Engagement of the Mannose Receptor by Tumoral Mucins Activates an Immune Suppressive Phenotype in Human Tumor-Associated Macrophages

Tumor-Associated Macrophages (TAMs) are abundantly present in the stroma of solid tumors and modulate several important biological processes, such as neoangiogenesis, cancer cell proliferation and invasion, and suppression of adaptive immune responses. Myeloid C-type lectin receptors (CLRs) constitute a large family of transmembrane carbohydrate-binding receptors that recognize pathogens as well as endogenous glycoproteins. Several lines of evidence demonstrate that some CLRs can inhibit the immune response. In this study we investigated TAM-associated molecules potentially involved in their immune suppressive activity. We found that TAMs isolated from human ovarian carcinoma samples predominantly express the CLRs Dectin-1, MDL-1, MGL, DCIR, and most abundantly the Mannose Receptor (MR). Components of carcinomatous ascites and purified tumoral mucins (CA125 and TAG-72) bound the MR and induced its internalization. MR engagement by tumoral mucins and by an agonist anti-MR antibody modulated cytokine production by TAM toward an immune-suppressive profile: increase of IL-10, absence of IL-12, and decrease of the Th1-attracting chemokine CCL3. This study highlights that tumoral mucin-mediated ligation of the MR on infiltrating TAM may contribute to their immune suppressive phenotype.


Introduction
Among tumor-infiltrating leukocytes, Tumor-Associated Macrophages (TAMs) constitute a major subset [1][2][3]. While the presence of T lymphocytes in tumor stroma is usually correlated with more favourable prognosis of cancer patients [4,5], in most studies the density of TAM is associated with rapid tumor progression [6][7][8][9]. TAMs are poorly cytotoxic against neoplastic cells and may actually favour tumor cell survival and proliferation by actively producing growth factors for cancer and endothelial cells. They are also a major source of proteolytic enzymes that degrade the extracellular matrix thus favouring the invasion of neoplastic cells [9,10]. Further, TAM contributes to the evasion of tumors from immune control by producing immune-suppressive cytokines such as IL-10 and TGFβ [2,9].
Our group proposed that TAMs are M2-like polarized macrophages [11]. Along a conventional definition, macrophages activated in the presence of inflammatory mediators (e.g., LPS) and Th1 cytokines (e.g., IFNγ) are defined as M1 or classically activated macrophages. These effectors have high cytotoxic functions, produce immunestimulatory cytokines, and are important cells for the defense against intracellular pathogens and transformed cells [11][12][13]. On the other hand M2 macrophages, or alternatively activated macrophages, differentiate in microenvironments rich in anti-inflammatory mediators or Th2 cytokines (e.g., IL-10, IL-4, IL-13); they have high scavenging activity, produce several growth factors, activating the process of tissue regeneration, and suppress adaptive immune responses [2,12,14,15]. While these activities are of extreme importance during wound healing to return to the homeostatic state, in the context of a growing tumor they are deleterious to the host. Indeed several studies have found a strong correlation between high numbers of TAM, number of vessels, and lower disease-free survival [9,10,[16][17][18].
In this study we investigated TAM-related mechanisms of immune escape and considered C-type lectin receptors (CLRs) as interesting candidates. CLRs are a large family of structurally related transmembrane receptors which, by virtue of their carbohydrate-recognition domains, bind high affinity sugar moieties present on the surface of pathogens as well as of endogenous glycoproteins [19,20]. Together with the Toll-like receptor (TLR) family, CLRs expressed on myeloid cells of the innate immunity constitute the major system to sense the outer world [21][22][23]. Myeloid CLRs are subdivided into two major families: the first including the mannose receptor (MR, DEC206), ENDO180, DEC205, and PLA2 receptor, and the second including DC-SIGN, Dectin-1, Langerin, DCIR, MGL, and BDAC2 [24,25].
Although the majority of studies have investigated the role of CLRs in the recognition, internalization, and clearance of pathogens through activation of innate immunity, other studies have clearly demonstrated that at least some receptors elicit anti-inflammatory/immune-suppressive responses, raising the hypothesis that pathogens may exploit CLR binding and internalization ability to overcome innate immunity and survive within the host [22,24,25]. For instance, M. tuberculosis binds DC-SIGN and the MR and inhibits the production of IL-12 [26][27][28]. We previously reported in mono-DC that cross-linking of the MR with an agonist antibody increased IL-10 production leading to inhibition of IL-12 and defective Th1 differentiation [27]. Activation of BDAC2, expressed by plasmacytoid DC, as well as DCIR, downregulates the production of IFNα [29,30]. DCIR also interferes with GM-CSF signalling [31]. The receptor MGL recognizes the isoform CD45RB expressed by effector memory T cells and negatively influences T cell receptor signalling [32]. Thus, several lines of evidence point to a role of at least some CLRs in the restriction of inflammatory reactions and in homeostasis preservation [24,25].
Previous studies on CLRs have been mainly performed with in vitro generated macrophages and DC or with in vivo mouse models of diseases. The aim of this study is to investigate the expression of CLRs in TAM. Here we show that TAM isolated from human ovarian carcinoma samples predominantly expressed Dectin-1, MDL-1, MGL, DCIR, and most abundantly the MR. Experiments demonstrated that the MR recognizes ligands present in carcinomatous ascites and tumoral mucins such as CA125 and tumor-associated glycoprotein-(TAG-) 72. Upon mucin-engagement of the MR, TAMs secrete higher levels of IL-10 and lower levels of the T cell attracting chemokine CCL3. Thus, tumoral mucin-mediated activation of the MR on TAM triggers an immune-suppressive response which likely contributes to tumor immune evasion.

