Growth Hormone Reprograms Macrophages toward an Anti-Inflammatory and Reparative Profile in an MAFB-Dependent Manner

Growth hormone (GH) is produced and secreted by somatotropic cells of the anterior pituitary gland in mammals and is a main regulator of postnatal growth and development. Although initially implicated in somatic growth control, GH is a pleiotropic hormone with myriad functions, including anabolic actions on muscle and bone and catabolic effects on adipose tissue (1). It also promotes fluid retention in the kidney (2), exerts metabolic effects on the liver (3), triggers sexual maturation (4), and induces insulin resistance (5). Many of these effects are mediated through insulin-like growth factor (IGF)-1, which is produced in response to GH in liver and other tissues (6).

Studies both in vitro and in vivo have also demonstrated a role for GH in immune regulation. GH receptor (GHR) is expressed by various leukocyte subsets (7), including B and T cells (8) and macrophages (9)^, and GH has been shown to mediate thymic development (8), promote T cell engraftment in SCID mice (10), improve B cell responses and Ab production (11, 12), and modulate the activity of NK cells (13) and macrophages (9). In myeloid cells, GH stimulates the proliferation of RAW 264.7 macrophages (14) and functions as a human macrophage–activating factor (15). Beneficial effects of GH administration have been reported in autoimmunity, as it alters tolerization mechanisms through activation of regulatory T cells and modulation of Th17 cell plasticity, reduces type I diabetes development (16), and contributes to improve collagen-induced arthritis symptoms (17). In inflammatory diseases, GH limits mucosal inflammation in experimental colitis (18) and has protective effects in patients with active Crohn’s disease (19, 20).

Tissue-resident macrophages are a heterogeneous and highly plastic cell population whose effector functions are dependent on their ontogeny and the surrounding tissue microenvironment. Their functional versatility enables them to play critical roles in both the initiation and resolution of inflammatory and immune responses. Reflecting this functional versatility, macrophages primed by GM-CSF (GM-Mϕ) or macrophages primed by M-CSF (M-Mϕ) exert opposite functions in terms of T cell stimulation, inflammatory cytokine production, tumor cell growth, and methotrexate sensitivity (21–25). Indeed, GM-Mϕ and M-Mϕ are considered as pro- and anti-inflammatory macrophages, respectively, as they display specific transcriptional profiles resembling those of macrophages from the synovia of rheumatoid arthritis patients or tumor-associated macrophages, respectively (26, 27).

We report in this study that GH downmodulates the transcriptional and cytokine profile of proinflammatory GM-Mϕ by reducing, in a PI3K-dependent manner, the secretion of activin A, a protein involved in tilting the balance of macrophages toward a proinflammatory phenotype (28), and upregulating the expression of MAFB, a transcription factor that promotes the anti-inflammatory polarization of macrophages (29, 30). In line with these in vitro results, we demonstrate that GH-overexpressing transgenic (GHTg)–mice show improved remission of inflammation and mucosal repair during recovery from dextran sodium sulfate (DSS)–induced acute colitis. Indeed, together with possible effects on other immune cells, the polarization of gut macrophages from GHTg mice was shifted toward an anti-inflammatory/reparative profile during the recovery phase. Overall, our results point to GH as a regulatory factor for macrophage polarization under physiological and pathological settings.

Mice transgenic for bovine GH under the control of the phosphoenolpyruvate carboxykinase promoter were on a C57BL/6J background (31) and were maintained by repeated backcrosses to C57BL/6J females. A total of 40 GHTg mice and 40 control littermates (10–14 wk old) were subjected to the DSS-induced experimental model of colitis, using matched sex ratios in each experiment. Mice were handled according to national and European Union guidelines, and experiments were approved by the Animal Experimentation Ethics Committee of the Madrid regional government (PROEX 316- 15).

Acute colitis was induced with 3% DSS (w/v, molecular mass 36,000–50,000 Da; MP Biomedicals, Solon, OH) in drinking water for 5 d, followed by a recovery period of an additional 9 d as indicated. Mice were monitored for body weight and signs of distress, diarrhea, and rectal bleeding and (when required) were killed. The colon was removed, washed with PBS, and opened longitudinally for analysis.

