Fractalkine Receptor CX₃CR1 Is Expressed in Epithelial Ovarian Carcinoma Cells and Required for Motility and Adhesion to Peritoneal Mesothelial Cells

Epithelial ovarian carcinoma (EOC) is a relatively rare but highly lethal gynecologic malignancy that claims the lives of more than 14,000 women in the United States and more than 140,000 worldwide (1). EOC is seldomly diagnosed at early stages, when the rate of survival is nearly 90%. Instead, most patients are diagnosed with EOC at late metastatic stages (FIGO stages III–IV; ref. 2). Treatment of advanced and metastatic EOC cases is clinically difficult. The current standard of care includes surgery and chemotherapy with platinum-based drugs (3). However, many EOC cases frequently recur and become insensitive to these therapies (4, 5). Improvement of the existing approaches and development of new therapies to be used, either alone or in combination, with the existing treatments are necessary to prolong and improve patient survival.

Metastatic EOC spreads throughout the peritoneum largely by sloughing off of the ovary and subsequently attaching to and invading the organs and tissues of the peritoneal cavity. Dissemination through blood and lymphatic vessels, however, is minimal (2, 6). Malignant EOC cells are shed from the ovarian surface and later attach to the mesothelial layer outlining the peritoneum (2, 7–10). This spreading pattern suggests that the local microenvironment of the peritoneal cavity may contain signals that support homing of the EOC metastasis. Patients with EOC frequently accumulate large volumes (0.5–4 L) of peritoneal ascitic fluid. This fluid is a rich source of growth factors, extracellular matrix proteins, chemokines, and other factors that may support and promote peritoneal metastasis (11–13).

The development of new approaches to treat metastatic ovarian carcinoma and the identification of novel pathways for drug targeting could be bolstered by the identification of highly efficacious treatments against molecules involved in metastasis. The goal of this study was to identify methods to prevent or retard metastatic dissemination. Given that identifying new drug target pathways is our priority, we took a focused approach to ascertain the molecules that are most critical for promoting metastasis and can be targeted therapeutically. Chemokine receptors encompass a subclass of G protein–coupled receptors (GPCR), which in turn represent a preferred class of drug targets (14). Chemokine receptors show promise for development as therapeutic targets because they have been successfully targeted in the clinic (15) and play a role in homing of metastatic cells to their niches (16–19); however, beyond CXCR4, their role in the development and progression of EOC is largely unknown (11, 20, 21). It is possible that ovarian cancer antichemokine receptor antagonists could be useful in preventing metastasis in patients with early-stage disease, as well as in restricting metastatic spread in patients presenting with advanced stages of malignancy.

We were interested in determining whether chemokine networks that have not yet been implicated in EOC could be important for progression of this malignancy. Preliminary reverse transcriptase PCR (RT-PCR) array data in our laboratory suggested that CX₃CR1, a member of the chemokine family of GPCRs, is expressed in cell lines propagated from acute EOC samples. CX₃CR1 is a member of the 7-transmembrane CX₃C chemokine receptor group family (22). CX₃CL1, or fractalkine, is the major, highly specific ligand that activates CX₃CR1 (23, 24). This robust ligand specificity suggests that CX₃CR1 could be an attractive target for therapeutic interventions. Interestingly, it has been suggested that CX₃CL1 can also activate another surface receptor, a receptor tyrosine kinase termed epidermal growth factor receptor (EGFR), to stimulate cellular proliferation (25). It has also been reported that poor prognoses and EOC progression correlate with the accumulation of CX₃CR1-positive immunosuppressive leukocytes (26). Because CX₃CR1-positive immunosuppressive leukocytes are attracted to the peritoneal milieu to suppress the native immune system, these findings may suggest that the CX₃CR1-positive EOC cells that are also present in the peritoneum may escape clearance by the native immune system. Escape from immune surveillance and use of immune cell secretions for tumor progression are common features of metastatic tumors (27–31). However, in colon cancer expression of CX₃CL1 correlated with a better prognosis and associated with high densities of T lymphocytes (32).

Here, we show that CX₃CR1 expression is nearly absent in the normal ovarian surface epithelium and that its expression accumulates during the course of tumorigenesis. We also report that CX₃CL1 is present in the ascites of women with EOC and benign gynecologic conditions. Furthermore, we show that a functional interaction between CX₃CL1 and CX₃CR1 facilitates cell migration and cell–cell adhesion between EOC cells and peritoneal mesothelial cells. CX₃CL1 is also required for EOC cell proliferation. Because cell migration, proliferation, and peritoneal seeding are crucial for the formation and development of EOC metastasis, our data suggest that fractalkine signaling in EOC may contribute to metastatic progression.

