Author contributions: H.L and R.X: collection and assembly of data, data analysis and interpretation, and manuscript writing; T.X, Z.Z, S.L., and Z.L: collection and assembly of data and data analysis and interpretation; Z.C. and B.Z.: conception and study design, administrative support, and manuscript writing. H.L. and R.X. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS August 7, 2012.
The concept of cancer stem cells (CSCs) proposes that solely CSCs are capable of generating tumor metastases. However, how CSCs maintain their invasion and migration abilities, the most important properties of metastatic cells, still remains elusive. Here we used CD133 to mark a specific population from human ovarian cancer cell line and ovarian cancer tissue and determined its hyperactivity in migration and invasion. Therefore, we labeled this population as cancer stem-like cells (CSLCs). In comparison to CD133− non-CSLCs, chemokine CCL5 and its receptors, CCR1, CCR3, and CCR5, were consistently upregulated in CSLCs, and most importantly, blocking of CCL5, CCR1, or CCR3 effectively inhibits the invasive capacity of CSLCs. Mechanistically, we identified that this enhanced invasiveness is mediated through nuclear factor κB (NF-κB) activation and the consequently elevated MMP9 secretion. Our results suggested that the autocrine activation of CCR1 and CCR3 by CCL5 represents one of major mechanisms underlying the metastatic property of ovarian CSLCs. STEM CELLS2012;30:2309–2319
Growing evidence reveals that only a minority of cells within the tumor have the ability to propagate; this rare fraction of self-renewing and tumor-initiating cells are termed cancer stem cells (CSCs) or cancer stem-like cells (CSLCs). The first conclusive evidence for CSCs was reported in 1997 by Dick and coworkers , who isolated CSCs from patients with acute myeloid leukemia. Subsequently, CSCs have been identified in a variety of solid tumors, including brain cancer , breast cancer , colon cancer , ovarian cancer , pancreatic cancer , and prostate cancer . It has been established that CSCs can generate tumors through their self-renewal and multidifferentiation capacity. Furthermore, CSCs are proposed to be the cause of relapse and metastasis . However, how CSCs maintain their invasion and migration capabilities, the most important properties of metastatic cells, still remains to be explored.
Previous studies have indicated that chemokines, through interaction with their specific receptors, are involved in cancer metastasis. Recently, increasing evidence has shown that the interaction of chemokines with their receptors plays an important role in maintaining the metastastic capacity of CSCs. This includes a role for CXCR1/interleukin 8 interleukin-8 receptor antagonist (IL-8RA) in promoting breast CSCs self-renewal and invasion ; a subpopulation of migratory CD133+CXCR4+ CSCs is essential for tumor metastasis in human pancreatic cancer ; highly metastatic MDA-MB-231 cells expressing higher levels of CXCR4 have a larger proportion of CSCs than CXCR4-negative cells . Collectively, these findings suggest that CSCs may indeed be involved in metastatic tumor formation, and, furthermore, the metastatic potential of CSCs may be controlled, at least partially, by chemokine-mediated signaling.
Ovarian cancer is the most lethal gynecological cancer and ranks as the fifth most common cause of cancer-related death in the world. About 70% of patients are diagnosed with either stage III or IV ovarian cancer, the latter characterized by distant metastases , so there is a urgent need for a better understanding of the mechanisms of ovarian carcinoma spreading. Although some evidence implicates ovarian CSCs as the putative cells of origin for ovarian tumorgenesis [13, 14], how ovarian CSCs maintain their invasion and migration abilities has not been established. In this study, we show that the binding of CSLC-produced CCL5 with CCR1 and CCR3 on ovarian CSLCs leading to NF-κB activation and MMP-9 upregulation is the main mechanism underlying the metastatic property of ovarian CSLCs.
MATERIALS AND METHODS
Generation and Culture of Ovarian CSLCs from the A2780 Cell Line
The ovarian cancer cell line A2780 was obtained from the American Type Culture Collection. We generated ovarian CSLCs according to a previously published protocol [5, 15, 16]. Briefly, after dissociation with trypsin (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), single cells were cultured in low attachment plates under stem cell conditions: 5 μg/ml insulin (Sigma), 0.4% bovine serum albumin (Sigma, St. Louis, MO, http://www.sigmaaldrich.com/sigma-aldrich/home.html), 20 ng/ml human recombinant epidermal growth factor (Invitrogen), and 10 ng/ml basic fibroblast growth factor (Invitrogen, Carlsbad, CA, http://www.invitrogen.com.cn).