When needed, infiltrating immune cells were obtained by enzymatic digestion of the intestinal tissues. Briefly, tissue samples were incubated (30 min, 37°C) in 10 ml HBSS with 0.4 mg/ml Dispase (Life Technologies) and 10,000 U/ml penicillin/streptomycin. FBS was added to 5% (v/v), and tissue debris was removed by sequential filtering through 100, 70, and 40 μm cell strainers. Cells were collected by centrifugation (150 × g, 10 min, 4°C). Colon cells were stained with combinations of fluorescence-labeled Abs to the cell surface markers CD45, CD11b, F4/80, Gr1, CD11c and CD86 and analyzed in a Gallios Flow Cytometer (Beckman Coulter, Brea, CA). The profiles obtained were analyzed with FlowJo software (BD Biosciences); leukocytes were gated as CD45⁺ cells. F4/80⁺ cells from mouse colon samples were isolated using anti-F4/80 MicroBeads UltraPure (Miltenyi Biotech) (∼80% purity).

Full-length colon samples were longitudinally divided into two halves for histological and transcriptional analysis. Inflammation and crypt damage were scored on formalin-fixed, paraffin-embedded colon sections stained with H&E. The scoring scheme (32) included inflammation severity (0, none; 1, mild; 2, moderate; and 3, severe), inflammation extent (0, none; 1, mucosa; 2, submucosa; and 3, transmural), crypt damage (0, none; 1, basal one-third damage; 2, basal two-thirds damage; 3, crypt loss, surface epithelium present; and 4, crypt and surface epithelium loss), and percentage of colon involvement (0, 0%; 1, <25%; 2, 25–50%; 3, 50–75%; and 4, >75%). Histological evaluation was performed by two independent investigators in a double-blinded fashion.

For analysis, formalin-fixed, paraffin-embedded colon sections were stained with anti–arginase-1 (BD Biosciences, San Jose, CA) and K13-a anti–inducible NO synthase (iNOS) (Novus Biologicals, Littleton, CO) Abs and immunodetected with Alexa 488–labeled goat anti-rabbit IgG (Molecular Probes, Eugene, OR). Slides were mounted for fluorescence with ProLong Gold reagent with DAPI (Invitrogen, San Diego, CA) and visualized on a TCS SP5 Confocal Microscope (Leica Microsystems, Wetzlar, Germany).

Human PBMCs were isolated from buffy coats of healthy donors by density gradient centrifugation using endotoxin-tested (<0.12 endotoxin units/ml) Ficoll Paque Plus (GE Healthcare, Uppsala, Sweden), according to standard procedures. Monocytes were purified from PBMCs by magnetic cell sorting using CD14 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Monocytes (>94% CD14⁺ cells) were cultured at 0.5 × 10⁶ cells per milliliter for 7 d (37°C, 5% CO₂) in RPMI 1640 medium supplemented with 10% FCS, sodium pyruvate and l-glutamine, and containing GM-CSF (1000 U/ml) or M-CSF (10 ng/ml) (both from ImmunoTools, Friesoythe, Germany) to generate GM-Mϕ and M-Mϕ, respectively. Cytokines were added every 2 d. To evaluate the direct effect of GH on monocytes, cells were incubated with recombinant human GH (rhGH) (1 μg/ml Genotonorm; Pfizer, New York, NY; for 7 d, 37°C, 5% CO₂). To determine the effect of GH on differentiated GM-Mϕ and M-Mϕ, cells were treated with rhGH (1 μg/ml) in fresh medium for 24 h. To evaluate the effect of IGF-1 on differentiated GM-Mϕ, cells were treated with rhIGF-1 (100 ng/ml) for 24 h, or it was added at the same time as GM-CSF. For signaling analysis, GM-Mϕ and M-Mϕ were stimulated with rhGH (1 μg/ml) for the indicated times. Where appropriate, GM-Mϕ and M-Mϕ were treated with LY294002 (10 μM, 60 min) prior to rhGH addition. Functional polarization of GM-Mϕ and M-Mϕ was determined by quantification of cytokine (TNF-α, IL-10, IL-6) production. Murine GM-Mϕ and M-Mϕ bone marrow–derived macrophages were generated using human M-CSF (10 ng/ml; ImmunoTools) or murine GM-CSF (20 ng/ml; PreproTech), respectively, as previously described (21, 33).

RNA sequencing (RNAseq) was performed on RNA obtained from three independent human samples of untreated or GH-treated (1 μg/ml, 24 h) GM-Mϕ, using the BGISEQ-500 platform (BGI, Hong Kong, China), with 20 million reads and 100 paired-end sequencing reads. Statistical analysis of RNAseq data were performed following previously described procedures (29). Differential gene expression analysis by DSeq2 revealed no statistically significant differences between the experimental conditions. RNAseq data have been deposited in the Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra) under accession no. PRJNA555143. Gene-set enrichment analysis (GSEA) was performed using data sets available at the Web site (http://software.broadinstitute.org/gsea/index.jsp) (34), as well as data sets containing the genes differentially expressed by either human M-Mϕ (anti-inflammatory gene set) or GM-Mϕ (proinflammatory gene set) (GSEA68061) (28, 35), and the genes specifically downregulated in MAFB-silenced human M-Mϕ (GSE84622) (29).

