CD133+ ovarian cancer stem-like cells promote non-stem cancer cell metastasis via CCL5 induced epithelial-mesenchymal transition

Abstract

Cancer stem cells (CSCs, also called cancer stem-like cells, CSLCs) can function as "seed cells" for tumor recurrence and metastasis. Here, we report that, in the presence of CD133+ ovarian CSLCs, CD133- non-CSLCs can undergo an epithelial-mesenchymal transition (EMT)-like process and display enhanced metastatic capacity in vitro and in vivo. Highly elevated expression of chemokine (C-C motif) ligand 5 (CCL5) and its receptors chemokine (C-C motif) receptor (CCR) 1/3/5 are observed in clinical and murine metastatic tumor tissues from epithelial ovarian carcinomas. Mechanistically, paracrine CCL5 from ovarian CSLCs activates the NF-κB signaling pathway in ovarian non-CSLCs via binding CCR1/3/5, thereby inducing EMT and tumor invasion. Taken together, our results redefine the metastatic potential of non-stem cancer cells and provide evidence that targeting the CCL5:CCR1/3/5-NF-κB pathway could be an effective strategy to prevent ovarian cancer metastasis.

Keywords: NF-κB; cancer stem like cells; chemokine (C-C motif) ligand 5; epithelial-mesenchymal transition; non-cancer stem like cells.

Figure 1. CSLCs promote NCSLC metastasis

(A) 2×10⁴ RFP-labeled NCSLCs were seeded on the upper well (above the culture inserts) and cultured with either unlabeled NCSLCs or GFP labeled CSLCs at a 1:2, 1:4, 1:8, or 1:16 CSLC:NCSLC ratio, and the number of RFP+ NCSLCs that migrated was quantified. (B) Similar to (A), 2×10⁴ NCSLCs were plated on the upper well and indirectly co-cultured with NCSLCs or CSLCs in the lower chambers and the number of NCSLCs that migrated (left) or invaded (right) was quantified. (C) Similar to (B), NCSLCs were indirectly co-cultured with NCSLCs or CSLCs obtained from specimens from three different ovarian cancer patients and the number of NCSLCs that migrated or invaded was quantified. (D) NCSLCs were plated in the upper well in the presence of CM from high-density CSLCs (supernatant from 5×10⁵ CSLCs in 1 ml of medium), CM from low-density CSLCs (supernatant from 2.5×10⁴ CSLCs in 1 ml of medium), or control medium (medium incubated overnight without CSLCs) and the number of cells that migrated or invaded was quantified. (E) Wound healing assay for NCSLCs and NCSLCs co-cultured with CSLCs (ratio = 1:16). The migration into the gap at each time point was calculated and is shown in the graphs in the right panels. Scale bar = 50 μm. (F) Experimental model for investigating the effect of CSLCs on the metastasis of NCSLCs in vivo. (G-I) Biofluorescent images of the body (G top) and organs (G, bottom) of tumor-bearing mice. Scale bar = 75μm. The numbers of metastases per mouse are also shown (H). The fluorescent signal intensity was calculated and is shown in the graphs (I). The error bars represent the means ± standard deviation (SD) (n=6). Each experiment was repeated at least three times. *p<0.05, ** p<0.01, *** p<0.001.