Magnetic Sorting of Ovarian CSLCs from Primary Cancer Tissues
All studies were performed with the protocols approved by the institutional review board of the Third Military Medical University. The three tumors used in this study (designated ovca1, 2, 3) were categorized as stage III serous adenocarcinomas. CSLCs were isolated by magnetic bead sorting (MACS) using the MidiMACS system. Briefly, single cell suspensions were obtained by enzymatic dissociation of tumor samples, via incubation with 300 units/ml of both collagenase and hyaluronidase for 2.5 hours at 37°C, red blood cells were lysed, and washed twice with phosphate buffered saline (PBS). 1 × 107 single cells were then incubated with microbeads conjugated with anti-CD133/1 (AC133, mouse IgG, cell isolation kit, Miltenyi Biotech, Auburn, CA, http://www.miltenyibiotec.com/) for 15 minutes on ice, washed twice with ice-cold PBS, and sorted with a MACS column (MiltenyiBiotec, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com). Quality of sorting was determined by flow cytometry with an antibody against CD133/2 (293C3-allophycocyanin (APC), Miltenyi Biotech, Auburn, CA, http://www.miltenyibiotec.com/).
Flow Cytometric Analysis
For flow cytometric analysis, tumor spheres were dissociated into single cells, washed and incubated with APC-conjugated monoclonal antibody (mAb) specific for human CD133/1 ([AC133, mouse IgG1, Miltenyi] or isotype-matched control mAb [mouse IgG1, APC]) for 30 minutes on ice. Cells were washed twice, and CD133 expression was assessed by flow cytometry according to the manufacturer's instructions; data were analyzed using CellQuest software.
Immunofluorescence analysis was performed on 8-μm-thick frozen sections, or spheroids (CD133+) cells. Spheres were deposited by cytospin onto glass slides, fixed with ice-cold 4% paraformaldehyde for 15 minutes at 37°C, and blocked with normal serum for 20 minutes at room temperature before incubation with one or more specific antibodies against CD133 (Abcam, Cambridge, U.K., http://www.abcam.com), CCL5, CCR1, CCR3, CCR5, MMP-9, or NFκB (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com), overnight in the dark at 4°C. After three washes, slides were then stained with fluorescein isothiocyanate (FITC)-conjugated anti-rabbit antibodies or Cy3-conjugated anti-mouse antibodies (Abcam, Cambridge, http://www.abcam.com). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI). Stained cells were visualized with an Olympus confocal microscope.
RNA Extraction and Reverse Transcription-Polymerase Chain Reaction Analysis
For reverse transcription-polymerase chain reaction (RT-PCR), total RNA was extracted from ovarian CSLCs, differentiated spheroid cells, or A2780 cells using TRIZOL reagent according to the manufacturer's instructions and then reverse-transcribed using random hexamers to generate cDNA. The PCR reaction was carried out in 25 μl reactions with 10 pmol primers (gene-specific primer sequences are shown in Supporting Information Table 1). Cycle parameters for Oct-4, Nanog, CD133, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) amplification were: 30 seconds at 94°C, 30 seconds at 60°C, and 60 seconds at 72°C for 30 cycles. The amplified products were electrophoresed on 1.5% agarose gels and visualized by ethidium bromide staining.
Assessment of Sphere Differentiation
To induce differentiation of ovarian cancer sphere-forming cells, spheres were trypsinized into single-cell suspensions and seeded into collagen-coated dishes under a standard differentiating condition: Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) without growth factors. Cell morphology was assessed 11 days after plating using a phase-contrast microscope. The differentiation of ovarian cancer spheres was further evaluated by RT-PCR analysis, to demonstrate the loss of stem cell markers, as described above, and immunochemistry was used to detect the expression of ovarian cancer epithelial markers CA125 in differentiated cells.
Gene Expression Profiling Using qPCR Array
Total RNA was isolated as described above. First strand cDNA was prepared using SuperscriptIII RNase Reverse Transcriptase (Invitrogen). Expression of 384 inflammation-associated genes was analyzed using the Human Inflammatory Response and Autoimmunity PCR Array (PAHS-3803E, SABiosciences, Frederick, http://www.sabiosciences.com/) according to the manufacturer's protocol and a previously published protocol .
Gene expression levels were determined using the Ct (ΔΔCt) method, with β-actin as a reference gene, and fold changes were calculated by comparing the expression in CD133+ CSLCs and CD133− non-CSLCs derived from A2780 cells.
