Comparison of polarization properties of human adult microglia and blood-derived macrophages

Authors


Abstract

Both microglia, the resident myeloid cells of the CNS parenchyma, and infiltrating blood-derived macrophages participate in inflammatory responses in the CNS. Macrophages can be polarized into M1 and M2 phenotypes, which have been linked to functional properties including production of inflammation association molecules and phagocytic activity. We compare phenotypic and functional properties of microglia derived from the adult human CNS with macrophages derived from peripheral blood monocytes in response to M1 and M2 polarizing conditions. Under M1 conditions, microglia and macrophages upregulate expression of CCR7 and CD80. M2 treatment of microglia-induced expression of CD209 but not additional markers CD23, CD163, and CD206 expressed by M2 macrophages. M1-polarizing conditions induced production of IL-12p40 by both microglia and macrophages; microglia produced higher levels of IL-10 under M1 conditions than did macrophages. Under M2 conditions, microglia ± LPS produced comparable levels of IL-10 under M1 conditions whereas IL-10 was induced by LPS in M2 macrophages. Myelin phagocytosis was greater in microglia than macrophages under all conditions; for both cell types, activity was higher for M2 cells. Our findings delineate distinctive properties of microglia compared with exogenous myeloid cells in response to signals derived from an inflammatory environment in the CNS. © 2012 Wiley Periodicals, Inc.

INTRODUCTION

The CNS contains distinct populations of innate immune cells under both physiologic and pathologic conditions. Under physiologic conditions, microglia are the dominant myeloid cell found within the CNS parenchyma. This cell originates from the yolk sac, populates the CNS early in development, and has an extended life span with little turnover (Ginhoux et al.,2010). Myeloid cells of hematopoietic origin populate the perivascular space and have a relatively rapid rate of repopulation. Under inflammatory and injury conditions, hematogenous myeloid cells access the CNS parenchyma and are referred to as macrophages. Currently there are no distinct lineage markers that can distinguish microglia and infiltrating macrophages, although differences in relative expression of lineage markers such as CD45 are used. Both microglia and macrophages can be present at sites of inflammatory lesions in the CNS as directly demonstrated in the EAE model (Taupin et al.,1997). Boven et al. (2006) showed that myelin-laden myeloid cells in MS lesions express antiinflammatory rather than proinflammatory molecules, but they did not distinguish between macrophages and microglia.

Previous comparative studies have demonstrated that human microglia differ from hematogenous myeloid cells found in the CNS with respect to immunoregulatory properties and their response to environmental stimuli. When microglia derived from the human adult CNS were exposed in vitro to conditions that generated mature dendritic cells from blood monocytes, they activated an antiinflammatory rather than a proinflammatory cytokine program and inhibited allogenic T cell responses (Lambert et al.,2008). These microglia differed from blood monocyte-derived macrophages with regard to their migratory responses to chemokines and cyclic nucleotides (Lambert et al.,2010).

Studies of peripheral myeloid cells (monocytes, macrophages) demonstrate that similar to the Th1/Th2 T cell polarization axis, these cells can acquire differential inflammation-related phenotypes, termed M1 and M2 (Gordon,2003). The M1 phenotype occurs in response to proinflammatory cytokines (IFNγ, TNF) as well as pathogen-associated molecular patterns (PAMPS) such as lipopolysaccharide (LPS) (Fairweather and Cihakova,2009; Van Ginderachter et al.,2006). In context of immunoregulatory relevant properties, M1 macrophages express higher levels of the costimulatory molecules CD80 and CD86, resulting in efficient antigen presentation capacity. M1 macrophages also upregulate levels of TLR2, TLR4, Fc receptors (CD16, 32, 64), and the chemokine receptor CCR7. M1 cells produce large amounts of proinflammatory and Th1/17 inducing cytokines including IL-12, IL-23, and tumor necrosis factor (TNF). The M2 phenotype was initially induced using IL-4 (Stein et al.1992), although additional M2-polarization paradigms including both IL-4 and IL-13 have been described (Gordon,2003; Sica et al.,2006). Microglia as well as astrocytes have been shown to act as endogenous sources of IL-4 and IL-10 (Ponomarev et al.,2005), suggesting that myeloid cells in the CNS are potentially exposed to these polarizing factors. M2 cells produce the antiinflammatory cytokines IL-10 and TGF-β and low amounts of IL-12, and are associated with the dampening of immune responses and the promotion of tissue remodeling and repair (Schwartz,2010). M1 and M2 cells further differ with regard to receptor profiles that impact their interaction with their microenvironment. The surface marker profile of M2 macrophages includes expression of CD23 (Fcε-RII), the scavenger receptors CD163 and CD204, the mannose receptor CD206, and CD209 (DC-SIGN), (Mantovani et al.,2004; Martinez et al.,2006; Van Ginderachter et al.,2006). M2 macrophages have been found to be more efficient than M1 cells to phagocytose opsonized targets (Leidi et al.,2009). Rodent-based studies indicate that microglia may be more efficient than macrophages for receptor-based phagocytosis of myelin (Mosley and Cuzner,1996) whereas others suggest that the phagocytic activity of microglia may be limited compared with blood-borne macrophages (Neumann et al.,2009).