Mouse samples were collected in RNAlater, and total RNA was extracted with TriReagent (both from Sigma-Aldrich, St Louis, MO). Total RNA from cultured cells was extracted using the Nucleospin RNA/Protein Kit (Macherey-Nagel, Düren, Germany). RNA was retrotranscribed using SuperScript II Reverse Transcriptase (Invitrogen), and real-time quantitative PCR (RT-qPCR) analysis was performed with Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). Triplicate samples were quantified using the ABI Prism HT7900 sequence detection system (Applied Biosystems). Oligonucleotides for selected genes were designed employing the Roche Universal ProbeLibrary Assay Design Center (Table I). Assays were performed in triplicates and results normalized according to the expression levels of TBP and GAPDH; for murine samples, results were normalized using Actb expression levels. Results were obtained using the 2^−∆Ct comparative threshold method for quantification.

Supernatants from untreated and GH-treated GM-Mϕ or M-Mϕ were evaluated for the presence of cytokines and growth factors using commercially available ELISA kits for activin A and CCL2 (Quantikine immunoassay; R&D Systems, Minneapolis, MN) and for IL-6, IL-10, and TNF-α (Human ELISA Max Standard; BioLegend, San Diego, CA) following the protocols supplied by the manufacturers.

GHR expression was determined in human samples using an anti-GHR mAb (36). The incubations were done in the presence of 20 μg/ml human IgG to prevent biding through the Fc portion of the Abs. Infiltrating immune cells from mouse intestinal tissues were phenotypically characterized using anti-CD45, anti-CD19, anti-CD11b, anti-Gr1, anti-F4/80, anti-CD11c and anti-CD86 mAb (BioLegend), as described previously (37). Analysis was performed on a Gallios Flow Cytometer (Beckman Coulter).

PBMCs were isolated from healthy donors as described and depleted of CD14⁺ cells using specific MicroBeads (Miltenyi Biotec) and magnetic cell sorting. Allogenic cells were then labeled with 5 μM Cell Trace Violet (CTV; Invitrogen) for 20 min at 37°C and washed twice with RPMI 1640 medium containing 10% FCS. CTV-labeled cells were seeded onto 96-well plates containing GM-Mϕ (1 or 2 × 10⁵ cells per well) or M-Mϕ (1 × 10⁴ cells per well) in the presence or absence of rhGH (1 μg/ml, 24 h, 37°C). Cells were cocultured in RPMI 1640 medium with 10% FCS (10 d) and supplemented with rhGH (1 μg/ml) 24 h before cell proliferation evaluation. As negative controls, we used isolated CTV-labeled cells in the presence or absence of rhGH. CTV-labeled cells exposed to PMA (100 ng/ml, 24 h) were used as positive control. Cell viability and proliferation (dye dilution) were determined on CD4⁺ cells using a Gallios Flow Cytometer (Beckman Coulter). The percentage of dividing cells was calculated using FlowJo (Tree Star, Ashland, OR).

Listeria monocytogenes coupled to GFP (Listeria-GFP), kindly provided by Dr. E. Veiga (Centro Nacional de Biotecnología/Consejo Superior de Investigaciones Científicas, Madrid, Spain), was grown in brain-heart infusion medium (overnight, 37°C), diluted, recovered at midlogarithmic growth phase (OD_600nm = 0.5), and washed in PBS before use.

Cells were infected with Listeria-GFP as described (38). Briefly, 30 min (37°C, 5% CO₂) after inoculation, gentamicin (100 μg/ml) was added (60 min; 37°C) to cultures to eliminate extracellular bacteria. Cells were then extensively washed, fixed with paraformaldehyde (4% w/v, 20 min, room temperature) and seeded (30 min, 37°C) onto poly-L-lysine–coated plates (20 μg/ml, 10 min, room temperature; Sigma-Aldrich) prior to their incubation with wheat germ agglutinin (Thermo Fisher Scientific, Waltham, MA) and DAPI (Sigma-Aldrich) for 5 min at 4°C. Cell staining was evaluated by confocal microscopy (Zeiss Axiovert LSM 510-META) and images were analyzed and quantified using ImageJ software (National Institutes of Health).