Quantitative Real-Time PCR
Relative quantification of gene expression was performed with a TaKaRa PCR Kit (TaKaRa, Tokyo, Japan, http://www.takara.com.cn/) and carried out in triplicate with an ABI 7300 Prism Sequence Detection System (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). The reaction conditions were as follows: 95°C for 30 seconds, followed by 35 cycles of: 95°C for 5 seconds, 60°C for 34 seconds, and 72°C for 45 seconds. The relative gene expression levels were calculated using the comparative Ct (ΔΔCt)method, with GAPDH as a reference gene. The primers used for RT-PCR are shown in Supporting Information Table 1.
Enzyme-Linked Immunosorbent Assay
Supernatants from 1 × 105 CD133+ CSLCs and CD133− cells from primary cancer tissues and A2780 cells were collected after 48 hours in culture and stored at −20°C. The concentration of CCL5 and CXCL16 in each supernatant was measured using an ELISA kit (R&D Systems, Minneapolis, MN, http://www.rndsystems.com) according to the manufacturer's protocol. Absorbance at 450 nm was measured by a microplate reader (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Each measurement was performed in triplicate. Enzyme-linked immunosorbent assay (ELISA) for NF-κB activity was performed in accordance with the manufacturers' instructions (NF-κB/p65Active ELISA assay, Cayman Chemical, USA, http://www.caymanchem.com).
Migration and Invasion Assay
The migration assay was performed using 24-well culture inserts with a porous polycarbonate membrane (8.0 μm, Millipore, Billerica, MA, http://www.millipore.com). For the Matrigel invasion assay, filters were precoated with 30 μl Matrigel (BD Biosciences) for 3 hours. The migration and invasion assays were performed according to a previously published protocol . Briefly, 1 × 104 cells in 200 μl of serum-free medium were added to the upper chamber, and basal serum-free medium or medium with different concentrations of one or more of anti-CCL5, -CCR1, -CCR3, -CCR5 antibody (BD Biosciences), recombinant human CCL5 (R&D Systems), anti-MMP-9 mAB (clone 6-6B, Oncogene Science), or pyrrolidine dithiocarbamate (PDTC, an inhibitor of NFκB) was placed in the lower chamber. The plates were incubated for 24 hours at 37°C in 5% CO2; cells that did not migrate or invade through the pores were removed by a cotton swab. Cells on the lower surface of the membrane were examined and counted under a microscope. Each experiment was repeated at least three times.
CCL5 short hairpin RNA (shRNA) oligonucleotides were obtained commercially (Neuron Bio, Shanghai, China). Briefly, a stem-loop structure oligonucleotide containing a CCL5-target sequence 5′-GTGTGTGCCAACCCAGAGA-3′ was cloned under the control of the human U6 promoter in lentiviral vectors, which also contained a green fluorescent protein (GFP) reporter. The control shRNA had no inserted loop structure. The day before transduction, cells were dissociated into single ones, then transduced with shRNA according to the manufacturer's protocol. After 48 hours, the medium was replaced, and cells were harvested for additional experiments. The efficiency of knockdown by CCL5-shRNA was determined by RT-PCR and ELISA.
In Vivo Xenograft Experiments
Female severe combined immunodeficient (SCID) mice were purchased from the Chinese Academy of Medical Sciences (Beijing, People's Republic of China). Mice were housed and maintained in laminar flow cabinets under specific pathogen-free conditions. For xenograft experiments, ovarian CSLCs and non-CSLCs were counted, resuspended in 200 μL PBS, and simultaneously injected into the left and right flank, respectively, of a 4-week-old mice using a limiting dilution assay . Groups of mice were injected with CSLCs at 101, 1 × 102, 1 × 103, 1 × 104, and 1 × 105 or non-CSLCs at 1 × 102, 1 × 103, 1 × 104, 1 × 105, and 1 × 106 . Engrafted mice were inspected biweekly for tumor appearance by visual observation and palpation. For tumor metastasis experiments, 1 × 105 CCL5-shRNA or vector transduced CSLCs, in 200 μL PBS, were injected intraperitoneally into SCID mice. PBS alone, without any cells, was used as a control. Metastasis formation was monitored using bioluminescence imaging. The tumor xenograft, liver, spleen, kidney, and bowel of each mouse were harvested for further evaluation. Mouse care and use was performed in accordance with local ethical guidelines.