We elected to use previously published protocols to generate M1 and M2 myeloid cells (Leidi et al.,2009). M1 cells were generated in vitro using GM-CSF treatment and then further activated with IFNγ/LPS. The M2 cells as generated in our study in the presence of M-CSF and activated using IL-4/IL-13 (Leidi et al.,2009) are now referred to as M2a. The purpose of the current study was to use these specific polarization protocols to compare adult human and fetal CNS-derived microglia with blood-derived macrophages with regard to their capacity to become polarized into either the M1 or M2 phenotype. We assessed phenotypic properties of the cells using expression of two M1 surface markers (CD80, CCR7) and four M2 markers (CD23, CD163, CD206, CD209). We characterized functional properties of these cell types with regard to cytokine expression profiles by measuring the production of the proinflammatory cytokine IL-12p40 and the antiinflammatory cytokine IL-10 and capacity to phagocytose nonopsonized myelin. The capacity of human microglia to survive in vitro in the absence of growth factors further allowed comparisons of polarization responses with basal cell properties.

MATERIALS AND METHODS

Microglia Isolation and Cell Culture

Adult microglia were isolated from a mixture of white and gray matter of temporal lobe brain tissue, from seven individual patients undergoing surgery for nontumor-related intractable epilepsy. The tissue provided was outside of the suspected focal site. Mean age of patients was 44.0 years (range of 25–60) with an average seizure duration of 22.4 years (range 2–46), with four of the patients having undergone previous resections. All patients were previously prescribed multiple (>3) antiseizure medications. These studies were approved by our institutional review board. Brain tissue was processed as described (Fedoroff,1992; Lambert et al.,2008). Tissue was obtained in pieces <1 mm3 and treated with DNase (Roche, Nutley, NJ) and trypsin (Invitrogen, Carlsbad, CA) for 30 min at 37°C. Following dissociation through a nylon mesh, the cell suspension was separated on a 30% Percoll gradient (GE Healthcare, Piscataway, NJ) at 400g for 30 min. Glial cells (oligodendrocytes and microglia) were collected from underneath the myelin layer, washed and then plated in tissue-culture-treated vessels. Floating oligodendrocytes were washed off on the subsequent day and the remaining adherent microglia were collected with trypsin and 2 mM EDTA. Cells were plated at 1 × 105 cells mL−1 in MEM (Sigma) containing 5% FBS, penicillin/streptomycin and glutamine (all from Invitrogen). The proportion of microglia in the culture, determined using CD11c staining by flow cytometry, was ≥90%. Histological sections were made from the CNS material used to isolate microglia.

Human fetal CNS tissue (cerebral hemispheres) was obtained from the human fetal tissue repository (Albert Einstein College of Medicine, Bronx, NY), following approved institutional and Canadian Institutes for Health Research (CIHR) guidelines. Cells were isolated as previously described (D'Souza et al.,1995; Lambert et al.,2008). Fetal brain tissue (gestational age 14–20 weeks) was minced and treated with DNase/trypsin. Tissue was then dissociated through a nylon mesh and cells were plated at 7 × 106 cells mL−1 in high glucose DMEM with 5% FBS, penicillin/streptomycin and glutamine. After 10–14 days in culture, floating microglia were harvested and plated at 1 × 105 cells mL−1.

To generate M1 microglia, cells were treated with GM-CSF (5 ng mL−1, PeproTech, Rocky Hill, NJ) for 5 days followed by 1 h activation with IFNγ (20 ng mL−1, Invitrogen) and 48 h activation with LPS (serotype 0127:B8, 1 μg mL−1, Sigma). To generate M2 microglia, cells were treated with M-CSF (25 ng mL−1, PeproTech) for 5 days followed by 48 h activation with IL-4 (20 ng mL−1, Invitrogen) and IL-13 (20 ng mL−1, PeproTech).