Figure 2. CSLCs enhance NCSLC metastasis through CCL5

(A) Confocal microcopy of NCSLCs and NCSLCs co-cultured with CSLCs (generated from A2780 cells) for 24 h and then stained with FITC-labeled anti-CCR1, anti-CCR3, and anti-CCR5 antibodies and CY3-labeled anti-CCL5 antibody. The cell nuclei were counterstained with DAPI. Scale bar = 25μm. (B and C) Transwell migration assay and matrigel invasion assay for NCSLCs derived from A2780 cells (B) and SKOV3 cells (C) in the presence of increasing concentrations of rhCCL5 (2.5-10 ng/ml). *p<0.05 for migration, #p<0.05 for invasion. (D) Wound-healing assay for NCSLCs derive from A2780 cells or SKOV3 cells in the presence or absence of rhCCL5 (5 ng/ml). The migration into the gap after 24 h was calculated and is shown in the graphs in the right panels. Scale bar =50μm. *p<0.05. (E) Similar to (B&C), NCSLCs generated from three ovarian cancer patients were treated with rhCCL5 (5 ng/ml), or left untreated, and the number of cells that invaded was quantified. *p<0.05. (F) Transwell migration assay and matrigel invasion assay for NCSLCs with or without CSLCs (generated from A2780 cells) plated in the lower wells in the presence or absence of anti-CCL5 antibody at 1-10 ng/ml. (G) Similar to (F), invasion assay for NCSLCs co-cultured with CSLCs (generated from specimens from three different ovarian cancer patients) in the presence or absence of anti-CCL5 antibody. (H) Similar to (F), NCSLCs co-cultured with CSLCs (generated from A2780 cells) in the presence or absence of anti-CCR1, anti-CCR3, or anti-CCR5 antibody (5ng/ml). (I) Similar to (H), NCSLCs co-cultured with CSLCs (generated from specimens from three different ovarian cancer patients in the presence or absence of anti-CCR1, anti-CCR3, and anti-CCR5 antibodies (5ng/ml). *p< 0.05 for the comparison between NCSLCs alone and NCSLCs co-cultured with CSLCs. #p< 0.05 for the comparison between NCSLCs co-cultured with CSLCs and NCSLCs co-cultured with CSLCs in the presence of indicated antibody. The error bars represent the means ± standard deviation (SD). Each experiment was repeated at least three times.

Figure 3. CSLC-produced CCL5 enhances NCSLC metastasis in vivo

(A) Experimental model to investigate whether the promoting effect of CSLCs on the metastasis of NCSLCs is mediated by CCL5. RFP-labeled NCSLCs (2×10⁶) alone, or in combination with CSLCs (1×10⁵) transduced with lentivirus carrying GFP-shRNA or CCL5-shRNA, were injected intraperitoneally into SCID mice (6 mice/group). (B) Metastasis formation was examined by biofluorescence imaging. Fluorescent images of the whole body (top) and organs (bottom) of tumor-bearing mice are shown. (C-E) The fluorescent signal intensity for the indicated regions was calculated and is shown in the graphs. *p< 0.05 for the comparison between NCSLCs alone and NCSLCs in combination with CSLCs. #p< 0.05 for the comparison between NCSLCs with CSLCs transduced with lentivirus carrying GFP-shRNA and NCSLCs with CSLCs transduced with lentivirus carrying CCL5-shRNA. The error bars represent the means ± standard deviation (SD) (n=6).

Figure 4. CCL5 and CCR1/CCR3/CCR5 expression is correlated with tumor invasiveness

(A) Representative example of immunohistochemical staining for CCL5, CCR1, CCR3, and CCR5 in ovarian paracancerous tissue, ovarian primary cancer tissue, and ovarian metastasis tissue samples from the same ovarian cancer patient (n=20, paired patient specimens). Scale bar = 25μm. (B) The staining intensity in these sections was quantified according to histological scoring. The error bars represent the means ± standard deviation (SD) (n=20).

Figure 5. CSLCs induce EMT of NCSLCs

(A and B) Phase-contrast images of NCSLCs cultured in the presence or absence of CSLCs, or conditioned medium (CM) from high-density CSLCs, for 48 h. The NCSLCs and CSLCs were generated from A2780 cells (A) or ovarian cancer patients (B). Magnification, 200×. C and D, EMT marker expression in NCSLCs cultured in the presence or absence of CSLCs. The markers were analyzed by Western blot (left) and qRT-PCR (right). β-actin was employed as a loading control for western blot and as a reference gene in qPCR. *p<0.05. The NCSLCs and CSLCs were generated from A2780 cells (C) or ovarian cancer patients (D). (E and F) EMT marker expression in NCSLCs treated with CM from high-density CSLCs, or co-cultured with CSLCs, or cultured with control medium. The markers were analyzed by immunofluorescence, and the nuclei were counterstained with DAPI. Scale bar=75 μm. The NCSLCs and CSLCs were generated from A2780 cells (E) or ovarian cancer patients (F). Scale bar=25μm. The error bars represent the means±standard deviation (SD). Each experiment was repeated at least three times.