Isolation of Human Tumor-Associated Macrophages (TAMs).
Having obtained an informed consent, we collected carcinomatous ascites and/or tumor samples from 27 patients with histologically confirmed ovarian tumors. TAMs were isolated from ascites by density Ficoll, and Percoll gradients (Lonza, Italy) as described in [37]; TAMs from solid tumors were isolated by enzymatic digestion and Ficoll gradient [38] and were further purified by plastic adherence (RPM1 1640 w/o FBS, 1h, 37 • C); adherent cells were usually 80-90% CD68+ macrophages as assessed by flow cytometry.

Transcriptional
Profile Analysis. TAMs from 7 different patients (5 from carcinomatous ascites and 2 from solid tumors) were used for the transcriptional profiling experiments. TAMs were either immediately processed or after 18-h stimulation with LPS (100 ng/mL) (Sigma, Italy) or IL-10 (20 ng/mL) (Peprotech,Italy) (for 4 TAM preparations). Macrophages from the peritoneal free-fluid of nontumoral patients (ovarian cysts) were collected during surgery from 12 different patients, centrifuged over Ficoll and immediately processed for RNA (purity >90%). Total RNA was extracted from 5×10 6 cells using Trizol (Invitrogen Life Technologies), retrotranscribed and prepared for GeneChip hybridization as previously described [38]. Each TAM preparation was independently tested. Macrophages from nontumoral patients were pooled to reach the minimum necessary amount for 1 GeneChip. Fragmented cRNA was hybridized to HG-U133 Plus 2.0 GeneChips (Affymetrix) and then washed and scanned according to manufacturer's guidelines. Expression measures were computed using Robust Multiarray Average (RMA). Statistical differences were assessed by a moderated t-test analysis performed using a Limma bioconductor package, and resulting P-values were adjusted using the Benjamini and Hochberg step-up method for controlling the False Discovery Rate (FDR). Genes were defined as regulated when characterized by a fold of induction ≥2 and an FDR Pvalue ≤.05. Computations were conducted using the R statistics programming environment (http://www.r-project.org/).

Phenotype Analysis.
Tumor macrophages were analysed by flow cytometry on FACS Canto (BD Bioscience, Milan, Italy). Cells were first incubated with PBS 1% HS (30 minutes 4 • C) to block FcγR, washed and resuspended in FACS buffer (PBS 0.5% BSA, 0.05% NaN 3 ). PE-mouse antihuman CD14 (clone M5E2) was purchased from (BD Pharmingen, Italy). Three of mouse anti-human MR/CD206 were used with identical results. Clone 19.2 was from BD Pharmingen; clone PAM-1 was previously characterized [27]; clone WE458 was in-house generated and selected for reactivity against MRtransfected CHO cells.

