Phenotypic and Functional Delineation of Murine CX₃CR1⁺ Monocyte-Derived Cells in Ovarian Cancer^,

Abstract

Ovarian tumor progression is marked by the peritoneal accumulation of leukocytes. Among these leukocytes, an immunosuppressive CD11b⁺CD11c⁺ population has been identified in both human and ovarian tumors. The use of transplantable models of murine ovarian tumors has demonstrated that this population promotes ovarian tumor growth, whereas elimination of this population has been shown to inhibit ovarian tumor progression. Despite the demonstrated importance of these cells to ovarian tumor progression, the mechanisms by which these cells are recruited to the peritoneal tumor are largely unknown. Therefore, this study analyzes the mechanisms these cells use to migrate to the peritoneum with the goal of therapeutically blocking their recruitment and subsequent immunosuppressive activity. Recent studies have identified that CX₃CR1, Gr-1, and CCR2 delineate phenotypic and functional murine monocyte subsets. Here, we report that CX₃CR1^loGr-1^hi cells dominate the population of peritoneal CD11b⁺ leukocytes early in murine tumor development; however, the CX₃CR1^hi population of cells present in the peritoneum dramatically increases in both total numbers and percentage during tumor progression. Functional analyses reveal that both of these CX₃CR1 subsets are immunosuppressive to naive CD8⁺ and CD4⁺ T-cell responses. Importantly, we demonstrate that CCR2 is a critical functional facilitator of leukocyte recruitment to the ovarian tumor microenvironment, and its genetic deletion results in a reduced tumor burden compared with wild-type mice. These results demonstrate that subsets of immunosuppressive leukocytes are recruited to the ovarian tumor environment through the CCR2 pathway, which offers a viable therapeutic target to inhibit their migration to the tumor site.

Introduction

Peritoneal ovarian cancer recruits a prolific influx of leukocytes to the tumor microenvironment [1,2]. Previous analyses have revealed that high levels of leukocyte infiltration into the tumor microenvironment are associated with a poor clinical prognosis [3–6]. Emerging data on the function of these leukocytes now demonstrate that they are co-opted by the tumor to foster tumor progression by supporting angiogenesis and metastasis, functioning as myeloid-derived suppressor cells (MDSCs) to suppress antitumor T-cell responses, and to alter cytokine expression to create a tolerogenic milieu [3,7–18]. Recent data from our laboratory and from others, using independent methods, have demonstrated the importance of the CD11c⁺ leukocytes to ovarian cancer by showing that their selective depletion therapeutically inhibits the ID8 murine model of peritoneal ovarian cancer [2,12,19]. These tumor-infiltrating CD11c⁺ cells are also F4/80⁺CD11b⁺ and represent a population of cells variably referred to as vascular leukocytes, tumor-associated macrophages, and immature dendritic cells [2,10,14,20,21]. Despite the demonstrated importance of these cells to ovarian tumor progression, the mechanisms by which these cells are recruited to the peritoneal tumor are largely unknown.

Recent data indicate that the massive influx of these CD11c⁺ cells are derived from the peritoneal macrophage pathway and are reminiscent of elicited macrophage recruitment after induced peritoneal inflammation [22]. In support of this was the finding that these F4/80⁺CD11c⁺ cells supplant the canonical F4/80⁺CD11c^- peritoneal macrophage population and that these cells continue to express CD115 [10]. This led us to hypothesize that the CD11c⁺ leukocytes are monocyte-derived cells that are recruited to the peritoneal ovarian tumor microenvironment through mechanisms by which peritoneal macrophages are elicited and, importantly, that inhibition of recruitment of these leukocytes would retard tumor progression.

In these studies, we have used the CX₃CR1/CCR2/Gr-1 classification of murine monocytes to define the CD11b⁺ ovarian tumor-recruited monocyte-derived cells both by phenotype and by function. Murine monocyte subsets have recently been delineated based on a high or a low expression of the fractalkine receptor, CX₃CR1. These subsets are further divided by expression of CCR2 and Gr-1 into “inflammatory” (CX₃CR1^loCCR2^hiGr-1^hi) or “resident” (CX₃CR1^hiCCR2^loGr-1^lo) populations based on their preferential recruitment to sites of inflammation or noninflamed tissue and relative life span in tissue [23]; human monocytes have been analogously divided into the CX₃CR1^loCD14^hi and CX₃ CR1^hiCD14^loCD16^hi subsets, respectively [23,24]. We show that after an initial recruitment of CX₃CR1^lo monocytes, the tumor microenvironment is subsequently and progressively dominated by a higher percentage and number of CX₃CR1^hi leukocytes. Our previous data demonstrated that the tumor-associated CD11b⁺ leukocytes are highly immunosuppressive to T-cell activity and that eradication of the CD11b⁺SRA⁺ population of leukocytes inhibited tumor progression [10,19]. Here, we provide the first functional analyses of the relative immunosuppressive activities of the two CX₃CR1 subsets and demonstrate that both subsets are immunosuppressive to naive T-cell responses. Because clinical ovarian tumors and ID8 murine ovarian tumor cells express high levels of the CCR2 ligand MCP-1 [1,25–27], and CCR2 mediates the recruitment of monocyte/macrophages to peritoneal inflammation in response to MCP cytokines [28], we tested the role of CCR2 in the context of ovarian tumor recruitment of leukocytes. Here, we demonstrate that CCR2 is a critical functional facilitator of leukocyte recruitment to the ovarian tumor microenvironment and, correspondingly, that loss of CCR2 function (CCR2^-/-) is sufficient to substantially reduce numbers of infiltrating leukocytes and to concomitantly inhibit ovarian tumor progression. These studies provide new insight into the origins and phenotypes of tumor-associated leukocyte populations, the mechanisms that they use to migrate to the site of tumor growth, and the tumor dependence on leukocyte presence and function.

