Distinctive Effects of GM-CSF and M-CSF on Proliferation and Polarization of Two Major Pulmonary Macrophage Populations

Abstract

Granulocyte-macrophage colony-stimulating factor (GM-CSF) is required for alveolar macrophage (AM) development shortly after birth and for maintenance of AM functions throughout life, while macrophage CSF (M-CSF) is broadly important for macrophage differentiation and self-renewal. However, the comparative actions of GM-CSF and M-CSF on AMs are incompletely understood. Interstitial macrophages (IMs) comprise a second major pulmonary macrophage population. However, unlike AMs, IM responses to CSFs are largely unknown. Proliferation, phenotypic identity, and M1/M2 polarization are important attributes of all macrophage populations, and in this study, we compared their modulation by GM-CSF and M-CSF in murine primary AMs and IMs. CSFs increased the proliferation capacity and upregulated anti-apoptotic gene expression in AMs, but not IMs. GM-CSF, but not M-CSF, reinforced the cellular identity, as identified by surface markers, of both cell types. GM-CSF, but not M-CSF, increased the expression of both M1 and M2 markers exclusively in AMs. Finally, CSFs enhanced the IFN-γ- and IL-4-induced polarization ability of AMs, but not IMs. These first data comparing effects on the two pulmonary macrophage populations in general demonstrate that the activating actions of GM-CSF and M-CSF on primary AMs are not conserved in primary IMs.

Introduction

Granulocyte-macrophage colony-stimulating factor (GM-CSF) and macrophage CSF (M-CSF) were originally defined as growth factors generating hematopoietic lineages from the bone marrow, including monocyte-derived macrophages that are essential for tissue maintenance and immunity . However, normal maintenance of certain tissue macrophage populations, including alveolar macrophages (AMs), is now recognized to be largely independent of recruited monocytes (1–3), shifting focus instead to understanding the actions of CSFs on resident tissue macrophages.

Unlike other tissue macrophages, AMs require GM-CSF for their development shortly after birth. Moreover, mice deficient in this cytokine or its functional receptor show impaired capacity to clear pulmonary surfactant, manifested as pulmonary alveolar proteinosis. GM-CSF is also necessary for the phagocytic ability of AMs and for intact immune responses against bacterial and fungal infections (4–6). Lack of functional M-CSF has been shown to result in very low AM numbers during postnatal development, and as adults these mice developed spontaneous emphysema associated with enhanced secretion of matrix metalloproteinases and derangement of elastin fiber deposition (7). Thus, both GM-CSF and M-CSF play an important role in AM differentiation and proliferation (2, 8, 9). At baseline, levels of GM-CSF and M-CSF are low in the lung but increase during inflammation (10). It is thus possible that changes in pulmonary levels of these CSFs during an inflammatory event could affect the functions of AMs and possibly other pulmonary resident and recruited macrophages.

Of the various subsets of pulmonary macrophages described, two populations predominate in the healthy lung and these are distinguished by their tissue localization and their expression of surface markers (11). The aforementioned AMs are located within the alveoli and airways; their localization makes them easy to obtain by lavage for study, and highlights their potential to serve as important sentinels of the respiratory mucosal surface (12). Indeed, while they can rapidly initiate antimicrobial responses, they also have the capacity to restrain immune and inflammatory processes (13, 14). AMs express high levels of sialic acid binding Ig-like lectin F (Siglec-F) and integrin alpha X (CD11c) (15). In contrast, interstitial macrophages (IMs) reside in the interstitial space of the alveolar walls and peribronchial tissue. Their isolation therefore requires digestion of lung tissue (16, 17). In addition, their function is less apparent, although their anatomic localization positions them to interact with dendritic cells as well as interstitial mesenchymal cells. IMs are characterized by high expression of integrin alpha M (CD11b) and Cx3c chemokine receptor 1 (Cx3cr1) (11, 16, 18, 19). AMs are known to be long-lived cells derived from embryonic precursors that replenish by local proliferation (2, 3). However, it is still unclear whether IMs are likewise of embryonic origin, derived from circulating monocytes, or both, and whether they too are self-renewing. Under certain experimental conditions, it has been shown that AMs can be replaced by IMs (17, 20).

Additionally, the functions of macrophages can be further shaped by signals present in their local environment. For instance, exposure to pro-inflammatory cytokines such as interferon-gamma (IFN-γ) increases the microbicidal activity (sometimes known as the M1 phenotype) of macrophages (21, 22). By contrast, interleukin (IL)-4 treatment confers a macrophage phenotype (sometimes termed M2 phenotype) associated with a wide range of physiologic and pathologic processes including tissue repair, parasite killing, and potentiation of allergic airway diseases (23) .