Endocytosis Assay.
Mannose receptor-mediated endocytosis was measured as the cellular uptake of FITC-dextran and quantified by flow cytometry. Approximately 2 × 10 5 cells per sample were incubated in media containing FITCdextran (1 mg/mL) (molecular weight 40,000; Sigma) over a period of 60 min. After incubation, cells were washed twice with phosphate-buffered saline (PBS) to remove excess dextran and fixed in cold 1% formalin. Endocytosis was expressed as fluorescence intensity, calculated as mean fluorescence intensity of positive cells at 37 • C-mean fluorescence intensity of positive cells at 4 • C.

Immunohistochemistry.
Human surgical samples of ovarian tumors were immediately frozen in OCT after surgical collection. Sections were stained with anti-CD206 mAbs, followed by a goat antimouse secondary antibody (EnVision horseradish peroxidase rabbit/mouse, DakoCytomation). After a diaminobenzidine reaction (Liquid DAB + Substrate Chromogen System, Dako Cytomation), sections were counterstained with hematoxylin (Mayer, DIAPATH).

Statistical Analysis. Prism software (GraphPad) and
Microsoft Excel were used for all statistical analyses. Student's t tests were used to determine statistically significant differences between experimental groups. P < .05 was considered to be statistically significant.

C-Type Lectin Receptor Gene Expression in Human TAM.
To study the expression of CLRs in TAM we interrogated our Affymetrix database performed with 7 different populations of purified TAM isolated from human ovarian carcinoma (5 from carcinomatous ascites and 2 from solid tumors). CLR gene expression levels from TAM of solid tumors or from ascitic fluids were similar and were considered together. The most expressed CLR genes were the mannose receptor (Mrc1, CD206), Dectin1, DCIR MDL-1, and MGL-1 ( nontumoral patients. As the amount of free-fluid and the cellular content is usually very small, we pooled the samples from 12 different subjects who underwent surgery for nonneoplastic diseases (ovarian cysts) and analyzed with the same GeneChips (Affymetrix) used with TAM. Figure 1 shows that the relative expression of the eleven CLRs analyzed was similar between TAM and normal tissue macrophages, though some differences were noted: Mrc1, MDL-1, and MGL-1 were higher in TAM, while Dectin1 and DCIR and DCL-1 were higher in normal macrophages.
On the basis of these findings we further investigated the mannose receptor (MR) in human TAM.

Phenotype and Functional Activity of the MR in Human TAM.
To check for protein expression, 12 different preparations of human TAM were purified from the ascitic fluid of patients with ovarian carcinoma and tested by flow cytometry. Figure 2(a) shows the results of each individual preparation as percentage of MR (CD206) and of CD14, used as a pan-myeloid marker. MR expression was variable and ranged from 17% to 72% (median value 39%). Such heterogeneity is likely due to the fact that the MR is an endocytic receptor that continuously shuttles from the cell membrane to the early endosome compartments.
Immunohistochemistry of surgical samples of human ovarian cancer was performed with two different anti-CD206 mAb. Macrophages infiltrating the tumor stroma showed strong reactivity (Figure 2(b)); these results confirmed that the MR is expressed both by ascitic fluid macrophages and by TAM infiltrating solid tumors.
To evaluate MR ability to internalize soluble particles, we incubated TAM with FITC-Dextran, a known ligand of MR. TAM rapidly internalized FITC-Dextran over 60 period, with a kinetic similar to that of normal macrophages differentiated in vitro with M-CSF (Figure 3(a)). Receptor specificity was checked by pretreating cells with a blocking anti-MR mAb, which resulted in significant inhibition of internalization (Figure 3(b)). Pretreatment of TAM with tumoral ascites (33% v/v) reduced by 50% FITC-Dextran   endocytosis, suggesting that ascitic fluids contained putative ligand(s) of the MR. To have further proof of this, we incubated TAM with ascitic fluids prior to staining with anti-CD206 mAb and analyzed in flow cytometry. Figure 4(a) shows that tumoral ascites did induce the internalization of the MR from the surface of TAM and of normal macrophages: the percentage of surface MR decreased by 60-80% while that of CD14 was unaffected (Figure 4(b)). It is established that the MR can bind to MUC1 mucin and to the tumoral mucin TAG-72 [35,36,39,40]. We therefore tested the mucin CA125 that is specifically associated with ovarian cancer. Pretreatment of TAM with TAG-72 or CA125 decreased MR expression, indicating that also CA125 is able to engage the MR and to induce its internalization (Figure 4(c)).
Notably, both unconjugated Dextran and the other MR ligands tested did not completely block endocytosis or inhibit receptor expression, most likely because-as mentioned above-MR has high recycling ability. In support of this, we noticed that in vitro culture of TAM for 24 hours in the absence of ascitic fluid (i.e., out of the original microenvironment), resulted in higher MR levels compared to TAMs that were immediately tested after isolation (not shown).