Materials and Methods

Mice

Female C57Bl/6 mice were purchased from the National Cancer Institute (Fredricksburg, MD). CCR2 knockout mice [29] and CX₃CR1-GFP [30] mice were purchased from Jackson Laboratories (Bar Harbor, ME). All animal experiments were approved by the Dartmouth Medical School Institutional Animal Care and Use Committee.

Cells and Antibodies

ID8 cells transduced with Vegf-A and Defβ29 (referred to as ID8 within this manuscript) and ID8-Vegf cells transduced with green fluorescent protein (ID8-C3) were obtained from Dr. Jose Conejo-Garcia (Dartmouth Medical School). Cells were generated and maintained as previously described [2,31]. Anti-mouse Fc-Block and anti-mouse CD11c (HC3) were purchased from BD Biosciences (San Jose, CA). Anti-mouse F4/80 (BM8), anti-mouse CD3 (145-2C11), anti-mouse Ly-6 (Gr-1) (RB6-8C5), anti-mouse CD45 (30-F11), and anti-mouse CD11b (M1/70) were purchased from eBioscience (San Diego, CA). Anti-mouse CCR2 (E68) was purchased from Novus Biologicals (Littleton, CO) and Zenon AlexaFluor 647 labeling kit was purchased from Invitrogen (Carlsbad, CA). Anti-mouse CD45 (30-F11) was purchased from BioLegend (San Diego, CA).

RNA Isolation and Reverse Transcription-Polymerase Chain Reaction

Total RNA was isolated from FACS-sorted CX₃CR1 subsets isolated out of blood and ascites from tumor-bearing mice using RNeasy mini columns (Qiagen, Valencia, CA). RNA was quantified through spectrophotometric analysis, and samples were treated with TURBO DNase (Ambion, Austin, TX) to eliminate DNA contamination. RNA was then used as template for cDNA synthesis using a Sensiscript RT kit (Qiagen) and oligo DT primers. Polymerase chain reaction (PCR) analysis was carried out using SYBR Green Supermix (Bio-Rad, Hercules, CA) with primers designed to amplify products specific for Ly6C (Gr-1) mRNA. Primers for Ly6C were as follows: reverse 5′-ACTTACCCAGCAGGGGCTAT-3′ and forward 5′-GCCTCTGATGGATTCTGCAT-3′. Cycling conditions were initial denaturation for 2 minutes at 94°C and 34 cycles for 45 seconds at 94°C, 1 minute at 55°C, and 1 minute at 72°C, and a final extension for 2 minutes at 72°C. Reverse transcription-polymerase chain reaction (RT-PCR) for β-actin [32] was carried out in parallel; amplifications were also performed in the absence of RTas a negative control. Polymerase chain reaction products were analyzed by gel electrophoresis.

Tumor and Leukocyte Isolation

Ovarian tumors were generated by intraperitoneal (i.p.) injection of 5 x 10⁶ ID8 or ID8-C3 cells as previously described [2]. At the indicated time points, ascites and blood were harvested from mice, and the plasma and cellular fractions were collected. Red blood cells were removed from the cellular fraction using ACK lysis buffer (0.15 M NH₄Cl, 1.0 mM KHCO₃, 0.1 mM EDTA). Cells were resuspended in 0.5% BSA in PBS or medium for subsequent analysis or sorting. FACS sorting: Cells from mouse ascites and blood were resuspended in 0.5% BSA in PBS and incubated with Fc blocking antibody (clone 2.4G2) before staining with the indicated antibodies. Flow cytometry and cell sorting were done on a FACS Calibur, FACS Canto, or FACS Aria. Flow data were analyzed using Cellquest (BD Biosciences) and FlowJo 8.8.4 software (FlowJo, Ashland, OR). Plasma collected from tumor samples was analyzed for cytokine and chemokine content using the luminex mouse 23-plex panel by the Immune Monitoring Laboratory at Dartmouth Medical School (Norris Cotton Cancer Center).