Cell lysates were prepared in RIPA buffer (50 mM HEPES, 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 10 mM NaF, 5 mM EDTA, 0.1 mM NaPO₄, 5% ethylene glycol) supplemented with 10 μg/ml of aprotinin and leupeptin, 10 μM sodium orthovanadate, and 1 mM PMSF. Lysates (50 μg) were subjected to electrophoresis and transferred to nitrocellulose membranes (GE Healthcare, Freiburg, Germany). After blocking with 5% BSA in TBS buffer (100 mM Tris 1 HCl [pH 8] and 1.5 M NaCl), membranes were incubated with anti-MAFB (BioLegend), anti-pGSK3α/β, -AKT, -pAKT, (Cell Signaling Technology, Danvers, MA), anti-ERK1/2 or anti-pERK1/2 (Santa Cruz Biotechnology, Santa Cruz, CA), or anti-vinculin (Sigma-Aldrich) Abs, followed by the corresponding HRP-conjugated secondary Ab.

For comparison of means, and unless otherwise indicated, statistical significance of the data were evaluated using paired Student t test and two-way ANOVA (Sidak multiple comparations test). The p value < 0.05 was considered significant (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Statistical parameters used in the GSEA analysis were as previously described (34).

Previous reports have demonstrated that GH modulates immune (16, 17) and inflammatory (17–20) responses. Because GH has also been shown to modulate adipocyte differentiation in a macrophage-dependent manner (9), we sought to determine whether GH affects macrophage polarization during inflammatory responses. To this end, we initially made use of a murine model of DSS-induced acute colitis in which GH limits pathological conditions (18). We observed that both GHTg and wild-type (WT) mice survived 5 d of treatment with 3% DSS. We did not conduct survival analysis during the recovery phase as euthanasia was performed at the designed time points to cover the kinetics of immune cell infiltration, but we did not observe differences in mortality in any of the groups of mice maintained until the end of the experiment (9 d). Along the 5-d DSS treatment, both GHTg and WT mice were positive for occult blood (Hemoccult test) with no significant differences between the two groups of mice (Fig. 1A). Likewise, weight loss (Fig. 1B) and histological scores for severity of inflammation were similar in both groups (Fig. 1C, upper panel). Lastly, we also detected comparable levels of destruction of the crypt structure, a disturbed epithelial layer and massive infiltration of inflammatory cells in mucosal and submucosal layers in both mouse strains (Fig. 1C, middle [WT] and bottom [GHTg]; for higher resolution images of these micrographs, see Supplemental Fig. 1A [WT] and Supplemental Fig. 1B [GHTg]), confirming the lack of significant differences between the two groups of mice during disease onset.

However, although both groups of mice showed similar relative weight loss during disease onset, the recovery phase was significantly quicker in GHTg mice in terms of changes in body weight (Fig. 1B). Histological analysis of samples harvested 7 d after the DSS treatment (DSS + 7) (Fig. 1D, upper, Fig. 1E; Supplemental Fig. 1C [WT] and Supplemental Fig. 1D [GHTg] for higher resolution images), further supported the role of GH in enhancing tissue repair after DSS-induced colitis based on the extent and severity of inflammation, crypt damage, and percentage of affected colon. To assess the state of macrophage polarization at the recovery phase, we determined the expression of conventional proinflammatory (iNOS) and anti-inflammatory/reparative (arginase 1) macrophage markers in the recovering tissue. Immunohistological analysis of DSS + 7 mice revealed that the proportion of arginase-1⁺ cells infiltrating the colon was higher in GHTg mice than in WT littermates, with the latter group showing an increase in the number of iNOS⁺ cells compared with GHTg mice (Fig. 1D, bottom). Of note, histological scores were also significantly different between GHTg and WT littermates at this stage, with a lower grade of mucosal destruction and inflammation in the colon of GHTg mice (Fig. 1E). Indeed, flow cytometry analysis of the infiltrating macrophages of the intestinal tissues revealed a higher percentage of anti-inflammatory macrophages (CD45⁺F4/80⁺Gr1⁺CD11b⁺ CD86^high) in GHTg mice (0.72 ± 0.13) than in controls (0.28 ± 0.09) (Table II). We also found a higher CD4⁺/CD8⁺ ratio in GHTg than in control animals (0.74 versus 0.38) in the recovery phase (DSS + 7), whereas the opposite occurred during inflammation (DSS + 1 d, GHTg 0.24 versus controls 0.68) (Table II). Notwithstanding the possibility of a GH-mediated effect on other immune cells, which might also contribute to improved recovery from DSS-induced colitis in GHTg mice, our results indicate a GH-triggered change in the polarization state of macrophages within the inflamed tissue.