While CSFs have been shown to influence AM proliferation and polarization, it is still unclear whether the actions of GM-CSF and M-CSF are distinct or redundant. Moreover, little is known about the effects of these CSFs on IMs. Therefore, our current study focused on understanding the influence of these two distinct CSFs on proliferation, cellular identity, and polarization in autologous murine AMs and IMs.

Materials and Methods

Animals

Specific pathogen-free male C57BL/6 mice aged 6–8 weeks were purchased from The Jackson Laboratory. The mice were housed in groups of 5 and they had ad libitum access to water and food. All methods were carried out in accordance with relevant national and local guidelines and regulations regarding the use of experimental animals and with approval of the University of Michigan Committee for the Use and Care of Animals.

Isolation of AMs and peritoneal macrophages (PMs)

Mice were sacrificed and lung lavage was performed as described previously (24) by gently flushing the lungs with 800-μl aliquots of ice cold PBS through a cannula inserted in the trachea until a return volume of 3 mL was recovered. PMs were obtained as described (24, 25) by performing peritoneal cavity lavage using 10 mL of ice cold PBS. Lung cells and peritoneal cells were pelleted by centrifuging for 10 min at 300 x g (4°C), resuspended in RPMI 1640 medium (Gibco), and cultured at a density of 0.5 × 10⁶ cells/mL. After overnight culture of lung and peritoneal cells, loosely adherent cells were washed off. Macrophage purity of both adherent populations has previously been reported >90% (14, 24); flow cytometric analysis for expression of AM markers (Siglec-F, F4/80, CD11c and CD11b) is shown in Figure 1 and expression of PM markers (F4/80, CD11c and CD11b) is shown in Supplemental Figure 1.

Figure 1.

(A) AM and IM isolation strategy. (B) IM and (C) AM phenotypic characterization for expression of F4/80 (general macrophage marker), Siglec-F, CD11c and CD11b by flow cytometry. (D) Basal expression of CD11c, (E) basal expression of Siglec-F, (F) basal expression of CD11b, and (G) basal expression of Cx3cr1 in AMs and IMs relative to basal expression in AMs. **p<0.01 and ****p<0.0001 using Student’s t test.

Isolation of IMs from lung digests

After lung lavage, the lungs were removed and IMs were isolated according to previously published methods, with modifications (16, 26, 27). Briefly, lungs were transferred into petri dishes containing RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 0.7 mg/mL collagenase A (Roche) and 10 μg/mL DNAse (Roche). Lungs were minced and incubated for 45 min at 37°C on a shaker. After digestion, the cell aggregates were dispersed by repeated passage through a syringe and filtered through a 40 μm cell strainer (BD Biosciences) to obtain single cell suspensions. A 2-min incubation with 0.8% ammonium chloride lysis buffer (1 mL per lung) was performed to lyse contaminating erythrocytes. Lysis was stopped by addition of excess RPMI 1640 medium supplemented with 10% FBS. After centrifugation (300 x g, 10 min, 4°C), lung cells were resuspended in PBS containing 2 mM EDTA and 0.5% FBS and counted. Lung cells were then stained with magnetic bead-conjugated anti-CD11b antibodies (Miltenyi Biotec) followed by magnetic separation according to manufacturer’s protocol. CD11b-positive lung cells were collected, counted, and cultured at a density of 0.5 × 10⁶ cells/mL. Overnight culture allowed macrophages to adhere while loosely adherent cells were washed off. IM purity of adherent cells has previously been reported to be at least 90% (26).

Flow cytometry

After overnight culture, AMs, IMs, and PMs were harvested using trypsin (Gibco-Invitrogen) and kept at 4°C during all further steps. Cells were pelleted by centrifuging for 10 min at 300 x g, and resuspended in PBS containing 2% FBS, to block for 1 h. Next, DAPI (for dead cell exclusion) and primary antibodies: anti-mouse F4/80-Alexa Fluor 700 (diluted 1:100; Biolegend), anti-mouse Siglec-F-PE (1:100; Biolegend), anti-mouse CD11b-FITC (1:100; Biolegend), anti-mouse Cx3cr1-APC (1:100; Biolegend) and anti-mouse CD11c-Brilliant Violet 510 (1:100; Biolegend) were added to cells and incubated for 1 h in the dark. After staining, cells were washed twice, and kept in 2% FBS-containing PBS in the dark until flow cytometric analysis. Fluorescent staining was measured on an LSRFortessa cell analyzer (BD Biosciences), and data were analyzed using FlowJo v10 software (Tree Star, Inc.).