Immunosuppression Assays

A modified Lyons-Parish assay (described previously [10]) was used to assess T-cell proliferation. Sorted cell populations from naive and tumor-bearing mice were cocultured at a 1:10 ratio with 10⁶ naive splenocytes obtained by passing spleen through a 70-µm cell strainer (BD Biosciences). Single-cell suspensions were treated with ACK lysis buffer to remove red blood cells and were resuspended in 2.5 µM carboxyfluorescein succinimidyl ester (CFSE) in Hank's balanced salt solution. Splenocytes were incubated at 37°C for 10 minutes and washed twice with HBSS. Plated splenocytes were stimulated with 1 µg of anti-CD3 (145-2C11), and culture supernatants were collected after 72 hours and analyzed for interferon gamma (IFN-γ) production using the murine DuoSet ELISA (R&D Systems, Minneapolis, MN). Splenocytes from the Lyons-Parish assay were stained for either CD4 or CD8, and CFSE dilution was determined by FACS analysis. T-cell proliferation was quantitatively assessed with the FlowJo Division Index, which is the average number of divisions a cell within the population has undergone. In indicated experiments, cell populations were incubated with 1 mM of the ARG1 inhibitor hydroxy-nor-l-arginine (nor-NOHA; Calbiochem, San Diego, CA) for 60 minutes at 37°C and washed three times with HBSS before coculturing with splenocytes.

Cytotoxicity Assays

Cultured ID8 cells were CFSE-labeled using the same protocol as above and cultured in the presence or absence of CD11b⁺ leukocytes (isolated through a positive selection using CD11b magnetic beads (Miltenyi Biotec, Auburn, CA)) derived from ID8 tumor ascites. The cells were cocultured 1:1 overnight and assayed for viability using the Mitoprobe DiIC₁(5) Assay Kit (Molecular Probes, Eugene, OR). Cytotoxicity was additionally quantified by measuring the release of CFSE from the tumor cells, as measured by fluorescence within the media.

Results

Analyses of CX₃CR1-Expressing Leukocyte Subsets during Ovarian Tumor Progression

We previously showed that the predominant population of immunosuppressive leukocytes in the ascites of ID8 ovarian tumor-bearing mice expresses CD11b, CD11c, and F4/80 and is derived from the monocyte/macrophage lineage [10]. With the recent creation of a transgenic mouse that expresses GFP from the CX₃CR1 (fractalkine receptor) locus, murine monocytes and monocyte-derived cells have been divided into two major subsets based on CX3CR1-GFP expression: the “inflammatory” CX₃CR1^loGr-1^hi and “resident” CX₃CR1^hiGr-1^lo populations [23]. The use of this mouse allowed us to analyze the subsets of monocyte-derived cells within the ovarian tumor ascites and to define their relative contributions to the tumor microenvironment. To assess both the absolute and relative numbers of these monocyte-derived subsets within the ovarian tumor ascites during ID8 tumor progression, cells from the blood and peritoneum were collected at weekly intervals and analyzed for CX₃CR1-GFP and Gr-1 expression by flow cytometry. No appreciable differences were observed in the percentages of the Gr-1^lo or Gr-1^hi CX₃CR1-GFP-positive monocytes in the blood among naive mice, mice with early tumor development (day 7), and mice with later-stage (day 28) tumors and discernable ascites (Figure 1A). However, there was a massive accumulation of CX₃CR1^hiGr-1^lo cells, both in terms of percentage (Figure 1C) and in total quantity (Figure 1D) over the course of tumor progression in the ascites. A predominance of the human CD14^hiCD16^lo population, the analogous subset to the murine CX₃CR1^hi subset, has previously been reported in clinical ascites, indicating that the ID8 mouse model recapitulates human disease [23,24,33]. However, in the context of inflammation, this was somewhat unexpected because the Gr-1^hi, rather than the Gr-1^lo, population has been reported to preferentially recruit to sites of inflammation [23]. To confirm that the tumor-associated CX₃CR1-expressing cells are within the previously characterized CD11b⁺ leukocyte population, we stained for CD11b. The CX₃CR1-GFP-expressing cells in the blood and ascites of both naive and tumor-bearing mice indeed express CD11b (Figure 1B). This demonstrates that the CX₃CR1^hiGr-1^lo subpopulation of cells is predominant among the CX₃CR1-expressing cells in the murine tumor environment and is the majority of the CD11b⁺ leukocyte population recruited during tumor development.

Figure 1.

