Androgen and androgen receptor as enhancers of M2 macrophage polarization in allergic lung inflammation

Abstract

Allergic asthma is a disease initiated by a breach of the lung mucosal barrier and an inappropriate Th2-inflammatory immune response that results in M2 polarization of alveolar macrophages (AM). The number of M2 macrophages in the airway correlates with asthma severity in humans. Sex differences in asthma suggest that sex hormones modify lung inflammation and macrophage polarization. Asthmatic women have more M2 macrophages than asthmatic men, while androgens have been used as an experimental asthma treatment. Here, we demonstrate that although androgen (DHT, dihydrotestosterone) reconstitution of castrated mice reduced lung inflammation in a mouse model of allergic lung inflammation, it enhanced M2 polarization of AM. This indicates a cell-specific role for androgens. Dihydrotestosterone also enhanced IL-4-stimulated M2 macrophage polarization in vitro. Using mice lacking androgen receptor (AR) in monocytes/macrophages (AR^floxLysMCre), we found that male but not female mice exhibited less eosinophil recruitment and lung inflammation due to impaired M2 polarization. There was a reduction in eosinophil-recruiting chemokines and IL-5 in AR-deficient AM. These data reveal an unexpected and novel role for androgen/AR in promoting M2 macrophage polarization. Our findings are also important for understanding pathology in diseases promoted by M2 macrophages and androgens, such as asthma, eosinophilic esophagitis, and prostate cancer, and for designing new approaches to treatment.

Introduction

Allergic asthma is a chronic inflammatory disease of the lungs characterized by an aberrant Th2 inflammatory immune response against an otherwise innocuous stimulus. Th2 cytokines, such as interleukin (IL)-4 and IL-13, and chemokines, enhance inflammation, mucus production, airway constriction, and inflammatory immune cell recruitment in the lungs. This Th2 environment contributes to asthma symptoms in humans and in mouse models of allergic lung inflammation. In the non-inflamed lung, alveolar macrophages (AM) are the most abundant immune cells in the alveolar space and one of the first cell types in contact with the allergenic stimulus. Th2 inflammation polarizes AM to an M2 phenotype (1) and accumulation of M2-polarized AM in the lung correlates with asthma severity (2). Alveolar macrophages secrete eosinophil-recruiting chemokines and cytokines during allergic lung inflammation (3). Lung eosinophilia is a hallmark of allergic asthma, and eosinophils enhance airway hyperresponsiveness and mucus production (4). Furthermore, human and mouse M2 AM in the asthmatic lung produce excess remodeling and inflammatory factors that decrease lung function and aggravate asthma severity (5).

Asthma outcome in humans and animal models is influenced by biological sex. The incidence and severity of asthma is greater in women than in men (6) and in female compared to male mice (7, 8). However, before puberty, boys are more likely to have asthma than girls (9). This change in the incidence of asthma suggests that sex hormones play a key role in affecting the disease. Previously we had demonstrated that AM from female mice more highly expressed genes characteristic of M2 polarization than male macrophages. This was shown both in a mouse model of allergic lung inflammation in vivo and following IL-4 stimulation in vitro. We showed that estrogen (E2) augmented M2 gene expression and this was dependent on estrogen receptor alpha (ERα). The incidence of asthma between adult men and women suggests that the increase in male sex hormones after puberty may have the opposite effect to E2 and control asthma.

Testosterone and other androgens, such as dihydrotestosterone (DHT), have broad immunoregulatory effects that suppress immune responses (10). Male mice produce less Th2 cytokines and specific immunoglobulin (Ig) E, and have fewer lung lymphocytes after ovalbumin (OVA) sensitization and challenge in the asthma model (8). Castrated mice have increased OVA-induced eosinophil and lymphocyte infiltration in the bronchoalveolar lavage fluid [BALF; (11)]. Experimental use of androgens improved asthma symptoms in up to 88% of women with premenstrual asthma (12). The mechanisms by which androgens improve asthma symptoms or severity are poorly understood. Androgens induce relaxation of airway smooth muscle (13) and Laffont et al. (14) showed recently that androgens decrease proliferation of group 2 innate lymphoid cells (ILC2) in the lung. However, the role of androgens in macrophage polarization and function in the lung is unknown. Given the positive correlation between M2 macrophages in the lung and asthma severity, and the lower incidence of asthma in men than in women, we hypothesized that androgens, unlike E2, suppress M2 polarization of macrophages. Therefore, in this study, we aimed to determine the role of androgens in macrophage polarization.

Materials and Methods

Animals

Six-week-old C57BL/6J wild-type (strain number 664) and LysMCre (strain number 004781 - B6.129P2 Lyz2^tm1(cre)Ifo/J) mice were purchased from Jackson Laboratories, Bar Harbor, ME. Castrated male C57BL/6 mice (3 weeks old) were purchased from Charles River, Germantown, MD. AR^flox mice on a C57BL/6 background, generated by Karel De Gendt (15), were obtained from the European Mutant Mouse Archive (Rome, Italy), were re-derived by the Johns Hopkins Transgenic Core and bred in our facilities. All mice were housed in the Johns Hopkins University mouse facility, and all experiments were conducted under a protocol approved by the Johns Hopkins Animal Care and Use Committee.

Genotyping of AR^flox and AR^floxLysMCre mice

Genomic DNA from the AR^flox and AR^floxLysMCre mice was isolated from ear tissue using the REDExtract-N-Amp Tissue PCR Kit Protocol (Sigma-Aldrich, St Louis, MO) according to the manufacturer’s instructions. Primers used for genotyping were AR28 AGCCTGTATACTCAGTTGGGG (16), AR29 AATGCATCACATTAAGTTGATACC (16), LysMCre Mutant CCCAGAAATGCCAGATTACG, LysM wild-type TTACAGTCGGCCAGGCTGAC, and LysM Common CTTGGGCTGCCAGAATTTCTC. PCR products were resolved on a 3 % agarose gel, and DNA bands were visualized by ethidium bromide staining.