Macrophage Generation

To obtain monocytes, PBMCs collected from venous blood of healthy volunteers (age range 19–50 years) were separated on a Ficoll density gradient (GE Healthcare). CD14+ monocytes were positively selected to >95% purity by MACS using anti-CD14 microbeads (Miltenyi Biotec, Auburn, CA), then plated at 5 × 105 cells mL−1 in 24- or 6-well plates in RPMI with 10% FBS, penicillin/streptomycin and glutamine. For immunocytochemistry, cells were plated at 2.5 × 104 cells per well in 14-well poly-L-lysine-coated chamberslides (VWR, Mississauga, ON, Canada).

To generate monocyte-derived macrophages, monocytes were treated for 5 days with human recombinant GM-CSF (5 ng mL−1) for M1 generation or M-CSF (25 ng mL−1) for M2 generation. M1 cells were activated for 1 h with IFNγ (20 ng mL−1) and 48 h with LPS (100 ng mL−1), and M2 cells were activated for 48 h with IL-4 (20 ng mL−1) and IL-13 (20 ng mL−1).

Immunocytochemistry

To assess cell morphology, cells were fixed in 2% PFA (Sigma) for 10 min at room temperature, permeabilized in cold acetone for 10 min, and blocked for 10 min with HHG (1 mM HEPES, 2% horse serum, 10% goat serum, Hanks balanced salt solution). Cells were incubated for 30 min with Abs against the myeloid cell marker CD68 (8 μg mL−1, Dako) followed by goat anti-mouse Cy3 (Jackson ImmunoResearch, West Grove, PA), as well as Alexa Fluor 488 phalloidin to visualize the actin cytoskeleton (1:40, Invitrogen), and the nuclear stain Hoechst (Invitrogen). Slides were mounted with Fluoromount-G (SouthernBiotech, Birmingham, AL), and stains were visualized using fluorescent microscopy (Leica Microsystems, Wetzlar, Germany) and OpenLab imaging software (PerkinElmer, Waltham, MA). Concentration-matched isotype controls (mouse IgG1, Sigma) showed low nonspecific staining (not shown).

Phenotypic Analysis

Cultured myeloid cells were collected using cold PBS (macrophages) or warm PBS with 2 mM EDTA (microglia) and scraping. Cells were blocked overnight with 3 μg mL−1 normal mouse IgG (Caltag Laboratories, San Francisco, CA) and 10% human serum. Cells were stained using specific or isotype-matched Abs directly conjugated to FITC, PE, PerCP, or allophycocyanin fluorochromes. Abs for CD23, CD80, CD163, CD206, CD209, CCR7, and all isotype controls were from BD Biosciences (Mississauga, ON, Canada). Cells were fixed in 1% formaldehyde and acquired using a FACSCalibur flow cytometer (BD Biosciences); data were analyzed using FlowJo software (Tree Star, San Carlos, CA). Dead cells were gated out of all analyses. Mean values of forward scatter (FSC) were used as an indication of cell size.

RNA Isolation, Reverse Transcription, and Real-Time PCR Gene Array

Total RNA was isolated from myeloid cells by lysis in TRIzol (Invitrogen) followed by isolation using the Qiagen RNeasy mini kit following manufacturer's instructions. RNA samples were treated with DNase (Qiagen, Germantown, MD). Reverse transcription (RT) and cDNA generation were performed using random hexaprimers (Roche) and the Moloney murine leukemia virus-RT enzyme (Invitrogen) at 42°C. RNA samples from three separate macrophage or microglia donors were assessed using the human dendritic and antigen presenting cell 84-gene PCR array (SABiosciences, Qiagen). PCR reaction cycling was performed according to the ABI PRISM 7000 Sequence Detection System default temperature settings (2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C, 1 min at 60°C). Analysis of five separate housekeeping genes (ACTB, GAPDH, RPL13A, HPRT1, and B2M) showed no significant differences across replicates, and average expression of these genes was used to normalize the data. Fold changes in gene expression were calculated using the ΔΔCt method according to manufacturer's instructions.

Cytokine Production

Cytokine levels were measured in supernatants collected from myeloid cells. IL-10 and IL-12p40 (BD Biosciences) were measured in duplicate wells by ELISA according to the manufacturer's instructions.