ID8 ovarian tumor-bearing mice preferentially accumulate CX₃CR1⁺Gr-1^lo monocyte-derived cells in the peritoneal tumor microenvironment. Blood and peritoneal monocyte phenotypes were analyzed over a time course of ID8 tumor progression in CX₃CR1^+/GFP mice. (A, B) FACS plots from blood and peritoneal samples from naive and tumor-bearing (7 and 28 days after tumor cell injection, as indicated) mice as assessed by CX₃CR1 expression (GFP) and stained for Gr-1 (A) and CD11b (B). (C) Percentages of CX₃CR1 subsets as determined by FACS from peritoneum of early (7 days) and late (28 days) tumor-bearing mice. (D) A weekly time course analysis of the total number of peritoneal CX₃CR1⁺Gr-1^lo cells after ID8 tumor cell injection as determined by FACS. In panels (C) and (D), the SD of n ≥ 4 mice is shown. Statistical significance (**P < .01) was determined by Student's t test.

CX₃CR1^lo Gr-1hi Monocytes from the Blood Convert to Gr-1^lo Cells within the Tumor Ascites

The observation that the predominant CX₃CR1 population that accumulates in the ascites of tumor-bearing mice was Gr-1^lo whereas previous studies have shown that Gr-1^hi monocytes preferentially recruit to inflammation in the peritoneum and dominate early monocyte recruitment to the tumor-bearing peritoneum seemed somewhat surprising. To determine whether the accumulation of the CX₃CR1^hiGr-1^lo leukocyte population within the ascites of tumor-bearing mice was exclusively from selective recruitment of this monocyte population in the blood or whether other monocyte subsets could be converted to this phenotype, CX₃CR1^hiGr-1^lo (Figure 2A) and CX₃CR1^loGr-1^hi (Figure 2B) blood monocyte subsets were sorted and independently adoptively transferred i.p. to ID8 tumor-bearing mice. Two days after transfer, ascites from recipient mice were harvested, stained for Gr-1, and analyzed by FACS for donor GFP⁺ cells. CX₃CR1-GFP-positive cells from recipient mice were almost entirely Gr-1^lo regardless of donor subset (Figure 2). This demonstrates that CX₃CR1^lo monocytes decrease the expression of Gr-1 upon arrival in the peritoneal tumor environment. Interestingly, the transferred monocyte subsets maintain their CX₃CR1 expression differences relative to one another as measured by mean fluorescence intensity (Figure 2C), albeit both sets of cells had slightly higher levels of CX₃CR1-GFP compared with their corresponding populations in naive blood. This increase in CX₃CR1 expression by monocytes transferred into the tumor microenvironment corresponds with recent findings by Green et al. [34], which showed increased CX₃CR1 expression after cellular stimulation. To determine whether the loss of cell surface Gr-1 expression on the CX₃CR1^lo subset was reflective of decreased gene expression, we FACS-sorted the Gr-1^hi subset from the blood and Gr-1^lo subset from ascites of tumor-bearing mice (Figure 2B) and performed RT-PCR to compare relative RNA expression of Gr-1. The ascitic cells express lower levels of Gr-1 RNA than those isolated from the blood (Figure 2E). Thus, the decreased surface staining of Gr-1 is likely due to a down-regulation of cellular expression of Gr-1.

Figure 2.

CX₃CR1^loGR-1^hi monocytes lose Gr-1 expression in the peritoneal tumor microenvironment. (A) Gr-1^loCX₃CR1^hi and (B) Gr-1^hiCX₃CR1^lo blood leukocyte subsets from naive mice were FACS-sorted (left panels), and 10⁵ cells from each population were transferred in parallel into tumor-bearing mice by i.p. injection. Two days after injection, peritoneal cells were recovered from recipient mice and analyzed by FACS for Gr-1 expression on the GFP⁺ population (right panels). (C) The mean fluorescence intensity of GFP expression of the CX₃CR1⁺ subsets, as indicated, from the blood of naive mice (closed bars) and cells recovered from tumor-bearing recipient mice after adoptive transfer (open bars) (n ≥ 3, SD shown). (D) As a negative control, a tumor-bearing mouse that did not receive an adoptive transfer of GFP⁺ cells was analyzed. (E) Gr-1^hi blood and Gr-1^lo peritoneal CX₃CR1^lo cells were sorted and analyzed by RT-PCR for Gr-1 mRNA expression. β-Actin expression was used as an expression-positive control; PCR was performed on samples without RT (as indicated) as a negative control for DNA contamination.

