Alveolar macrophages of GM-CSF knockout mice exhibit mixed M1 and M2 phenotypes
© Dalrymple et al.; licensee BioMed Central Ltd. 2013
Received: 11 April 2013
Accepted: 29 August 2013
Published: 17 September 2013
Skip to main content
© Dalrymple et al.; licensee BioMed Central Ltd. 2013
Received: 11 April 2013
Accepted: 29 August 2013
Published: 17 September 2013
Activin A is a pleiotrophic regulatory cytokine, the ablation of which is neonatal lethal. Healthy human alveolar macrophages (AMs) constitutively express activin A, but AMs of patients with pulmonary alveolar proteinosis (PAP) are deficient in activin A. PAP is an autoimmune lung disease characterized by neutralizing autoantibodies to Granulocyte-Macrophage Colony Stimulating Factor (GM-CSF). Activin A can be stimulated, however, by GM-CSF treatment of AMs in vitro. To further explore pulmonary activin A regulation, we examined AMs in bronchoalveolar lavage (BAL) from wild-type C57BL/6 compared to GM-CSF knockout mice which exhibit a PAP-like histopathology. Both human PAP and mouse GM-CSF knockout AMs are deficient in the transcription factor, peroxisome proliferator activated receptor gamma (PPARγ).
In sharp contrast to human PAP, activin A mRNA was elevated in mouse GM-CSF knockout AMs, and activin A protein was increased in BAL fluid. Investigation of potential causative factors for activin A upregulation revealed intrinsic overexpression of IFNγ, a potent inducer of the M1 macrophage phenotype, in GM-CSF knockout BAL cells. IFNγ mRNA was not elevated in PAP BAL cells. In vitro studies confirmed that IFNγ stimulated activin A in wild-type AMs while antibody to IFNγ reduced activin A in GM-CSF knockout AMs. Both IFNγ and Activin A were also reduced in GM-CSF knockout mice in vivo after intratracheal instillation of lentivirus-PPARγ compared to control lentivirus vector. Examination of other M1 markers in GM-CSF knockout mice indicated intrinsic elevation of the IFNγ-regulated gene, inducible Nitrogen Oxide Synthetase (iNOS), CCL5, and interleukin (IL)-6 compared to wild-type. The M2 markers, IL-10 and CCL2 were also intrinsically elevated.
Data point to IFNγ as the primary upregulator of activin A in GM-CSF knockout mice which in addition, exhibit a unique mix of M1-M2 macrophage phenotypes.
Activin A, a pleiotrophic cytokine belonging to the transforming growth factor-beta (TGF-β) superfamily, is synthesized by many cell types throughout the body [1, 2]. The molecular structure is a disulphide-linked, homodimeric glycoprotein composed of two inhibin βA chains. Activin A was first recognized as an endocrine factor, but is now known to be essential to developmental and repair processes, and total ablation is neonatal lethal . Contrasting regulatory roles have been cited for Activin A in inflammation . Human monocytes synthesize activin A upon stimulation with classical M1 macrophage activation inducers such as GM-CSF, LPS, and IFNγ [5, 6]. Exposure of GM-CSF treated macrophages to anti-Activin A reduces M1 markers and enhances alternative M2 phenotype markers such as IL-10 . Activin A also inhibits monocyte production of IL-1β and enhances IL-1 receptor antagonist production . Interestingly, in severe asthma, activin A may be elevated in serum, and data from animal models suggests that activin A may suppress T helper 2 (Th2) mediated allergic responses . Collectively these observations suggest multifunctional roles for activin A in inflammatory processes.
Maintenance of lung homeostasis is a complex process dependent upon a network of interacting cells and cytokines. GM-CSF is required for alveolar macrophage (AM) function and pulmonary homeostasis . In genetically altered mice homozygous for a disrupted GM-CSF gene (GM-CSF knockout), hematopoiesis is normal but there is accumulation of excess lung surfactant [11, 12]. This surfactant pathology mirrors that of human PAP, an autoimmune disease characterized by high levels of autoantibody to GM-CSF [12–14]. Aerosolized GM-CSF resolves the pulmonary pathology of GM-CSF knockout mice, thus demonstrating that surfactant homeostasis can be influenced by local administration of GM-CSF to the respiratory tract .
Previously we reported that healthy human AMs synthesize activin A in response to GM-CSF but AMs of patients with PAP are deficient in activin A . In addition, PAP AMs are deficient in the nuclear transcription factor, Peroxisome Proliferator-activated Receptor, (PPARγ), a regulator of lipid and glucose metabolism that is restored by GM-CSF treatment . PPARγ has also been shown to be a negative regulator of inflammation [18, 19]. Interestingly, alveolar macrophages of GM-CSF knockout mice are also deficient in PPARγ . The role of activin A in the lung has not been established. Because of the phenotypic similarities between human PAP and the GM-CSF knockout mouse, this study was undertaken to investigate activin A regulation in the lung. Initially, it was hypothesized that activin A might be impaired in GM-CSF knockout mice based upon previous data from PAP studies .