Stimulation of macrophages with growth factors and cytokines

Fresh RPMI 1640 (0% FBS) in the absence or presence of GM-CSF or M-CSF (both 50 ng/mL, Peprotech) was added to adherent macrophage cultures for 48 h to assess effects on gene expression and for 5 days to assess effects on proliferation. For priming and polarization experiments, cells were pretreated with GM-CSF, M-CSF, or medium alone for 24 h, after which fresh medium or medium containing IFN-γ or IL-4 (both 10 ng/mL, Peprotech) was added for another 24 h to effect polarization.

Fluorescent microscopy

Isolated AMs were seeded on a 4-chamber slide (Nunc) at a density of 30,000 cells per well. After overnight culture loosely adherent cells were washed off and AMs were stimulated with GM-CSF or M-CSF (both 50 ng/mL) for 48 h. To visualize morphologic changes, cells were stained with 0.1 μM CFSE dye (Sigma) for 8 min and then washed with PBS. AMs were then mounted in SlowFade Gold Antifade reagent with DAPI (Molecular Probes). Images were acquired on a Nikon Eclipse E600 microscope. To visualize CSFE staining and DAPI-labeled nuclei, cells were excited at 488 nm and 360 nm, respectively. Representative images of CSFE and DAPI staining were acquired at 20x magnification and merged using NIS-Elements Software.

Proliferation assay

Lavaged cells from lung or peritoneum or CD11b-positive lung cells were seeded in 96-well plates with black bottom at a density of 15,000 cells per well with 200 μl RPMI 1640 medium containing 2% FBS. On the next day, loosely adherent cells were washed off and fresh medium was added containing either GM-CSF or M-CSF (both 50 ng/mL). After 5 days of CSF stimulation, culture medium was removed and 100 μl/well of fluorescent DNA-binding dye (CyQUANT NF Cell Proliferation assay kit, ThermoFischer Scientific) was added and incubated for 45 min at 37°C. Fluorescence measurements were made using a microplate reader with excitation at 485 nm and emission at 530 nm. The average fluorescence of five replicate wells was calculated for each condition per experiment. Proliferation in each treatment condition was expressed relative to fluorescence measurements of unstimulated wells that were set at 100 %.

RNA isolation and quantitative RT-PCR

AMs and IMs were suspended in 700 μl TRIzol reagent (ThermoFischer Scientific) and RNA was extracted using the RNeasy Micro Kit (Qiagen) according to manufacturer’s instructions and converted to cDNA. Levels of mRNA were assessed by quantitative RT-PCR performed with a SYBR green kit (Applied Biosystems) on an ABI Prism 7300 thermocycler (Applied Biosystems). Expression of CD11c, CD11b, Siglec-F, Cx3Cr1, chitinase-like 3 (YM1), inducible nitric oxide synthase (iNOS), cyclin B1, cyclin D1, Polo-like kinase-1 (Plk-1), spleen focus forming virus proviral integration oncogene (PU.1), GATA binding protein 6 (GATA-6), baculoviral IAP repeat-containing 5 (Birc5 or survivin), B cell lymphoma 2 (Bcl-2), apoptotic peptidase activating factor 1 (Apaf1), Fas, IL-12, interferon regulatory factor (IRF) 4, IRF5, and resistin like alpha (FIZZ1) was assessed (sequences of primers used can be found in Supplemental Table 1). Relative gene expression was determined by the ΔCT method, and β-actin was used as a reference gene.

ELISA

Macrophage culture supernatants were collected after 48 h stimulations. Cell- and apoptotic body-free supernatants were concentrated using 3 kD Amicon Ultra exclusion filters (MilliporeSigma). Levels of mouse YM1/Chitinase 3-like 3 and mouse IL-12p40 were determined in concentrated supernatants by DuoSet ELISA (R&D Systems) following the instructions of the manufacturer.

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM). To determine normality of the data a D’Agostino & Pearson omnibus normality test was used. Data were log transformed to fit a normal distribution when not normally distributed. Differences between groups were tested using a one-way ANOVA followed by Sidak’s test for multiple comparisons, or by paired Student t test, as appropriate, using Prism 7.0 (GraphPad Software). p-values below 0.05 were considered to be statistically significant.

Results

Phenotypic characterization of resident pulmonary and peritoneal macrophages.

