IL-34 Induces the Differentiation of Human Monocytes into Immunosuppressive Macrophages. Antagonistic Effects of GM-CSF and IFNγ

IL-34 is a recently identified cytokine that signals via the M-CSF receptor and promotes monocyte survival. Depending on the environment, monocytes can differentiate into macrophages (Mφ) or dendritic cells (DC). A wide spectrum of Mφ and DC subsets, with distinct phenotypes and functions, has been described. To date, the phenotype of monocytes exposed to IL-34 remains unexplored. We report here that IL-34 induces the differentiation of monocytes into CD14high CD163high CD1a− Mφ (IL-34-Mφ). Upon LPS stimulation, IL-34-Mφ exhibit an IL-10high IL-12low M2 profile and express low levels of the costimulatory molecules CD80 and CD86. IL-34-Mφ exhibit poor T cell costimulatory properties, and have potent immunosuppressive properties (decrease of TCR-stimulated T cell proliferation). For all the parameters analyzed, IL-34-Mφ are phenotypically and functionally similar to M-CSF-Mφ. IL-34 appears as efficient as M-CSF in inducing the generation of immunosuppressive Mφ. Moreover, the generation of IL-34-Mφ is mediated through the M-CSF receptor, is independent of endogenous M-CSF consumption and is potentiated by IL-6. In an attempt to identify strategies to prevent a deleterious M2 cell accumulation in some pathological situations, we observed that IFNγ and GM-CSF prevent the generation of immunosuppressive Mφ induced by IL-34. IFNγ also switches established IL-34-Mφ into immunostimulatory Mφ. In conclusion, we demonstrate that IL-34 drives the differentiation of monocytes into immunosuppressive M2, in a manner similar to M-CSF, and that IFNγ and GM-CSF prevent this effect.