Tissue microarray (TMA) slides containing specimens of normal ovary and primary and distant metastatic ovarian carcinoma (catalog no. OV208), specimens of normal ovary (catalog no. OV806), specimens of serous cystadenoma (catalog no. OV603), and specimens of multiple types of ovarian carcinoma (catalog no. OV1002) were obtained from US Biomax. cDNA arrays prepared from normal ovary and primary ovarian carcinoma specimens were obtained from Origene Technologies. CX₃CL1 was obtained from RayBiotech. Human collagen type I, type III, bovine serum albumin (BSA), fura-2 AM, and ATP were purchased from Sigma-Aldrich. WST-1 proliferation kits were obtained from Takara Bio. CX₃CR1 siRNA, control siRNA (#A), a goat anti-human CX₃CR1 antibody (K-13), a rabbit anti-human EGFR antibody, a mouse anti-human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody, a rabbit anti-human proliferating cell nuclear antigen (PCNA) antibody, and total cell lysates from THP-1 and A431 were obtained from Santa Cruz Biotechnology. DharmaFECT was obtained from Dharmacon. Matrigel was obtained from BD Biosciences. Transwell cell migration chambers with 8-micron pores were purchased from Corning. AG1478 was obtained from Cayman Chemicals. Rabbit polyclonal CX₃CR1 (catalog no. ab8020 and ab8021) antibodies and normal rabbit IgG were obtained from Abcam. A fluorescein isothiocyanate (FITC)-conjugated secondary anti-rabbit IgG was purchased from Invitrogen. Mouse monoclonal anti-human CX₃CL1 was procured from R&D Systems, and rabbit anti-human CX₃CL1 was purchased from BioVision. Mouse monoclonal anti-human CD44 and mouse monoclonal anti-β1–integrin (MAB2253) were obtained from Millipore. Vectastain ABC and 3,3′—diaminobenzidine (DAB) kits were obtained from Vector Laboratories.

Two human ovarian carcinoma cell lines that originated from malignant cells in the ascites fluid and presented a serous histotype, SKOV-1 and A2780, were obtained from the National Cancer Institute (NCI) Tumor Cell Repository (Detrick, MD). These cell lines were cultured as directed by the manufacturer for no longer than 20 consecutive passages. Another human ovarian carcinoma cell line originating from malignant cells in the ascites fluid and presenting a serous histotype, Caov-3, was obtained from Dr. M.S. Stack (University of Missouri-Columbia, MO) and propagated in minimal essential media supplemented by 10% FBS for no longer than 15 consecutive passages. The human ovarian carcinoma cell line with a serous histotype and originating from a primary tumor, IGROV-1, was obtained from the NCI Tumor Cell Repository (Detrick, MD) and cultured as directed by the manufacturer for no longer than 20 consecutive passages. Two borderline EOC cell lines (HuIOSBT-1.5 and HuIOSBT-3.3) and 2 immortalized human ovarian surface epithelial cell lines (IOSE29 and IOSE80) were obtained from Dr. N. Auersperg through the Canadian Ovarian Tissue Bank. The immortalized human ovarian surface epithelial cell line HOSE11-12 was obtained from Dr. G.S.W. Tsao (The University of Hong Kong, PRC) and propagated in minimal essential media supplemented with 10% FBS for 5 consecutive passages. The immortalized human ovarian surface epithelial cell line T1074 was obtained from Applied Biological Materials and propagated as directed by the manufacturer. The normal (not immortalized) human ovarian surface epithelial cell line HOSEpiC was purchased from ScienCell Research Laboratories and maintained in the conditions suggested by the manufacturer between passages 1 and 2. The human immortalized peritoneal mesothelial cell line LP-9 was obtained from the Coriell Aging Cell Repository and cultured as indicated by the manufacturer for 5 to 8 passages. All cell lines were routinely assessed for cellular morphology and average doubling time. All cell lines were propagated from stocks originally obtained from cell banks and individual investigators and have been stored in aliquots for future use. Each aliquot was further propagated for no longer than 20 consecutive passages or 4 months, whichever came first.

Real-time PCR was carried out using the ABI Prizm system (Applied Biosystems) according to the manufacturer's instructions. The following primers were used for detection of CX₃CR1 mRNA in normal ovary and primary EOC samples (obtained from Origene Technologies): CX3CR1FORW 5′-CACAAAGGAGCAGGCATGGAAG-3′; CX3CR1REV 5′-CAGGTTCTCTGTAGACACAAGGC-3′. Primers for beta-actin detection (ACTNB) were supplied with the cDNA sample panel (Origene). The iTaq SYBR Green Supermix (Bio-Rad) was used for quantitative PCR as a double-stranded DNA-specific fluorophore. The PCR reaction conditions were as follows: initial denaturation for 10 minutes at 95°C followed by 40 cycles of 94°C for 15 seconds and 60°C for 30 seconds. To determine the specificity of the PCR primers, melting curves were collected by heating the products to 95°C, cooling them to 65°C, and slowly melting them by increasing the temperature 0.5°/sec up to 95°C. Primers for mRNA detection of our genes of interest were designed using the Primer3 software according to the requirements for real-time RT-PCR oligonucleotide primers.