Mouse allergic sensitization and challenge

We used the protocol of Wang et al. (17) for OVA-induced allergic lung inflammation. Briefly, 8-week-old mice were sensitized by intraperitoneal (IP) injection with 100 μg OVA (Sigma-Aldrich) emulsified in 100 μL of Imject Alum (Thermo Scientific; Grand Island, NY) on days 1 and 6. Next, mice were exposed to an airway challenge with nebulized OVA (1 % in PBS) for 40 min on days 12 and 14. On day 16, mice were anesthetized with an IP injection of 2.5% Avertin (Sigma-Aldrich) to collect BALF, serum, and lung tissue for further analysis. For the DHT reconstitution experiments, we implanted DHT-releasing (Sigma-Aldrich) or placebo pellets subcutaneously into castrated mice on day 7 of the OVA protocol to avoid interfering with T-cell priming.

Bone marrow macrophage (BMM) and AM culture

Bone marrow macrophages from 5- to 6-week-old male and female C57BL/6 mice were grown as published previously (18). Briefly, bone marrow cells were seeded in phenol red-free αMEM containing 40 ng/mL of recombinant mouse M-CSF (Gemini Bio-Products, West Sacramento, CA) and cultured for 10 days prior to experimentation. Bone marrow macrophages grown in M-CSF were used for all experiments, except for Figure 2B-2D as noted below.

Figure 2. DHT enhances M2 gene expression, YM1 and FIZZ1 production in BMM.

Bone marrow from female and male C57BL/6 mice was cultured for 10 days with M-CSF (40 ng/mL) to obtain BMM. (A) Validation of the specificity of anti-AR antibody. AM from female and male mice were cultured for 24 h in the presence or absence of the AR-degrading drug, ASC-J9 (10 μM), left panel. AR expression in BMM from C57BL/6 female and male mice, middle panel. Comparison of AR expression in BMM and AM, right panel. (B) BMM were cultured for 10 days with 40 ng/mL of GM-CSF or M-CSF in 6-well plates, cells were collected and viability determined. F4/80⁺ cells were selected in the live population. Dot plots of CD11c⁺Siglec F⁺ cells from F4/80⁺ cells cultured with GM-CSF or M-CSF. (C) Analysis of double-positive CD11c⁺Siglec F⁺ macrophages with GM-CSF or M-CSF (D) Analysis of AR⁺ cells from F4/80⁺ cells cultured with GM-CSF or M-CSF. (E) BMM were pretreated overnight with the indicated concentrations of DHT and then stimulated with IL-4 (1 ng/mL) for 48 h. Expression of the indicated M2 genes was determined by qPCR using the 2^-ΔΔCT method, and compared to the amount of mRNA in the male IL-4 sample (= 100%). (F) Expression of YM1 and FIZZ1 protein was determined by Western blot in the same BMM supernatants after 48 h of IL-4 stimulation. Equal volumes of supernatant were loaded in each lane. Densitometry of the specific bands was normalized to the amount of protein in male IL-4 (= 100%). Results are representative of four independent in vitro experiments. †p < 0.05, ††p < 0.01, ††p < 0.001 for the comparison between DHT pretreatment and IL-4 stimulation alone; *p < 0.05; **p < 0.01, ***p < 0.001 for the comparison between females and males in that specific treatment.

For DHT pretreatment experiments, cells were incubated with indicated concentrations of DHT overnight prior to stimulation for 48 h with 1 ng/mL IL-4 (R&D Systems, Minneapolis, MN) in phenol red-free αMEM with 10 % charcoal-stripped FBS and 40 ng/mL mouse M-CSF.

For development of macrophages with an “AM-like” phenotype (Figure 2B-2D), bone marrow cells were collected as described and seeded onto 6-well plates and cultured in αMEM containing 40 ng/mL of recombinant mouse M-CSF or GM-CSF (Gemini Bio-Products) for 10 days prior to experimentation.

For differentiation of dendritic cells (DC), bone marrow cells were collected as described and seeded onto 10 cm dishes and cultured in αMEM containing 20 ng/mL of recombinant mouse GM-CSF (Gemini Bio-Products) and 40 ng/mL of recombinant mouse IL-4 (R&D Systems, Minneapolis, MN). After 10 days of culture, non-adherent cells were collected for flow cytometric analysis.

Alveolar macrophages were enriched from BALF as previously reported (18). Briefly, BALF cells were grown in αMEM for 2 h at 37°C in 5% CO₂. Alveolar macrophages were isolated by adhesion in culture, and eosinophils and lymphocytes were removed by washing with cold PBS.

For AR degradation, cells were incubated with ASC-J9 (10 μM; Advanced ChemBlocks Inc., Burlingame, CA) for 24 h before surface and intracellular staining for flow cytometry.

For the flutamide inhibition of AR experiments, BMM cultured for 10 days were pretreated with 10 μM flutamide (Sigma-Aldrich) for 2 h before adding DHT (10 nM). After 18 h, cells were stimulated with 1 ng/mL IL-4 (R&D Systems, Minneapolis, MN) for 48 h in αMEM with 40 ng/mL mouse M-CSF. After this time, cells were lysed for mRNA analysis by quantitative PCR as described below.

Proliferation of alveolar macrophages by bromodeoxyuridine (BrdU) incorporation.

