The Complement C3a Receptor Contributes to Melanoma Tumorigenesis by Inhibiting Neutrophil and CD4⁺ T Cell Responses

This article is featured in In This Issue, p.4421

Melanoma is the most aggressive form of skin cancer, with 232,000 new cases and 55,000 deaths reported globally in 2012. Despite recent advances in treatment, sustained responses remain elusive and the 3-y survival rate for metastatic melanoma remains ∼15% (1). There is therefore an urgent need for new therapeutic strategies for advanced melanoma.

Inflammation is known to contribute to the growth of many tumors, and the complement system is an important mediator of inflammation. An essential part of the innate immune system, the complement system is comprised of plasma and membrane-bound proteins whose overall function is to regulate inflammation, facilitate immune defense mechanisms, and maintain tissue homeostasis (2). Complement activation leads to the production of the complement peptides C3a and C5a and the membrane attack complex (C5b-9), through which the complement system exerts many of its effects (3).

The detection of complement activation products, including C3a and C5a in tumor tissue (4), is evidence that the complement system is activated in response to tumor cells. Several studies have identified a role for C5a in promoting tumor growth (5–9). However, despite reports of upregulated serum C3a levels in many cancers, including breast, colorectal, and esophageal cancer (10, 11), the role of C3a in tumor growth has not been investigated. C3a binds to the G protein–coupled receptor C3aR, which is expressed by cells of myeloid origin (including neutrophils, mast cells, monocytes, macrophages, and dendritic cells) as well as nonimmune cells in lung, liver, muscle, and other tissues. Activation of C3aR can induce both pro- and anti-inflammatory effects, depending on the cell type and disease context (12). Therefore, the aim of the present study was to determine whether C3aR signaling regulates tumor growth.

We demonstrate that the development and growth of syngeneic primary murine B16-F0 melanomas are retarded in the absence of C3aR signaling, and that the antitumor effects of C3aR inhibition are linked to a reduction in tumor-associated macrophages, with concomitant increases in tumor-infiltrating neutrophils and CD4⁺ T lymphocytes. We also demonstrate protective effects of C3aR deficiency/inhibition in BRAF^V600E mutant melanoma, colorectal, and breast cancer models. These results suggest the potential of C3aR as a novel therapeutic target for melanoma, and possibly other intractable tumors.

The murine melanoma B16-F0 (ATCC CRL6322), mammary carcinoma 4T1 (ATCC CRL-2539), and macrophage J774A.1 (ATCC TIB-67) cell lines were obtained from the American Type Culture Collection (Manassas, VA). Murine SM1WT1 (BRAF^V600E mutant) (13) and MC38 colon carcinoma (14) cell lines were obtained from Prof. Mark Smyth (QIMR Berghofer Medical Research Institute, Brisbane, Australia). B16-F0, SM1WT1, and MC38 cells were cultivated in high-glucose DMEM (Invitrogen, Auckland, New Zealand) and 4T1 and J774 cells were cultivated in RPMI 1640 supplemented with 10% heat-inactivated FCS (Moregate, Brisbane, Australia) and penicillin/streptomycin at 37°C in a humidified atmosphere of 5% CO₂ in air.

C3aR knockout (C3aR^−/−) mice were obtained from Rick Wetsel (University of Texas–Houston) (15) and back-crossed onto C57BL/6J for at least 15 generations. C57BL/6J mice were obtained from Animal Resources Centre, Western Australia, and female BALB/c mice were from Monash Animal Services (Melbourne, Australia). All mice were housed in the University of Queensland Biological Resources animal facility and used at 6–8 wk of age. All procedures were approved by the University of Queensland Animal Ethics Committee Guidelines and conformed to the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (8th Ed., 2013).