Introduction
Circulating monocytes are precursors that exhibit some effector functions [1,2]. They can differentiate into a variety of tissueresident macrophages (MQ), dendritic cells (DC), and osteoclasts [3]. Macrophages play a central role in tissue homeostasis, host defense, inflammation, and tissue repair. They also participate to tissue remodeling in ontogenesis and to the control of metabolic functions [4]. Macrophages, characterized by their plasticity and heterogeneity [4,5,6], are highly sensitive to their environment and, according to the signals encountered (microorganisms, damaged self, cytokines), may acquire distinct functional phenotypes and exhibit opposite activities (i.e. pro-versus antiinflammatory, immunosuppressive versus immunostimulatory) [4]. Pathophysiological situations are usually associated with changes in MQ phenotypes and functions and, in some of them, MQ with different phenotypes coexist [4].
Reflecting the Th1/Th2 dichotomy, two states of MQ polarization, called M1 and M2, which represent extremes of a spectrum of activation states, have been described [6]. The M1 phenotype results from the stimulation of MQ with interferon-c (IFNc), alone or with microbial products (mainly Toll-like receptor [TLR] agonists) or cytokines (such as tumor necrosis factor-a [TNFa]). M1 cells produce high levels of IL-12 (IL-12 high ) and low levels of IL-10 (IL-10 low ), proinflammatory cytokines, nitrogen and oxygen intermediates, and have potent microbicidal and tumoricidal activities [4,7,8]. M2 cells, also called ''alternatively activated macrophages'', were initially described as cells generated in the presence of IL-4 or IL-13. M2, characterized by an IL-12 low IL-10 high phenotype, have potent phagocytic properties and are involved in tissue remodeling, tumor progression, and immune suppression. Other subsets of IL-10 high IL-12 low M2 cells (called M2-like), that differ phenotypically from IL-4-MQ, have also been characterized [6,7,9,10]. However, this classification does not reflect the extreme plasticity of MQ and the continuum of MQ subsets (from M1 to M2) that can exhibit characteristics shared by more than one MQ population. Another classification has thus been proposed, based on three different homeostatic activities: classically activated MQ (involved in host defence and antitumoral immunity), wound healing MQ (production of extracellular matrix components) and regulatory MQ (immune regulation and resolution of inflammation) [11].
IL-34 was identified by functional screening of a library of secreted proteins, based on its ability to support human monocyte survival and to promote, with the same efficiency as macrophage colony-stimulating factor (M-CSF), the formation of the colonyforming unit-macrophage (CFU-M) in human bone marrow cell cultures [12]. Murine IL-34 also promotes peritoneal MQ survival and bone marrow progenitor cell proliferation [13]. IL-34 is a dimeric glycoprotein which presents no amino-acid sequence homology with M-CSF, although human IL-34 secondary structure prediction suggests the existence of four a-helical bundles, as observed for human M-CSF [14,15,16]. The M-CSF receptor (M-CSF-R), also called colony-stimulating factor 1receptor, c-fms or CD115, has been identified as an IL-34 receptor [12]. IL-34 mRNA is expressed in various tissues and is most abundant in the spleen [12]. The spatiotemporal expression patterns of IL-34 and M-CSF mRNA differ, suggesting that IL-34 and M-CSF may have non-redundant and complementary roles [17,18]. In agreement with these observations, two recent studies have evidenced the role of IL-34 in the homeostasis of Langherans cells and microglia [19,20].
When used in combination with receptor activator of nuclear factor kappa-B ligand (RANKL), IL-34 can replace M-CSF for human and murine osteoclast differentiation [21,22]. In vivo, administration of IL-34 reduces bone mass [21,22] and the expression of IL-34 in an M-CSF-specific manner rescues osteopetrotic deficiencies in M-CSF-deficient mice [17]. However, studies suggest that IL-34 and M-CSF may also have nonoverlapping properties [12]. IL-34 and M-CSF differ in their ability to induce the production of the chemokines MCP-1 and eotaxin by human monocytes [13]. They also differently affect the migration of a murine myeloid cell line [13], while their ability to promote the survival of murine MQ or the growth of murine bone marrow progenitors is identical [13]. In support, it has been suggested that IL-34 and M-CSF may interact with distinct regions of M-CSF-R and activate different signaling pathways [13,16]. The potential physiopathological roles of IL-34 remain poorly described. IL-34 has been reported overexpressed in rheumatoid arthritis [23] and involved in RA-associated osteoclastogenesis [24]. In breast cancer, some antitumoral drugs induce the production of both IL-34 and M-CSF by mammary epithelial cells, which may favor the recruitment of MQ into the tumor microenvironment [25].
Taken into account the ability of IL-34 to promote human monocyte survival and the central role played by MQ in physiological and pathological situations, the objective of this study was to analyze the phenotype and function of human MQ generated in the presence of IL-34. We report that IL-34 gives rise to immunosuppressive MQ with an IL-10 high IL-12 low phenotype and demonstrate that granulocyte-macrophage colony-stimulating factor (GM-CSF) and IFNc prevent their generation, while IL-6 pushes them toward a more pronounced phenotype.

Cell purification and culture
Peripheral blood mononuclear cells (PBMC) were isolated with Lymphocyte Separation Medium (PAA Laboratories, Pashing, Austria). CD14 + monocytes and CD4 + T cells were sorted from PBMC by positive magnetic sorting (Miltenyi Biotec). CD4 + memory T cells were sorted from PBMC by negative depletion followed by a positive magnetic sorting of CD45RO + T cells (both from Miltenyi Biotec). Myeloid cells were cultured in medium consisting of RPMI 1640 medium (Lonza, Verviers, Belgium) supplemented with 10% fetal calf serum (PAA Laboratories), 2 mM L-glutamine, 1 mM sodium pyruvate, 0.1 mM non essential amino acids, 10 mM Hepes, 100 U/ml penicillin and 100 mg/ml streptomycin (all from Lonza). CD4 + T cells were cultured in serum-free X-VIVO-20 medium (Lonza).

Cytokine quantification
IL-10 and IL-12 were quantified using commercial ELISA from Diaclone (Dijon, France) and IL-34 using a commercial ELISA from R&D Systems.

Cell viability analysis
The viability of cells was evaluated by annexin V labeling. Data were acquired using a FACScalibur and analyzed with the FlowJo software. Results are expressed as a percentage of annexin V 2 viable cells.