The above results led us to hypothesize that GH might directly modulate macrophage polarization. To address this issue, we switched to the analysis of human monocyte–derived proinflammatory (GM-Mϕ) and anti-inflammatory (M-Mϕ) macrophages. Human GHR was found on monocytes, proinflammatory GM-Mϕ, and anti-inflammatory M-Mϕ, as demonstrated by flow cytometry using a GHR-specific mAb (36) (Supplemental Fig. 2A). Because the expression of the major GH effector molecule IGF-1 (39) is restricted to anti-inflammatory M-Mϕ (28), we evaluated the influence of GH on macrophage polarization by comparing the transcriptome of GM-Mϕ untreated or exposed to 1 μg/ml of rhGH for 24 h (Fig. 2A–C). GSEA of the RNAseq data revealed a significant positive enrichment of M-Mϕ-specific genes, but not GM-Mϕ-specific genes (28, 35), following GH exposure (Fig. 2D). We validated these findings by RT-qPCR analysis, which showed that GH treatment of GM-Mϕ significantly upregulated the expression of genes associated with the M-Mϕ anti-inflammatory profile (HMOX1, STAB1, IGF1, and FOLR2), whereas the GM-Mϕ–specific genes CCR2, MMP12, and EGLN3 were significantly downregulated (Fig. 3A). The effect of GH was specific for GM-Mϕ, as it failed to modify the transcriptome of M-Mϕ (Fig. 3B) or monocytes (Supplemental Fig. 2B). Overall, these data revealed that GH modulates the human macrophage gene signature, favoring the expression of genes specifically linked to anti-inflammatory M-Mϕ. To correlate these observations with data from GHTg mice, we derived murine GM-Mϕ and M-Mϕ from bone marrow progenitors using GM-CSF or M-CSF, respectively, followed by GH exposure. RT-qPCR analysis showed that GH treatment of murine GM-Mϕ significantly upregulated the expression of genes associated with the M-Mϕ anti-inflammatory profile (Arg1, Cd206, and Ym1), whereas it downregulated the expression of GM-Mϕ–associated inflammatory markers (Nos2, Inhba, and Tnf) (Fig. 3C). Thus, these data confirmed that GH also modulates the mouse macrophage gene signature, favoring the acquisition of anti-inflammatory markers.

As macrophage polarization is essentially defined by the prevailing cytokine secretion profile, we assessed the ability of GH to influence the cytokine signature of GM-Mϕ and M-Mϕ. To do this, we cultured both macrophage populations in the absence or presence of rhGH for 24 h and then measured the levels of TNF-α, IL-6, and IL-10 in the corresponding culture supernatants. Results showed that, compared with control cultures, GH exposure for 24 h triggered a significant decrease of TNF-α and a significant increase of IL-10 in GM-Mϕ supernatants, whereas the levels of IL-6 were unaffected (Fig. 4A). By contrast, and in accordance with the gene profile of M-Mϕ, GH treatment failed to modify the cytokine signature of anti-inflammatory macrophages (Fig. 4A). Because M-Mϕ secrete CCL2 (28), we also evaluated CCL2 levels in the culture supernatant of untreated or rhGH-treated GM-Mϕ or M-Mϕ and found that GH also induced a significant increase in the levels of CCL2 in GM-Mϕ supernatants (Fig. 4A). Altogether, these results indicate that GH treatment reprograms human macrophages, as it impairs the phenotypic and functional polarization of GM-Mϕ and causes a shift toward an anti-inflammatory profile.

Activins are pluripotent growth and differentiation factors of the TGF-β superfamily (40). Like TGF-β, activins exert immunostimulatory and immunosuppressive functions (41). Activin A is considered a crucial modulator of inflammatory responses (42) and, in fact, activin A skews macrophage polarization by promoting the acquisition of a proinflammatory phenotype (28). We thus measured the levels of activin A in the culture supernatants of untreated or rhGH-treated GM-Mϕ and M-Mϕ and observed that GH treatment led to a significant decrease in activin A in GM-Mϕ cultures (Fig. 4B), whereas it did not modify activin A secretion from M-Mϕ (Fig. 4B). These data agree with the decrease in the mRNA levels of the activin A-encoding gene (INHBA) (43) detected in rhGH-treated GM-Mϕ (Fig. 4C). Thus, our results suggest an active role of GH in downregulating activin A expression and secretion by GM-Mϕ, which would ultimately contribute to shape the transcriptional and functional profile of these macrophages.

The physiological effects of GH are partially mediated by IGF-1 produced in the liver (44). IGF-1 is nonetheless produced by several cell types (45), including M-Mϕ (46). We thus evaluated whether GH-induced GM-Mϕ repolarization might be mediated by IGF-1. We found that treatment of fully differentiated proinflammatory GM-Mϕ with IGF-1 had no effect on their characteristic gene expression profile (Supplemental Fig. 3A, 3B), or on activin A secretion (Supplemental Fig. 3C), suggesting that the polarizing action of GH is independent of IGF-1 production.