Homozygous C3aR^−/− mice or wild-type (WT) controls (age and sex matched) were injected s.c. with B16-F0 melanoma (2.5 × 10⁵ cells/mouse), SM1WT1 melanoma (1 × 10⁶ cells/mouse), or MC38 colon carcinoma (1 × 10⁶ cells/mouse); female BALB/c mice were injected into the mammary fat pad with 4T1 mammary carcinoma (5 × 10⁵ cells/mouse). For some experiments, mice received daily injections of either a nonpeptide C3aR antagonist (SB290157; VDM Biochemicals, Bedford Heights, OH; 1 mg/kg/d i.p.), the cyclic hexapeptide C5aR1 antagonist (PMX53; produced in-house; 1 mg/kg/d) (16), or vehicle only (0.9% saline or 5% glucose solution) control, commencing once tumors became palpable (approximately day 7 after tumor cell injection). Mice were euthanized at day 14 (or once the largest tumor reached ∼150 mm²), blood was collected by cardiac puncture, tumors, spleen, draining lymph nodes (inguinal, axillary, and brachial), and femurs were excised, and tumors were weighed. For “survival” studies, the time taken for each tumor to reach a predetermined size (150 mm²) was recorded.

WT and C3aR^−/− mice were injected i.p. (0.1 mg/mouse) with anti-Ly6G Ab (1A8, selective for neutrophils) (17) or isotype control (rat IgG2a, 2A3; both from Bio X Cell, West Lebanon, NH) 24 h prior to s.c. injection of B16-F0 cells, then every 3 d until the largest tumor reached 150 mm². This regimen was based on a serum half-life for rat IgG2a Abs of 7.5 d (18), and it was shown in preliminary trials to deplete >95% of neutrophils from the peripheral blood. Mice were then euthanized, and blood and tumor tissue were removed.

Cryostat sections (0.5 μm) of tumor tissue were incubated with fluorophore-conjugated Abs to mouse CD45, Ly6G, F4/80, and CD3 (all from BioLegend, San Diego, CA). Nuclei were stained with Hoechst 33342 (Invitrogen) and visualized using an Olympus BX-61 microscope and a Retiga EXi cooled CCD camera.

Single-cell suspensions of bone marrow, blood, tumor, spleen, and lymph node were prepared for surface staining with Abs to mouse CD45, CD25 F4/80, CD11b, Ly6G, Gr-1, CD3, CD4, and CD8a (BioLegend). Staining for regulatory T cells (Tregs) was carried out using a FoxP3 Fix/Perm kit (BioLegend). For Th1, Th2, and Th17 staining, cells were stimulated with PMA (50 ng/ml; Sigma-Aldrich) and ionomycin (10⁻⁶ mol/l; Sigma-Aldrich) in the presence of brefeldin A (5 μg/ml; BioLegend) for 4 h at 37°C. After surface staining, cells were fixed and permeabilized for staining with Abs to intracellular cytokines IL-4, IL-17A, or IFN-γ (BioLegend). Cells were analyzed on an Accuri C6 flow cytometer (BD Biosciences, Franklin Lakes, NJ) and data analysis was performed with FlowJo software (Tree Star, Ashland, OR). Gating strategies for different leukocyte populations are illustrated in Supplemental Fig. 1.

Excised tumor tissue was stored in RNAlater (Qiagen, Hilden, Germany), and then total RNA was extracted using an RNeasy mini kit (Qiagen). Quantity and quality of total RNA were determined by a NanoDrop ND1000 (Thermo Fisher, Waltham, MA) and bioanalyzer using an RNA 6000 NanoChip (Agilent Technologies, Santa Clara, CA). cDNA was prepared from 400 ng total RNA using an RT² first-strand kit (Qiagen) and then analyzed on a mouse cancer inflammation and immunity crosstalk RT² Profiler PCR array (Qiagen). Gene expression levels were normalized to a panel of three housekeeping genes (Actb, Gapdh, and Hsp90ab1) and data were expressed as fold change calculated by the 2^−ΔΔCt method.

Snap-frozen tumor tissue was pulverized with a mortar and pestle, and then cytoplasmic extracts were prepared (cell lysis kit from Merck Millipore, Darmstadt, Germany). Cytokine levels in tumor tissue extracts (0.25 mg protein/well) and plasma (undiluted) from the same animals were determined by multiplex ELISA (mouse cytokine/chemokine panel MPMCYTOMAG70K14; Millipore) or conventional ELISA (human/mouse TGF-β1 ELISA kit; eBioscience, San Diego, CA).

Each experiment was performed a minimum of three times, and results were expressed as mean ± SD. Tumor growth curves were analyzed by a permutation test and survival studies by log-rank Mantel–Cox test. PCR array data were analyzed by a Student t test based on the 2^−ΔΔCT values for each gene in WT versus C3aR^−/− mice (n = 6/group). All other data were analyzed by a Mann–Whitney U test or Kruskal–Wallis test followed by a Dunn multiple comparisons test. For all tests, a p value < 0.05 was considered statistically significant.