The strategy employed for isolation of AMs and IMs, based on published protocols (16, 24, 26, 27), is depicted in Figure 1A and is briefly described in Materials and Methods. As expected, AMs showed high baseline expression of cell surface markers Siglec-F and CD11c (Figure 1C-E) and IMs showed high baseline expression of CD11b (Figure 1B and F) (16). AMs expressed low CD11b (Figure 1C and F) and IMs expressed no Siglec-F (Figure 1E). Consistent with the literature (16), expression of Cx3cr1 was observed in IMs but not in AMs (Figure 1B and G). Expression of CD11b and CD11c was also observed in PMs (consistent with previous reports (28)), but their level of CD11b expression was significantly higher than IMs and their CD11c expression was significantly lower than AMs (Supplemental Figure 1A-B). PMs did not express Siglec-F or Cx3cr1 (Supplemental Figure 1A-B). Consistent with previous reports on IMs (17), we observed expression of TLR9 in IMs but not in AMs and PMs (Supplemental Figure 1C). We also determined the expression of GATA6, a transcription factor implicated in the functional development of PMs but which is known to be minimally expressed in AMs (29). As expected, GATA6 expression was very high in PMs and absent in AMs; expression was modest but significant in IMs (Supplemental Figure 1D). A summary of relative gene expression patterns in AMs, IMs and PMs is presented in Table 1.

Table 1:

Summary of the relative expression of markers in AMs, IMs and PMs

Gene expression (mRNA)	AMs	IMs	PMs
CD11c	+++	−	−
Siglec-F	+++	−	−
CD11b	+	++	+++
Cx3cr1	−	+++	−
TLR9	−	+++	−
IRF5	++	+++	++
GATA6	−	+	+++
iNOS	−	+	+
YM1	+++	+	−

[− (no expression); +, ++ and +++ are arbitrary estimated expression levels]

Differential effects of CSFs on lung macrophage phenotype.

The relative expression profile of pulmonary macrophage markers (CD11b, CD11c, Siglec-F, TLR9 and Cx3cr1) is depicted in Figure 2A for AMs and Figure 2B for IMs. For the subsequent figures, we set the basal expression of each gene in AMs as 1 and assessed the changes in expression in IMs or PMs under various culture conditions relative to that value. Next, we assessed the effect of 48-h treatment with CSFs on cell surface phenotypic markers. GM-CSF stimulation increased the expression of CD11c and CD11b in both AMs and IMs (Figure 2C and 2E), but the enhancement of CD11c was stronger in AMs whereas that of CD11b was stronger in IMs; by contrast, M-CSF stimulation had no effect on these phenotypic markers. Neither CSF altered the expression of Siglec-F in AMs (Figure 2D). Interestingly, GM-CSF reduced, while M-CSF potentiated, the expression of Cx3cr1 in IMs (Figure 2F). GM-CSF also reduced the expression of TLR9 in IMs, while M-CSF did not affect it (Figure 2G). Consistent with previous reports (30, 31), we also observed changes in macrophage morphology in response to CSFs. As shown in Supplemental Figure 2A-C, some AMs acquired a spindle-like morphology in response to GM-CSF but a multinucleated giant cell appearance in response to M-CSF. Both IMs and PMs exhibited a mostly round morphology at baseline, and in response to CSFs, IMs changed their shape to spindle-like morphology. However, CSFs stimulation showed no effect on morphology of PMs (data not shown).

Figure 2.

(A-B) Basal expression of CD11c, Siglec-F, TLR9 and Cx3cr1 in (A) AMs relative to basal expression of CD11b in AMs, and (B) in IMs relative to basal expression of Siglec-F in IMs. #p<0.05 using a one-way ANOVA followed by Sidak’s multiple comparisons test. (C-F) Effects of 48 h-treatment with GM-CSF and M-CSF on expression of (C) CD11c, (D) Siglec-F, (E) CD11b, (F) Cx3cr1, and (G) TLR9 in AMs and IMs relative to basal expression in AMs. #p<0.05 compared to AM control and *p<0.05, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

AMs but not IMs or PMs proliferate in response to CSFs.

We next evaluated the mitogenic activity of both CSFs in all three macrophage populations. As shown in Supplemental Figure 2D, both CSFs increased the proliferation of AMs in a dose-dependent manner. A near maximal effect of CSFs on AM proliferation was achieved with a dose of 50 ng/mL (Figure 3A). Interestingly, GM-CSF-induced proliferation resulted in an increase in the numbers of both round and spindle-shaped macrophages that were described in Figure 2B-D. Unexpectedly, the concomitant addition of M-CSF impaired the proliferation ability of AMs observed in response to GM-CSF alone (Supplemental Figure 3E). By contrast, IMs failed to proliferate in response to either CSF (Figure 3B and Supplemental Figure 2E). A similar failure of proliferation in response to CSFs was noted for PMs (Supplemental Figure 2F).