Analyses of the Tumor-Associated CX₃CR1⁺ Cellular Subsets to Mediate T-cell Immunosuppression

Previous work has shown that the CD11b⁺ leukocyte population that accumulates in the ascites suppresses T-cell activation and IFN-γ production in response to mitogen stimulation [10]. The use of the CX₃CR1-GFP mouse allowed us to further divide the CD11b⁺ population into distinct subsets, which may differ significantly in function and action within the tumor environment, and to ask whether the observed immunosuppression is specific to one CD11b⁺ leukocyte subset associated with the tumor or a summation of the multiple subsets. Because the CX₃CR1-GFP-positive cells in the tumor environment are CD11b⁺ and contribute to the massive leukocyte infiltration within the ascites (Figure 1), we analyzed the relative ability of the CX₃CR1 subsets to functionally suppress naive T-cell responses. CX₃CR1hi- and CX₃CR1^lo-expressing cells from the ascites of tumor-bearing mice were FACS-sorted and independently cocultured with naive CFSE-labeled splenocytes. The T cells were subsequently stimulated, and the CD4 and CD8 splenocyte populations were analyzed for CFSE dilution as an indicator of proliferation. Both CX₃CR1^hi and CX₃CR1^lo populations from tumor ascites significantly suppressed naive CD4 (Figure 3A) and CD8 (Figure 3B) T-cell proliferation. However, in the absence of any coculture or when cultured with CX₃CR1 subsets from the peritoneum of naive mice, both CD4 (Figure 3D) and CD8 (Figure 3E) T cells proliferated robustly. As a complementary assay to confirm the ability of the CX₃CR1 subsets to inhibit T-cell proliferation, the culture supernatants were analyzed for IFN-γ as a measure of splenocyte T-cell activation. Both CX₃CR1 subsets from tumor-bearing mice were able to significantly (P < .01 for CX₃CR1^lo; P < .005 for CX₃CR1^hi) and substantially suppress IFN-γ production (Figure 3C) when compared with splenocytes cultured alone or with CX₃CR1 subsets from naive mice (Figure 3F). These results demonstrate that both subsets of CX₃CR1 cells in the ascites are capable of suppressing endogenous T-cell responses and contribute to the immunosuppressive, tumor-supporting environment in the peritoneum.

Figure 3.

Tumor-associated CX₃CR1 subsets immunosuppress naive CD4⁺ and CD8⁺ T-cell responses. CX₃CR1^lo and CX₃CR1^hi tumor-associated leukocytes from the ascites of ID8 tumor-bearing CX₃CR1^+/GFP mice were sorted by FACS. CX₃CR1 -hi and -lo subsets were independently assayed for immunosuppressive capability by a 72-hour coculture with CFSE-labeled, anti-CD3-stimulated, naive splenocytes. CFSE dilution was used to calculate division indices for CD4⁺ (A) or CD8⁺ (B) T cells. (C) To confirm the CFSE analyses, supernatants from stimulated naive splenocytes cocultured without (-) or with CX₃CR1⁺ subsets of cells (as indicated) were collected and IFN-γ production was measured by ELISA. (D–F) As negative controls, the CX₃CR1⁺ peritoneal subsets from naive mice were FACS-sorted and cocultured as in panels (A) to (C) to analyze their ability to suppress CD8 (D) and CD4 (E) proliferation as well as IFN-γ production (F). Statistical significance is shown (**P < .01, ***P < .005).

Previous work analyzing the immunosuppressive mechanism of the entire population of CD11b⁺ leukocytes from the ascites of tumor-bearing mice demonstrated that arginase-1 expression is required by these leukocytes to suppress T-cell responses [10]. To determine the role of arginase-1 in the immunosuppression mediated by the individual tumor-associated CX₃CR1 cellular subsets, FACS-sorted subsets from ID8 ascites were tested for their ability to inhibit naive splenic T-cell proliferation after treatment with the specific arginase-1 inhibitor, nor-NOHA [35,36]. Suppression assays were performed as in Figure 3 with untreated or nor-NOHA-treated CX₃CR1 cells, using T-cell proliferation as an experimental end point to quantitatively measure suppression of T-cell responses. Consistent with the data from Figure 3, both CX₃CR1 subsets potently inhibited T-cell proliferation. However, preincubation with nor-NOHA was sufficient to alleviate both CX₃CR1^lo and CX₃CR1^hi immunosuppression on CD4⁺ (Figure 4A) and CD8⁺ (Figure 4B) T-cell responses. This demonstrates that arginase-1 activity is required for functional immunosuppression of T-cell responses by both CX₃CR1 subsets and that blockade of arginase-1 activity abrogates this effect.

Figure 4.