Recent reports indicate that MAFB plays an essential role in human macrophage anti-inflammatory polarization (29, 30). MAFB is known to constrain the ability of M-CSF to instruct myeloid cell proliferation, promotes macrophage differentiation (47), and controls the acquisition of an anti-inflammatory transcriptional profile of M-Mϕ (29). We thus performed GSEA on the GH-treated GM-Mϕ RNAseq data and observed a significant positive enrichment of MAFB-regulated genes (Fig. 5A). Indeed, Western blot analysis showed that GH treatment increased the level of MAFB protein in GM-Mϕ, whereas no changes were observed in the already elevated expression of MAFB in M-Mϕ (29) (Fig. 5B, left). In line with these results, and with the previous demonstration that MAFB protein stability is controlled by GSK3β-mediated phosphorylation and subsequent proteasomal degradation (48), we found a significant increase in the inactivating serine 9 phosphorylation of GSK3β (49) in rhGH-treated GM-Mϕ but not in rhGH-treated M-Mϕ (Fig. 5B, right). The relevance of these results was supported by the increase in the expression of MAFB-dependent genes (IL-10, CCL2) and the decreased levels of the proinflammatory gene CLEC5A, whose expression is inhibited by MAFB (29), in GH-treated GM-Mϕ (Fig. 5C). These observations indicate that GH modifies the polarization state of GM-Mϕ likely through the inhibition of GSK3β, which would block MAFB degradation and result in the upregulation of MAFB-dependent genes and the subsequent acquisition of an anti-inflammatory profile.

Kinetic analysis indicated that MAFB expression in GM-Mϕ was evident ∼1 h after exposure to GH and that GH-induced serine 9 phosphorylation of GSK3β remained elevated 4 and 24 h after treatment (48, 49) (Fig. 6A, Supplemental Fig. 4). To dissect the molecular mechanisms underlying these events, and based on previous data on the regulation of GSK3β activity (50, 51), we tested the initial signaling events triggered by GH in GM-Mϕ. We detected GH-mediated phosphorylation of ERK1/2 and AKT after 30 min and 1 h, respectively (Fig. 6A, Supplemental Fig. 4). Of note, blocking PI3K activity in GM-Mϕ with LY294002 (52) inhibited the GH-mediated phosphorylation of GSK3β and the increase in MAFB expression (Fig. 6A) and activin A secretion (Fig. 6B). These data are in accord with a previous report describing a role for AKT in GSK3β inactivation (49). Correspondingly, LY294002 pretreatment of GM-Mϕ abrogated the GH-mediated upregulation of CCL2, FOLR2, IGF1, and IL-10 and the downregulation of CLEC5A and IHNBA (Fig. 6C). Overall, these results demonstrate the involvement of a PI3K-GSK3β-MAFB axis in the GH-mediated polarization of GM-Mϕ.

We next examined whether the phenotypic, transcriptional, and signaling effects of GH on GM-Mϕ influenced the effector functions normally mediated by this macrophage subtype. As innate immune cells, macrophages efficiently phagocytose pathogens and stress-inducing agents (53), with M-Mϕ displaying a higher phagocytic activity than GM- Mϕ (22). We thus tested whether GH treatment altered the phagocytic properties of GM-Mϕ by coculturing untreated and rhGH-treated GM-Mϕ or M-Mϕ with Listeria-GFP for 30 min, and determining bacterial uptake by confocal microscopy. Results showed that bacterial uptake was significantly higher in rhGH-treated GM-Mϕ than in untreated cells, whereas this effect was absent in M-Mϕ (Fig. 7A).

Macrophages are also Ag-presenting and T cell–activating cells (54), and GM-Mϕ display a considerably higher T cell–activating capacity than M-Mϕ (55). To determine whether GH modifies the T cell–activating ability of GM-Mϕ, we cocultured untreated and rhGH-treated GM-Mϕ with allogeneic T cells for 10 d and examined T cell proliferation by evaluation of dye dilution using flow cytometry (Fig. 7B). Results showed that rhGH-treated GM-Mϕ promoted a significantly lower proliferation of allogenic T cells than untreated GM-Mϕ (Fig. 7C). These results indicate that GH-mediated repolarization of GM-Mϕ not only affects phenotypic and transcriptional parameters, but also negatively affects effector functions that characterize proinflammatory GM-Mϕ (T cell activation) while favoring the acquisition of functions preferentially associated with anti-inflammatory M-Mϕ (bacterial capture and uptake).