Intrinsic factors that could potentially affect activin A levels were subsequently investigated in GM-CSF knockout mice. Macrophage colony-stimulating factor (M-CSF) has been reported to be upregulated in GM-CSF knockout mice . Examination of M-CSF in the current study, however, indicated no effect on activin A in vitro in either wild-type or GM-CSF knockout AMs (data not shown). Elevated IFNγ has been reported in lungs of GM-CSF knockout mice  therefore intrinsic levels of IFNγ were examined. IFNγ mRNA expression was significantly (p < 0.005) elevated in GM-CSF knockout BAL cells compared to wild-type controls (Figure 1C). Immunocytochemistry of GM-CSF knockout BAL cells confirmed mRNA results and indicated markedly increased expression of intracellular IFNγ protein compared to wild-type (Figure 1D).
Because activin A is intrinsically elevated in PPARγ deficient GM-CSF knockout mice but severely decreased in PPARγ deficient human PAP patients , it appeared unlikely that PPARγ would exert a direct effect on activin A. Observations made elsewhere  also found no evidence of a PPARγ effect on activin A. We have shown, however, that IFNγ is elevated in macrophage-specific PPARγ knockout mice and significantly reduced after in vivo restoration of PPARγ via a lentivirus vector . We utilized this approach to determine whether PPARγ restoration in GM-CSF knockout mice might reduce IFNγ and thereby reduce activin A. Results supported this action. Ten days post intratracheal inoculation of lentivirus reagents into GM-CSF knockout mice, BAL cell mRNA expression of both IFNγ and activin A was significantly reduced in animals receiving lentivirus-PPARγ compared to controls receiving lentivirus-eGFP (p < 0.05) (Figure 3C).
Summary: comparison of macrophage activation regulatory factors in human pulmonary alveolar proteinosis (PAP) patients and GM-CSF knockout mice
Associated macrophage phenotype
GM-CSF knockout mice
Deficient protein, not mRNA 
mRNA - not elevated (comparable to healthy controls)
Undetectable in human alveolar macrophages (unpublished observation)
Elevated IFNγ has been reported previously in the BAL fluids of GM-CSF knockout mice . Our previous studies also found elevated IFNγ expression in macrophage-specific PPARγ knockout mice . Restoration of PPARγ via lentivirus vector in these mice greatly diminished IFNγ expression . In the current study, similar results were seen after PPARγ–lentivirus treatment of GM-CSF knockout mice. Such findings suggest that the PPARγ deficiency present in GM-CSF knockout mice may contribute to elevated IFNγ. GM-CSF has been shown to be a critical upregulator of PPARγ [31, 34]. The total lack of GM-CSF in knockout mice may maintain an extreme PPARγ deficiency which is ineffective at repressing inflammatory mediators such as IFNγ. In human PAP, IFNγ levels are not increased despite PPARγ deficiency, furthermore, GM-CSF is not totally absent . The primary etiology of PAP is considered to be an autoimmune response to GM-CSF in the form of high levels of circulating, neutralizing autoantibody to GM-CSF . It is also possible that additional regulatory mechanisms are present in human lung to help prevent IFNγ buildup in PAP.
The varying characteristics of activated macrophages have led to attempts to categorize activation phenotypes [35–39]. The M1 phenotype is characterized by production of microbial or IFNγ-triggered molecules such as iNOS and IL-12. GM-CSF has been cited as an inducer of M1 phenotypes while M-CSF has been shown to induce the M2 alternative activation phenotype in which IL-10 or TGFβ may be produced [7, 40]. We have shown that M-CSF is elevated in GM-CSF knockout mice  and in human PAP  which might suggest the presence of an M2 macrophage phenotype (see Table 1). Interestingly, PPARγ, which is deficient in GM-CSF knockout mice, is also a major driver of the M2 phenotype . It has been pointed out however, that macrophage phenotypes were defined by carefully controlled in vitro conditions which may be vastly different from the in vivo milieu . Thus the juxtaposition of both IFNγ and M-CSF in the lungs of GM-CSF knockout mice could produce the novel combination of macrophage activation phenotypes illustrated by elevated M1 (iNOS, CCL5, IL-6) and M2 (IL-10, CCL2) markers (Table 1). Other IFNγ-inducible pro-inflammatory mediators (chemokines CXCL9, CXCL10, and CXCL11) have been noted in the lungs of GM-CSF knockout mice . Previously, we found that MMP-2, a matrix metalloproteinase associated with M-CSF and alternative M2 activation, is also elevated in GM-CSF knockout BAL cells .