TMA slides were deparaffinized by baking at 60°C for 2 hours and rehydrated by incubation in xylene, 100% ethanol, 95% ethanol, 70% ethanol, and PBS, pH 7.4, for 5 minutes each. Peroxidase activity was inhibited with H₂O₂. Antigen retrieval was carried out by a 15-minute incubation in 1 mmol/L EDTA (pH 8.0) at 95°C. Prior to primary antibody staining [1:50 dilution (Abcam catalog no. ab8021), 1 hour at room temperature] nonspecific binding was blocked by incubation with 10% goat serum for 1 hour. The biotin-conjugated goat anti-rabbit secondary antibody was used at a 1:200 dilution for 30 minutes at room temperature. The Vectashield ABC kit was used as directed by the manufacturer, and tissues were incubated for 45 minutes at room temperature. The DAB reagent was prepared as instructed by the manufacturer and applied to tissues on TMA slides for 2 to 10 minutes until brown color developed. TMAs were stained with hematoxylin, dehydrated in 50%, 70%, 95%, and 100% ethanol, and mounted with Permount. Pancreatic cancer tissue was used as a positive control. Staining was evaluated by A.A.K.-B. and L.R., who were both blinded to the experimental outcomes of the study. Staining was scored on the basis of the intensity and percentage of positive cells. Intensity of staining was “0” for negative samples, “1” for weakly positive samples, “2” for moderately positive samples, and “3” for highly positive samples. Overall scores were derived as the intensity score multiplied by the percentage. Staining was assessed separately in the cytoplasm and the membrane.

Cells (1 × 10⁶ per tube) were harvested with trypsin/EDTA, resuspended in 100 μL of ice-cold PBS supplemented with 10% fetal calf serum and 1% sodium azide, and fixed in a 1% paraformaldehyde solution in PBS for 15 minutes on ice. Two micrograms of rabbit anti-human CX₃CR1 antibody (Abcam, ab8021) was added to the cells. For negative controls, the cells were incubated with either 2 μg of anti-rabbit IgG antibody or no primary antibody. Cells were incubated for 1 hour on ice in the dark with agitation following washing and resuspension in 400 μL ice-cold PBS. Two micrograms of goat anti-rabbit FITC-conjugated IgG (Invitrogen) was added to the cells and incubated for 1 hour on ice in the dark with agitation. The cells were washed and resuspended in 400 μL of ice-cold PBS supplemented with 2% BSA and 1% sodium azide. Labeled cells were analyzed using a Becton Dickinson LSR I flow cytometer or an Accuri C6 flow cytometer on the same day.

Western blotting analysis was used to detect the expression of CX₃CR1 and β-tubulin in normal (not immortalized) ovarian surface epithelial cells, immortalized ovarian surface epithelial cells, and EOC cell lines. This procedure was carried out as previously described (33–35). Antibodies were used at the following dilutions: 1:500 rabbit anti-human–CX₃CR1 (Abcam, ab2080) in a 1× non-animal blocker solution (Vector Laboratories) or 1:200 goat anti-human CX₃CR1 (Santa Cruz Biotechnology, K-13) in 3% BSA in a solution of 50 mmol/L TBS, pH 7.4, 150 mmol/L NaCl, and 0.05% Tween-20 (TBST; Sigma), 1:200 mouse anti-human GAPDH (Santa Cruz Biotechnology) and 1:200 mouse anti-human–β-tubulin (Developmental Studies Hybridoma Bank, Iowa) in 3% BSA in TBST. Immunoreactive bands were visualized with an anti-(rabbit-IgG)-peroxidase, anti-(goat-IgG)-peroxidase, or anti-(mouse-IgG)-peroxidase (Sigma; 1:1,000 in 3% BSA in TBST), and enhanced chemiluminescence was read using Chemidoc (Bio-Rad) and Bio-Rad Chemidoc ImageReader software.

Intracellular free calcium concentrations were measured using digital video microfluorimetry as previously described (11). Briefly, SKOV-3 cells were plated on collagen I–coated glass coverslips, rinsed with HEPES buffer, exposed to 10 μmol/L fura-2 AM (Molecular Probes), and then incubated for 30 minutes at room temperature. Complete dye deesterification was achieved by removing the fura-2 AM, rinsing and maintaining the cells in the dark for 30 minutes. Glass coverslips were mounted on the stage of a Zeiss AxioObserver inverted epifluorescence microscope equipped for digital fluorescence microscopy. Fluorescence was digitally monitored at 520 nm after excitation at 340 nm (bound Ca²⁺) and 380 nm (free Ca²⁺). Ratios of F340/F380 were collected before and during treatment with 20 nmol/L CX₃CL1 using Zeiss software, exported in Excel software and plotted.

Cells were cultured on glass coverslips to 50% to 70% confluence, fixed, permeabilized with 0.01% Triton X-100 and blocked in goat serum. Rabbit anti-human-CX₃CR1 (Abcam, ab8021), rabbit anti-human-CX₃CL1 (BioVision), and rabbit anti-human PCNA (Santa Cruz Biotechnology) antibodies were used at a dilution of 1:50. Cells were incubated with the primary antibodies for 1 hour at room temperature (22°C). The cells were then incubated with secondary Alexa 488–conjugated or Alexa 555–conjugated anti-rabbit antibodies (1:500) for 1 hour at room temperature in the dark. 4′,6-Diamidino-2-phenylindole (DAPI) was added to the secondary antibody solution to a final concentration of 10 μg/mL 10 minutes prior to the end of the incubation period. Cells were washed, air dried, and mounted on glass slides using ProlongGold (Invitrogen). Fluorescent imaging was done using a Zeiss AxioObserverD.1 fluorescent microscope.