To determine whether C3a–C3aR signaling influences melanoma development and growth, we first compared the growth of B16-F0 melanoma cells in C3aR^−/− and WT mice. As shown in Fig. 1A, the rate of tumor growth in C3aR^−/− mice was significantly retarded compared with WT mice. Excised tumor weights at termination (day 14) were also markedly smaller in C3aR^−/− mice (23% of control; Fig. 1B, 1C). Furthermore, survival of C3aR^−/− mice was increased by 70% to 26.6 ± 1.4 d, compared with 15.7 ± 0.7 d for the WT controls (Fig. 1D). These results suggest that C3aR signaling by host cells promotes B16-F0 tumor growth.

As the downstream activation product C5a has previously been implicated for a role in promoting tumor growth (6–9), we next investigated whether the tumor inhibitory effects observed in C3aR^−/− mice could be augmented by C5aR1 antagonism. For this experiment, WT and C3aR^−/− mice were injected with B16-F0 melanoma cells, and then once tumors became palpable (day 8), they were administered daily s.c. injections of C5aR antagonist (PMX53) or vehicle alone. As previously shown for other tumor types, C5aR antagonism alone inhibited melanoma growth in WT mice (Fig. 1E, 1F). However, the reduction in tumor growth observed in C3aR^−/− mice was not augmented by C5aR inhibition, indicating that the contribution of C3aR signaling to melanoma growth is at least as potent as C5aR signaling, and that the inhibition of either receptor is sufficient to impair tumor growth.

The potent antitumor effects observed in C3aR^−/− mice prompted us to determine the potential of C3aR therapeutic inhibition. As shown in Fig. 2A, tumor progression was also dramatically retarded by daily i.p. injection of a C3aR antagonist (C3aRA; SB290157), commencing day 7 after tumor induction. Excised tumor weights at day 14 were also significantly smaller in the C3aRA-treated group (Fig. 2B, 2C). Similarly, survival rates were significantly improved by treatment with C3aRA (Fig. 2D). Taken together, these data suggest C3aR as a therapeutic target for melanoma.

Because C3aR is known to be expressed by immune cells (predominantly myeloid cells, including neutrophils, monocytes, and macrophages) (19), we next determined whether the lack of C3aR alters tumor-infiltrating leukocyte populations. Immunofluorescence staining of s.c. B16-F0 tumor tissue from both C3aR^−/− and WT (Fig. 3A) mice demonstrated a band of (CD45⁺) leukocytes encapsulating the tumors, as well as infiltrating the tumor tissue. Many tumor-infiltrating leukocytes were identified as neutrophils (Ly6G⁺) or macrophages (F4/80⁺); T lymphocytes (CD3⁺) were also evident. Whereas neutrophils appeared to be increased in C3aR-deficient mice, macrophages appeared to be reduced.

Quantitation of tumor-infiltrating leukocyte populations by flow cytometry (Fig. 3B) showed that the proportion of total leukocytes (CD45⁺) within tumor tissue was not significantly different in C3aR^−/− mice compared with WT mice (5.8 ± 2.7 versus 4.7 ± 2.9%); the proportions of monocytes (CD11b⁺Ly6C⁺) were also similar. Whereas tumor-associated macrophages were reduced in C3aR-deficient mice, the contribution of neutrophils (CD11b⁺Ly6G⁺) was significantly higher (16.6 ± 8.1% of total leukocytes) than in WT mice (9.7 ± 4.8%), a result that is in accord with our prior finding that C3aR deficiency causes the egress of neutrophils from bone marrow (20). We also investigated whether the increase in tumor-associated (CD11b⁺Ly6G⁺) neutrophils might be associated with differences in activation status (21, 22), but found no significant difference in expression of activation markers CD62L or CXCR4 (not shown). The Gr-1 Ab commonly used to identify myeloid-derived suppressor cells (MDSC) recognizes a common epitope of Ly6G and Ly6C (23), and thus should recognize both granulocytic and monocytic MDSC. However, the percentages of CD11b⁺Gr-1⁺ cells were very similar to those for CD11b⁺Ly6G⁺ cells, suggesting that most Gr-1⁺ cells are granulocytic; this was confirmed in later neutrophil depletion experiments (see Fig. 5). Because it is not possible to distinguish mature neutrophils from granulocytic MDSC using the Gr-1 Ab, most subsequent analyses used the separate Abs for Ly6G and Ly6C to identify monocytes and neutrophils.