To investigate the pathophysiological relevance of the GH-induced MAFB-mediated switch in macrophage polarization, we evaluated the polarization state of colon-infiltrating recovery-phase macrophages from mice subjected to DSS-induced colitis. In agreement with immunohistochemistry results, analysis of colon samples from DSS + 7 mice revealed a higher expression of Arg1 and significantly diminished levels of Nos2 in GHTg mice (Fig. 8A). We then isolated the infiltrated F4/80⁺ cells at DSS + 7 d (Fig. 8B) and observed a significant increase of Arg1, Cd206, and Ym1, three specific markers of anti-inflammatory macrophages (56, 57), in cells from GHTg mice compared with control littermates, and a reduced expression of Nos2, Tnf, and Inhba (Fig. 8C). These data corroborate the switch in macrophage polarization toward an anti-inflammatory phenotype observed in GHTg mice. Accordingly, although a direct effect of GH on other immune cells cannot be discarded, the presence of macrophages with a more reparative and less proinflammatory phenotype correlates with the improved recovery from DSS-induced colitis in GHTg mice. Overall, these results support the in vivo relevance of the macrophage-repolarizing ability of GH, and further illustrate the capacity of GH to drive macrophage polarization to a more anti-inflammatory/reparative phenotype.

GH modulates immune function through direct effects on immune cells that express the GHR. For example, in macrophage-specific GHR-knockout mice, the loss of GH action exacerbates inflammation in adipose tissue (58). We previously showed that transgenic expression of GH ameliorates inflammation in diabetic NOD mice (16) and reduces the incidence of arthritis in a collagen-induced murine model (17). Similarly, GH attenuates the inflammatory burden in inflammatory bowel disease (59) and reduces NF-κB activation in colitis (18). A preliminary study in humans revealed that GH might be a beneficial treatment for Crohn’s disease (20). Although some of these observations have been associated with GH resistance processes, they might also be related to a potential role of GH modulating the immune system and favoring anti-inflammatory responses.

The phenotypic and functional heterogeneity of macrophages is essential to create the appropriate extracellular environment for immune responses to occur. Macrophages are highly plastic and can adopt proinflammatory or anti-inflammatory/resolving properties depending on microenvironmental signals, and the deregulation of this balance leads to chronic inflammatory diseases (60). Modulation of macrophage polarization is thus a potential target for therapeutic intervention in these chronic pathologies (61). The results of the current study reveal that GH does not promote monocyte polarization, but repolarizes the GM-Mϕ phenotype to an anti-inflammatory phenotype. Consistent with this, GSEA of RNAseq data demonstrated a significant enrichment of M-Mϕ–specific genes in rhGH-treated GM-Mϕ, some of them also confirmed by RT-qPCR analysis (HMOX1, IL-10, and IGF1), whereas the expression of others associated with GM-Mϕ were reduced (INHBA, MMP12, and EGLN3). The reprogramming effect of GH on human macrophages was also seen in murine GM-Mϕ, whose treatment with GH resulted in upregulation of M-Mϕ–associated anti-inflammatory genes (Arg1, Cd206, and Ym1) and reduction of GM-Mϕ-specific inflammatory genes (Nos2, Inhba, and Tnf). Functional in vitro analysis showed that GH reduces TNF-α release by GM-Mϕ and concomitantly increases their ability to secrete IL-10. Likewise, we also detected CCL2 secretion by rhGH-treated GM-Mϕ. By contrast, GH failed to alter TNF-α, IL-10, or CCL2 production in M-Mϕ. Further studies are needed to determine whether these findings are a consequence of a partial effect of GH on M-Mϕ, a dosage-mediated effect, or other unknown cofactors required to complete the repolarization process.

Initially characterized as inducers of follicle-stimulating hormone production, activins are now recognized to regulate the growth of numerous cell types, for example, by contributing to the maintenance of pluripotency in embryonic stem cells, and exerting antitumorigenic effects (41). Moreover, activin A is a crucial modulator of inflammatory responses (42), and its expression is upregulated in response to inflammatory mediators (62, 63). Activin A, which is encoded by the INHBA gene, induces arginase A expression in murine peritoneal macrophages and is important for the differential gene expression profiles and effector functions exhibited by proinflammatory macrophages (28). We observed that, in vitro, rhGH treatment of GM-Mϕ reduces the expression of INHBA and activin A, pointing to a possible role for GH in triggering GM-Mϕ repolarization.