Figure 3.

(A) Proliferation of AMs after 5 days of stimulation with either GM-CSF or M-CSF (as % of control). (B) Proliferation of 5-day GM-CSF- and M-CSF-stimulated IMs (as % of control). #p<0.05 compared to AM control and *p<0.05, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

CSFs upregulate the expression of cell cycle-related genes in AMs but not in IMs.

Since cell proliferation is associated with induction of genes involved in the cell cycle, we next assessed the expression of several such genes, including cyclin B1, cyclin D1 and Plk-1, in response to CSFs. In parallel with proliferative responses observed in Figure 3A, GM-CSF, and to a lesser degree M-CSF, induced the expression of cyclin B1, cyclin D1 and Plk-1 in AMs (Figure 4A-C). However, no such induction was observed in IMs or PMs (Figure 4A-C and Supplemental Figure 3A-B), again in parallel with the absence of proliferation. Of note, the basal expression levels of these cell cycle-related genes in IMs as well as in PMs were higher than in AMs, despite the absence of basal proliferation in these latter two populations. Unexpectedly, however, stimulation with CSFs reduced the basal expression of cell cycle-related genes in IMs and PMs (Figure 4A-C and Supplemental Figure 3A-B) – again directionally opposite to their actions in AMs. The transcription factor PU.1 promotes differentiation and proliferation of myeloid progenitors. In addition, it has also been implicated in certain GM-CSF-induced actions in AMs in vivo (4). However, we observed no changes in PU.1 expression after stimulation with GM-CSF or M-CSF in AMs or IMs (Figure 4D). Since GATA6 has also been implicated in proliferative renewal of tissue resident macrophages (32), we measured its expression in response to CSFs. Consistent with data shown in Supplemental Figure 1G, we observed no GATA6 expression in AMs and its expression was not altered by CSFs (Figure 4E). However, expression of GATA6 in IMs and PMs was reduced by CSFs (Figure 4E and Supplemental Figure 3C), as was also noted for cell cycle genes.

Figure 4.

(A) Expression of Cyclin B1, (B) Cyclin D1, (C) Plk-1, (D) PU.1, and (E) GATA6 in AMs and IMs stimulated for 48 h with either GM-CSF or M-CSF (relative to expression in AM control). #p<0.05 compared to AM control. *p<0.05 and ****p<0.0001, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

AMs but not IMs show increased expression of anti-apoptotic genes in response to CSFs.

To assess the influence of CSFs on determinants of macrophage survival, the expression of anti-apoptotic genes was determined in AMs and IMs. As shown in Figure 5A and 5B, GM-CSF and to a lesser extent M-CSF increased the expression of survivin and Bcl-2 in AMs. No significant change in expression of survivin or Bcl-2 was seen in IMs. We also measured the expression of pro-apoptotic genes Fas and Apaf-1, and as shown in Figure 5C and D, no effect of CSFs was observed on pro-apoptotic gene expression in AMs or IMs. CSFs also failed to up-regulate the expression of another key apoptotic gene, Caspase 3 (data not shown).

Figure 5.

(A) Expression of pro-apoptotic survivin, (B) pro-apoptotic bcl-2, (C) anti-apoptotic Fas, and (D) anti-apoptotic Apaf-1 in AMs and IMs stimulated for 48 h with either GM-CSF or M-CSF (relative to expression in AM control). #p<0.05 and ##p<0.001 compared to AM control.

GM-CSF reinforces baseline polarization status in AMs but not in IMs.

Since macrophage polarization plays an important role in a panoply of host responses, we assessed the ability of CSFs to initiate macrophage polarization as determined by expression of iNOS (a marker for M1 macrophages) and YM1 (a marker for M2 macrophages) in the three populations of macrophages. GM-CSF, but not M-CSF, increased the basal expression of both iNOS (Figure 6A) and YM1 (Figure 6B) in AMs. In comparison to AMs, both IMs and PMs had high levels of iNOS expression at baseline. However, iNOS expression in IMs, but not PMs, was reduced by both CSFs (Figure 6A and Supplemental Figure 3D). Compared to AMs, basal YM1 expression was not observed in either IMs or PMs and the expression was not altered by CSFs (Figure 6B and data not shown for PMs).

Figure 6.

(A) Expression of M1 marker iNOS, and (B) M2 marker YM1 in AMs and IMs stimulated for 48 h with either GM-CSF or M-CSF (relative to expression in AM control). #p<0.05 compared to AM control. *p<0.05, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

CSFs prime AMs but not IMs for cytokine-induced polarization.