AR^flox and AR^floxLysMCre mice were injected IP with 1 mg of BrdU (BD Biosciences, San Jose, CA) on days 1, 3, and 5. Also, 500 ng rmIL-4 was delivered intratracheally (i.t.) on day 5 to all mice, to stimulate proliferation. On day 8 mice were anesthetized with an IP injection of 2.5% Avertin (Sigma-Aldrich) and BALF was collected as previously. Alveolar macrophages were enriched from BALF as previously reported (18). Bromodeoxyuridine incorporation was measured as an index of AM proliferation as indicated in the kit and analyzed by FACS.

Quantitative PCR (qPCR)

Messenger RNA from AM or BMM was collected in RLT buffer and processed by using the RNeasy mini kit (Qiagen, Valencia, CA). Quantitative PCR was carried out with the indicated primers in a 7500 Fast Real Time PCR instrument from Applied Biosystems (Grand Island, NY). The primer sequences used for M2 genes have been published previously (19). For qPCR analysis, we obtained AM by 2 h adherence on 6-well culture plates. Non-adherent cells were removed after this time by washing with PBS.

Fluorescence activated cell sorting (FACS)

Cells collected from the BALF were stained as previously published (18) for LIVE/DEAD, CD11c-PECy7 (clone N418), Ly6G-APCCy7 (clone 1A8), Ly6C-PerCPCy5.5 [clone HK1.4; (BioLegend, San Diego, CA)], Siglec F-BV421 [clone E50–2440; (BD Biosciences, San Jose, CA)]. For intracellular YM1 (STEMCELL Technologies, Vancouver, BC) or AR (Cell Signaling Technology, Danvers, MA) detection, cells were stained with a secondary PE-anti-rabbit antibody (BioLegend). For the DC stain, CD209a-PE (clone MMD3), CD80-PE (clone 16–10A1), and CD86-AlexaFluor700 [clone GL-1; (BioLegend)], were used. Stained cells were analyzed with a CytoFLEX flow cytometer (Beckman-Coulter, Brea, CA), courtesy of the Anesthesiology and Critical Care Medicine Flow Cytometry Core. CytEXPERT software v.2.0 (Beckman-Coulter) was used for analysis.

Dihydrotestosterone and placebo pellets

Dihydrotestosterone-releasing and placebo pellets were implanted subcutaneously in the backs of anesthetized mice on day 7 of the OVA protocol for allergic lung inflammation. The pellets, which consisted of 1 cm-long standard silicone tubing (Helix Medical, Carpinteria, CA), contained compressed DHT (Sigma-Aldrich, St Louis, MO) or placebo, and were sealed with A-100 medical silicone adhesive (Factor II, Inc., Lakeside, AZ). Pellets were prepared in our laboratory and disinfected before implantation.

As an indicator of DHT release, mice were weighed before pellet implantation and at the end of the experiment at sacrifice. Seminal vesicles were also collected and weighed at the end of the experiment as a bioassay for the systemic release of DHT. Differences in weight among groups were calculated and plotted.

Western blotting

Equal volumes of denatured BALF from the different experimental groups were loaded onto 4–20% Criterion TGX polyacrylamide gels (Bio-Rad, Hercules, CA) for electrophoresis under reducing conditions. Proteins were transferred onto PVDF membranes (Bio-Rad, Hercules, CA), and probed with antibodies against YM1 (STEMCELL Technologies) and FIZZ1 (Abcam, Cambridge, MA). Secondary antibodies conjugated to HRP were used to detect target proteins, which were visualized with the ECL method (Bio-Rad). Chemiluminescence from the blots was captured using KwikQuant Imager (Kindle Bioscience, LLC) and target band intensities were measured using the Image Studio Lite software (LI-COR Biosciences, NE).

ELISA

We determined levels of CCL24, YM1, and CCL5 in BALF samples by ELISA (R&D Systems, Minneapolis, MN) using the manufacturer’s protocols. For total IgE quantification, we collected blood by cardiac puncture and centrifuged it to obtain serum. Total serum IgE was measured with the OptEIA kit from BD Biosciences. We determined the concentration of testosterone (IBL, Minneapolis, MN) in mouse serum according to the manufacturer’s instructions.

Histology

Lungs sections were obtained as previously reported (18, 20) and stained with Periodic acid–Schiff (PAS), hematoxylin and eosin (H&E), and Masson’s Trichrome by the Molecular and Comparative Pathobiology Histology Laboratory at the Johns Hopkins University School of Medicine. Samples were number-coded to reduce experimenter bias and were counted by an independent examiner blinded to the experimental groups. Cells that were positive for PAS were counted in five randomly chosen alveoli from each sample and normalized to the pixel circumference of each selected alveolus. The data were then represented as PAS⁺ cells / 1000 pixels. For the H&E analysis, multiple images of alveolar spaces were taken for each mouse. Cell infiltration was scored on a scale of 0 to 3. Examples of different inflammation grades and details can be seen in Fig. S1 C. The infiltration scores for each image were averaged per mouse.

Statistical analysis

GraphPad Prism software (La Jolla, CA) was used for statistical analysis and graph generation. Each experiment was carried out independently at least three times. Mouse experiments contained 3–5 mice per group and were repeated either two or three times, as indicated. Statistical significance was measured by using a parametric Student t test, and a p value < 0.05 was considered statistically significant. Data are plotted as the mean ± SEM.

Online supplemental material

Fig. S1 shows the weight of seminal vesicles, the concentration of serum testosterone after castration with and without DHT reconstitution, and the histologic analysis of cellular infiltration of the mouse lungs, which was scored on a scale of 0–3 with H&E stain. Fig. S2 shows MHC-II and CD86 expression in macrophages and dendritic cells. Fig. S3 shows quantification of gene expression in IFN-γ-stimulated BMM pretreated with DHT, the effect of the AR antagonist, flutamide, on DHT enhancement of IL-4-stimulated M2 macrophage polarization in BMM, and the quantification of gene expression in BMM from AR^flox and AR^floxLysMCre male mice stimulated with the indicated concentrations of IL-4 for 48 h. Fig. S4 shows a schematic summary of AR-mediated enhancement of M2 macrophage polarization during allergic lung inflammation.