Analysis of chemokine and chemokine receptor mRNA expression
IL-34-MQ, M-CSF-MQ and GM-CSF-MQ were stimulated or not for 6 h with 200 ng/ml LPS. After cell lysis using the Trizol reagent (Life technologies, Saint Aubin, France), total RNA were extracted using the RNeasy Micro kit according to the manufacturer recommendation (Qiagen, Hilden, Germany) and reverse transcribed using the Superscript II reverse transcriptase (Life technologies). The expression of CCL1, CCL2, CCL5, CCL18, CCL17, CCL22, CCL24, CXCL9, CXCL10, CXCL11, CXCL16, CX3CL1, CCR1, CCR2, CCR8, CX3CR1, CXCR1, and CXCR2 mRNA was analyzed by qPCR using the iQ SYBR Green Supermix (Biorad, Marnes-la-Coquette, France); primer sequences are available upon request. Specific gene expression was calculated using the 2 2DDCT method using GAPDH as calibrator. The transcriptional profiles were evaluated in four donors. Omics Explorer 2.3 software was used for hierarchical clustering and principal-component analysis (PCA) (Qlucore, Lund, Sweden). Hierarchical clustering of both samples and transcripts was performed using average linkage method and the Euclidean metric, with each variable being normalized to mean 0 and variance 1.

Inhibition of T cell proliferation and mixed lymphocyte reaction
In T cell proliferation assays, 10 5 allogeneic CD4 + CD45RO + memory T cells, previously incubated with 3 mM CFDA-SE (Molecular Probes, Carlsbad, CA), were cultured in X-VIVO-20 medium supplemented with 20 U/ml IL-2, in 96-well flatbottomed plates, previously coated with 1 mg/ml anti-CD3 mAb (clone OKT3; ATCC, Manassas, VA), without or with 2610 4 allogeneic LPS-activated MQ subsets. In MLR experiments, 10 5 CFDA-SE stained CD4 + T cells were cultured, as described above, in non-coated plates without or with 2610 4 allogeneic LPSactivated MQ subsets, in the presence of 20 U/ml IL-2. At day 7, cells were stained with APC-labeled anti-CD4 mAb for gating on T cells and the proliferation of CD4 + T cells was measured by analyzing CFDA-SE dilution. Data were subsequently analyzed using the FlowJo Proliferation Platform software v7.6.5 (Tree Star, Ashland, OR).

Statistical analysis
Data are shown as mean 6 SD and were analyzed by the Mann-Whitney test, with p,0.05 being considered significant.

IL-34 induces the differentiation of human monocytes into macrophages
In order to investigate the role of IL-34 on human monocyte differentiation, highly purified peripheral blood monocytes were cultured with IL-34 for 5 days. Preliminary experiments showed, as previously reported [12,13], that IL-34 promotes monocyte survival; in our experimental conditions, the maximal survival was obtained with a concentration of 50 ng/ml IL-34 (data not shown). The phenotype of cells cultured with IL-34 was analyzed by flow cytometry and their ability to produce IL-10 and IL-12 was assessed by ELISA. IL-34 induced monocyte differentiation into cells exhibiting MQ characteristics, as evidenced by the expression of CD14 and CD163 (Fig. 1A), but not dendritic cell features (absence of CD1a and no induction of CD83 upon LPS stimulation; data not shown). These differentiated cells were called IL-34-MQ. As controls, the macrophage differentiation factors GM-CSF (GM-CSF-MQ) and M-CSF (M-CSF-MQ) also gave rise to CD1a 2 CD14 + CD163 + cells ( Fig. 1A and data not shown) [7,10,26]. IL-34-MQ and, as previously reported [10], M-CSF-MQ, expressed higher levels of CD14 (2.5-fold increase) and CD163 (25-fold increase) than GM-CSF-MQ (Fig. 1A). IL-34 and M-CSF were equivalent in maintaining the level of CD14 (Fig. 1B, left panel) and in increasing the expression of CD163 (Fig. 1B, right panel) during the 5-days differentiation process. Interestingly, the induction of CD163 and the expression of CD14 were homogenous on the whole cell population (Fig. 1B), suggesting that IL-34 and M-CSF induce the differentiation of monocytes into a MQ subset. In contrast, GM-CSF decreased the expression of CD14 and did not induce the expression of CD163 (Fig. 1B, right  panels). Finally, the expression of MHC-I, MHC-II, CD54, CD80, CD86, B7-H1, ILT3 and CD206 was similar on IL-34-MQ and M-CSF-MQ and was different to the one exhibited by GM-CSF-MQ (Table 1).
These results show that IL-34 supports the differentiation of human peripheral blood monocytes into MQ.
We have previously demonstrated that IL-6 favors M-CSF consumption by monocytes and, in the presence of M-CSF, gives rise to cells with a more pronounced M2 phenotype [10,28]. Similarly, cells cultured with IL-34 and IL-6 (called IL-34+IL-6-MQ) express higher levels of CD163 (Fig. 2B) and ILT3 (Fig. 2E) than IL-34-MQ. Interestingly, the concentrations of IL-34 were reduced in the supernatants of cells cultured with IL-34 plus IL-6 compared to cells cultured with IL-34 alone (Fig. 2F). This results, added to the fact that no IL-34 mRNA was detected during the 5days culture with IL-34, in the absence or presence of IL-6 (data not shown), suggests that IL-6 enhances the consumption of IL-34 during the differentiation process.
Previous studies suggested that M1 and M2 exhibit different chemokine and chemokine receptor patterns [6,29,30,31]. PCA of the transcriptional profiles evidenced a relationship between samples, according to the polarization, and revealed the absence of segregation between IL-34-MQ and M-CSF-MQ (Fig. 3A). Moreover, IL-34-MQ and M-CSF-MQ have lower levels of the  (Fig. 3B). We also observed that the transcripts encoding the chemokine receptors CCR2 and CXCR1 were lower in IL-34-MQ and M-CSF-MQ than in GM-CSF-MQ (Fig. 3B). Stimulation of the MQ with LPS did not modify this dual expression (Fig. 3). The expression of the mRNA encoding the other chemokines and chemokine receptors were equivalent in the three MQ subsets (Fig. 3B).
Together, these results show that IL-34-MQ exhibit a CD14 high CD163 high CD80 low CD86 low M2 phenotype and that IL-6 pushes them into cells with a more pronounced phenotype.