Next, we assessed the ability of CSFs to prime macrophages for subsequent cytokine-induced polarization. Briefly, macrophages were pretreated with or without GM-CSF or M-CSF for 48 h followed by stimulation with IFN-γ (M1-polarizing cytokine) or IL-4 (M2-polarizing cytokine). As shown in Figure 7A, IFN-γ induced the expression of iNOS in both AMs and IMs. The increment in IFN-γ-induced iNOS expression was markedly increased when AMs were pre-treated with GM-CSF or M-CSF, but no such potentiation by CSF priming was observed in IMs. We extended this analysis to expression of IRF5, the transcription factor responsible for M1 polarization. As shown in Figure 7B, IFN-γ induced the expression of IRF5 in AMs, consistent with previous reports (33). GM-CSF pretreatment potentiated the increment in IRF5 expression in AMs in response to IFN-γ stimulation. Unlike GM-CSF, M-CSF failed to potentiate IRF5 expression. IFN-γ stimulation did not increase the expression of IRF5 in IMs and we found no potentiation by pre-treatment with CSFs (Figure 7B). A similar priming effect on IFN-γ-induced IL-12 expression in AMs was observed with GM-CSF, but not with M-CSF pre-treatment (Figure 7C). Although IFN-γ stimulation induced the expression of IL-12 in IMs, the fold increase was lower than AMs. Moreover, in IMs, CSFs failed to show any priming effect on IFN-γ-induced IL-12 expression (Figure 7C). We utilized ELISA to quantify IL-12 protein levels (Figure 7D); these data substantiate the ability of both GM-CSF and M-CSF to prime AMs for M1 polarization in response to IFN-γ stimulation.

Figure 7.

(A) Expression of M1 markers iNOS, (B) IRF5, and (C) IL12 in AMs and IMs pre-treated for 24 h with either GM-CSF or M-CSF followed by 24-h stimulation with IFN-γ (relative to expression in AM control). (D) Secretion of IL-12p40 by AMs and IMs pre-treated for 24 h with either GM-CSF or M-CSF followed by 24-h stimulation with IFN-γ. #p<0.05 compared to AM control. *p<0.05, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

We next evaluated the effect of priming with CSFs on IL-4-induced expression of M2 markers YM1, FIZZ1 and IRF4. As shown in Figure 8A, the induction of IL-4-induced expression of YM1 in AMs was reduced after pre-treatment with CSFs. In contrast, pre-treatment of IMs with CSFs increased the IL-4-induced expression of YM1. In AMs, IL-4-induced FIZZ1 expression was also modestly reduced by CSFs, but this did not reach statistical significance (Figure 8B). Whereas IL-4 stimulation failed to enhance FIZZ1 expression in IMs, it did so following M-CSF pre-treatment (Figure 8B). In parallel with the transcriptional role of IRF5 in M1 polarization, IRF4 is necessary for M2 polarization (34). However, the effects of GM-CSF priming on IRF4 expression were discrepant from those effects seen on YM1 (Figure 8C). We utilized ELISA to quantify protein levels of secreted YM1 (as shown in Figure 8D); these data further confirm the ability of both GM-CSF and M-CSF to prime IMs for M2 polarization in response to IL-4 stimulation. In AMs, however, YM1 protein levels were discrepant from those effects seen on YM1 expression as CSF pre-treatment did not reduce IL-4-stimulated secretion of YM1.

Figure 8.

(A) Expression of M2 markers YM1, (B) FIZZ1, and (C) IRF4 in AMs and IMs pre-treated for 24 h with either GM-CSF or M-CSF followed by 24-h stimulation with IL-4 (relative to expression in AM control). (D) Secretion of YM1 by AMs and IMs pre-treated for 24 h with either GM-CSF or M-CSF followed by 24-h stimulation with IL-4. #p<0.05 compared to AM control. *p<0.05 or p<0.1, using a one-way ANOVA followed by Sidak’s multiple comparisons test.

Overall, the enhanced iNOS expression and IL-12 production in response to IFN-γ, and the diminished YM1 expression (but not secretion) in response to IL-4 after priming AMs with CSFs suggest that both GM-CSF and M-CSF predispose AMs towards M1 polarization. On the other hand, unaltered iNOS expression and IL-12 production in response to IFN-γ, and higher YM1 secretion in response to IL-4 after priming IMs with CSFs suggests that both GM-CSF and M-CSF predispose IMs towards M2 polarization. A summary of the priming effects of both CSFs on M1 and M2 polarization markers in both pulmonary macrophage populations are summarized in Table 2.

Table 2:

Summary of the priming effects of both CSFs on M1 and M2 polarization markers in both pulmonary macrophage populations.