Whereas the proportions of B lymphocytes (not shown) were similar in tumors from C3aR^−/− and WT mice, total T lymphocytes (CD3⁺) were significantly higher in the absence of C3aR, mainly due to an increase in CD4⁺ T lymphocytes; CD8⁺ T lymphocytes were not significantly different. Analysis of CD4⁺ T cell subsets showed that the proportions of Tregs (CD4⁺CD25⁺Foxp3⁺) were similar, but Th1 (CD4⁺IFN-γ⁺), Th2 (CD4⁺IL-4⁺), and Th17 (CD4⁺IL-17A⁺) subsets were all significantly higher in tumors from C3aR^−/− mice (Fig. 3B).

To investigate how C3aR signaling influences the systemic immune response to the tumor, we next compared leukocyte subpopulations in bone marrow, blood, draining lymph nodes, and spleen from tumor-bearing WT and knockout mice (Fig. 4). Flow cytometric analysis of bone marrow from C3aR^−/− mice (Fig. 4A) revealed a significant reduction in the proportions of monocytes (CD11b⁺Ly6C⁺) but no significant difference in neutrophils (CD11b⁺Ly6G⁺). Although blood monocyte and T lymphocyte populations were similar in WT and C3aR^−/− mice, the percentage of neutrophils was significantly lower in knockout mice (Fig. 4B), possibly reflecting the increased sequestration of neutrophils to the tumor site. The draining lymph nodes, important sites for presentation of tumor Ag to T lymphocytes, showed no significant differences in neutrophils or monocytes. However, in accord with the tumor data, macrophages were significantly lower and the proportions of total T lymphocytes, Th1, Th2, and Th17 cells were higher in C3aR^−/− compared with WT mice (Fig. 4C). In spleens, the proportions of monocytes and macrophages were lower, but neutrophils were significantly higher in the absence of C3aR. The proportions of total (CD3⁺) T lymphocytes and CD4⁺ cells were significantly higher, with Th1 and Th17 subsets increased (Fig. 4D); in this case splenic Tregs were also higher.

Treatment of WT mice with the C3aRA (SB290157) induced changes in tumor-infiltrating populations similar to those observed in knockout mice (Supplemental Fig. 2). As shown in Supplemental Fig. 2A, the percentage of tumor-infiltrating (CD11b⁺Ly6G⁺) neutrophils was increased by C3aRA treatment, whereas the proportion of tumor-associated macrophages was reduced; the percentages of CD3⁺ (total) and CD4⁺ T lymphocytes were increased, as were Th1, Th2, and Th17 subsets. Both monocytes and neutrophils were significantly lower in bone marrow from C3aRA-treated mice (Supplemental Fig. 2B), suggesting that inhibition of C3aR signaling aids sequestration of these cells into the circulation. In contrast to knockout mice, C3aRA-treated mice showed an increase in blood neutrophils, but a reduction in Tregs (Supplemental Fig. 2C). Within draining lymph nodes, the percentages of T lymphocytes (CD3⁺, CD4⁺, and CD8⁺) were all significantly increased by C3aRA treatment; however, only the Th2 subset was increased (Supplemental Fig. 2D). Spleens from C3aRA-treated mice showed similar trends, with total (CD3⁺) and CD4⁺ T lymphocytes increased, along with Th2 and Treg cells; in this case there was a significant reduction in CD8⁺ cells (Supplemental Fig. 2E). The effects of C3aR deficiency and antagonism on leukocyte populations in tumor-bearing mice are summarized in Table I. The minor differences between C3aR-deficient and C3aRA-treated mice may be due to the development of compensatory mechanisms in knockout mice.