EOC cells were cultured to 80% confluence and transfected with siRNAs using DharmaFECT (ThermoScientific) according to the manufacturer's instructions.

Inserts with 0.8-micron porous membranes were bottom-coated with Matrigel (diluted 1:100) for 1 hour at 37°C, rinsed, and dried. Next, 5,000 EOC cells in a final volume of 300 μL were seeded onto the inserts. The inserts were placed in 24-well plates filled with serum-free minimal essential media in the presence or absence of CX₃CL1. Cells were allowed to migrate for 5 hours at 37°C and 5% CO₂. Migration was terminated by removing the nonmigrated cells from the inside of the inserts. Cells that did migrate through the membranes were fixed, stained, and counted.

Ascites samples from 6 patients with benign (n = 3) or malignant (n = 3) ovarian disease were obtained from ProteoGenex and analyzed for CX₃CL1 using Human CX₃CL1 Quantikine ELISA (R&D Systems) according to the manufacturer's instructions.

The human-derived peritoneal mesothelial cell line LP-9 was cultured in 96-well plates to near confluence. SKOV-3 cells were cultured in monolayers and labeled with fluorescent DiO (Invitrogen) according to the manufacturer's instructions prior to the adhesion assays. They were subsequently released from monolayers with trypsin and resuspended in serum-free cell culture media. When needed, SKOV-3 cells were transiently transfected with either control or CX₃CR1-specific siRNAs and used in adhesion assays between 48 and 72 hours from the start of transfection. For the assay, 10,000 DiO-labeled SKOV-3 cells were seeded onto LP-9 monolayers (in sextuplicate for each condition) and allowed to adhere for 5 hours at 37°C and 5% CO₂. Subsequently, the monolayers were washed 2 times with PBS and fixed in a methanol-containing cell fixative. When required, LP-9 cells were pretreated with CX₃CL1 blocking antibodies for 30 minutes, and SKOV-3 cells were pretreated with either normal anti-rabbit IgG, anti-CD44, or anti-β1–integrin blocking antibodies for 30 minutes at 37°C and 5% CO₂. Adherent cells showing a round cell morphology were visualized by green fluorescent signals using a Zeiss fluorescent microscope. The DiO-labeled cells were then counted, averaged, and characterized as a percentage of adherent cells.

Tissue culture–treated 48-well plates were precoated with 10 μg/mL human collagen type I, 10 μg/mL human collagen type III, Matrigel (diluted 1:100), or 1% BSA for 1 hour at 37 °C. The plates were subsequently rinsed with PBS and air dried. Next, 10,000 SKOV-3 cells were seeded (in triplicate for each condition) in coated wells and allowed to adhere for 1 hour at 37°C and 5% CO₂. This seeding was followed by 2 washes with PBS, fixation in a methanol-containing cell fixative, and staining. Cells were counted, averaged, and plotted.

Cells were seeded in tissue culture–treated 96-well plates and allowed to attach for 6 hours following overnight serum starvation (0% serum) at 37°C in 5% CO₂. Cells were plated at a density that yielded confluence below 25% at the beginning of the treatment. Cells were seeded in triplicate for each condition. To assess the proliferative effect of CX₃CL1 on EOC cells, cells were treated with 5, 15, 25 nmol/L CX₃CL1, or 25 nmol/L CX₃CL1 with 5 μmol/L AG1478 for 24 hours. Cells treated with complete media containing 10% FBS were used as a positive control. Cells treated with serum-free media served as a negative control. WST-1 was used according to the manufacturer's instructions and incubated with the cells for 1 hour following measurement of OD₄₄₀. A baseline OD₄₄₀ reading of the mixture of media and WST-1 containing no cells was subtracted from the experimental values.

The mechanisms of malignant transformation and metastatic progression of EOC are poorly understood. Current hypotheses have proposed the ovarian surface epithelium and the fimbriated edge of the fallopian tube as putative cellular origins of the carcinoma (36, 37). It has recently been hypothesized that these 2 epithelia are part of a transitional epithelium that shares a common origin and is prone to neoplastic transformation due to incomplete differentiation (38). Another unique feature of EOC that is particularly pertinent to this study is the pattern of metastatic dissemination, which is largely intraperitoneal. The chemokine axis has been associated with homing of malignant cells into their metastatic niches. We hypothesized that the fractalkine axis could play a role in the peritoneal dissemination of EOC. Therefore, we evaluated CX₃CR1 expression in normal and pathologic ovarian surface epithelia.