All the data from quantitative assays were expressed as the mean ± SD. Statistical analyses were performed using the independent-samples t test or one-way ANOVA. The difference was considered statistically significant when p < 0.05. All statistical analyses were carried out with SPSS 13.0 software.
Identification and Characterization of Ovarian CSLCs
Self-renewal, expression of stem cell markers, multidifferentiation capacity, and tumor initiation are proposed to be key characteristics of CSLCs. Using serum-free culture selection, we thus characterized ovarian CSLCs obtained from ovarian cancer cell line and stage III serous adenocarcinomas patients based on these four properties. We found that a typical spheroid contained more than 100 viable cells. These oligo-clones of cells have been serially passaged for many generations without any visible sign of senescence, for at least 18 months to date (Supporting Information Fig. 1A). Our findings are consistent with previous results from Lee et al. , in which they used the same selection method to enrich CSCs and demonstrated that prolonged in vitro passages are the product of an outgrowth of a cell clone(s). Accumulating evidence indicates that CD133 is a key molecular marker to identify many types of CSLC, including ovarian CSLCs [2, 4, 20, 21]; thus we assessed the percentage of CD133+ cells after serum-free selection using FACS analysis and immunostaining. We found that less than 5% of cells expressed CD133 before serum-free culture selection, however, the proportion of CD133+ cells gradually increased with passage generation. After the fourth passage, more than 90% of the spheroid cells, which are suitable for the anchorage-independent growth, were positive for CD133 (Supporting Information Fig. 1B, 1D). In addition to CD133, it has been demonstrated that CSLCs express other stem cell markers, including Oct-4 and Nanog [5, 22]. RT-PCR analysis showed that our CD133+ spheres expressed high levels of CSC markers, such as Oct-4 and Nanog, while spheres cultured under differentiating conditions showed minimal expression of stem cell markers (Supporting Information Fig. 1C). To investigate the differentiation capacity of CSLCs, we performed a standard differentiation assay and found that spheroid cells adhered to plates and acquired an epithelial morphology similar to the A2780 cancer cells. These cells expressed the ovarian cancer epithelial markers CA125 after culture in differentiating conditions for 11 days (Supporting Information Fig. 1E). Additionally, in vivo xenograft experiments, we found that the injection of CD133+ CSLCs, as few as 103 per mouse, was able to consistently generate tumor xenografts; in the same moue, 105 CD133− cells was far less tumorigenic (Supporting Information Fig. 1F; Supporting Information Table 2). Furthermore, we also characterized CD133+ cells from primary ovarian cancer tissues in the same fashion and found that these cells also possessed all stem cell properties as tested above (Supporting Information Fig. 2; Supporting Information Table 3). Collectively, our results demonstrate that the CD133+ cell enriched through serum-free selection are genuine ovarian CSLCs: they carry standard CSLC marker and tumor stem cell function; their characteristics are consistent with previous reports [2, 4–6, 9]; and, even after 1 year with culture condition, their “stemness” and tumorogenic capacity are well maintained.
Ovarian CSLCs Carry Stronger Metastatic Capacity Than Non-CSLC Ovarian Cancer Cells
The very traits that are used to define CSCs, self-renewal and tumor-initiating ability, appear to be essential factors of any successful metastasis formation. Recently, a subset of pancreatic CSCs was observed to metastasize to bone after orthotopically transplanted in athymic mice . Also, breast CSCs were shown to be able to invade through Matrigel. These data suggested that CSCs are capable of initiating metastasis, or, at least, that in comparison to non-CSCs, these cells display enhanced metastatic potential [9, 23, 24]. To determine whether our ovarian CSLCs have elevated invasive and migratory capacities, Matrigel invasion assays and transwell migration assays were performed. Our experiments showed that in comparison to the mixed population or CD133− population, CD133+ CSLCs derived from A2780 cells (Fig. 1A) or enriched from primary cancer tissues (Fig. 1B) had a two- to threefold fold advantage in Matrigel invasion and transwell migration. These results strongly supported our hypothesis that CSLCs have enhanced invasive and migratory properties in comparison to non-CSLCs in ovarian cancer.