The current findings extend our previous studies examining pulmonary mechanisms operative in human PAP and the GM-CSF knockout mouse. It is clear that pathways of activin A regulation may utilize GM-CSF or IFNγ as stimulatory factors. In the GM-CSF knockout mouse, lack of GM-CSF may restrict production of sufficient PPARγ to control inflammation. The persistent elevation of both M-CSF and IFNγ may influence AMs to express characteristics of both M1 and M2 phenotypes. The current data emphasize the plasticity of alveolar macrophages in assuming a unique activation phenotype when regulatory pathways become dysfunctional.
Animal studies were conducted in conformity with Public Health Service (PHS) Policy on humane care and use of laboratory animals and were approved by the institutional animal care committee. The GM-CSF knockout mice were generated by Dr. Glenn Dranoff and have been previously described . Controls consisted of C57BL/6 wild type mice obtained from Jackson Laboratory (Bar Harbor, ME). BAL cells and fluids were obtained from 8-12 week-old GM-CSF knockout mice and age and gender matched wild-type C57BL/6 controls as previously described . Briefly, cytospins of BAL cells were stained with a modified Wright-Giemsa stain for differentials. A minimum of 100 cells was scored for each lavage. Mean (± SEM) BAL cells from C57BL/6 mice were composed of 98 ± 1% macrophages and 2 ± 1% lymphocytes; GM-CSF knockout BAL cells were composed of 91 ± 2% macrophages and 5 ± 1% lymphocytes. For in vitro studies, BAL cells were plated at 150,000 cells/well in 48-well plates as previously described . Recombinant murine IFNγ was obtained from R&D Systems. Neutralizing anti-IFNγ and control antibodies were purchased from BD Biosciences. For all experiments a minimum of 3 sets of pooled BAL cells from 3-5 mice were used except where indicated.
The protocol was approved by the East Carolina University Institutional Review Board and written informed consent was obtained from all patients and control subjects. Healthy control subjects had no history of lung disease and were not on medication. PAP subjects were recruited from patients undergoing routine clinical evaluation. The diagnosis of idiopathic PAP was confirmed by histopathological examination of material from open lung or transbronchial biopsies as previously described . Alveolar macrophages were derived from bronchoalveolar lavage (BAL) obtained by fiberoptic bronchoscopy as previously described . Differential cell counts were obtained from cytospins stained with a modified Wright’s stain. For PAP patients, the mean BAL cell percentages (means ± SEMs) were: alveolar macrophages, 83 ± 9%, and lymphocytes, 10 ± 5%. Healthy control values were: alveolar macrophages, 93 ± 2% and lymphocytes, 7 ± 2%. For in vitro culture, BAL cells were plated into 24-well plates (300,000 alveolar macrophages per well) or chamber slides (60,000 cells/well) as previously described .
Total RNA was extracted from BAL cells or cultured alveolar macrophages and analyzed by Q-PCR as previously described . RNA specimens were analyzed in duplicate using primer-probe sets for activin A, IL-10, iNOS, CCL2, CCL5, IL-6, IFNγ and GAPDH as previously described . Data were normalized to GAPDH and expressed as fold change in mRNA expression compared to controls values as previously described .
The self-inactivating lentivirus expression vector used here has been described previously . Construction of the lentivirus-PPARγ (lenti-PPARγ) and control lentivirus construct has also been described in detail [20, 25]. Control consisted of a lentivirus vector expressing Enhanced Green Fluorescent Protein (eGFP) (lenti-EGFP). Animals received 50 ug of lentivirus vector in 50 μl PBS or PBS alone (sham) by intratracheal instillation. After 10 days, five animals per group were lavaged, BAL differential counts were obtained and RNA was extracted.
Activin A or follistatin proteins (pg/ml) in BAL fluids or conditioned media from cultured alveolar macrophages were quantified by ELISA according to the manufacturer’s instructions (Serotec, Raleigh, NC; R&D Systems, Minneapolis, MN).
Immunocytochemistry for IFNγ was carried out on cytospin samples from freshly isolated BAL cells using rat anti-mouse IFNγ (Santa Cruz Biotechnology,1:100) followed by goat anti-rat IgG (Invitrogen) as described . Slides were counter-stained with DAPI (Invitrogen) to allow nuclear localization.
Data were analyzed by student’s t-test using Prism software (GraphPad). Values from treated cells were compared to untreated. Significance was defined as p ≤ 0.05.
Granulocyte macrophage colony stimulating factor
Pulmonary alveolar proteinosis
Peroxisome proliferator activated receptor
Inducible nitric oxide synthetase.
This work was funded by North Carolina Biotechnology Center FRG-1013 and NIH RO1 AI164153 to MJT.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.