IL-34-MQ are immunosuppressive
Macrophages participate to the control of T cell responses [4,6,10]. We have thus evaluated the ability of IL-34-MQ to affect T cell proliferation. In agreement with studies showing that IL-10 high IL-12 low M2 are immunosuppressive [4], we observed, in mixed lymphocyte reaction (MLR) assays, that IL-34-MQ exhibitd lower T cell costimulatory properties than GM-CSF-MQ (Fig. 4A). Moreover, IL-34-MQ suppressed TCR-dependent T cell proliferation more efficiently than GM-CSF-MQ (Fig. 4B), showing that they present potent immunosuppressive properties.
These results show that IL-34 drives the differentiation of monocytes into immunosuppressive MQ.

IL-34 induces M2 via CD115 and independently of M-CSF
M-CSF is produced by monocytes and induces their differentiation into M-CSF-MQ by acting via CD115 [10,32]. We therefore evaluated whether the ability of IL-34 to induce monocyte differentiation into M2 may also occur through autocrine secretion of M-CSF. A neutralizing anti-M-CSF mAb did not affect the ability of IL-34 to induce the generation of IL-34-MQ, as evidenced by the maintenance of CD14 expression (Fig. 4C, left panel), the increase of CD163 expression (Fig. 4C, right panel) and the acquisition of an IL-10 high /IL-12 low phenotype (Fig. 4D). The expression of CD80 and CD86 by IL-34-MQ was not modulated by an anti-M-CSF mAb (data not shown). In parallel, we analyzed the expression of M-CSF during the course of monocyte differentiation into IL-34-MQ. No M-CSF was detected in the culture supernatants collected from D1 to D5 and the expression of the transcript encoding M-CSF was not modulated by IL-34 during the 5-days culture (data not shown). In contrast, monocytes exposed to M-CSF plus an anti-M-CSF mAb rapidly died in culture (Fig. 4C&D), a result in agreement with the fact that M-CSF is a survival factor for monocytes/ macrophages.
In agreement with the fact that IL-34 and M-CSF signal via CD115, the c-fms tyrosine kinase inhibitor GW2580 [12] impaired the survival of monocytes cultured with IL-34 or M-CSF (Fig. 4C-D). Finally, we observed that the viability of cells cultured with IL-34 and M-CSF was similar from D1 to D5 (Fig. 4E and data not  shown).
Collectively, these results demonstrate that IL-34 switches monocytes into M2 by acting through CD115, in an M-CSFindependent manner.
These results show that GM-CSF and IFNc counteract the ability of IL-34 to generate immunosuppressive MQ. IFNc skews established IL-34-MQ cells from an immunosuppressive to an immunostimulatory IL-12 high IL-10 low phenotype Macrophages are characterized by their plasticity, and dynamic changes in MQ polarization may occur in some pathologies [4]. We previously showed that cytokines that prevent the generation of M-CSF-MQ may also induce their reversion into immunostimulatory M1 [34]. We thus evaluated whether GM-CSF and IFNc may also revert the phenotype of established IL-34-MQ. Upon IFNc exposure, IL-34-MQ recovered the capacity to produce high levels of IL-12 (Fig. 5D), and to express CD80 and CD86 in response to LPS (Fig. 5E), while they expressed lower levels of IL-10 ( Fig. 5D), CD14 and CD163 (Fig. 5F) than IL-34-MQ. IL-34-MQ, exposed to GM-CSF for 3 days, recovered the ability to express CD80 and CD86 in response to LPS (Fig. 5E) but remained unable to produce IL-12 (Fig. 5D). The ability of GM-CSF and IFNc to modulate the expression of these markers was dose-dependent, significant at 2 or 20 ng/ml, depending on the parameters analyzed, and maximal at the highest concentration tested (50 ng/ml) (Fig. S3).
In addition to confirming the plasticity of MQ [6,34], these results show that IFNc switches established IL-34-MQ cells into MQ with an immunostimulatory phenotype.