Mucin-Mediated Ligation of the MR Modulates Cytokine
Production in Human TAM. We previously reported in dendritic cells (DCs) that activation of the MR with an agonist mAb or with MUC1 induced a regulatory/immunosuppressive phenotype with a switch of cytokine production characterized by low IL-12 and high IL-10 [27]. Hence, we tested cytokine production of mucin-treated TAM. Figure 5(a) shows that all tested MR ligands (TAG-72, CA125, and anti-CD206) induced a significant increase of IL-10 in TAM as well as in normal macrophages. By contrast, IL-12 secretion was strongly decreased in normal macrophages ( Figure 5(a)). TAMs, as already reported [41], are unable to produce IL-12 even after optimal stimulation with LPS and IFNγ ( Figure 5(a)).
Macrophages and DC are a major source of chemokines which importantly amplify the immunological network by recruiting immunocompetent cells at tumor tissues. We investigated the production of the chemokine CCL3, which recruits Th1 and cytotoxic effector lymphocytes. Figure 5(b) shows that TAG-72 mucin strongly inhibited the secretion of CCL3 by TAM and by in vitro generated macrophages and mono-DC.
Overall these results demonstrate that the MR expressed by TAM recognizes endogenous ligands present in the tumor microenvironment, including the ovarian cancer specific mucin CA125. Mucin-induced MR engagement modulates the cytokine secretion of TAM toward an immunesuppressive phenotype: increase of IL-10, absence of IL-12 and decrease of CCL3. This cytokine profile is likely to contribute to tumor immune escape.