Tumor-associated CX₃CR1⁺ leukocyte subsets require arginase-1 to immunosuppress naive CD4⁺ and CD8⁺ T-cell responses. CX₃CR1^lo and CX₃CR1^hi tumor-associated leukocytes from the ascites of ID8 tumor-bearing CX₃CR1^+/GFP mice were sorted by FACS. Sorted CX₃CR1 -hi and -lo subsets were incubated in the presence or absence of 1 mM nor-NOHA and subsequently independently assayed (as indicated) for immunosuppressive capability by a 72-hour coculture with CFSE-labeled naive splenocytes. White bars indicate unstimulated splenocytes, whereas black bars indicate splenocytes stimulated during the coculture with anti-CD3 antibody. CFSE dilution was used to calculate division indices for (A) CD4⁺ or (B) CD8⁺ T cells. Statistical significance is shown (**P < .01, ***P < .005).

Genetic Deletion of CCR2 Reduces Tumor-Infiltrating Leukocytes within the Ascites

In a previous work, we provided evidence that the CD11b⁺ leukocytes that accumulate in the ascites of tumor-bearing mice are recruited through the monocyte/macrophage pathway [10]. Work by Boring et al. [29] established that monocyte/macrophage migration to inflammation in the peritoneum requires CCR2 signaling, and furthermore, human and murine (including the ID8 model) ovarian cancers produce the CCR2 ligand MCP-1 [1,25–27]; our Luminex analyses supported the previous reports from human ovarian ascites and the ID8 ovarian tumor model [25–27,37,38] with the finding that ID8 ascites consistently contains robust MCP-1 concentrations between 758 and 4140 pg/ml. This evidence, along with our observation that the CX₃CR1^loGr-1^hi monocyte population, known to express CCR2, can undergo conversion to the cells found in the ascites, led us to hypothesize that CCR2 expression is critical to the migration of leukocytes to the tumor site and that blocking this pathway could reduce accumulation of tumor-promoting leukocytes.

FACS analyses revealed that both blood (Figure 5A) and peritoneal (Figure 5B) murine CD11b⁺ leukocytes exhibited surface expression of CCR2, and the CD11b⁺ cells were the predominant population that expressed CCR2; leukocytes from CCR2^-/- mice were stained in parallel as a negative control (Figure 5C). To test the functional contribution of CCR2 to the recruitment of peritoneal tumor-infiltrating leukocytes, wild-type (C57BL/6) and CCR2 knockout mice were injected i.p. with ID8 ovarian tumor cells, and after tumor progression, the subsequent leukocyte populations within the peritoneal ascites were analyzed by FACS. CCR2 knockout mice exhibited significantly reduced numbers of leukocytes, confirmed by F4/80⁺ (Figure 5D) and CD11b⁺ (Figure 5E) expression, within the peritoneal tumor microenvironment in comparison to wild-type mice. Thus, CCR2 functions to recruit leukocytes, shown to have a tumor-promoting phenotype and function ([10] and Figure 3), to the ovarian tumor site. These findings are consistent with reports from other systems that have shown involvement of CCR2 signaling in the recruitment of anti-inflammatory cell types, including MDSCs and tumor-associated macrophages [39,40]. The evidence that CCR2 is involved in recruiting leukocytes with potential to support tumor development led us to test whether CCR2 ablation would influence the development and growth of tumors.

Figure 5.

Loss of CCR2 inhibits leukocyte recruitment to ID8 peritoneal ascites. Cells from the blood (A) and peritoneum (B) of wild-type (C57Bl/6) mice were stained for CD11b and CCR2 and analyzed by FACS. As a control, CCR2 expression on the CD11b⁺ cells from WT (gray histograms) mice was compared with the respective cells from CCR2^-/- (empty histograms) mice (A and B, inlaid). (C) CCR2^-/- cells were used as a control for antibody specificity and demonstrate that CCR2 is primarily expressed on the CD11b⁺ cells (blood sample shown). (D, E) ID8 tumor-bearing wild-type and CCR2^-/- mice were analyzed in parallel by FACS for F4/80⁺ and CD11b⁺ leukocyte infiltration. A comparison of wild-type and CCR2^-/- peritoneal F4/80⁺ (D) and CD11b⁺ (E) leukocyte population numbers 6 weeks after tumor cell injection. The statistical significance (**P < .01, ***P < .005) of the CCR2^-/- deficit was determined by Student's t test (WT n = 10, CCR2^-/- n = 9).