Discussion
IL-34 signals via M-CSF-R and promotes human monocyte survival [12,13,17]. Different observations suggest that IL-34 and M-CSF might have non-redundant and complementary roles [13,17,18,19,20]. As MQ participate to the control of inflammation, tissue homeostasis and metabolism, to identify the role of IL-34 on monocyte differentiation may have potential therapeutic applications, especially when the relative expression of IL-34 and M-CSF in pathological situations will be established. To date, the phenotype of the human myeloid cells exposed to IL-34 remains unexplored. We show here that (i) IL-34 gives rise to immunosuppressive MQ with a phenotype similar to the one induced by M-CSF, (ii) that GM-CSF and IFNc prevent their generation and, (iii) that IFNc skews established IL-34-MQ into immunostimulatory IL-12 high IL-10 low M1. Previous studies have reported that M1 and M2 exhibit different profiles of chemokines and chemokine receptors [29,30,31]. We observed that IL-34-MQ and M-CSF-MQ are indistinguishable, based on selected chemokines and chemokine receptors. Interestingly, GM-CSF-MQ express higher levels of the chemokines CCL1, CCL17, CCL22, CCL24 and CX3CL1, and of the chemokine receptors CCR2 and CXCR1, than IL-34-MQ and M-CSF-MQ. These molecules may constitute, in addition to the reliable markers IL-10 and IL-12, a molecular signature allowing discriminating M1 and M2.
Although it has been suggested that M-CSF and IL-34 may differentially modulate the production of some cytokines by monocytes [13], our results show that the effects of IL-34 and M-CSF on monocyte differentiation (in term of phenotypes and functions) are superimposable, a result consistent with data from others showing that human IL-34 and M-CSF are equivalent in inducing the proliferation of a mouse macrophage cell line transfected with the human M-CSF-R [17]. Our results also support a previous study showing that IL-34 was as effective as M-CSF in inducing the survival/proliferation of human monocytes and the formation of CFU-M [12]. Although our results do not allow excluding the existence of another receptor for IL-34, distinct from M-CSF-R, as previously suspected [17,18], we demonstrate, in agreement with others [12,13], that IL-34 acts on human monocytes via the M-CSF-R. Moreover, we also report that IL-34 acts on monocytes independently of endogenous M-CSF. Reciprocally, although human monocytes express IL-34 mRNA, we failed in detecting IL-34 secretion by monocytes in response to M-CSF, suggesting that M-CSF may also act independently of endogenous IL-34. Finally, in support of our results showing that IL-34 mimics the activity of M-CSF on the differentiation of human monocytes, it has been reported by others that IL-34 can replace M-CSF for osteoclast generation [21,22].
IL-34 and M-CSF mRNA are expressed in different tissues and differences in their spatiotemporal expression have been reported [17,18,19,20]. Moreover, no compensatory IL-34 mRNA expression has been observed in M-CSF-deficient mice [17]. Both these observations suggested that IL-34 and M-CSF have complementary and independent roles. More precisely, IL-34 mRNA levels are higher than M-CSF mRNA in skin, salivary glands, and in most areas of the brain [17,35,36]. Based on our results, it is tempting to speculate that, in these organs, IL-34, rather than M-CSF, may contribute to the generation of immunosuppressive M2, able to tune inflammation and to promote tissue repair. Interestingly, microglia and bone marrow-derived MQ are tightly down-regulated to avoid excessive and detrimental immune responses. Our results suggest that IL-34, expressed by neurons and meningeal cells [37], may contribute to the generation and/or to the maintenance of the phenotype of brain-resident myeloid cells.