We continued our study to examine mechanisms that support the superior invasive and migratory capacities of CSLCs. Several studies have demonstrated that chemokines and chemokine receptors are involved in tumor metastasis; thus, we hypothesized that chemokines and their receptors might contribute to the regulation on CSLCs' migration and invasion. We first examined the expression profile of various chemokines and their receptors in ovarian CSLCs and non-CSLCs, both derived from A2780 cells, using a qPCR Array. We found that the expression of several chemokines and chemokine receptors was much higher in CSLCs than in non-CSLCs (Supporting Information Fig. 3); interestingly, many of these molecules (CCL5, CCR1, CCR3, CXCL16) are known to play significant roles in tumor metastasis. To determine whether primary CSLCs also express these chemokines and their receptors, we performed real-time PCR, ELISA, and flow cytometry on CSLCs and non-CSLCs isolated from primary ovarian cancer tissues. We found that several chemokines, including CCL5 and CXCL16, and, chemokine receptors, including CCR1, CCR3, CCR5, and CXCR6, have higher expression in primary CSLCs than that in non-CSLCs; consistent with the results from A2780 derived cells (Fig. 2A–2C; Supporting Information Fig. 3).
CCL5 Is Required for Invasion and Migration of CSLCs
CCL5 and its receptors, including CCR1, CCR3, and CCR5, have been previously reported to play an important role in tumor invasion and metastasis [25, 26]. Interestingly, with cells from the A2780 origin, we found that CSLCs have an elevated level of expression of not only all three CCL5 receptors but also CCL5 itself. In addition, we examined the expression level of CCL5, CCR1, CCR3, and CCR5 within primary ovarian cancer tissues with isolated CSLCs or in situ. CCL5, CCR1, CCR3, and CCR5 are expressed by CSLCs both in vitro (Fig. 2D) and in vivo (Fig. 2E). Furthermore, confocal microscopy revealed colocalization of CCL5 with CCR1, CCR3, or CCR5 on the same cell surface by immunostaining (Fig. 2D), suggesting the possibility that autocrine CCL5 signaling promotes invasion and migration of CSLCs.
To determine the functional importance of CCL5 expression by CSLCs, we took two loss-of-function approaches to test its impact on tumor cell invasion and migration: one is to use function blocking antibody specific to CCL5; the other is to use shRNAs against CCL5 expression. CD133+ or CD133− cells, isolated from primary tumor tissues or derived from A2780 cells, were placed in the upper well and cultured with or without a purified monoclonal neutralizing antibody against human CCL5, which was added to the lower chamber. For CCL5 knockdown experiments, CD133+ or CD133− cells, generated from A2780 cells, were transduced with CCL5-shRNA or control shRNA and placed in the upper well. The efficacy of our shRNA treatment was tested by ELISA: we are able to reduce the level of CCL5 expression by more than fourfold, without any adverse effects on cell viability observed (Supporting Information Fig. 4). The invasion and migration assays showed that both the invasive and migratory capacities of CSLCs, either derived from A2780 cells or isolated from primary ovarian cancers, were significantly decreased by anti-CCL5 antibody blockade (Fig. 3A) or CCL5-shRNA knockdown (Fig. 3D). The addition of anti-CCL5 antibody decreased the number of invasive cells in a dose-dependent manner (Fig. 3B). Pretreatment with 10 ng/ml anti-CCL5 mAb in the lower chamber completely abolished the migration and invasion of CSLCs (Fig. 3B). Surprisingly, in contrast to its effects on CSLCs, anti-CCL5 mAb treatment or CCL5-shRNA knockdown did not have any effect on the migration capacity of CD133− non-CSLCs (Fig. 3A, 3C, and 3D). Moreover, after treatment with rhCCL5, both of the CD133+ and CD133− cells significantly enhanced their migration and invasion capacity in a dose-dependent manner, although CD133− cells to a lesser extent at each dosage in comparison to CD133+ cells (Fig. 3E, 3F). Additionally, our profiling results have also identified the upregulation of another chemokine-receptor pairs in CSLCs: CXCL16 and CXCR6 (Supporting Information Fig. 3A–3C); However, anti-CXCL16 mAb treatment did not have any inhibitory effects on the invasive capacity of CSLCs (Supporting Information Fig. 3D). Collectively, these data indicate that the migration and invasion capacity of ovarian CSLCs, but not non-CSLCs, was predominantly mediated by CCL5 signaling in an autocrine fashion.