Given the data showing an increase in tumor-infiltrating neutrophils in C3aR-deficient and antagonist-treated mice, we proposed that the influx of neutrophils into the tumor may tip the balance toward an effective antitumor response. To determine whether neutrophils actively contribute to the tumor inhibition observed in the absence of C3aR signaling, we used an mAb (anti-Ly6G; 1A8) to deplete these cells (17). The efficacy of neutrophil depletion was confirmed by flow cytometric analysis of blood from WT and C3aR^−/− mice (Fig. 5A). As shown in Fig. 5B, neutrophil depletion slowed tumor growth in WT mice, although the reduction in excised tumor weight was not significant (Fig. 5C). Conversely, the lack of neutrophils reduced the tumor inhibitory effects observed in C3aR^−/− mice, such that excised tumor weight was not significantly different from that of WT mice treated with the isotype control Ab (2A3; Fig. 5C). Flow cytometric analysis (Fig. 5D) showed that neutrophil depletion of C3aR^−/− mice led to a significant reduction in tumor-infiltrating CD4⁺ T lymphocytes, with the proportion of CD4⁺ cells having returned to levels similar to those of WT mice treated with the isotype control Ab; in contrast, the proportion of CD4⁺ cells appeared to be increased in neutrophil-depleted WT mice, although this was not significant. Interestingly, anti-Ly6G treatment also removed most CD11b⁺Gr-1⁺ cells (MDSC) in tumors from both WT and C3aR^−/− mice, thus providing support for our suggestion that in this model most cells recognized by the anti–Gr-1 Ab are granulocytic rather than monocytic (see Fig. 3A). These results suggest that Ly6G⁺ myeloid cells regulate the CD4⁺ T cell response to tumor, and that C3aR signaling determines the functional status of Ly6G⁺ cells within the tumor (immunosuppressive or immunoenhancing).

To determine how C3aR signaling regulates the inflammatory milieu within the tumor microenvironment, we first examined expression of a panel of genes associated with cancer inflammation and immunity. The results indicated upregulated expression of genes for NF-κβ, TLRs 2, 3, and 4, the Th2 cytokine IL-10, as well as chemokines CCL22 and CXCL10 (IP-10) and receptors CXCR4, CCR5, and CCR7 in tumor tissue extracts from C3aR-deficient mice compared with WT mice (Supplemental Table I). We also examined plasma and tumor tissue extracts for a panel of cytokines and chemokines previously implicated as regulators of the antitumor response. As shown in Fig. 6, plasma G-CSF was significantly lower in C3aR^−/− mice whereas IL-1β was higher; there were no significant differences in plasma chemokine levels. Cytokine levels were slightly increased in tumor tissue from C3aR^−/− mice, although these were not significant; this included a 2-fold elevation in IL-10, which was also upregulated at the mRNA level (Supplemental Table I). Although tumor tissue extracts from C3aR^−/− mice showed a general trend toward higher chemokine levels, only the T cell chemokine CCL5 (RANTES) was significantly increased, a result in accord with the PCR array data that show upregulation of its cognate receptor CCR5.

Finally, to determine whether C3aR signaling has similar effects on other tumor types, we investigated the growth of two other tumor models in C3aR^−/− mice: the murine melanoma cell line SM1WT1, which bears the BRAF^V600E mutation harbored by ∼50% of human melanomas (13), and the colon cancer cell line MC38. Similar to our results with B16-F0 melanoma, the growth of these tumors was significantly slowed in C3aR^−/− mice compared with WT mice (Fig. 7A, 7C). Excised tumor weights at termination were also significantly smaller in C3aR^−/− mice (91 and 58% reductions for SM1WT1 and MC38 tumors, respectively; Fig. 7B, 7D). C3aR antagonist treatment of BALB/c mice harboring 4T1 mammary carcinomas also demonstrated significant inhibition of tumor growth (Fig. 7E, 7F). Overall, these results suggest the potential of C3aR as a therapeutic target for a range of cancers.

The complement system is an essential part of the innate immune system, regulating inflammation, facilitating immune defense mechanisms, and maintaining tissue homeostasis (24). Complement is now thought to be a key component of cancer-related inflammation, and a recent study has suggested that exacerbated inflammation in the tumor microenvironment promotes genetic instability (25). The canonical proinflammatory mediator C5a has been widely implicated for a role in tumor growth, with both tumor-promoting and tumor-inhibitory effects reported, depending on tumor type, immune status of the host, and C5a levels (6–9, 26). Although these and other studies (27) have shown that tumor growth is reduced in the absence of C3 (which generates both C3a and C5a), the tumor-promoting effects have been attributed solely to C5a. Until now, there have been no studies investigating the role of the upstream activation fragment C3a in tumor growth.

