantly, whereas such enhancement did not observed in CRC cells co-cultured with NFs. We then applied CD44 shRNA and anti-CD44 antibody to explore the role of CD44 in the adhesion induced by CC-CAFs. The effect of CD44 shRNA was shown in Fig. 2b, CRC cells adhesion induced by CC-CAFs was substantially reduced after shRNA transfection or anti-CD44 antibody treatment. To determine whether CC-CAFs have impact on the migration of CRC cells, the migration and invasion of tumor cells were evaluated in the presence or absence of CC-CAFs-CM, more migrated cells were observed when CRC cells were subjected to CC-CAFs-CM treatment. But the count of migrated cells was decreased after CD44 shRNA transfection (Fig. 2c).

To identify underling cytokine which promotes adhesion effect of CC-CAFs-CM, we cultured CC-CAFs or NFs in serum-starved F12 for 24 h, then determined the level of HGF using ELISA. As shown in the Fig. 3a, HGF in the CC-CAFs-CM was 3000 pg/ml which was about ten folds to supernatant of NFs or BM-MSC (bone marrow-derived MSC). A possible precursor of CC-CAFs. No HGF was detected in SW48, Interestingly, SDF-1 which has been widely reported existing in supernatant of CAFs did not detected in supernatant of CC-CAFs or NFs. CD44 expression of CRC cells was evaluated by immunofluorescence (Fig. 3b), a significantly up-regulation of CD44 was observed when CRC cells were treated by CC-CAFs-CM, the result was confirmed by flow cytometry analysis (Fig. 3c), and when 50% diluted CC-CAFs-CM was used to treat tumor cells, expression of CD44 was decreased compared with 100% CM which indicated that the CD44 up-regulation induced by CC-CAFs-CM was dose-dependent. To exclude possibility that this effect was clinical sample-specific, CAFs derived from 2 different patents was used to co-culture with CRC cells. As shown in Fig. 3d, CAFs from two patients all significantly increased CD44 expression of CRC cells. Time course experiment showed that CD44 expression on SW48 cells reached maximum levels (Fig. 3e) within 12 h of CC-CAFs-CM treatment, and the expression was reduced after 24 h of incubation. Similar time course can be observed when recombinant HGF was used to induce up-regulation of CD44 (Fig. 3f). To further confirm the role of HGF in the CC-CAFs-induced adhesion, recombinant HGF was used. Treatment with recombinant HGF enhanced adhesion or migration of CRC cells in dose-dependent manner (Fig. 3g), and when anti-HGF antibody was added to CC-CAFs-CM, significant reduction in the adhesion and migration of CRC cells can be observed (Fig. 4b).

Taken together, these data showed that the effect of the CC-CAFs on adhesion and migration of CRC cells was dependent on HGF secretion.

Underling mechanism was investigated by western blotting. As shown in Fig. 4a, CC-CAFs-CM resulted in increased phosphorylation of MET, AKT and expression of CD44 in colorectal cancer cells. After the application of anti-HGF antibody, AKT, MET, activation and up-regulation of CD44 was partly inhibited, same result can be observed when c-MET inhibitor SU11274 was used, and the inhibitor blocked CC-CAFs-CM induced adhesion or migration significantly (Fig. 4b). These data indicated that CC-CAFs induced activation of HGF/MET/AKT result in up regulation of CD44 and subsequently adhesion or migration.

To confirm the effect of CC-CAFs on tumor metastasis and the role of CD44 in vivo, we injected SW48 cells alone, CC-CAFs alone, SW48 cells mixed with CC-CAFs or SW48-shCD44 mixed with CC-CAFs via spleen capsule and tail vein. The mice were sacrificed 6–8 weeks later (n = 7). Compared with SW48 cells alone, co-injection of SW48 cells and CC-CAFs increased the number and size of tumor nodules in liver and lung and significantly enhance weight of lung and liver. But CC-CAFs-induced enhancement of metastasis was decreased when CD44 was knocked down, the injection of CC-CAFs alone did not result in formation of metastatic tumors (Fig. 5a, b). To conform the effect of CC-CAFs on adhesion of CRC cells in vivo, metastasis assay was performed. Compared with SW48-GFP, co-injection of SW48-GFP and CC-CAFs enhance the count of tumor cells adhered to endothelial cells of lung within 48 h significantly and knock down of CD44 can decrease the effect (Fig. 5c). These results demonstrated that CC-CAFs enhanced CD44-dependent adhesion and metastatic potential in vivo.