CCR2 Deletion Inhibits ID8 Ovarian Tumor Growth and Progression

Elimination of various tumor-associated leukocyte populations through a variety of methods has been reported to be therapeutically efficacious. Particularly relevant to the current studies is that SRA-targeted killing of CD11b⁺F4/80⁺ leukocytes within the peritoneal ovarian tumor ascites was previously reported to result in a significant reduction in tumor burden [19]. Because the loss of CCR2 expression inhibits the recruitment of these CD11b⁺ monocyte/macrophages to the murine ascites and because CD11b⁺ leukocytes have been shown to be critical contributors to tumor progression in a variety of tumor models, we tested the effect of loss of host CCR2 expression on ID8 ovarian tumor progression. CCR2^-/- and wild-type mice were injected in parallel with GFP-expressing ID8 cells (referred to as ID8-C3 cells), and the tumors and ascites were subsequently harvested 7 weeks later and quantitatively analyzed for tumor burden by measurement of total cellularity (after red blood cell lysis), GFP⁺ cell number, and peritoneal wall metastases. Ex vivo analyses revealed that the CCR2 knockout mice had reduced total cellularity in their peritoneal ascites compared with their wild-type controls (Figure 6A). Importantly, the loss of host CCR2 expression resulted in a reduction of tumor burden within the ascites, as assessed by FACS analyses for GFP⁺ tumor cells (Figure 6B). As a complementary and confirmatory assay to comparatively assess tumor burden and to ensure that reduced tumor burden in the ascites was not concomitant with increased tumor deposition on the peritoneal wall, we also quantitatively analyzed the accumulation of metastases on the peritoneal walls. After peritoneal lavage and ascites harvest, the skin of the peritoneal walls of the wild-type and CCR2^-/- tumor-bearing mice was measured for GFP (tumor) fluorescence. CCR2 knockout mice had fewer metastases than wild-type mice both quantitatively (Figure 6C) and qualitatively (Figure 6D), with a ∼40% reduction in solid metastatic tumor burden on the peritoneal walls. These data indicate that CCR2 is required for normal leukocyte recruitment and is an integral part of peritoneal tumor progression and growth. In addition, these results specifically identify CCR2 as a potential route of therapeutic intervention for peritoneal ovarian tumors and, in the broader picture, support an emerging body of data that reduction of the number of tumor-associated leukocytes from the microenvironment is an efficacious and viable therapeutic option.

Figure 6.

CCR2 knockout mice have reduced ovarian tumor burden. Wild-type and CCR2^-/- mice were injected in parallel with ID8-C3 (GFP-expressing) tumor cells, and the ascites (A, B) and peritoneal wall (C) samples were subsequently harvested. CCR2^-/- mice exhibit a significant deficit in the total numbers of cells (after red blood cell lysis) in the ascites (A) and the number of the GFP⁺ tumor cells (B) within the ascites relative to wild-type mice. (C) Peritoneal wall samples were collected from the WT and CCR2^-/- tumor-bearing mice, and GFP fluorescence was imaged using a Xenogen imaging system. (D) The mean GFP fluorescence per pixel from the samples in (C) was quantified. SDs and statistical significance (*P < .05, **P < .01) are shown.

Discussion

The presence of tumor-associated leukocytes is emerging as a vital requirement for tumor progression. Increased leukocyte infiltration of tumors has been reported to correlate with poor clinical prognosis, and correspondingly, recent studies in a variety of models have demonstrated that reduction or elimination of leukocyte populations in the tumor microenvironment can inhibit the growth of tumors [12,19,35,39,41–43]. The predominant leukocyte population in ovarian cancer is the CD11b⁺CD11c⁺ leukocytes, which are immunosuppressive to naive T-cell responses [10]. However, the relation of these cells to circulating monocyte subsets and the mechanisms by which these leukocytes are recruited to the tumor site have not previously been defined. The observation that the predominant ovarian tumor-associated CD11b⁺CD11c⁺ leukocyte population is derived from the CD115⁺CD11b⁺F4/80⁺ peritoneal macrophage pathway [10] led us to hypothesize that these tumor-associated leukocytes are recruited from the blood monocyte populations and that we could use the mice that express CX₃CR1-GFP, with the recent delineation of monocyte-derived cells into CX₃CR1 -hi and -low populations, to determine the relative presence and the functional contributions of these populations within the ovarian tumor environment.

With the use of the murine ID8 ovarian tumor model, we demonstrate that both CX₃CR1^lo and CX₃CR1^hi monocyte-derived cells are present in the peritoneum during tumor progression. The CX₃CR1^lo cells dominate the population of peritoneal CD11b⁺ leukocytes early in tumor development; however, the CX₃CR1^hi population of cells present in the peritoneum dramatically increases in both total numbers and percentage and overtakes the CX₃CR1^lo population during tumor progression. The observation that the CX₃CR1^hi population dominates as tumors develop is consistent with findings by Auffray et al. [44] that CX₃CR1^hi monocytes respond to tissue damage and preferentially differentiate into macrophages. In addition, we demonstrate that the CX₃CR1^lo Gr-1^hi monocytes lose Gr-1 expression and increase CX₃CR1 expression upon adoptive transfer to the peritoneum, a phenomenon also reported by Sunderkotter et al. [45] who observed the ability of some Gr-1^hi monocytes to convert to a Gr-1^lo phenotype. Likewise, the ability of CX₃CR1^lo cells to convert to CX₃CR1^hi cells within the ovarian tumor microenvironment is consistent with the previous finding that CX₃CR1^lo monocytes can convert to CX₃CR1^hi monocyte/macrophages with an anti-inflammatory phenotype in a muscle injury model [46]. Thus, our data support that “inflammatory” monocytes recruited to the tumor site can convert their phenotype and contribute to the growing CX₃CR1^hi population observed during tumor progression.