We report that IL-34 induces the generation of MQ which a phenotype and functions similar to the ones of human tumorassociated macrophages (TAM) [10]. TAM accumulate in various solid tumors, and, in most of them, their density is correlated with poor prognostic [4,5,10]. In addition to immunosuppressive properties, TAM are trophic for tumors and favor angiogenesis, dissolution of connective tissues and metastasis [4,5,6,10]. In additional experiments, we have also observed that IL-34-MQ produce similar levels of TGFb, MMP9, and VEGF than M-CSF-MQ (personal data). TAM derive from circulating monocytes that differentiate locally into M2 in response to tumor microenviron-mental factors, such as M-CSF and IL-6 [10]. Recently, DeNardo and coll reported that, in breast cancer, some antitumoral drugs induce the production of both IL-34 and M-CSF by mammary epithelial cells, suggesting that, in some circumstances, therapeutic strategies may favor tumor immune escape [25]. Based on these observations, it appears crucial to analyze, in various tumors, the relative expression of M-CSF and IL-34 by tumoral and tumorassociated stromal cells, in order to determine whether IL-34 could participate to tumor-induced immunosuppression and to determine the relative roles of IL-34 and M-CSF in different types of tumors.
While M2 maintain tissue homeostasis and prevent excessive immune responses, their accumulation in some pathologies, such as in tumor, is detrimental. Strategies to prevent cell generation and/or to reverse the immunosuppressive properties of M2 may increase the efficacy of anti-tumor immunotherapies. Interestingly, we observed that IFNc (i) switches established IL-34-MQ into immunostimulatory M1 and, (ii) prevents the generation of IL-34-MQ. IFNc is a potent activator of myeloid cells inducing the generation of immunostimulatory M1 and, as previously observed, preventing the generation of TAM-like in ovarian cancer [34]. In addition, GM-CSF was able to prevent IL-34-MQ generation. Recently, by analyzing the transcriptional profiles of GM-CSFand M-CSF-induced MQ, it has been observed that the expression of the M-CSF-induced genes was counteracted by the addition of a low dose (1 ng/ml) of GM-CSF [38]. We also observed that GM-CSF (used at 1 to 20 ng/ml), prevented the generation of M-CSF-MQ (personal unpublished data). Supporting these observations, GM-CSF suppresses, in a murine myeloid cell line, the expression of the M-CSF-R mRNA [32] through the transcriptional activation of a ribonuclease degradation system [27]. In contrast, IFNc does not modulate M-CSF-R mRNA expression by monocytes [39]. As IL-34 signals via M-CSF-R, this result may explain why GM-CSF prevents the effect of IL-34 on monocytes while GM-CSF appears less potent than IFNc in reverting the immunosuppressive phenotype of IL-34-MQ, although we cannot exclude that this could be achieved after a prolonged time of exposure.
In conclusion, our results identify IL-34 as a novel factor involved in the control of MQ polarization. By acting through the M-CSF-R, IL-34 switches monocytes into immunosuppressive MQ. IL-34 may thereby play a central role in immune homeostasis, especially in the brain and skin where IL-34 is constitutively expressed. Our results also suggest that, in order to reverse tumor-associated immunosuppression mediated by myeloid cells, strategies based on the inhibition of M-CSF-R should be privileged to prevent the effects of both IL-34 and M-CSF.

Supporting Information
Text S1 Supporting materials and methods.