		AMs			IMs
Cytokine stimulation	Gene expression (mRNA)	No CSF (baseline)	GM- CSF	M-CSF	No CSF (baseline)	GM- CSF	M-CSF
IFN-γ	iNOS	++	++++	++++	+	+	+
	IL12	++	++++	+++	+	+	+
	IRF5	++	+++	++	+	+	+
IL-4	YM1	++++	++	++	+	++	++
	FIZZ1	+++	++	++	++	++	+++
	IRF4	+	++	+	+	+	+

[+, ++ , +++ and ++++ are arbitrary estimated expression levels]

Discussion

AMs and IMs represent two major subpopulations of macrophages in the lung. Although AMs have been studied extensively over the last six decades, the characteristics and functions of IMs are much less well studied and understood (35). This report sought to characterize the effects of GM-CSF and M-CSF on a variety of pertinent responses of resident murine AMs and IMs. Although both CSFs have been shown to promote proliferation of AMs (9, 36), the broader functional effects of M-CSF on AMs remain unexplored and neither CSF has been studied in IMs. Responses examined herein included changes in phenotypic marker expression, morphology, determinants of cell number including proliferation and apoptosis, and M1/M2 polarization. Our studies reveal a variety of actions of both CSFs on both macrophage populations, making it clear that both AMs and IMs express receptors and the capacity for functional signaling in response to both GM-CSF and M-CSF. The specific response patterns observed, however, varied by CSF and by macrophage population, and include responses that were redundant as well as distinct. An overall conclusion is that CSFs generally promoted the functional attributes examined in AMs, while either not affecting or actually suppressing them in IMs.

AMs and IMs exhibited the expected mutually exclusive expression profiles of their characteristic surface markers: AMs were CD11c and Siglec-F positive, while IMs were CD11b, Cx3cr1, and TLR9 positive. GM-CSF increased the basal expression of CD11c in AMs, consistent with a previous report of this occurring in PMs which had been adoptively transferred into the lungs, and reduced CD11c expression in AMs derived from GM-CSF knockout mice (37). However, such an intensification of basal Siglec-F expression by GM-CSF treatment of AMs was not observed. Likewise, while GM-CSF potentiated the basal expression of CD11b in IMs, a contradictory effect was seen on expression of Cx3cr1. Thus, GM-CSF appears to reinforce the already characteristic phenotypes of AMs and IMs, as defined by certain, but not all, markers. M-CSF did not exert such effects.

AMs are known to be extraordinarily long-lived in both mice and humans (38, 39), but the mechanisms for their survival are not fully understood at the molecular level. Previous studies have described anti-apoptotic effects of GM-CSF on AMs in mouse models of pulmonary pneumococcal (40) and influenza infections (41). Therefore, we assessed the effects of CSFs on apoptosis-related gene expression in AMs and IMs. Both GM-CSF and M-CSF increased expression of the anti-apoptotic genes survivin and Bcl-2 in AMs. We also sought to determine if CSFs directly inhibited apoptosis susceptibility of AMs, but we were unable to replicate reported apoptotic responses elicited by stimuli such as tumor necrosis factor-alpha and lipopolysaccharide (42). Thus, we can only speculate that induction of anti-apoptotic genes may explain the contribution of CSFs to the long-term survival of AMs in the lung (38).

Compared to AMs, knowledge of the physiologic and pathophysiologic roles of IMs is very limited and no prior reports on effects of CSFs on IMs are available. Therefore, we can only speculate about the functional implications of our IM data. Because of their high-level baseline expression of Cx3cr1, IMs but not AMs may migrate in response to the Cx3cr1 ligand, fractalkine (43). In view of the opposing actions of the two CSFs on expression of this chemokine receptor, GM-CSF would be expected to inhibit, while M-CSF would be expected to enhance, the fractalkine-induced migration of IMs. Likewise, the marked down-regulation by GM-CSF of TLR9 in IMs would be predicted to suppress macrophage activation and antigen presentation (44), These predictions will require dedicated functional studies.