The Effect of CCL5 on CSLC Invasion is Mediated by CCR1 and CCR3
It is well-established that the biological effects of CCL5 can be mediated through three CC chemokine receptors—CCR1, CCR3, and CCR5. In addition to increased expression of CCL5, we also detected enhanced expression of all three receptors in CSLCs. We next examined which of these receptors is/are important for the invasion and migration capacities of CSLCs. We treated ovarian CSLCs separately with blocking antibodies specific to CCR1, CCR3, or CCR5. Interestingly, all blocking antibodies significantly suppressed migration and invasion of ovarian CSLCs, whether from A2780 cell line (Fig. 4A) or primary ovarian cancer tissues (Fig. 4E), while CCR5 blocking had less dramatic effects than CCR1 and CCR3 blocking (Fig. 4A). We also verified that the blocking of CCR1 or CCR3 reduces the CD133+ cells in a dose-dependent manner (Fig. 4B, 4C). Moreover, we found that, at the lower dosage range of antibody treatments, combination of anti-CCR1 and anti-CCR3 has a moderate advantage than treating cells with each individual antibody (Fig. 4D).
CCL5 Plays an Important Role in Maintaining the Metastatic Capacity of CSLCs In Vivo
The importance of CCL5 signaling in the migration and invasion of CSLCs in vitro suggests that CCL5 might mediate metastasis of CSLCs in vivo. To investigate whether CCL5 is crucial for CSLCs metastasis, we performed a xenograft experiment, in which CCL5-shRNA or control shRNA transduced CSLCs were injected intraperitoneally into nude mice. CSLCs expressing CCL5-shRNA showed less propensity for metastasis compared with vector transduced CSLCs. Knocking down of CCL5 by shRNA in CSLCs significantly decreased the number of metastatic nodules in the liver, bowel, and spleen in tumor-bearing mice, when compared with vector-shRNA transduced CSLCs (Fig. 5A, 5C, 5D). Furthermore, mice injected with CSLCs transduced with CCL5-shRNA showed a moderate, but, significant reduction on weight loss (Fig. 5B). Collectively, these data suggest that CCL5 signaling in CSLCs is important in its metastasis in vivo.
The Superior Migration and Invasion of CSLCs Is Dependent on the CCL5-CR1/CCR3-NF-κB-MMP-9 Axis
Ligation of chemokines to their receptors induces cell migration via activation of intracellular signaling pathways and the consequent alterations in gene expression. Matrix metalloproteinases (MMPs) are enzymes that degrade the extracellular matrix and play a pivotal role in metastatic processes. MMP-9 has been implicated in many types of cell translocation, including invasion of tissues by tumor cells, T cells, and dendritic cells [26, 27]. Furthermore, it has been postulated in previous studies that MMP-9 is involved in CCL5-directed cell migration [27, 28, 29]. We thus hypothesized that MMP-9 may be involved in CCL5/CCR1- and/or CCR3-mediated CSLC migration and invasion. We first determined the expression of MMP-9 mRNA in CSLCs using qPCR. We found that the expression of MMP-9 in CSLCs, from both A2780 cells and primary cancer tissues, was significantly higher than that in non-CSLCs (Fig. 6A). Moreover, we detected the production of MMP-9 protein by CSLCs in situ using immunofluorescence staining of CD133 and MMP-9 with ovarian cancer tissue sections. Inside the tumor tissue, the production of MMP-9 was not restricted to CD133+ cells, although CD133+ cells are in general associated with strong MMP-9 staining (Fig. 6C). Consistently, MMP-9 mRNA expression was also decreased by anti-CCL5, -CCR1 or -CCR3 blocking antibody, or CCL5-shRNA knockdown in CSLCs (Fig. 6B). In addition, blocking MMP-9 function with specific antibody could also significantly decrease the invasion capacity of CSLCs (Fig. 6D). Taken together, our results suggest that MMP-9 upregulation might be an important mechanism underlying CCL5/CCR1 and/or CCR3-mediated ovarian CSLC invasion.