Results

DHT reconstitution of castrated male mice reduces cell recruitment to the bronchoalveolar space but enhances M2 polarization of AM

To determine the role of androgens in lung inflammation, particularly on the polarization of AM, we evaluated lung inflammation in an OVA model of allergic lung inflammation in castrated male mice (Fig. 1A). Mice were castrated at 3 weeks old, prior to sexual maturation, to avoid long-term remodeling effects of sex hormones, as has been demonstrated for E2, on chromatin (18, 21). At 7 weeks of age before the allergen challenge with aerosolized OVA, mice were implanted with pellets that either released DHT, a potent and non-aromatizable metabolite of testosterone (22), or placebo. Pellets were implanted after the last sensitization to minimize hormone effects on the development of T-cell responses, as the timing of ovariectomy/E2 withdrawal affects outcomes in allergic lung inflammation (23). To confirm the systemic activity of DHT, we compared body mass of the placebo- and DHT-implanted groups (24). Both groups reconstituted with DHT exhibited significantly greater weight gain than the placebo groups (Fig. 1B). As another indicator of the systemic activity of DHT, we compared the weight of the seminal vesicles and quantified serum testosterone and DHT in the mice from the different experimental groups. As expected, DHT pellet implantation induced an increase in the weight of the seminal vesicles and an increase in serum testosterone compared with the placebo-implanted mice (Fig. S1 A and B).

Figure 1. Androgen reconstitution of castrated C57BL/6 wild-type mice with DHT-releasing pellets decreases allergic lung inflammation.

C57BL/6 male mice were castrated at 3 weeks of age. At 7 weeks, allergic lung inflammation was induced with the OVA protocol. (A) Schematic of the timeline for OVA-induced allergic lung inflammation and pellet implantation. IP, intraperitoneal; OVA, ovalbumin; PBS, phosphate-buffered saline; Neb, nebulization. On day 16, mice were anesthetized and BALF was collected. The left lobe from each mouse lung was inflated with 10% formalin and paraffin-embedded for sectioning. (B) Percent change in mouse body weight on day 7 after OVA or PBS. (C) Immunohistochemical analysis of lung tissue from DHT- and placebo-implanted mice. Mucus production was measured by Periodic acid–Schiff (PAS) staining. Cell infiltration was analyzed by hematoxylin and eosin (H&E) staining. Images were acquired with a 20X objective. Periodic acid–Schiff⁺ quantification is plotted as PAS⁺ cells / 1000 pixels. Hematoxylin and eosin graphic represents cell infiltration scored on a scale of 0 to 3 (0 = no inflammation, 1 = light infiltration in only a few areas, 2 = moderate cell infiltration around <50% of the airways and vessels, and 3 = dense cell infiltration around >50% of the airways and vessels. (D) ELISA quantification of total IgE in serum from the different implanted groups. (E) Gating strategy for BALF cells analyzed by FACS. Total cells (left panel), live cells (middle panel), and eosinophils (LIVE/DEAD^-CD11c^-Siglec F⁺) and AM (LIVE/DEAD^-CD11c⁺Siglec F⁺) (right panel). (F) Number of total live cells in BALF from DHT- and placebo-implanted mice. Each data point represents one mouse. (G) Quantification of eosinophils (left panel) and AM (right panel) in BALF. (H) YM1 quantification in BALF by ELISA. (I) Analysis of intracellular YM1 in AM by FACS. The change in mean fluorescence intensity (MFI) of YM1 (MFI target – MFI isotype) is shown, with female data pooled from our other experiments here and previously. Results are representative of three different experiments. n = 9–12 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001.

Histological analysis of lung sections from these mice showed no differences in mucus production in airway epithelial cells after DHT reconstitution (Fig. 1C, top). Though cell infiltration was observed in both OVA groups in the periphery of alveolar spaces by H&E staining, no difference was found with DHT treatment (Fig. 1C, bottom). Total IgE in serum was quantified to examine the effect of DHT on Ig production from B cells. The DHT-OVA group produced less total IgE than did the placebo-OVA group (Fig. 1D).

Next, we analyzed the cellularity, composition, and polarization status of AM in the BALF. We focused on AM in particular because only AM showed changes in macrophage polarization from our previous work in this model (18) . Live CD45⁺ cells were identified as AM (CD11c⁺Siglec F⁺), eosinophils (CD11c^-Siglec F⁺), monocytes (Ly6C⁺Ly6G^-), and neutrophils [(Ly6C^lowLy6G⁺); Fig. 1E]. Alveolar macrophages were also highly autofluorescent, CD64⁺, CD11b^-, Ly6C^-, Ly6G^-, CD24^- and MHC-II^low cells (data not shown). As expected, the total number of cells in the BALF was significantly increased of the placebo-OVA group (Fig. 1F). Dihydrotestosterone reconstitution reduced BALF cellularity in the OVA group, consistent with a regulatory role for androgen in suppressing inflammation in the lung [(13); Fig. 1F]. Eosinophils were the major cell type recruited into the BALF after challenge (Fig. 1G, left). Dihydrotestosterone reconstitution reduced eosinophils in the BALF after OVA challenge (Fig. 1G, left) and there was no change in the number of AM (Fig. 1G, right).