We also found that GH triggers MAFB upregulation via a PI3K-dependent pathway. MAFB is a member of the MAF family of transcription factors and is involved in many cellular functions, such as control of lymphangiogenesis (64), or differentiation of pancreatic α and β cells (65, 66). Within the murine myeloid lineage, MAFB restricts the ability of M-CSF to trigger myeloid cell proliferation (67)^, and also critically determines the acquisition of the anti-inflammatory transcriptional and functional profiles in human macrophages (29, 30). Simultaneously, it negatively regulates genes associated with proinflammatory polarization (29, 30). Our GSEA of RNAseq data indicated a significant enrichment of MAFB-specific genes in rhGH-treated GM-Mϕ. In the current study, GH reduced activin A secretion and promoted MAFB upregulation, and thus supporting the role of GH in GM-Mϕ reprogramming. In accord with these results, activin A was shown to suppress MAFB expression in a pancreatic islet cell line (68). MAFB stabilization is controlled by GSK3β (48), an inhibitory serine/threonine kinase that is inactivated by phosphorylation at serine-9 (49). We found that these GH-mediated processes were blocked in cells pretreated with LY294002, a classical PI3K inhibitor, suggesting a role for p-AKT in this signaling event. Activated AKT has been associated with phosphorylation of GSK3β and inhibition of its activity (69), indicating that GSK3β is a substrate for AKT, as was described several years ago (70). Nonetheless, the effect triggered by GH was not completely abolished by LY294002 treatment. Although several mechanisms might be behind this effect, our results indicate that GH also triggered ERK1/2 activation, a pathway that has also been implicated in GSK3β inactivation (71, 72).

The polarization state of macrophages not only conditions the secreted cytokine profile, but also affects other cellular processes such as phagocytosis, production of reactive oxygen species (22), and capacity to trigger allogenic T cell proliferation. As expected, bacterial capture ability was higher in M-Mϕ than in GM-Mϕ (22), and we found that GH treatment increased the bacterial capture and microbial killing properties of GM-Mϕ. In addition, GH treatment significantly limited the ability of GM-Mϕ to promote allogeneic T cell proliferation. All of these data confirm the role of this hormone in repolarizing GM-Mϕ.

Available evidence supports the use of GH to treat inflammatory bowel diseases, including Crohn disease and ulcerative colitis (73). Although in most of the cases the observed effects have been associated with metabolic changes triggered by GH (3), a role for GH in enhancing survival, remission of inflammation, and mucosal repair during recovery has been linked to a local decrease of IL-1β production in a murine model of DSS-induced colitis in GHTg mice (74). This in vivo data suggest that GH reduces inflammation by altering the cytokine profile secreted by immune cells. We thus analyzed the in vivo relevance of GH-mediated GM-Mϕ reprogramming by comparing the phenotype of macrophages infiltrating the intestinal tissue of GHTg and WT littermates after DSS-induced colitis. As reported (74), GH did not influence the susceptibility of mice to acute DSS-induced colitis, but instead improved the remission of inflammation and mucosal repair during recovery. Importantly, GHTg mice presented a high proportion of infiltrating M-Mϕ (arginase 1⁺) when compared with WT littermates, which showed GM-Mϕ (iNOS⁺) infiltration. Flow cyometry analysis of these isolated macrophages confirmed the data. In GHTg mice, we found higher numbers of anti-inflammatory macrophages (CD45⁺F4/80⁺Gr1⁺CD11b⁺CD86^high) than in controls. In addition, the infiltrating F4/80⁺ cells in GHTg mice showed higher expression of Arg1, Cd206, and Ym1, three genes whose expression marks murine anti-inflammatory macrophages (56, 57), whereas the expression of inflammatory genes (Nos2, Inhba, and Tnf) was reduced.

It has been previously described that IGF-1 secreted by satellite cells, myofibers, fibroblasts, endothelial cells, and inflammatory cells, plays an important role in muscle regeneration (75), with the authors of this study proposing an autocrine role for IGF-1 in modulating murine macrophage polarization to an anti-inflammatory phenotype. This observation correlates with in vitro analyses showing that both murine and human macrophages express significant quantities of IGF-1 propeptides, which are upregulated by stimulation with so-called M2-polarizing stimuli, including IL-4, IL-13, or M-CSF (76, 77). Our in vitro studies clearly showed an increase of IGF-1 production by GM-Mϕ treated with GH, but we excluded an effect of IGF-1 on GM-Mϕ reprogramming, per se. It is nonetheless known that, in vivo, IGF-1 influences the magnitude of tissue inflammation by redirecting epithelial cells to a phagocytic phenotype (78).

In conclusion, our findings provide strong support for a role of GH in reprograming inflammatory macrophages to an anti-inflammatory phenotype, which contributes to dampening the inflammatory microenvironment where the immune response takes place.

The authors have no financial conflicts of interest.