The present study used both C3aR-deficient mice and C3aR therapeutic inhibition to identify a hitherto unknown role for C3a–C3aR signaling in regulating tumor growth in murine models of melanoma and colon and breast cancer. The potential of C3aR as a therapeutic target is highlighted by the demonstration that C3aR antagonism potently inhibits the growth of established B16-F0 melanomas. Note that C3aR agonist and off-target activity has been reported for the antagonist (C3aRA; SB290157) used in this study (28). However, the reduction in tumor growth following SB290157 treatment is similar to that observed in C3aR^−/− mice. Thus, although we cannot rule out other off-target effects, the drug is unlikely to be exerting C3aR agonistic effects in this model. Moreover, our demonstration that the antitumor effect observed in C3aR^−/− mice is not augmented by treatment with the C5aR antagonist PMX53 implies that in this model the tumor-promoting effects of C3a are at least as potent as those of C5a, and that the inhibition of either receptor alone is sufficient to impair tumor growth. Furthermore, our results suggest that the tumor-promoting effects of C3aR signaling are mediated via the host immune system, with flow cytometric data showing changes in tumor-infiltrating leukocyte populations in the absence of C3aR signaling. In contrast to C5aR, which has been reported to act via induction of MDSC and inhibition of CD8⁺ CTL responses (6), we show that C3aR deficiency/antagonism leads to a reduction in tumor-associated macrophages, along with increased neutrophil and CD4⁺ T cell populations.

Macrophages are an essential component of tumor-promoting inflammation, and reduced tumor growth in C3-deficient mice has recently been associated with a reduction in tumor-associated macrophages (25). However, this same group showed that the reduction in macrophages was not sufficient to completely ablate tumor susceptibility, suggesting the involvement of other cells or mechanisms. Our demonstration that neutrophil depletion significantly reverses the tumor inhibitory effects observed in C3aR-deficient mice implicates neutrophils as major contributors to the antitumor response. Neutrophils are generally thought to promote tumor cell proliferation, angiogenesis, and metastasis via the production of chemokines, cytokines, and reactive oxygen species; they also contribute to MDSC populations (29). Unfortunately, there are currently no Abs available that can effectively discriminate granulocytic MDSC from other neutrophil populations, but the demonstration that most MDSC within the tumor are removed by treatment with anti-Ly6G Ab indicates that in this model, most of the MDSC are granulocytic. However, in addition to their contribution to tumor-promoting MDSC populations, neutrophils have been shown to exert efficient antitumor activity (30) and inhibit metastatic seeding (31). Indeed, there is evidence that, similar to macrophages, neutrophils can acquire different activation states in response to distinct tumor-derived signals (21), with tumor-associated neutrophils shown to have a unique transcriptional profile that differentiates them from naive bone marrow neutrophils and granulocytic MDSC (32). Thus, the effects of neutrophils on tumor growth may depend on their activation status, such that a shift from acute to chronic inflammation within the tumor microenvironment, with concomitant changes in cytokine expression, converts antitumor effectors to tumor-promoting cells (30). This hypothesis is supported by recent studies showing that neutrophils at the early stages of tumor development are more cytotoxic toward tumor cells (33) and capable of stimulating effective T cell responses (34), but at later stages develop a protumor phenotype.

Although we found no differences in neutrophil expression of activation markers CD62L or CXCR4, gene expression analysis of tumor tissue from C3aR^−/− mice showed upregulation of genes associated with inflammation, including Tlr2, 3, and 4, Nfkb, Il10, Ccl22, Cxcl10, Ccr5, Ccr7, and Cxcr4. In accord with the recent study by Downs-Canner et al. (27), we also showed upregulation of the CCR5 ligand, CCL5 (RANTES) at the protein level, along with a slight, but not significant, increase in the Th2 cytokine IL-10. Increased CCL5 levels have been associated with unfavorable outcomes in some cancers (35), but they predict survival in others (36), regulating antitumor immunity (37) and synergizing with CXCR3 ligands to attract effector T cells into cutaneous metastases and inhibit tumor growth (38). Interestingly, we found a reduction in plasma G-CSF levels in C3aR-deficient mice, a result in line with previous studies showing an inverse correlation between G-CSF and neutrophil counts, and possibly due to negative feedback mechanisms at the transcriptional level (39).