CX₃CR1 expression in normal ovarian surface epithelium was assessed by 4 methods. Immunohistochemistry was used to detect CX₃CR1 protein expression in normal ovary specimens. CX₃CR1 expression was also measured in normal ovarian surface epithelial cells (HOSEpiC) using immunofluorescence staining, flow cytometry, and Western blot. The majority of the normal epithelial cells (21 cases of 22) were negative for membranous CX₃CR1 expression (Fig. 1A, Supplementary Table S1). Furthermore, 77% of the tested samples (17 of 22) lacked CX₃CR1 cytoplasmic staining (Supplementary Table S1). Samples were considered negative if the total score was below 100, weakly to moderately positive if the score was between 100 and 200, and strongly positive if the score was above 200. CX₃CR1 expression was not detectable in HOSEpiC cells by immunofluorescence staining, flow cytometry, or Western blot (Fig. 1B and C, Fig. 3A).

Next, we evaluated CX₃CR1 expression in immortalized ovarian surface epithelium. Interestingly, all of the immortalized ovarian surface epithelial cell lines tested (T1074, HOSE11-12, IOSE29, and IOSE80) were CX₃CR1 positive, as shown by flow cytometry and Western blot analysis (Fig. 2A, Fig. 3A). These findings suggest that induction of CX₃CR1 expression can occur very early in the tumorigenic transformation process.

Subsequently, we evaluated CX₃CR1 expression in specimens of a benign gynecologic condition, ovarian serous cystadenoma, and established cell lines originating from cases with borderline EOC. Flow cytometry also detected positive membranous expression of CX₃CR1 in the borderline EOC cell lines HuIOSBT-1.5 and HuIOSBT-3.3 (Fig. 2B). Immunohistochemical analysis of 30 cases of ovarian serous cystadenoma revealed that 83% of the specimens were CX₃CR1 positive in the epithelium (Fig. 2C, Supplementary Table S1). A small number of serous cystadenoma cases also showed extensive CX₃CR1 immunoreactivity. Because the clinical prognosis for these cases is excellent, the significance of this finding is unknown.

Furthermore, we utilized RT-PCR and immunohistochemistry to test the RNA and protein expression levels of CX₃CR1 in primary ovarian carcinoma specimens of various histotypes. Our findings revealed that 66% and 50% of the primary EOC cases with a serous histotype were CX₃CR1 positive by RT-PCR (19 of 29) and immunohistochemistry (12 of 24), respectively (Supplementary Tables S1 and S2, Fig. 3). Among mucinous EOC cases, 3 of 3 and 13 of 24 cases were CX₃CR1 positive as shown by RT-PCR and immunohistochemistry, respectively (Supplementary Tables S1 and S2, Fig. 3D). RT-PCR analysis identified 5 CX₃CR1-positive cases of endometrioid EOC of 8 tested (Supplementary Table S2). Notably, nearly all of the cases that received a negative classification by immunohistochemistry did contain some CX₃CR1-positive cells (Supplementary Table S1).

Ovarian carcinoma cell lines derived from primary EOC (IGROV-1) or ascites (Caov-3, SKOV-3, A2780) all showed CX₃CR1 expression, as shown by Western blot, immunofluorescence staining, and flow cytometry (Fig. 3). Quantification by ImageJ software revealed that the intensity of the CX₃CR1 signal in these samples was significantly greater than the background expression observed in pseudopodial protrusions of the SKOV-3 EOC cell line (Fig. 3A, graph). SKOV-3 cells responded to the addition of CX₃CL1 by a robust release of intracellular calcium, which is indicative of the ability of CX₃CR1 to support downstream signaling in these cells (Fig. 3B).

We tested specimens of metastatic EOC by immunohistochemistry and found that 64% of the specimens (7 of 11) were CX₃CR1 positive (Fig. 3D, Supplementary Table S1). However, nearly all of the samples that were considered negative by immunohistochemistry did contain some CX₃CR1-positive cells (Supplementary Table S2).

Together, our data suggested that CX₃CR1 expression may be induced very early in the course of the tumorigenic transformation of the ovarian surface epithelium.

CX₃CL1 is expressed in many cells and tissues, including the small and large intestine, ovary, and kidney (23, 24, 39, 40). CX₃CL1 exists in either a membrane-inserted form (23, 24) or a soluble form that has been cleaved to facilitate release (41). Peritoneal ascites is a rich source of many compounds, including growth factors, components of the extracellular matrix, and chemokines. Soluble CX₃CL1 is present in the low nanomolar range in the ascites of women with endometriosis (42). We therefore measured CX₃CL1 levels in the ascites of women with EOC and women with borderline gynecologic conditions. By ELISA, we found that the average concentration of CX₃CL1 in the ascites from stages III–IV EOC patients is 2.8 nmol/L. The level is approximately 10-fold lower, 0.18 nmol/L, in the ascites collected from patients with borderline gynecologic diseases (Table 1).