The massive recruitment of CX₃CR1 monocytes to the peritoneum in ovarian tumor-bearing mice indicated to us that these cells were likely promoting tumor growth rather than inhibiting it or being cytotoxic to the tumor cells (Figure W1). This is supported by the work of Hagemann et al. [21], which showed that macrophages are polarized to a tumor-associated phenotype upon exposure to ovarian cancer cells, and by the aforementioned work of Arnold et al. [46], which reported that CX₃CR1^lo monocytes can adopt an anti-inflammatory phenotype. Moreover, we have shown that the sum total of CD11b⁺ leukocytes from the peritoneum of tumor-bearing mice contribute to tumor growth, in part through suppression of T-cell immune responses [10,19]. Here, we use CX₃CR1-GFP mice to provide the first assessment of the relative immunosuppressive abilities of the two specific CX₃CR1 subpopulations by measuring each for their ability to functionally suppress naive T-cell responses. We demonstrate that both CX₃CR1^lo and CX₃CR1^hi leukocytes from the peritoneum of tumor-bearing mice potently suppress naive T-cell proliferation and activation. Interestingly, whereas the CX₃CR1^hi population of cells shares CD11b expression and the immunosuppressive properties of canonical murine MDSCs, it lacks the robust Gr-1 expression normally observed on MDSCs. However, the CD11b⁺CX₃CR1^lo monocytes in the blood of tumor-bearing mice express robust Gr-1 and we have now demonstrated that they lose expression of Gr-1 upon entering the peritoneal tumor microenvironment. These data indicate that MDSCs may decrease in Gr-1 expression upon entering the tumor environment yet retain their immunosuppressive properties. This may provide an explanation for why both CD11b⁺Gr-1^hi and CD11b⁺Gr-1^lo cells function similarly to immunosuppress T cells and, moreover, would indicate that analyses of MDSC populations solely by CD11b/Gr-1 staining may vary greatly depending on the stage of the tumor and the percentage of Gr-1^hi cells that have converted to Gr-1^lo expression.

One paradigm of tumor growth is that it results in low-level inflammation, likely the result of tissue invasion, angiogenesis, cytokine production, and immune cell infiltration [20,47,48]. An important chemokine that mediates the recruitment of immune cells to sites of inflammation, is known to be produced by human and murine ovarian cancer cells, and is a CCR2 agonist is MCP-1 [1,25–27]. This along with our previous observation that the CD11b⁺ cells recruited to ovarian tumors are likely the product of the peritoneal macrophage pathway led us to hypothesize that, analogous to the influx of CD11b⁺ leukocytes elicited by other forms of peritoneal inflammation, CCR2 may play a critical functional role in monocyte recruitment to the ovarian tumor microenvironment. If correct, this hypothesis would also support the previous observation that CCR2-expressing CX₃CR1^lo monocytes are recruited from the blood to peritoneal inflammation and other sites of inflammation. Here, we show that, in accord with other models of peritoneal inflammation, leukocyte recruitment to the peritoneal tumor microenvironment is dependent on CCR2 expression and loss of the receptor inhibits accumulation of the CD11b⁺ and F4/80⁺ cell population(s). Importantly, we demonstrate that prevention of CCR2-dependent cell recruitment to the peritoneum results in inhibition of ovarian tumor progression as assessed by the quantity of tumor cells within the ascites and in solid metastases upon the peritoneal wall (Figure 6). These findings identify CCR2 as a key mediator to the recruitment of immunosuppressive leukocytes to peritoneal ovarian cancer. Moreover, they support the emerging concept derived from a variety of tumor models that tumor-associated leukocytes are critical to tumor progression and that, correspondingly, inhibition or removal of these leukocytes can inhibit tumor growth.

In summary, these findings provide insight into the monocyte origin and the mechanisms of recruitment of ovarian tumor-associated leukocyte populations that provide immunosuppressive support for tumor growth and progression. Identification of the monocyte/macrophage pathway as a principal route of leukocyte recruitment to the tumor site has important implications for therapeutic intervention in the accumulation of these leukocytes and provides a number of potential approaches for blocking their supportive role in tumor development. Specifically, these studies implicate CCR2 as the predominant mechanism by which ovarian tumors recruit tumor-promoting monocyte-derived cells to the peritoneum. Our findings derived from the CCR2^-/- mouse lead us to propose that blockade of CCR2 activity in ovarian cancer may have therapeutic potential and that this strategy may improve the efficacy of other currently used treatment regimens.