In light of the recently identified role for C3a–C3aR signaling in retention of neutrophils in the bone marrow (20), we propose that C3aR inhibition provides a mechanism by which naive neutrophils are continuously sequestered into the tumor microenvironment, where they promote an antitumor response. This may include the release of cytokines and chemokines that recruit and activate other antitumor effector cells. Alternatively, C3aR signaling may regulate the immunosuppressive (tumor-promoting) activity of MDSC. Both mechanisms are consistent with previous reports showing that an efficient antitumor T cell response can be generated by altering the myeloid cell population within the tumor microenvironment (40).

Analysis of tumor-infiltrating T lymphocyte populations showed that C3aR deficiency/inhibition was associated with relative increases in CD4⁺ (Th1, Th2, and Th17) subsets, whereas the proportion of CD8⁺ T lymphocytes was unaltered. Although effective antitumor responses have been traditionally attributed to CD8⁺ CTL supported by Th1 cells (41), antitumor responses can occur independently of cytotoxic CD8⁺ T cells, mediated directly by CD4⁺ subpopulations (42, 43). Indeed, Th2 cells have been reported to play a critical role in the immune response to B16-F0 melanoma (44) and recruit tumoricidal myeloid cells into the tumor (45). The increase in tumor-infiltrating Th17 cells in the absence of C3aR signaling is in accord with a previous report by Lim et al. (46) who showed in a mouse model of allergic lung inflammation that C3aR-deficient mice had higher frequencies of both neutrophils and Th17 cells. It is also in agreement with reports showing a direct cross-talk between activated neutrophils and Th17 cells, with each cell type producing chemokines and cytokines capable of recruiting the other (47). However, the role of Th17 cells in tumor growth is controversial, with evidence for both antitumor cytotoxic T cell (48) and protumorigenic immunosuppressive responses (49).

In summary, to our knowledge, this study is the first to demonstrate a role for C3aR signaling in promoting tumor growth, and further that the effects of C3aR inhibition are at least as potent, if not more potent, than those of C5aR inhibition. The demonstration that the growth of established tumors can be arrested by C3aR antagonism highlights the therapeutic potential of C3aR-targeting drugs. Importantly, the demonstration that C3aR deficiency/antagonism is protective in murine models of BRAF^V600E mutant melanoma and colon and breast cancer suggests the broader potential of C3aR as a therapeutic target. The results indicate that the protumor effects of C3aR are mediated primarily via the host immune system, with C3aR deficiency/antagonism leading to alterations in tumor-infiltrating leukocyte populations, including neutrophils and macrophages. Although, similar to macrophages, neutrophils are generally thought to promote tumor growth, there is considerable evidence in the literature that they can be manipulated to exert effective antitumor activity. We propose that by altering myeloid cell populations within the tumor microenvironment, C3aR inhibition may change the inflammatory equilibrium, such that it is less favorable to tumor growth. This hypothesis is supported by the demonstration that tumor-infiltrating CD4⁺ T lymphocytes populations are also increased. Further studies are required to determine the precise mechanisms by which C3a–C3aR signaling influences the immune response to melanoma, in particular how C3aR deficiency/inhibition influences neutrophil (or MDSC) function, and whether the same mechanisms apply in other tumor types. Despite the clear evidence that C3a–C3aR signaling regulates tumor growth via effects on the host immune system, the potential for tumor-expressed C3aR to regulate carcinogenesis remains to be explored, particularly in light of the recent demonstration that C3a promotes epithelial–mesenchymal transition in ovarian cancer cells (50). The stage of tumor development and commencement of treatment may also be important, and thus future studies should determine the therapeutic benefits of targeting C3aR in advanced and metastatic cancers. Finally, because both C3a and C5a are required for effective tumor responses to radiotherapy (51), the potential for interactions with other therapeutic modalities should be investigated. Nevertheless, the results highlight the potential of C3aR as a therapeutic target for melanoma and other cancers.

We thank Dr. Geoff Osborne for advice on FACS analysis and Dr. Darryl Whitehead for histological advice.

The authors have no financial conflicts of interest.