NF-κB activation was also suggested to be necessary for the migration and invasion of human cancer cells [30, 31]. The ligation of CCL5 to its receptors can activate NF-κB, and upregulation of MMP-9 by NF-κB has been reported in previous studies [32, 33]. Therefore, we speculated that the observed upregulation of MMP-9 by CCL5 in ovarian CSLCs is dependent on NF-κB signaling. To investigate whether NF-κB activation is involved in CCL5-induced CSLC migration and invasion, we first used an ELISA-based assay to quantitate NF-κB binding to its response element. We found that NF-κB activity was significantly higher in extracts from CSLCs, from both A2780 cells and primary ovarian cancer, than in non-CSLCs (Fig. 6E). In addition, we performed three different assays to determine the causality between CCL5 and NF-κB activation: ELISA-based quantification of active NF-κB (Fig. 6F), Western blot-based quantification of NF-κB translocation (Fig. 6G), and immunofluorescence staining-based NF-κB nuclear enrichment (Fig. 6H). All assays demonstrated that NF-κB in CSLCs was partially inhibited when anti-CCL5, -CCR1 or -CCR3 antibody was added, or the cells were transduced with CCL5-shRNA. Furthermore, treatment of CSLCs with PDTC (an NF-κB inhibitor) also suppressed MMP-9 mRNA expression (Fig. 6B). Functionally, an invasion assay in the presence of PDTC revealed that treatment with PDTC significantly decreased the invasion of ovarian CSLCs in a dose-dependent manner (Fig. 6I). These data indicate that an NF-κB inhibitor not only downregulated MMP-9 expression in CSLCs but also significantly decreased CCL5-mediated CSLC migration and invasion. Therefore, we concluded that CCL5/CCR1 and/or CCR3 upregulates MMP-9 via activation of NF-κB, which subsequently enhances ovarian CSLC invasion and migration.
A growing body of evidence suggests that human cancers are diseases of stem cells. CSLCs are capable of persisting in tumors as a distinct population and can cause relapse and metastasis by giving rise to new tumors [34, 35]. In ovarian cancer, the role of CSLCs or CSCs in initiating tumor development has been previously examined [5, 13, 14]. However, to date, the exact surface markers to distinguish ovarian CSCs are still debatable. Some previous studies suggested CD133 as ovarian CSCs marker . Silva et al.  reported that the combination of aldehyde dehydrogenase (ALDH) and CD133 enhance the accuracy for stem cell identification. Here we used CD133+ as the single cellular marker to categorize ovarian tumor cell population. CD133+ cells enriched from primary cancer tissues or derived from A2780 cells were highly tumorigenic and displayed the capacity for self-renewal. These results indicated that CD133 may be used as an independent CSLC marker in ovarian cancer, consistent with previous reports in other type of tumors [10, 16, 21, 37–39]. Functionally, we show here that these CD133+ ovarian CSLCs have significantly stronger migratory and invasive capacity than CD133− tumor cells, which is also consistent with previous reports on CSCs from other types of cancer, such as breast and colorectal tumor [9, 11, 16]. A key characteristic of ovarian cancer is the high rate of distant metastases. Metastasis of ovarian cancer is a complicated multistep process, which is proposed to be related to the migration and invasion of ovarian CSLCs . Our study provided a molecular mechanism to explain how ovarian CSLCs gain their superior invasion and migration capabilities, the most important properties of metastatic cells.
It is clear from clinical studies that chemokines and chemokine receptors play a number of nonredundant roles in cancer metastasis. They are involved in several key steps of metastasis, including adherence of tumor cells to endothelium, extravasation from blood vessels, metastatic colonization, angiogenesis, and proliferation [40, 41]. Our data showed that ovarian CD133+ cells and CD133− cells have distinct patterns of chemokine and chemokine receptor expression. Several chemokines and chemokine receptors showed much stronger expression in CSLCs than in non-CSLCs; notably, many of these molecules have been reported to play important roles in tumor growth and metastasis in various cancers [42–46]. Previously, several groups have identified CXCR4+ CSCs as potential metastatic CSCs both in breast cancer and prostate cancer ; while the IL-8/CXCR1 axis has been reported to play an important role in breast CSCs invasion . In our case, increased expression of CXCR4 (2.88-fold) and IL-8 (4.81-fold) in ovarian CSLCs was also found, but the ligand (CXCL12, also called SDF-1) for CXCR4 and the receptor (CXCR1 and CXCR2) for IL-8 were both expressed at low levels (data not shown). These results suggested that other chemokines and their receptors may be responsible for the maintenance of the migration and invasion capabilities of ovarian CSLCs. Indeed, we found that CCL5 and its receptors (CCR1, CCR3, and CCR5) showed significantly higher expression in ovarian CSLCs than non-CSLCs; these findings and the known roles of these molecules in tumor metastasis led us to focus on the role of CCL5 in CSLC migration and invasion. CCL5 has been defined in several cancers as a pro-malignant factor associated with poor prognosis and enhanced migration through activation of the NF-κB pathway . However, the role of CCL5 in the development of ovarian cancer has not been established. With two independent loss-of-function approaches, we demonstrated that the reduction of CCL5 inhibited the invasion of ovarian CSLCs, but not non-CSLCs. We also examined the influence of exogenous CCL5 on invasion of CSLCs, we found that both of the CD133+ CSLCs and CD133− tumor cells respond to exogenous CCL5 to migrate and invade, although CD133− cells to a lesser extent at each dosage in comparison to CD133+ cells. We hypothesize that this difference is determined by the differential levels of the surface expression for all CCL5 receptors, CCR1, CCR3, and CCR5 are expressed at a higher level on the surface of CD133+ CSLCs than those non-CSLCs. Furthermore, we also demonstrated that autocrine CCL5 signaling plays a pivotal role in CSLC metastasis in vivo. Besides the autocrine pathway identified in this study, it has been previously shown that CCL5 can be produced by mesenchymal stem cells (MSCs) within tumor stroma and it plays a crucial role in breast cancer invasion . Furthermore, activated CD8+ T cells and other immune cells infiltrated in the tumor stroma could produce CCL5 as well . Therefore, potentially, within the primary tumor microenvironment, CCL5 produced by CSLCs, MSCs, and immune cells, all contribute to the tumor metastasis. However, the timing and the extent of their contribution remain to be determined.