Cell migration is essential for metastatic dissemination. Chemokine networks and CX₃CL1/CX₃CR1, in particular, have been implicated in the chemotactic migration of tumor cells to distant sites. The presence of soluble CX₃CL1 in the ascites fluid may indicate that this chemokine could serve as an attractive force that guides EOC cells to the peritoneum. Therefore, we tested the ability of CX₃CL1 to attract EOC cells in a Transwell cell migration assay. The addition of 5 nmol/L CX₃CL1 significantly and robustly stimulated the migration of both Caov-3 and SKOV-3 cells by 1.7- and 2.2-fold, respectively (Fig. 4A). Silencing of CX₃CR1 with specific siRNAs reduced the ability of SKOV-3 cells to migrate in response to CX₃CL1, to a level that was nearly identical to the control (no CX₃CL1; Fig. 4A). We previously reported a 1.5-fold increase in CXCL12/CXCR4-mediated invasion of EOC cells in the presence of CXCL12 (11). The CXCL12/CXCR4 axis is a well-established driver of metastatic dissemination in EOC (11, 20, 21); in comparison, the CX₃CL1/CX₃CR1 pathway makes an even greater contribution to cell migration.

We next tested the ability of SKOV-3 cells with silenced CX₃CR1 to migrate toward human ascitic fluid obtained from a patient with stage IV EOC (specimen #5, Table 1). Notably, cells with silenced CX₃CR1 displayed a 70% reduction in their ability to migrate compared with control cells expressing CX₃CR1 (Fig. 4B). CX₃CR1 siRNA treatment of SKOV-3 cells reduced the total CX₃CR1 expression to 70% (Fig. 4B).

CX₃CR1-negative HOSEpiC cells did not display altered migratory ability in the presence of CX₃CL1 (Fig. 4C). Interestingly, CX₃CR1-positive T1074 cells did not migrate significantly faster toward CX₃CL1 than controls (Fig. 4C). Together, these results show that CX₃CR1-positive EOC cells show an increased migration ability when stimulated with CX₃CL1 in vitro, whereas immortalized ovarian surface epithelium does not.

EOC metastatic dissemination occurs largely in the peritoneal cavity. Several molecular interactions, including integrin–ECM, CD44–hyaluronan, CA125–mesothelin, and L1–neuropilin-1 have been shown by both in vitro and in vivo models to facilitate peritoneal seeding of EOC cells. The CX₃CL1/CX₃CR1 axis has also been shown to promote cell–cell adhesion via the membrane-bound form of CX₃CL1 (22). Thus, we hypothesized that the CX₃CL1/CX₃CR1 interaction could foster EOC–mesothelial adhesion and, thereby, facilitate metastatic seeding. Prior to confirming this interaction, we determined CX₃CL1 expression in human-derived peritoneal mesothelial LP-9 cells. CX₃CL1 was expressed by the LP-9 cells and present on the cell surface (Fig. 5A). Because membrane-anchored CX₃CL1 is present on mesothelial cells, and CX₃CR1 is expressed on EOC cells, this interaction is feasible. To test the role of CX₃CL1/CX₃CR1 in EOC–mesothelial cell adhesion, we carried out in vitro adhesion assays by coculturing EOC and mesothelial cells. LP-9 cells were cultured to near complete monolayers. DiO-labeled EOC cells SKOV-3 were seeded on top of the LP-9 monolayers. We found that 90% to 93% of SKOV-3 cells adhered to the LP-9 monolayer (Fig. 5B, images 1–2, Supplementary Table S3). Interestingly, siRNA-mediated downregulation of CX₃CR1 resulted in an approximately 50% loss of EOC cell adhesion to the LP-9 cells (Fig. 5B, image 3, Supplementary Table S3). Moreover, progressive inhibition of CX₃CL1 by pretreating LP-9 monolayers with increasing doses of anti-CX₃CL1 antibodies (0.1–0 μg/mL) prior to the addition of EOC cells resulted in a 50% to 83% decrease in the ability of EOCs to adhere (Fig. 5B, images 5–7, Supplementary Table S3). Addition of a nonspecific IgG did not affect adhesion (Fig. 5B, image 4, Supplementary Table S3). In control experiments that tested the consequences of combined inhibition by anti-CD44 and anti-β1–integrin antibodies, the ability of SKOV-3 cells to adhere to LP-9 cells was reduced by 40% to 50% (Fig. 5B, images 8–9, Supplementary Table S3). Downregulation of CX₃CR1 or the addition of CX₃CL1 blocking antibodies to CD44- and β1–integrin blocking antibody treatment further exacerbated the combined effect and resulted in a nearly 70% reduction in SKOV-3–LP-9 adhesion (Fig. 5B, panels 10–11, Supplementary Table S3).

These data suggested that the CX₃CL1/CX₃CR1 axis plays a significant role in the adhesion of EOC cells to mesothelial cells.

After adhesion to mesothelial cells, EOC cells disrupt the cellular monolayer, expose extracellular proteins of the basement membrane, and further invade the submesothelial matrix, which is primarily composed of interstitial collagens type I and III (43). Chemokine networks and the CX₃CL1/CX₃CR1 axis, in particular, have been shown to promote cell–ECM adhesion (44, 45). Therefore, we hypothesized that this axis could play a role in EOC adhesion to proteins of the submesothelial matrix. To address this question, we tested the ability of SKOV-3 cells to adhere to Matrigel, collagen I, and collagen III. Surprisingly, addition of CX₃CL1 resulted in a slight, but significant, concentration-independent 1.5-fold increase in cell adhesion to Matrigel but not collagens I or III (results not shown). These findings indicate that CX₃CL1/CX₃CR1 plays a small role in promoting ECM adhesion in the initial steps of EOC cell interaction with the proteins of the basement membrane, but it is unlikely to affect adhesion to the deeper layers composed of the collagenous matrix.