To examine the effect of DHT on polarization of AM, we measured expression of intracellular YM1, a canonical M2 marker (25), in AM and of secreted YM1 in the BALF. YM1 is induced in AM (18) and in other cells, including lung epithelial cells (26), by allergic lung inflammation. Furthermore, we and others have reported previously that YM1 is differentially expressed in female and male mice (7, 8, 18). It is expressed more highly in AM from female compared to male mice during allergic lung inflammation (18). Dihydrotestosterone reconstitution reduced YM1 in the BALF from OVA-challenged mice (Fig. 1H), consistent with suppression of allergic inflammation by DHT. The higher expression of YM1 in BALF from placebo-implanted mice suggests that other cell types, possibly epithelial cells (26), express more YM1 in the absence of systemic androgen. Previously we have reported that E2 acting through ERα enhances M2 polarization in macrophages induced by IL-4. The increased production of YM1 observed in female macrophages and also induced by implantation with E2 indicated the importance of E2 in this process. Therefore, we analyzed how androgen reconstitution with DHT affected YM1 production in AM from castrated males and compared this with the YM1 production in intact male and female mice. We found that DHT reconstitution of castrated male mice resulted in the recovery of YM1 expression in AM (Fig. 1I, dark gray bar with diagonal stripes) to a similar degree as that measured in intact male mice (Fig. 1I, dark gray bar with vertical stripes). However, females from the OVA group had a higher intracellular expression of YM1 in AM (Fig 1I, black bar) compared to both intact and DHT-implanted mice from the OVA groups. This observation suggests that androgen/AR in males plays a similar role to E2/ERα in females in inducing expression of YM1 and M2 polarization, but that E2 has a stronger effect than androgen. Together, the YM1 data indicate that DHT may enhance the overall M2 macrophage polarization program but that DHT may have different effects on macrophages compared to other cell types.

Pretreatment with DHT enhances M2 gene expression, YM1 and FIZZ1 production in BMM

We asked whether DHT enhancement of M2 polarization in AM was a direct effect on macrophages. To address this, we used an in vitro system with BMM from male and female C57BL/6 mice. Although BMM will never completely recreate the phenotype of a tissue macrophage, they have been successfully and widely used in vitro to interrogate macrophage biology. Thus, BMM were differentiated from bone marrow for 10 days with macrophage colony-stimulating factor (M-CSF). Intracellular flow cytometry-based staining confirmed expression of androgen receptor (AR) in BMM after we validated the specificity of the AR antibody in AM from C57BL/6 female and male mice with an AR-degrading drug, ASC-J9 (Fig. 2A, left panel). Both male and female BMM expressed AR (Fig. 2A, middle panel), although male BMM expressed more (27). Interestingly, AM had higher expression (~10-fold more) of AR than did BMM (Fig. 2A, right panel). We speculated that this might be due to differences in the environmental cues in the bone marrow and the lungs. It has been demonstrated that GM-CSF is involved in the development of AM in the lungs (28, 29), hence we compared AR expression in BMM developed in either M-CSF or GM-CSF. As expected, GM-CSF not only induced higher expression of CD11c and Siglec F on F4/80⁺ BMM (Fig. 2B), but also induced ~25-fold more CD11c⁺ Siglec F⁺ double-positive “AM-like” macrophages than did M-CSF (Fig. 2C). The bone marrow (BM) cells that we grew in GM-CSF were not dendritic cells [DC; (30, 31)]. This was validated by comparing expression of CD86 and MHC-II on BM grown in M-CSF and GM-CSF to cells grown in GM-CSF + IL-4 [(32); Fig. S2]. GM-CSF induced much higher AR expression in BMM (38.86 ± 4.26% AR-positive) than did M-CSF (7.23 ± 2.72% AR-positive, Fig. 2D). Hence, the tissue environment in which resident AM develop affects the expression of AR and, likely, the responsiveness of tissue macrophages to androgens.

Next, we treated BMM with DHT, then stimulated them with 1 ng/mL IL-4 for 48 h. Expression of mRNA for the M2 macrophage genes chitinase 3-like 3 (Chi3l3), resistin-like molecule alpha (Retnla), arginase1 (Arg1), and matrix metalloproteinase-12 (Mmp12) was determined by quantitative PCR (qPCR). Dihydrotestosterone exposure increased IL-4-induced gene expression of Chi3l3, Retnla, and Arg1 in both male and female BMM (Fig. 2E). Dihydrotestosterone did not enhance IL-4-induced Mmp12 gene expression (Fig. 2E), suggesting that many, but not all, IL-4-induced M2 genes are enhanced by DHT. Moreover, no differences in IFN-γ-induced M1 macrophage gene expression were observed following DHT pretreatment and IFN-γ stimulation of BMM (Fig. S3 A).

Dihydrotestosterone pretreatment not only enhanced IL-4-mediated M2 gene expression but also increased IL-4-induced production of secreted YM1 and found in inflammatory zone 1 [(FIZZ1); Fig. 2F]. Thus, the enhancing effect of DHT on M2 polarization was present in both AM in vivo and BMM in vitro despite the differences in AR expression.

To demonstrate that DHT-induced augmentation of M2 gene expression was a result of AR activity in macrophages, we used flutamide, an AR antagonist, to block AR activity in BMM pretreated with DHT prior to IL-4 stimulation. We observed that pharmacological blockade of AR reduced the DHT-mediated increase in the IL-4-induced expression of Chi3l3, Retnla, and Arg1 in BMM to the amount measured with IL-4-stimulation alone (Fig. S3 B). This suggests that the augmentation of IL-4-induced M2 gene expression by DHT is mediated by the AR in BMM.

Cell recruitment to the BALF increases in the absence of AR

To further validate the role of AR in regulating macrophage polarization during allergic inflammation, we compared lung inflammation and AM polarization in AR^flox and AR^floxLysMCre male and female mice following OVA-induced allergic lung inflammation. Mice were genotyped for the AR gene 855 base pairs (bp), AR floxed gene (AR^flox 952 bp), and Cre (Fig. 3A, left). Expression of the zinc finger protein, Y-linked (Zfy) gene, an indicator of genetic sex, was determined as inconsistencies between external and genetic sex in this strain were reported (16). Phenotypic females were also genetic females (Fig. 3A). A significant decrease in AR protein was determined in AM from male and female AR^floxLysMCre mice (Fig. 3A, middle and right).