Phagocytosis Assay

Human myelin was isolated according to a previously described protocol (Norton and Poduslo,1973). Briefly, white matter obtained from surgical resections of the CNS was mechanically homogenized in 0.32 M sucrose and subjected to repeated sucrose density centrifugation and osmotic shocks to separate myelin from other cellular components. Myelin was found to be endotoxin free using the limulus amebocyte lysate test (Sigma). To evaluate myelin phagocytosis by myeloid cells, cells were plated in 14-well chamber slides and polarized as previously described. Myelin (25 μg mL−1) was added for an additional 24-h period. Cells were then fixed and stained as above, using antibodies against CD68 (1:10, mouse IgG1 conjugated to FITC, Dako) and MBP (1:200, mouse IgG2b, Covance, and a secondary anti-mouse IgG2b conjugated to Texas Red, Dako, 1:100). Cells from each donor were plated in triplicate and multiple fields of view (5–10) were counted. The proportion of cells with detectable MBP was used as the index for myelin ingestion.

Statistics

Statistical analyses were performed using Prism 5 (GraphPad Software). Comparisons of cell size, phenotype, and cytokine production across groups were analyzed by one-way or two-way ANOVA as appropriate and significant differences were decomposed using Tukey's or Bonferroni multiple comparison test. Probability values of <0.05 were considered to represent statistically significant differences.

RESULTS

Morphologic and Phenotypic Characterization of M1 and M2 Myeloid Cells

As an initial means to compare M1 and M2 macrophages and microglia, we examined the morphologic features of each of these cell types. Macrophages generated with GM-CSF but not further polarized as well as microglia under basal culture conditions were also examined. As shown in Fig. 1A, under M1 treatment conditions almost all macrophages adopt a rounded shape whereas under M2 conditions macrophages with an elongated bipolar morphology are readily identified. As regards microglia, these cells show evident process extension under basal conditions without apparent change under M1 or M2 conditions. To quantitate differences in the size of cells exposed to M2 vs. M1 conditions, we used flow cytometry analysis to calculate the ratio of mean FSC for M2:M1 cells. As illustrated for representative donors in Fig. 1B and as shown in Fig. 1C, the mean ratio for macrophages treated with M2 vs. M1 conditions was 1.51 ± 0.05, whereas the corresponding mean ratio was 1.11 ± 0.09 for adult microglia (P < 0.01 vs. macrophages) and 1.19 ± 0.06 for fetal microglia (P < 0.05). Mean FSC for M2-treated adult microglia did not differ significantly from basal microglia (data not shown).

Figure 1.

Morphologic characterization of M1 and M2 myeloid cells. (A) Human macrophages and adult microglia were treated with M1 or M2 polarizing conditions and stained for expression of the myeloid cell marker CD68 (red), the actin-binding phalloidin (green), and the nuclear marker Hoechst (blue). Macrophages generated in the presence of GM-CSF but not further polarized, as well as microglia under basal culture conditions, are shown (top panels). The majority of M1-treated macrophages show a rounded morphology whereas a significant proportion of M2-treated cells show process extension (arrows). Microglia under basal, M1 and M2 treatment conditions feature process bearing cells. (B and C) Comparison of size of M1 and M2 cells based on forward scatter (FSC) profiles. M1 and M2-treated macrophages, adult, and fetal microglia were subjected to flow cytometry analysis and assessed for differences in FSC profiles indicative of size. Because of different basal acquisition parameters required, FSC values cannot be compared across cell types. Results are shown as representative profiles (B) and quantified as ratio of M2:M1 geometric mean FSC value (C). Data are presented as mean ± SEM, with n = 3 donors per condition. *P < 0.05; **P < 0.01. MØ, macrophage; AM, adult microglia; FM, fetal microglia. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

We found no correlation in phenotypic properties of microglia in culture with histological features of the intact tissue from which these cells were derived. In all cases, there was no obvious evidence of significant astrogliosis or increased microglial activity when compared with control (autopsy-derived) tissue. Representative sections from each case are provided in Supporting Information Fig. 1.

Phenotypic Characterization of M1 and M2 Myeloid Cells by Flow Cytometry

M1 phenotyping

The majority of the M1 polarized macrophages expressed the M1 markers CD80 (83% ± 9%) and CCR7 (62% ± 12%), (Fig. 2A). These values were significantly greater than found on M2-treated cells (<15% of cells) and on ex vivo monocytes (<3%; data not shown), or on unpolarized macrophages treated with either GM-CSF (<20%) or M-CSF (<7%) (all values P < 0.01). The proportion of adult and fetal microglia that expressed CD80 and CCR7 following M1 polarization (Fig. 2A) was also significantly greater compared with M2 microglia and to unpolarized microglia (P < 0.05).