We observed proliferative actions of both GM-CSF and M-CSF on AMs, consistent with published literature (8, 9, 36), and with the recent recognition of self-renewal among AMs (2). At the molecular level, we further identified that these mitogenic actions of CSFs were associated with upregulation of cell cycle genes. Such increases would be envisioned to cooperate with increases in the anti-apoptotic genes noted above to promote survival and self-renewal. Since CSFs co-exist in the lung, we asked whether co-stimulation of AMs with GM-CSF and M-CSF exhibits additive or synergistic effects on proliferation. To our surprise, co-stimulation of AMs resulted in a level of proliferation that was less than that observed with GM-CSF alone and not significantly different from the control level. This suggests the potential for complex effects of CSFs on lung macrophages. As myriad other mediators present in the lung might likewise influence macrophage proliferative responses and interact with the actions of CSFs themselves, the potential for complex actions is substantial. Future studies are required to understand the integrated proliferative and other biological responses of macrophages in the complex milieu of the lung. CSFs have also been reported to promote proliferation in monocyte-derived macrophages (45, 46). Unexpectedly, however, both CSFs failed to induce proliferation in either IMs or PMs. It is thus evident that the effects of CSFs on proliferation cannot be generalized amongst all populations of tissue-resident macrophages. The failure of IMs to exhibit proliferative responses suggests that, unlike AMs, this population is more dependent on replenishment by recruitment of monocytes during homeostasis, consistent with recent reports (16, 17, 20). A number of functional differences between IMs and AMs have been recognized previously (17, 35), and in many of these contexts, IMs have been noted to more closely resemble PMs than AMs (47); this was indeed the case for proliferative responses noted in our studies. GATA6 is a key transcription factor implicated in self-renewal of PMs (32). Basal levels of GATA6 expression were undetectable in AMs, and its expression remained unchanged with CSF stimulation. Furthermore, forced overexpression of this transcription factor in AMs failed to promote proliferation (data not shown), consistent with the conclusion that GATA6 plays no role in CSF-induced proliferation of AMs. By contrast, the basal expression of GATA6 in IMs was high and comparable to that in PMs; CSFs did not further increase the expression of this transcription factor, but rather, inhibited it. These findings may help to explain the lack of proliferation effects of CSFs on IMs and PMs.

GM-CSF has been reported to polarize a variety of macrophages to a M1 phenotype, and M-CSF to a M2 phenotype (48, 49). However, contradictory findings have also been reported (50, 51). Confusion may stem from the use of macrophages from a variety of sources, including primary resident macrophages, inflammatory macrophages, monocyte-derived macrophages, and various macrophage cell lines. To better understand the effect of CSFs on polarization of pulmonary macrophages, we evaluated them both alone and in priming studies. In response to GM-CSF and M-CSF alone, we observed that only GM-CSF was capable of polarizing AMs. In fact, it led to induction of both iNOS and YM1 expression, rather than a distinct phenotype. Strikingly, the morphology of AMs changed in response to GM-CSF from predominantly round to a mixture of both round and spindle-shape. This morphology change did not seem to affect their proliferation ability as we observed the numbers of both shapes to increase in response to GM-CSF. Conflicting reports are published about the implication of spindle-like macrophages and whether these are polarized towards M1 (48) or M2 (31, 52). In our studies, however, we observed both spindle-like and round shaped macrophages in response to GM-CSF with enhanced expression of both M1 and M2 markers. Moreover, macrophages stimulated with polarizing cytokines such as IFN-γ (for M1) and IL-4 (for M2) resulted in spindle-shaped (48) and round-shaped (53, 54) macrophages respectively. Therefore, we interpret the mixed morphology observed with GM-CSF stimulation as a priming state without a specific polarization direction. The results of the priming studies were complicated and are therefore summarized them in Table 2. Together, the findings suggest that AMs, which at baseline have more M2 phenotypic characteristics, acquire more M1 phenotypic features when polarized by IFN-γ stimulation in the presence of CSFs. On the other hand, IMs, which at baseline have more M1 phenotype characteristics, acquire more M2 phenotypic features when polarized by IL-4 under the influence of CSFs. This is relevant for lung inflammatory diseases, such as asthma, in which high numbers of both M1 and M2 pulmonary macrophages are found and thought to contribute to the disease (53, 55–57). Blocking of elevated CSF levels in these lung diseases may serve as a therapeutic option to reduce pulmonary macrophage polarization. Although we have not herein explored functions of IMs other than proliferation, one could speculate, for example, that their proximity to fibroblasts within the interstitial space and their robust capacity for M2 polarization (with secretion of mediators such as TGF-β1) could facilitate fibrotic responses. This possibility requires future investigation.

In conclusion, this study is the first to evaluate, both at baseline and in response to CSFs, IM proliferation and polarization. It is also the first to directly compare these parameters between autologous IMs and their better-studied counterparts, AMs. Overall, we have found differences in responses of each macrophage population to the two CSFs, as well as differences between these two resident macrophage populations of the lung. Although descriptive in nature, our data provide a foundation for future studies, which will be necessary in order to determine the implications of these distinctive basal properties and responses to individual CSFs in disease states and models.