Figure 3. Bronchoalveolar cell recruitment is impaired in AR^floxLysMCre male mice in the OVA model of allergic lung inflammation.

(A) Genotyping of AR^flox and AR^floxLysMCre male and female mice. Zfy, AR (AR [855 bp], AR floxed gene [952 bp] and the excision product [404 bp]), and LysM (LysM and LysMCre) genes were analyzed by electrophoresis on a 3% agarose gel with ethidium bromide (left panel). AR measurement by intracellular FACS stain of AM from AR^flox and AR^floxLysMCre female (middle panel) and male (right panel) mice. (B) Total live cells (left panel), eosinophils (middle panel), and AM (right panel) recovered from the BALF of AR^flox and AR^floxLysMCre female and male mice on day 16 of the OVA protocol. (C) Immunohistochemical analysis of lung tissue from AR^flox and AR^floxLysMCre male mice. Mucus production was measured by Periodic acid–Schiff (PAS) staining. Cell infiltration was analyzed by hematoxylin and eosin (H&E) staining. Images were acquired with a 20X objective. PAS⁺ quantification is plotted as PAS⁺ cells / 1000 pixels. The H&E graph represents cell infiltration scored on a scale of 0–3, as in Figure 1H. Collagen deposition was analyzed by Masson trichrome stain. Concentration of (D) testosterone and (E) total IgE in serum were determined by ELISA. (F) Proliferation of AM by BrdU incorporation is shown as percent of BrdU⁺ cells in the total alveolar macrophage population. Results are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001.

Based on our results from DHT delivery in vivo (Fig. 1), we hypothesized that the absence of a functional AR in macrophages would result in increased BALF cellularity in the AR^floxLysMCre animals. Total cells in the BALF did not differ between AR^flox males and females (Fig. 3B, left). Contrary to expectations, the total number of cells in the BALF was lower in the OVA AR^floxLysMCre males (Fig. 3B, left), highlighting the importance of macrophages in allergic lung inflammation. The increased cellularity of the BALF in the WT mice was due to eosinophil recruitment (Fig. 3B), as before (Fig. 1G). However, no eosinophils were recruited in the AR^floxLysMCre males challenged with OVA (Fig. 3B, middle panel, filled squares). Furthermore, AR^floxLysMCre males had more AM than AR^flox males (Fig. 3B, right). We observed no differences in AM in the females (Fig. 3B, right), nor monocyte or neutrophil recruitment in any group (data not shown). Hence, the increased number of AM in the BALF of AR^floxLysMCre males was not due to recruitment of monocytes.

To determine if differences in eosinophils and AM in AR^flox and AR^floxLysMCre males affected pathology, we measured lung inflammation by histology. Periodic acid–Schiff staining showed significantly less mucus production in AR^floxLysMCre males after OVA exposure (Fig. 3C, left). AR^floxLysMCre males also had fewer inflammatory cells in the lung tissue, as shown by H&E staining (Fig. 3C, middle). We did not observe significant collagen deposition (Fig. 3C, right), typical of acute models of allergic inflammation (33).

The absence of AR in macrophages could potentially increase available testosterone as has been shown for the concentration of E2 in ERα-deficient mice (34–36). Serum testosterone in AR^flox and AR^floxLysMCre male mice was similar (1.315 ± 0.092 ng/mL and 1.14 ± 0.027 ng/mL, respectively; Fig. 3D). Thus, suppression of allergic lung inflammation in the AR^floxLysMCre males was not due to differences in testosterone. No difference in total IgE concentration was measured due to the absence of AR on macrophages (Fig. 3E). However, the female OVA groups had higher serum IgE than their male counterparts, as has been described previously (7).

Androgen receptor triggers androgen-dependent proliferation of cancer cells (37) but its effects on proliferation of other cell types is poorly understood. The elevated number of AM in AR^floxLysMCre mice and the unexpected effect of androgens on M2 polarization prompted us to compare proliferation of AM from AR^flox and AR^floxLysMCre mice. Proliferation of AM was low (~13%), consistent with low AM turnover and the long-lived nature of these cells (38). Similar proliferation of AM from AR^floxLysMCre and AR^flox mice was measured by BrdU incorporation (Fig. 3F).

Impaired eosinophil recruitment in BALF of AR^floxLysMCre males correlates with decreased chemokine levels in AM and BALF

To discern whether the decreased recruitment of eosinophils in AR^floxLysMCre males was due to diminished eosinophil-recruiting chemokines and eosinophil survival cytokines, we measured CCL24, CCL5, and IL-5 in the BALF from AR^flox and AR^floxLysMCre males. Chemokines and IL-5 were significantly lower in the AR^floxLysMCre BALF than in AR^flox BALF (Fig. 4A). However, no differences were observed in TNFα, IL-4, or IL-33, suggesting that AR regulates expression of some but not all cytokine and chemokine genes (data not shown).

Figure 4. Absence of AR results in decreased expression of chemokines and cytokines in AM of male mice.

BALF and AM were obtained from AR^flox and AR^floxLysMCre male mice on day 16 of the allergic lung inflammation model. (A) CCL24, CCL5, and IL-5 were measured in the BALF by ELISA. (B–E) BALF cells were seeded for 2 h in αMEM, and non-adherent cells were vigorously washed away. Relative gene expression in AM was analyzed by qPCR using the 2^-ΔΔCT method and compared to the amount of mRNA in the AR^flox male PBS sample (=1). (B) Ccl2 gene expression. (C) Ccl3, Ccl5, Ccl24, and Ccl11 gene expression. (D) Ccl17 and Ccl22 gene expression. (E) Tgfb and Il10 gene expression. Results are representative of two independent experiments. *p < 0.05, **p <0.01, ***p <0.001.