Figure 6. CCL5 mediates CSLC-induced EMT of NCSLCs

(A and B) EMT markers in NCSLCs alone, or NCSLCs co-cultured with CSLCs in the presence or absence of anti-CCL5 antibody at 5ng/ml. The expression of EMT markers was analyzed by Western blot (A) and immunofluorescence (B). β-actin was employed as a loading control. *p< 0.05 for the comparisons between NCSLCs alone and NCSLCs co-cultured with CSLCs. #p< 0.05 for the comparisons between NCSLCs co-cultured with CSLCs in the presence and absence of the anti-CCL5 antibody. The NCSLCs and CSLCs were derived from A2780 cells, and the ratio of CSLCs:NCSLCs was 1:16. (C) Similar to (A), EMT markers on NCSLCs alone, NCSLCs co-cultured with CSLCs or NCSLCs cultured in the presence of 5 ng/ml rhCCL5. The expression of EMT markers was analyzed by Western blot. **p<0.01. The NCSLCs and CSLCs were derived from A2780 cells, and the ratio of CSLCs/NCSLCs was 1:16. (D and E) Similar to (C), EMT markers on NCSLCs generated from SKOV3 cells (D) or ovarian cancer patients (E) in the presence or absence of 5 ng/ml rhCCL5. The expression of EMT markers was analyzed by Western blot. *p<0.05, **p<0.01. (F) EMT markers on NCSLCs generated from A2780 cells, SKOV3 cells, or ovarian cancer patients in the presence or absence of 5 ng/ml rhCCL5. The expression of EMT markers was analyzed by immunofluorescence. Scale bar=25 μm. (G) Confocal microscopy of ovarian cancer tissue sections stained in situ with anti-CCR1/CCR3/CCR5 antibodies and anti-vimentin antibody. The cell nuclei were counterstained with DAPI. Scale bar=25μm. The error bars represent the means±standard deviation (SD). Each experiment was repeated at least three times.

Figure 7. The NF-κB signaling pathway is involved in ovarian CSLC-mediated enhanced migration of NCSLCs

(A) Western blot analysis of the expression of the NF-κB p65 subunit in NCSLCs, NCSLCs co-cultured with CSLCs or NCSLCs directly stimulated with rhCCL5 (5 ng/ml). β-actin was employed as the loading control for the cytoplasmic fraction, and Lamin A was used to normalize the loading of the nuclear fraction. *p<0.05 for the comparisons between the cytoplasmic fractions. #p<0.05 for the comparisons between the nuclear fractions. (B) NCSLCs derive from A2780, NCSLCs directly stimulated with rhCCL5, and NCSLCs co-cultured with CSLCs in the presence or absence of anti-CCL5 antibody were stained with phospho-p65 antibody and then with Cy3-conjugated secondary antibody (red). The cell nuclei were counterstained with DAPI (left). The percentage of cells with nuclear staining of phospho-p65 was calculated from the positively stained cells (right). *p < 0.05 for the comparisons between NCSLCs alone and NCSLCs co-cultured with CSLCs or NCSLCs directly stimulated with rhCCL5. #p<0.05 for the comparisons between NCSLCs co-cultured with CSLCs in the presence or absence of anti-CCL5 antibody (5ng/ml). (C) NF-κB activity and nuclear localization in NCSLCs (generated from SKOV3 cells) and NCSLCs directly stimulated with rhCCL5 was measured by ELISA and immunofluorescence. *p<0.05. (D and E) NF-κB activity in NCSLCs, NCSLCs co-cultured with CSLCs and NCSLCs directly stimulated by rhCCL5 was measured by ELISA (D, cells generated from A2780; E, cells generated from three ovarian cancer patients). *p<0.05 for the comparisons between NCSLCs alone and NCSLCs co-cultured with CSLCs or NCSLCs directly stimulated with rhCCL5 (5 ng/ml). #p<0.05 for the comparisons between NCSLCs co-cultured with CSLCs in the presence or absence of anti-CCL5 antibody (5ng/ml). (F) The migration and invasion capabilities of NCSLCs derived from A2780, NCSLCs co-cultured with CSLCs or directly stimulated with rhCCL5 in the presence or absence of PDTC (0.5 μM). *p<0.05 for the comparisons between with or without PDTC. (G) Similar to (F), the invasion capabilities of NCSLCs, NCSLCs co-cultured with CSLCs (generated from three ovarian cancer patients) or directly stimulated with rhCCL5 in the presence or absence of PDTC (0.5μM). *p<0.05 for the comparisons between with or without PDTC. The error bars represent the means±standard deviation (SD). Each experiment was repeated at least three times.