Chemokines are effective inducers of cell proliferation (46). Proliferation plays a prominent role in the virulence of ovarian carcinoma. Metastatic progression of EOC is characterized by a widespread metastasis that far exceeds the size of the primary tumor. Furthermore, the size of the residual tumor after debulking surgery plays a significant role in determining patient survival (47). Our data and previously published results show that high levels of CX₃CL1 are present in the peritoneal milieu in the ascites of patients with various malignant and nonmalignant gynecologic diseases (Table 1; ref. 42). Therefore, we hypothesized that CX₃CL1 could induce the proliferation of EOC cells. We tested CX₃CL1 in a range of concentrations between 5 and 25 nmol/L and found that the majority of immortalized OSE, borderline, and serous EOC cell lines showed a significant proliferative response to CX₃CL1 stimulation. In several cell lines, including T1074, HuIOSBT-1.5, SKOV-3, and IGROV-1, the addition of CX₃CL1 completely rescued cellular proliferation to a level equal to or higher than the rate of cells stimulated with complete media containing 10% FBS (Fig. 6A and B). We also observed a strong induction of nuclear PCNA 24 hours after the addition of 25 nmol/L CX₃CL1 to serum-starved SKOV-3 cells (Fig. 6B). However, some cell lines, including the immortalized cell line IOSE80 and the borderline EOC cell line HuIOSBT-3.3, showed only slight changes in proliferation in response to CX₃CL1 (Fig. 6A). CX₃CL1 can partner with both EGFR and CX₃CR1 to induce cell proliferation (25, 40). Therefore, to assess the contribution of EGFR to CX₃CL1-dependent proliferation, we incubated each of our cell lines with the EGFR inhibitor, AG1478. Interestingly, CX₃CL1-dependent proliferation of all EOC cell lines tested was either inhibited completely or significantly reduced in the presence of AG1478 (Fig. 6A). For a more detailed understanding of the role of both CX₃CR1 and EGFR in CX₃CL1-induced proliferation of EOC cells, the receptors were transiently downregulated using specific siRNAs. A proliferation assay revealed that a 70% decrease in CX₃CR1 led to a nearly 50% reduction in CX₃CL1-dependent proliferation. Moreover, a 95% decrease in EGFR resulted in a nearly 90% reduction in CX₃CL1-dependent proliferation (Fig. 6C).

EOC is the most predominant type of ovarian carcinoma. However, ovarian tumors can also arise in stromal cells. Thus, to comprehensively evaluate CX₃CR1 expression in ovarian carcinoma, we evaluated specimens of Krukenberg tumors, yolk sac tumors, dysgerminomas, mature teratomas, Sertoli-stromal cell tumors, granular cell tumors, thecomas, and gynandroblastomas. Our data indicate that the majority of specimens were CX₃CR1 positive (Supplementary Fig. S1, Supplementary Table S4). Specifically, 7 of 10 Krukenberg, 8 of 8 yolk sac, 8 of 8 dysgerminoma, 13 of 13 mature teratomas, 1 of 1 Sertoli-stromal cell, 13 of 15 granular cell, 14 of 14 thecomas, and 1 of 1 gynandroblastoma tumor specimens were CX₃CR1 positive, suggesting that fractalkine signaling may affect the progression of all of these tumors. Importantly, these data show that cancers beyond those of epithelial origin may depend on fractalkine signaling.

The mechanisms underlying peritoneal dissemination of ovarian carcinoma are poorly understood. Ovarian carcinoma is a malignancy with an exceptionally high mortality rate (1), largely due to the lack of effective antimetastatic treatment approaches. A more detailed understanding of the mechanisms underlying the formation and development of EOC metastases could offer insights into how late stages of this disease might be effectively targeted. Proteins that are presented on the cell surface have consistently been considered attractive molecular targets for disease treatment. Our data suggest that a member of GPCR family, the chemokine fractalkine receptor, functions to support the prometastatic properties of EOC cells, which include migration, peritoneal adhesion, and proliferation. In addition, the fractalkine chemokine can support cell proliferation through another receptor, EGFR. Our data support the hypothesis that multiple types of ovarian carcinoma rely on fractalkine signaling for disease progression.