Figure 2.

Phenotypic characterization of M1 and M2 myeloid cells. M1 and M2 macrophages, adult, and fetal microglia were subjected to flow cytometry analysis and assessed for their expression of the M1 markers CD80 and CCR7 (A) and the M2 markers CD23, CD163, CD206, and CD209 (B). Basal adult and fetal microglia, as well as macrophages generated with M-CSF or GM-CSF but not further polarized, are also shown. (A) A significant increase was seen in CD80 and CCR7 on M1 cells vs. the corresponding M2 myeloid cell type. (B) Under M2 conditions, there was a significant increase in all surface markers on M2 vs. corresponding M1 macrophages. CD206 was significantly increased only on M2 vs. M1 fetal microglia; values were significantly lower in microglia than for M2 macrophages (###, P < 0.001). CD209 was increased on both adult and fetal M2 vs. M1 microglia; microglia values were significantly lower than for M2 macrophages (##, P < 0.01). Data are based on 3–12 experiments for each condition. MØ, macrophage; GM, GM-CSF; M, M-CSF; AM, adult microglia; FM, fetal microglia. *P < 0.05, **P < 0.01, ***P < 0.001 when corresponding M1 and M2 myeloid cells were compared.

M2 phenotyping

As shown in Fig. 2B, all the M2 markers (CD23, CD163, CD206, and CD209) were significantly upregulated following M2 treatment in macrophages compared with M1-treated macrophages (all values P < 0.05). Only CD206 was significantly increased on macrophages treated with either GM-CSF (62% ± 5%) or M-CSF (44% ± 5%) alone when compared with blood monocytes (2% ± 0.4%; data not shown) (P < 0.001).

In adult microglia, only CD209 was significantly increased under M2 conditions (37% ± 7%) compared with M1 conditions (P < 0.001). This value was significantly lower than for M2-treated macrophages (74% ± 6%, P < 0.01). For fetal microglia, both CD206 and CD209 were increased under M2 conditions when compared with M1 conditions (P < 0.05). Unpolarized adult and fetal microglia did not express any of the M2 markers.

Gene Expression Profiling

To determine regulation of gene expression in polarized myeloid cells, we utilized the 84-gene PCR “Dendritic cell and APC” array (SA Biosciences SuperArray). The raw data for gene expression, as expressed by average ΔCt, is provided in Supporting Information Fig. 2 (M2 myeloid cells) and Supporting Information Fig. 3 (M1 myeloid cells). In Fig. 3A,B, we present the listing of genes that were up- or downregulated by greater than fourfold in macrophages or microglia exposed to M1 vs. M2 polarizing conditions. We document genes that were differentially expressed when directly comparing across myeloid cell types under the same polarizing conditions. When we compared M1-treated myeloid cells, we found 12 genes overexpressed in macrophages and 14 overexpressed in microglia. Under M2 treatment conditions, five genes were overexpressed in macrophages, six were overexpressed in microglia. Three of the five genes upregulated in M2 macrophages (CD1A, B, and C) and five of the six genes upregulated in the M2 microglia (CCL8, CXCL10, CXCL12, HLA-DQB1, TLR2) were also relatively overexpressed when their corresponding M1 phenotypes are compared. These data suggest that there may be genes that reflect differences in the microglia vs. macrophage lineage rather than unique polarization responses.

Figure 3.

Gene expression profiling of M1 and M2 myeloid cells. (A and B) RNA collected from M1 and M2 macrophages and adult microglia was assessed using an 84-gene PCR array. Average ΔCt values, generated by comparing the average expression cycle of the gene of interest with the housekeeping genes, were used to calculate relative fold changes. Microglia and macrophages were compared under either M1 (A) or M2 (B) polarization conditions. Genes overexpressed for a given cell type under both M1 and M2 conditions are indicated in italics. (CE) Genes known to be differentially modulated under M1 and M2 conditions were used to confirm successful polarization. Shown are the expression profiles for the M1 marker CD80 (C) and the M2 markers CD209 (D) and CD23 (E). Data are presented as mean ± SEM, with n = 3 donors per condition. MØ, macrophage; AM, adult microglia; FM, fetal microglia.

Comparing M1 and M2 macrophages, we observed differential expression of 35 genes, with 74% overexpressed under M1 conditions and 26% overexpressed under M2 conditions (Supp. Info. Fig. 4). In microglia, we found differential expression of 44 genes, with 70% overexpressed under M1 conditions and 30% overexpressed under M2 conditions (Supp. Info. Fig. 5).