Discussion
Very little is known about the expression and functional role of CLRs in myeloid cells infiltrating tumors. This study demonstrates that human TAMs express a number of CLRs (e.g., Dectin-1, MDL-1, MGL, DCIR) and most prominently the MR/CD206. Other receptors were not significantly expressed (e.g., DC-SIGN, DEC205, Langerin), in line with their preferential localization on dendritic and Langerhans cells. ENDO180, which shares similarities with the MR, was also poorly represented, and this finding is consistent with its higher expression in fibroblasts [42]. We focused our attention on the MR. Although it has long been known that TAMs bear this receptor-and actually this evidence served as paradigm of their M2-like polarizationno functional characterization of MR-positive TAM has ever been provided.
We found that both TAMs from solid tumors and those from the ascitic fluid associated to advanced ovarian cancer, have high membrane expression of the MR. Levels of  expression were modulated by the tumor microenvironment, as components present in the ascitic fluid were able to induce receptor internalization.
The MR is one of the oldest CLR described in macrophages [43,44]. It is an endocytic and phagocytic receptor that binds carbohydrate moieties on several pathogens such as bacteria, fungi, parasites, and viruses and is therefore considered a Pattern Recognition Receptor (PPR). However, it has become increasingly clear that the MR is importantly involved in the silent clearance of circulating inflammatory molecules and degraded matrix components. Mice deficient in the MR do not show increased susceptibility to infections [45,46] but have elevated levels of lysosomal hydrolases and other glycoproteins which raise up during inflammation and tissue remodelling [47,48]. These in vivo experiments highlighted its important role in the clearance of unwanted molecules, especially for the MR localized at hepatic sinusoids.
Not only MR appears to be dispensable for pathogen clearance but also it can negatively modulate pathogenelicited immune responses. We and others have previously reported that MR-ligation with ManLAM from M. tubercolosis or with an agonist anti-MR antibody modulates cytokine production in human DC, with a shift from high to low IL-12, increased IL-10 levels, and defective Th1 immune responses [26,27]. These results have been confirmed in this study in tumor macrophages activated with agonist anti-MR mAbs. The mechanisms that account for the regulatory functions of the MR are not completely characterized. Unlike other CLRs, MR has no ITIM domain [25]. It has been shown that some CLRs may interfere with TLRs/NF-kB signalling [25,49], and the MR can indeed physically interact with TLR2 upon internalization [50]. In addition, Pathak et al. reported that mannan induced the upregulation of IRAK-M kinase, which was responsible for the decreased production of proinflammatory cytokines by inhibiting TLR-signaling [51,52].
A number of recent studies corroborated the hypothesis that the MR is implicated in the maintenance of homeostasis and tolerance. Macrophages cocultured with mesenchymal stem cells have high MR expression and produce IL-10 [53]. Royer   DC strongly impaired Th2 development [54]. Macrophages localized at sites where inflammation could be particularly harmful are usually strongly MR-positive (e.g., alveolar macrophages and brain microglia). Further, at the maternalfoetal interface the presence of immune cells is important to preserve tolerance as well as for active remodelling of uterine vessels. Decidual macrophages are a major source of IL-10 and IDO [55] and express high levels of the MR [40]. MR recognizes several endogenous ligands and acts as a bridge between innate immunity and homeostasis [25,56]. For instance, circulating hormones, like lutropin and thyrotropin are bound by the MR cystein-rich domain [57]. Collagen is another MR-ligand and the receptor may serve important scavenger functions [58].
In the context of a tumor microenvironment, where highly glycosylated molecules such as mucins are present [59] CLRs encounter several putative ligands. MGL and DC-SIGN recognize cancer-specific glycosylation changes of the mucin MUC1, in particular the carbohydrate sialyl Lewis X and the sialyl TN epitope [60]; MUC1 and TAG-72 bind also the MR [27,35,36]. We previously reported that mono-DCs differentiated in the presence of tumor cell-derived mucins have a tolerogenic/regulatory cytokine profile [34]. In the present study we extended this observation to tumor macrophages: TAMs bound and internalized both TAG-72 and the ovarian cancer-associated mucin CA125 via the MR, indicating a specific recognition by this receptor. Further, these mucins interfered with the LPS-induced production of IL-10 and of the chemokine CCL3. These results are in line with the observation that another tumoral mucin, the carcinoembryonic antigen (CEA) highly expressed by colon cancer cells, binds DC-SIGN on DC and induces increased secretion of IL-10 and IL-6 [61]. Hence, evidence is accumulating that CLR recognition of tumor glycans leads to the expression of the potent immunoregulatory cytokine IL-10. In the tumor microenvironment IL-10 has detrimental effects on immune responses as it promotes the polarization of M2 macrophages inhibits the differentiation 8 Clinical and Developmental Immunology of Th1 lymphocytes while favouring that of Treg [62]. Interestingly, a recent study showed that distinct TAM subsets can be distinguished on the basis of differential expression of MHC II molecules. TAMs characterized by MHC II low and suppressive activity on T cell proliferation have higher expression of the MR [63].
In addition, MR is expressed by endothelial cells of lymphatic vessels [64] and it has been demonstrated to be implicated in the dissemination of tumor cells along lymphatics [33]. Recently, Arteta et al. reported that MR-positive liver sinusoidal vessels also support hepatic metastasis of colon cancer cells by a mechanism that involves IL-1-induced upregulation of the MR [65].
Thus, while under physiological conditions the regulatory effect of CLRs on innate immunity cells is finalized to the preservation of homeostasis, in pathological conditions such as cancer, CLR activity may hamper the activation of a protective immune response and actually favour tumor spread.
In conclusion, we have demonstrated that the MR on human TAM can be engaged by mucins present in the tumor microenvironment. This interaction further enhances their immunosuppressive phenotype and can be considered as another mechanism of tumor immune evasion.