It is well-established that CCL5 mediates its biological activities through activation of the G-protein-coupled receptors CCR1, CCR3, and CCR5 . We observed expression of all the CCL5 receptors (CCR1, CCR3, and CCR5) in ovarian CSLCs. Quantitatively, CCR1 and CCR3 are more important than CCR5 in CCL5-mediated ovarian CSLC invasion and migration. Also, as proposed for CXCR4 for breast and prostate CSCs, our results suggest that, potentially, CCR1 and CCR3 could be used as markers for metastatic ovarian CSLCs. Furthermore, clinically, our results, together with Karnoub's findings , suggest that a variety of CCL5 analogs and CCR5 antagonists might have therapeutic potential in preventing or treating metastatic disease. This is an ongoing research to follow-up our current study.
We identified that the production of MMP-9 and the activation of transcription factor NF-κB are necessary for the migration and invasion of CD133+ CSLCs. There are NF-κB binding sites in the MMP-9 promoter region. We validated that the inhibition of NF-κB suppresses MMP-9 production and CCL5-CCR1/CCR3-mediated invasion and migration. Reciprocally, we also found that NF-κB activity can be partially inhibited by anti-CCL5, -CCR1, or -CCR3 blocking antibody or shRNA-mediated CCL5 knockdown. These data provided the first evidence that NF-κB-mediated MMP-9 upregulation might be the underlying mechanism of CCL5/CCR1- and/or CCR3-mediated ovarian CSLC invasion. Taken together, our findings indicated an attractive new avenue for therapeutics that may target metastatic CSLCs specifically, at least for ovarian cancer, via blocking the important signaling node in the CCL5/CCR1-CCR3/NF-κB/MMP-9 axis.
CSCs have the significant capabilities of metastasis and the previous studies mostly focused on the effects of factors mediated by tumor microenviroments, including chemokoines produced by inflammatory cells, such as macrophages and stromal cells. However, how CSCs maintain their invasion and migration abilities by itself has not been reported. In the present study, we firstly found ovarian CSCs either in vitro or in situ tumor not only produced a large body of CCL5, but also expressed abundant its receptors (CCR1, CCR3 and CCR5), indicating the maintenance of metastatic capabilities of ovarian CSCs might be regulated by autocrine CCL5. Most importantly, two loss-of-function of CCL5 and blocking of CCR1 or CCR3 effectively inhibits the invasive capabilities of ovarian CSCs, suggesting autocrine CCL5 could promote invasion and migration of ovarian CSCs. Finally, we show that the binding of self-production of CCL5 with CCR1 and CCR3 on ovarian cancer stem cells leading to NF-kappa B activation and MMP-9 upregulation is the main mechanism of metastatic property of ovarian cancer stem cells. In summay, our findings indicate that CCL5-CCR1/3 axis could be used as the therapeutic target to inhibiting CSCs-mediated ovarian metastasis.
We thank Dr. Symonds A.L.J. (Institute of Cellular and Molecular Science, University of London, U.K.) for reading and editing this manuscript. This work was supported by National Nature Science Foundation of China (No. 30901592 and No. 81071772), by the outstanding Youth Scientist Foundation of Chongqing (No. CSTC, 2008BA5035), by Century Excellent Talents in University by Ministry of Education of China (NCET-O8-0935), and by National Key Basic Research Program of China (973 program, No. 2010CB529404).
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.