Figure 4.

Cytokine production in M1 and M2 myeloid cells. M1 cells were generated with GM-CSF and activated using LPS ± IFNγ. (A) Macrophages and microglia under full M1 polarization (GM-CSF + IFNγ + LPS) produced significantly higher levels of IL-12p40 compared with cells treated with GM-CSF and stimulated with LPS alone (***P < 0.001). M1 macrophages and microglia also produced significantly higher levels of IL-12p40 compared with M2 cells (M-CSF + IL-4/IL-13) and M2 cells subsequently stimulated with LPS (***P < 0.001). (B) IL-10 production by macrophages was highest following M2 + LPS treatment. IL-10 production by microglia did not significantly differ between conditions. Levels of IL-10 produced by microglia under M1 conditions were significantly higher than for M1 macrophages (**P < 0.01). Macrophages generated with M-CSF or GM-CSF but not further polarized, as well as basal microglia, did not produce detectable levels of IL-12p40 or IL-10. Data are based on 3–6 experiments for each condition and are presented as mean ± SEM.

Figure 5.

Myelin phagocytosis by polarized myeloid cells. Following standard polarization protocols, macrophages and microglia were treated with purified human myelin (25 μg mL−1) for 24 h and then immunostained for MBP as illustrated in Panels A and B, respectively, and as quantified in Panel C. Adult microglia under all conditions (basal, M1, and M2) had greater myelin uptake than the macrophages (###P < 0.001 when comparing across cell types). Both M2 macrophages and microglia were more efficient at phagocytosing myelin than were their M1 counterparts (***P < 0.001 when comparing M1 and M2 treatment conditions in the same cell type). Inset in Panel B shows higher power (×25 vs. ×10) image of myelin within an M2 microglia cell body. Data are presented as mean ± SEM, with n = 3 donors per condition.

Validation of M1 polarization was observed by examining expression of the M1-associated marker CD80. Compared with M2 macrophages, changes in CD80 expression were 37-fold and 66-fold in M1 macrophages and M1 microglia, respectively (Fig. 3C). Similarly, validation of M2 polarization was observed using the M2-assocated markers CD209 and CD23. Compared with M1 macrophages, changes in CD209 expression were 18-fold and 32-fold in M2 macrophages and M2 microglia, respectively (Fig. 3D). Compared with M1 macrophages, changes in CD23 expression were 16-fold and 20-fold in M2 macrophages and M2 microglia, respectively (Fig. 3E).

Cytokine Production

To assess the functional consequences of polarization, the production of IL-12p40 and of IL-10 were determined as measures of pro- and antiinflammatory molecules, respectively (Fig. 4).

Both macrophages and microglia produced significantly higher levels (7409 ± 389 pg mL−1 and 8177 ± 845 pg mL−1, respectively) of IL-12p40 under full M1 polarizing conditions (Fig. 4A) compared with myeloid cells treated with GM-CSF and stimulated with LPS alone (P < 0.001) and compared with cells under M2 conditions where IL-12p40 was undetectable (P < 0.001). M2-treated myeloid cells subsequently activated with LPS produced ∼800 pg mL−1 (P < 0.001 compared with M1 conditions). Macrophages generated with M-CSF or GM-CSF but not further polarized, as well as basal microglia, did not produce detectable levels of IL-12p40 (data not shown).

In macrophages, IL-10 production (Fig. 4B) was highest in M2 cells subsequently activated with LPS (475 ± 150 pg mL−1). In microglia, M1 and M2 cells produced IL-10 at comparable levels (378 ± 68 and 275 ± 37 pg mL−1, respectively), even when M2 microglia were exposed to LPS (210 ± 120 pg mL−1). The levels produced by M1 microglia were significantly higher compared with M1 macrophages (151 ± 34 pg mL−1, P < 0.01). Macrophages generated with M-CSF or GM-CSF but not further polarized, as well as basal microglia, did not produce detectable levels of IL-10 (data not shown).

Phagocytosis of Myelin

To further compare functional properties of polarized myeloid cells, we assessed the proportion of such cells that had engulfed purified human myelin in a 24 h period. Results are illustrated in Fig. 5A,B and quantitated in Fig. 5C. Adult microglia under all conditions (basal, M1, and M2) showed greater myelin uptake than the macrophages (all values P < 0.001). Phagocytosis by M2 macrophages was significantly greater than by M1 macrophages (33% ± 3% vs. 15% ± 4%, P < 0.001), as was phagocytosis by M2 microglia compared with M1 microglia (75% ± 2% vs. 49% ± 8%, P < 0.001).