Because BALF is a mixture of cytokines and chemokines from different cellular sources, we determined chemokine gene expression specifically in AM by qPCR. Alveolar macrophages were enriched by adherence as previously reported [from ~10% to ~80% of the BALF cells from the allergic mice; (18)]. Ccl2, a monocyte-recruiting chemokine, was significantly reduced in AR-deficient AM from the OVA group (Fig. 4B). We also found significantly reduced expression of Ccl3, Ccl5, and the eotaxins, Ccl24 and Ccl11 (eosinophil-recruiting chemokines), in AM lacking AR (Fig. 4C, filled squares). A significant reduction in Ccl17 but not Ccl22 was found in AM from AR^floxLysMCre animals (Fig. 4D). However, the diminished M2 phenotype did not correlate with an increase in the “regulatory” cytokines, TGF-β and IL-10. In fact, expression of the mRNA for Tgfb and Il10 was higher in the AR-sufficient AM after OVA challenge (Fig. 4E), possibly a macrophage-mediated mechanism to suppress allergic inflammation. Therefore, AR deficiency in AM resulted in decreased expression of the typical chemokines that recruit eosinophils to the allergically inflamed lung.

AR promotes M2 polarization of AM

M2 polarization of AM during allergic lung inflammation induces eosinophil recruitment to the alveolar space (3) and expression of FIZZ1, ARGINASE 1, and YM1 (39). Given the diminished production of eosinophil-recruiting chemokines and cytokines by AM lacking AR, we tested the possibility that the AR deficiency diminishes polarization to the M2 phenotype.

First, we compared the amount of YM1 and FIZZ1 in BALF from AR^flox and AR^floxLysMCre males following allergic lung inflammation. YM1 and FIZZ1 in BALF were significantly induced by OVA challenge (Fig. 5A) but expression of both was lower in the AR^floxLysMCre males. The difference in YM1 protein in the BALF was confirmed by ELISA (AR^flox OVA group = 1,496.24 ± 158.9 ng/mL; AR^floxLysMCre OVA group = 803.37 ± 113.9 ng/mL). We also quantified the AM production of MMP-12 and MMP-9, proteins involved in tissue remodeling and associated with asthma exacerbations (40). Ovalbumin challenge induced significant MMP-12 in the BALF of both AR^flox and AR^floxLysMCre mice (Fig. 5B) but it was significantly less in AR^floxLysMCre mice. MMP-9 production was induced only in the AR^flox OVA-group (Fig. 5B). To determine if AM were responsible for these differences in BALF proteins, we analyzed M2 gene expression in AM. Quantitative PCR analysis revealed that expression of the M2 genes, Chi3l3, Retnla, and Arg1, was significantly lower in AM lacking AR (Fig. 5C). No differences were observed in Mrc1 expression, although this gene was also less robustly induced.

Figure 5. Impaired M2 polarization of AR-deficient alveolar macrophages.

BALF and AM were obtained from AR^flox and AR^floxLysMCre male mice after induction of allergic lung inflammation with OVA. (A) The amount of YM1 and FIZZ1 protein in the BALF was determined by Western blot. Densitometry of the specific bands was normalized to the amount in the AR^flox OVA group (= 100%). (B) MMP-12 and MMP-9 were measured in BALF by ELISA. (C) AM from BALF were isolated as described in Figure 4A and analyzed for expression of the indicated M2 genes by qPCR with the 2^-ΔΔCT method. Amount of mRNA was compared to that in the AR^flox PBS sample (= 1). (D) BMM cultured from AR^flox and AR^floxLysMCre male mice as described in Figure 2B were pretreated overnight with DHT at the concentrations shown and then stimulated with IL-4 (1 ng/mL) for 48 h. Relative gene expression was determined by qPCR with the 2^-ΔΔCT method and compared to the amount of mRNA in the IL-4 sample (= 100%) for each genotype (n = 3 independent experiments). Results are representative of three independent experiments. *p < 0.05, **p <0.01, ***p <0.001.

Next, we measured M2 polarization in BMM from male mice treated with or without DHT prior to IL-4 stimulation. Bone marrow macrophages from AR^flox and AR^floxLysMCre males exhibited no differences in M2 gene expression when stimulated with IL-4 alone (Fig. S3 C). However, treatment of BMM with increasing concentrations of DHT enhanced IL-4–induced gene expression of Chi3l3, Retnla, and Arginase1 in the AR^flox BMM but not in BMM lacking AR (Fig. 5D). Only the highest concentration of DHT (10 nM) induced a significant increase in the expression of Chi3l3 and Retnla in AR^flox BMM. This inability of DHT to enhance M2 gene expression in BMM lacking AR corroborated the decrease in M2 gene expression observed in AR^floxLysMCre mice after the OVA protocol (Fig. 5A). Together, these results indicate that AM are a major source of YM1, FIZZ1, and ARGINASE 1 in allergic lung inflammation and that AR is responsible for DHT augmentation of IL-4-induced M2 gene expression.

In summary, we show genetically and pharmacologically that AR is important for full polarization of M2 macrophages in male animals. Rather than suppressing M2 polarization as we hypothesized, AR activation by androgens resulted in enhancement of the IL-4-activated M2 macrophage polarization in vitro, while lacking AR in AM impaired polarization to the M2 macrophage phenotype in vivo. Similar to that which we previously described for E2 and ERα, here we describe a role for androgen and AR in enhancing M2 macrophage polarization in macrophages from male animals. Also, we demonstrate the importance of AR in macrophages in allergic lung inflammation in vivo. The diminished M2 profile of alveolar macrophages from male AR^floxLysMCre mice during allergic lung inflammation dramatically reduced the overall inflammatory profile of the lungs.