Peritoneal metastasis is a terminal stage of EOC progression. Nevertheless, the mechanistic details of mesothelial adhesion are poorly understood. Several molecular interactions between EOC cells and CD44, β1-integrin, L1, and CA125 as well as between mesothelial cells and hyaluronan, fibronectin, neuropilin-1, and mesothelin have been shown to contribute to peritoneal attachment both in vitro and in vivo (8, 48–51). Despite these findings, which show the importance of mesothelial adhesion, a complete understanding of all factors that play a role in EOC cell adhesion to the peritoneum has yet to be presented. Moreover, it is expected that differences in patient-specific expression levels and isoform heterogeneity of peritoneal adhesion-stimulating molecules (7, 52, 53) will diminish the extensive inhibition observed in both cell culture and xenograft models. Thus, it is imperative to elucidate other factors that contribute to mesothelial adhesion, which is a key initial step in the metastatic colonization of the peritoneal cavity. To this end, our data strongly suggest that the fractalkine network originating between CX₃CR1-positive EOC cells and CX₃CL1-positive mesothelial cells could play a role in cell–cell adhesion. There may be several potential mechanisms whereby this interaction bridges cell–cell attachment. A traditional mechanism utilized by rolling leukocytes involves interaction of surface expressed chemokine receptors with chemokines presented on the glycosaminoglycans chains of endothelial cells following integrin activation (22, 54). CX₃CL1, however, is a transmembrane protein, and it has been previously shown that CX₃CL1-mediated cell–cell adhesion can take place in the absence of integrin activation (22, 54). Our data support the latter mechanism as we have observed an additive negative effect on cell adhesion from simultaneous disruption of CX₃CL1/CX₃CR1-, β1-integrin-, and CD44-dependent modes of adhesion.

Our data indicate that 2 additional properties of the metastatic cell, migration and proliferation, are dependent on fractalkine signaling as well. We show that the CX₃CL1/CX₃CR1 pair plays a pivotal role in facilitating cell motility. This property may have implications for all stages of EOC metastatic progression. CX₃CL1 is present in the ascites fluid of EOC patients at levels that allow for productive CX₃CL1/CX₃CR1 signaling, which might induce shedding of the malignant cells and locoregional migration of the cells comprising growing metastatic lesions. Interestingly, our data, albeit limited by the number of specimens, indicate that the level of CX₃CL1 in the ascites fluid of patients with benign gynecologic diseases is an order of magnitude lower than that of the EOC ascites. These data suggest that although CX₃CR1 is expressed in transformed cells, CX₃CL1-mediated migration off of the ovarian surface is unlikely to occur due to the lower than optimal levels of the ligand, as the IC₅₀ and K_d of CX₃CL1 and CX₃CR1 binding were estimated at 2.4 and 1.7 nmol/L, respectively (55, 56). Thus, the benign tumors might harbor the potential for future activation of CX₃CL1/CX₃CR1-dependent migration.

This work also reveals a role for fractalkine signaling in EOC cell proliferation. Our data suggest that both EGFR and CX₃CR1 could be the primary receptors supporting CX₃CL1-dependent EOC cell proliferation. Proliferation plays a significant role in the progression of peritoneal metastasis from EOC. EOC is characterized by a vast uncontrolled spread of metastasis through the peritoneum. Moreover, because an estimated 48% of EOC cases are EGFR positive (57), and our studies show that the majority of EOC cases are CX₃CR1 positive, CX₃CL1 could be considered an important contributor to EOC cell proliferation.

Taken together, our data indicate that both fractalkine and its receptor may facilitate the metastatic progression of EOC by participating in migration, proliferation, and peritoneal adhesion. Hence, both fractalkine and its receptor may become useful therapeutic targets. Targeted inhibition of both membrane-bound and soluble CX₃CL1, as well as its receptor CX₃CR1, could result in impaired migration, adhesion, and proliferation. However, additional studies are required to validate our in vitro data and disclose the impact of these pathways (CX₃CL1/EGFR and CX₃CL1/CX₃CR1) on the development of metastasis in vivo. Interestingly, a possibility of the “cross-talk” between CXCL12/CXCR4 and EGFR, which may result in linking signals of cell proliferation in ovarian carcinoma, has been suggested before (58). Thus, it is interesting to speculate that activated CX₃CR1, which belongs to the family of chemokine receptors along with CXCR4, may activate EGFR-dependent proliferation using a similar “cross-talk” mechanism.

It has been proposed that the CX₃CL1/CX₃CR1 pathway could become a major therapeutic target due to its involvement in various pathologic conditions, such as inflammation, pain, and cancer (59). Although no anti-CX₃CL1 or anti-CX₃CR1 therapies are currently available, a few studies have shown the efficacy of several compounds in preclinical studies. For example, an analog of CX₃CL1 that has an inhibitory effect on CX₃CR1 during inflammation was recently discovered (60). A small molecule inhibitor of CX₃CR1, AZ12201182, was effective in abrogating mitosis and apoptosis (25). In another report, nicotinic acid was shown to effectively suppress fractalkine and other chemokines in adipocytes (61). Although there are cases in which various diseases were successfully alleviated by targeting chemokine receptors, challenges in drug design and efficacy, the existence of multiple receptors driving the disease, species differences, and relevance of the target to the human disease, among others, have also been reported (62). Thus, the clinical efficacy of future CX₃CR1-targeting agents remains to be confirmed.

No potential conflicts of interest were disclosed.

The authors thank Dr. Penzes for the critical reading of the manuscript.

This work was supported by the NIH grant 1R21CA160917 (to M.V. Barbolina).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.