DISCUSSION

In this report, we define distinct properties of human microglia compared with macrophages, which are known to access the CNS parenchyma during the course of neuroinflammatory responses. The classical M1 polarization state of myeloid cells has been linked with promoting inflammation (Weber et al.,2007), whereas the alternative M2 phenotype is antiinflammatory and promotes tissue repair (Kigerl et al.,2009). Ponomarev et al. showed in the murine system that the alternative activation of microglia required IL-4 and that lack of IL-4 in the CNS, but not the periphery, exacerbated symptoms of EAE (Ponomarev et al.,2007). Our selection of polarization conditions was based on recent reports regarding the optimal use of GM-CSF to initiate M1-polarized macrophages versus earlier work that exclusively used M-CSF (Sierra-Filardi et al.,2010). We also include fetal CNS-derived microglia in our studies as a model of more recently arrived population of microglia cells into the CNS.

We observed distinct morphologic differences between M1 and M2 macrophages, with M1 cells adopting a round morphology consistent with hyperactivity as suggested by others (Akagawa,2002). M2 cells appeared larger with bipolar processes, while flow cytometry analysis confirmed that M2 macrophages were significantly larger than M1 cells. In contrast, M1 and M2 microglia showed little differences, either by morphologic or flow cytometry criteria, when compared with each other or to unpolarized microglia under basal conditions. Under all conditions, microglia continue to express prominent processes, resembling M2 macrophages.

Using flow cytometry, our surface marker results distinguish microglia and macrophage responses to polarizing conditions. As M1 markers, we selected CCR7 and CD80. CCR7 is the chemokine receptor for CCL19 and CCL21, and is upregulated in the CNS during chronic EAE (Bielecki et al.,2007). CD80, a co-stimulatory molecule involved in T cell activation, is more highly expressed on perivascular myeloid cells than parenchymal cells under physiologic conditions, but is upregulated in the parenchyma under inflammatory conditions, including MS (Williams et al.,1994). These two M1 markers were highly expressed under M1 polarizing conditions on both macrophages and microglia. The M2 markers CD206 and CD209 are C-type lectins responsible for the uptake of complex carbohydrates (Saunders et al.,2009; Van Ginderachter et al.,2006). CD206 deficiency has been linked to increased susceptibility of CNS cryptococcus infections in mice (Dan et al.,2008). Expression of the scavenger receptor CD163 has been detected in the CNS (Galea et al.,2008). The Fcε receptor CD23 is known to be upregulated on monocytes in MS patients (Tsukada et al.,1994). Despite their process bearing morphology, adult M2 microglia express only CD209, the dominant molecule expressed by M2 macrophages. Unlike M2 macrophages, M2 microglia did not express detectable levels of CD23, CD163, or CD206. Increased expression of the M2 marker CD23 was detected at the RNA level in macrophages and microglia, suggestive of differential posttranscription control mechanisms. The fetal microglia feature a more intermediate phenotype, expressing both CD209 and low levels of CD206 under M2 conditions. Our findings that M2 macrophages, but not microglia, upregulate the characteristic M2 surface markers fits with the in situ observations of Boven et al., which showed that such markers were detected only on “foamy macrophages” in perivascular spaces and not in the parenchyma (Boven et al.,2006).

Our PCR array extended our analysis of M1 vs. M2 properties in macrophages and microglia. The array focused on genes involved in neuroinflammatory responses, and indicated that both macrophages and microglia show a greater induction of gene expression in response to M1 vs. M2 conditions. This is consistent with the previous report of Martinez et al. (2006) using a total microarray screen of macrophages. The majority of genes differentially regulated in M1 polarized macrophages were also observed in M1 microglia compared with their M2 counterparts. To further validate successful polarization, we examined the expression of known biological markers of M1 (CD80) and M2 (CD209 and CD23) cells. On the basis of expression of these genes under appropriate polarization conditions, we note that microglia and macrophages were responsive to both polarization paradigms.