Discussion

Allergic asthma is characterized by M2 polarization of AM that correlates with disease severity and eosinophil recruitment (2). Asthma affects more boys than girls before puberty, and more women than men during adulthood (6, 41). Similar sex differences have been observed in mice (7, 8). Generally, this suggest male and female sex hormones suppress and enhance allergic inflammation, respectively. Previously, we showed that E2 and ERα increase M2 macrophage polarization in vivo and in vitro (18). Other work has shown a suppressive role for androgens in ILC2 (14) and airway smooth muscle cells (13) in allergic lung inflammation. From this work and our E2 study, we hypothesized that androgens would suppress M2 polarization of macrophages. Our current study has uncovered a novel, cell-specific role in macrophages for androgens acting through the AR. As expected, androgens (DHT) down-modulated Th2 inflammation (eosinophil recruitment, IgE production) in the OVA model of allergic lung inflammation, supporting the idea that androgens act as global immune suppressors. However, analysis of the M2 macrophage marker, YM1, in vivo revealed greater M2 polarization (more intracellular YM1) in the DHT-treated mice and in AR-sufficient AM (Figure S4). This unexpected finding suggests that androgens may act differently in macrophages than in other immune cells in the setting of allergic inflammation.

Our in vitro analysis showed that DHT enhanced the expression of hallmark M2 genes (Chil3l3, Retnla, and Arg1) in BMM from both female and male mice, suggesting androgens are important immune modulators in both sexes. Taken with our previous work demonstrating a role for E2/ERα in M2 polarization, our findings here suggest both male and female sex hormones can polarize macrophages to the M2 phenotype. This is similar to earlier studies in B-cells that demonstrated that male and female sex hormones have the same effect on B-cell lymphopoiesis (42–44). Other studies demonstrate that estrogen can suppress the production of IL-2 by T-cells from both women and men (45) and that DHT can increase IL-10 gene expression in T-cells from male and female mice (46). However, physiologically available concentrations of male and female sex hormones in unmanipulated male and female animals would preclude M2 polarization by the opposite sex hormone and its receptor. Androgens are much lower in female animals than in males and therefore, androgen-/AR-augmented M2 polarization would be less important in females compared to males. Thus, the mechanisms by which M2 macrophage polarization is enhanced in vivo are different between females (estrogen-driven) and males (androgen-mediated). As shown in Fig. 1I, androgen reconstitution of castrated male mice with DHT recovered YM1 expression to the degree observed in intact males, while the absence of androgens impaired YM1 induction. Altogether with our previous report demonstrating a stronger YM1 expression in females than males (18) this indicates that, although both female and male sex hormones enhance M2 macrophage polarization, estrogen acts as a more potent M2 macrophage enhancer that androgens. Our study reinforces the need for different, sex-dependent approaches to asthma therapy.

Compensatory mechanisms have been described when either receptors or ligands are decreased (47, 48). Lack of a sex hormone receptor can cause supraphysiological elevations in that sex hormone as has been demonstrated for E2 in ERα-deficient mice (34–36). However, the concentration of testosterone in the serum of the AR^floxLysMCre and AR^flox mice was the same, indicating that the differences we saw were due solely to absence of the canonical AR, not to increased concentration of testosterone.

Another important difference observed between in AR^floxLysMCre and AR^flox male mice was more resident AM in the AR^floxLysMCre mice at steady-state. These data suggest that AR negatively regulates the number of AM in the lung, possibly by regulating proliferation or survival. Since BrdU incorporation into AM was similar, proliferation of AR-deficient AM was not higher than WT AM. Either increased survival or increased differentiation of monocytes to AM may explain the net difference in the number of AM. This contrasts with the role for androgens in driving cell proliferation in prostate cancer [PCa; (49)], and with suppression of IL-33-driven proliferation/survival of ILC2s in allergic lung inflammation (14). Thus, AR deficiency could increase cell survival or differentiation of AM from lung resident self-renewing or blood-derived monocytes, resulting in more AM in the AR^floxLysMCre male mice. We will address these questions in future work.

Recognizing the ability of androgens to enhance M2 polarization of macrophages could lead to new therapies for diseases, such as asthma and PCa, thought to be worsened by M2 macrophage polarization and sex hormones. For example, tumor-associated macrophages (TAM) are M2-like macrophages (50) that are pro-tumorigenic in PCa (51). Because high concentrations of testosterone are associated with PCa development (52), hormone therapy and androgen-deprivation therapy (ADT) have been used to control this disease (53). Our work suggests that ADT is beneficial not only in reducing the proliferation of prostate cancer cells but also in reducing M2 TAM polarization. The GM-CSF–induced enhancement of AR expression we observed in AM is also relevant to tumor immunology, as GM-CSF is used as an anti-cancer therapy (54). However, GM-CSF may also promote tumor progression and metastasis of colon adenocarcinoma (55), head and neck cancer cells (56), and colon cancer (57). Our data suggest that these results could stem from increased AR expression and polarization of TAMs to the M2 phenotype.

In summary, this work expands our understanding of macrophage polarization, revealing an unexpected role for androgens and AR. Androgens enhanced, rather than suppressed, M2 macrophage polarization. The M2-promoting role of DHT was, however, specific for macrophages, as DHT suppressed allergic inflammation overall. The increased M2 macrophage polarization induced by androgens was mediated by AR (Figure S4), as this effect was absent in macrophages from AR^floxLysMCre male mice. These results suggest possible mechanisms to counteract diseases promoted by M2 macrophages and affecting mainly men, such as eosinophilic esophagitis (58) and PCa (59). A deeper understanding the mechanisms by which androgen/AR enhances M2 macrophage polarization will help us to develop effective therapies against diseases in which M2 macrophages are important mediators.