Comparison of M1 macrophages to M1 microglia demonstrated that macrophages overexpressed CD1A, 1B, and 1C; expression was also increased in M2 macrophages compared with M2 microglia (Fig. 3A,B). CD1 molecules are a closely related group of MHC-like proteins involved in presentation of lipid-associated antigens (Strominger,2010). Interestingly, polymorphisms of CD1A have been linked to MS susceptibility (Caporale et al.,2011). M1 macrophages also overexpressed HLA-DM compared with M1 microglia. HLA-DM plays a role in MHC peptide loading and has been shown to bias T-cell differentiation towards the Th1 lineage (Pezeshki et al.,2011). Our data suggest that regardless of polarization state, macrophages may play a more significant role than microglia in antigen presentation, with M1 macrophages more likely to bias Th1 responses.

When comparing M2 microglia vs. M2 macrophages, multiple genes overexpressed in these microglia were also overexpressed in M1 microglia (compared with M1 macrophages). This finding may reflect differences in microglia vs. macrophage lineage rather than being attributed to polarization. The only marker overexpressed in M2 microglia (compared with M2 macrophages) that was not also overexpressed in M1 microglia was FCγR1A (CD64). Expression of this receptor has been previously correlated with phagocytic activity in macrophages (Leidi et al.2009). We now demonstrate increased expression of this cell surface receptor in M2 microglia, which also had increased phagocytic activity compared with macrophages (Fig. 5). In the context of MS, myelin debris has been shown to inhibit remyelination (Kotter et al.,2006), while debris clearance by myeloid cells has been shown to promote repair in EAE (Takahashi et al.,2007). This finding suggests that in response to CNS injury, microglia may be the dominant myeloid cell type responsible for clearing debris, including myelin, thereby promoting repair and regeneration.

As regards cytokine production, we detected IL-12p40 in M1 macrophages and microglia. As expected, IFNγ was an essential regulator of IL-12p40 (Trinchieri,1997). Under M2 conditions for both cell types, only very low levels of IL12p40 were detected. As regards IL-10, in the presence of LPS, M2 macrophages produced higher levels of this cytokine than did their M1 counterparts. Under M2 conditions alone, macrophages produced only low levels of IL-10 consistent with previous reports (Verreck et al.,2004). In contrast to macrophages, microglia produced comparable levels of IL-10 under all tested conditions (basal + LPS [data not shown], M1, M2 ± LPS). Under M1 conditions, microglia produced higher levels of IL-10 than did M1 macrophages, suggesting that under inflammatory conditions microglia are the likely source of this antiinflammatory cytokine. We did not examine other M2 induction regimens in this study such as exposure to IL-10 (M2c). Previously, we had observed that supernatants from Th2-polarized T cells augmented IL-10 production in monocytes but not in microglia; however Th2-treated microglia and monocytes were both able to induce Th2 differentiation of naïve T cells (Kim et al.,2004; Seguin et al.,2003). M2 markers were not examined in these studies.

Our data regarding myelin phagocytosis further indicate distinct properties of microglia compared with blood-derived macrophages. We observed that microglia displayed more phagocytic activity than macrophages under all conditions tested. As mentioned previously, others observed a similar pattern using opsonized myelin and rat-derived myeloid cells (Mosley and Cuzner,1996). We observed that phagocytosis of nonopsonized myelin by M2 polarized macrophages and microglia was more efficient than their M1 counterparts. Bruck et al. observed that TNFα can suppress phagocytosis of myelin by myeloid cells (Bruck et al.,1992); M1 polarized myeloid cells are a greater source of TNFα than are M2 cells (Mantovani et al.,2004).

Our combined phenotypic and functional data indicate distinct properties of human microglia compared with macrophages under polarization conditions. Although microglia do respond to both M1- and M2-inducing stimuli, they are more restricted in their capacity to adopt the M2 phenotype and cytokine profile as defined by macrophages. Such differences in response to environmental signals may have consequences as one aims to introduce therapies designed to modulate their profiles. Promoting M2 polarization, especially in the infiltrating myeloid cell subset that has been demonstrated to be more plastic than their resident microglia counterparts, could be a novel therapeutic target to control inflammatory responses in the CNS.

Acknowledgements

The authors thank Manon Blain, Ellie McCrea, and David Henault for technical assistance, Dr. Samuel Ludwin for providing control brain immunohistological sections and helpful discussions, Dr. Bradford Poulos, director of the Human Fetal Tissue Repository at Albert Einstein College of Medicine for providing human fetal tissue, as well as Dr. Jeffery Hall, Dr. André Olivier, and Dr. Kevin Petrecca, neurosurgeons at the Montréal Neurological Institute, for providing adult brain tissue. The authors have no financial or commercial conflicts of interest to disclose.

Ancillary