Local GM-CSF–Dependent Differentiation and Activation of Pulmonary Dendritic Cells and Macrophages Protect against Progressive Cryptococcal Lung Infection in Mice

The outcome of primary pulmonary infection with the encapsulated fungus Cryptococcus neoformans is profoundly shaped by the virulence of the organism and the host’s immune status (reviewed in Ref. 1). In humans, the importance of these host–fungal interactions is underscored by the clinical burden of infection when host defenses are compromised. Worldwide, 1 million new cases of C. neoformans infections occur each year in HIV-infected patients, resulting in 680,000 deaths (2), and cryptococcal infection is the second most common fungal infection in organ transplant patients (1). Although uncommon, clinically significant infections have been reported in seemingly immunocompetent hosts (3–5). Recent studies by Rosen et al. (6) and Saijo et al. (7) suggest that a subset of these patients have previously unidentified immune defects. Specifically, these authors identified autoantibodies against GM-CSF, a cytokine with pleiotropic immune effects (reviewed in Refs. 8, 9), in the serum and cerebrospinal fluid of a subset of patients with invasive cryptococcal infection for whom a predisposing immunocompromising condition had not previously been identified. Autoantibodies against GM-CSF or mutations in the GM-CSF receptor can lead to pulmonary alveolar proteinosis, a lung disease characterized by disrupted surfactant catabolism in alveolar macrophages (Mϕ) (10, 11) (reviewed in Ref. 12). Thus, the studies by Rosen et al. (6) and Saijo et al. (7) confirm and extend previously published associations between patients with pulmonary alveolar proteinosis and increased susceptibility to cryptococcal and other fungal infections (13–15). Despite this strengthened association between GM-CSF deficiency in humans and susceptibility to C. neoformans infections, the cellular and molecular mechanisms through which GM-CSF protects against progressive cryptococcal infections remain uncertain.

We previously demonstrated that lung Mϕ obtained from GM-CSF–deficient (GM^−/−) rats are poorly fungicidal (16). In a later study, we performed a comparative analysis of cryptococcal lung infection in wild-type C57BL/6 mice capable of producing GM-CSF (GM^+/+ mice) and GM^−/− mice (C57BL/6 genetic background) and showed that progressive fungal infection occurs in the absence of GM-CSF (17). Murine models have also shown a protective role for GM-CSF in host defense against other fungal pathogens, including Pneumocystis jiroveci, Histoplasma capsulatum, and Aspergillus fumigatus (18–22). These studies implicate that lack of GM-CSF induces alterations in adaptive immunity and Mϕ function, yet provide limited information about the number, immunophenotype, or microanatomic location of T and myeloid cells in response to infection.

Our more recent studies demonstrated that GM-CSF promotes the in vitro differentiation of Ly-6C^high inflammatory monocytes into CD11b⁺ dendritic cells (DC) and exudate Mϕ, a subset of primarily recruited lung macrophages expressing CD11b, which distinguishes them from resident CD11b⁻ alveolar Mϕ (23–25). Additional data demonstrated that CD11b⁺ DC interact with newly recruited lung T cells to influence adaptive immune responses and that exudate Mϕ are phagocytic and can become fungicidal. The objective of the current study was to further investigate the in vivo effects of GM-CSF on pulmonary DC and Mϕ in mice infected by the intratracheal (i.t.) route with a moderately virulent strain of C. neoformans (strain 52D). Our studies were performed using GM^+/+ mice (C57BL/6J mice), which develop persistent cryptococcal lung infection (17, 26, 27), and GM^−/− mice (C57BL/6J genetic background), which develop progressive lung infection (17). Our results show that GM-CSF promotes the differentiation, accumulation, activation, and alveolar localization of lung DC and Mϕ. These findings help define the critical role of GM-CSF and myeloid cells in pulmonary host defenses against fungal pathogens and enhance our understanding of cryptococcal infections in susceptible patient populations.

Wild-type (C57BL/6J) mice were obtained from Charles River Laboratory (Wilmington, MA). GM^−/− mice were originally developed (28) and provided by Dr. J. Whitsett (Children’s Hospital, Cincinnati, OH) and extensively backcrossed against C57BL/6J mice as previously described (17). For our studies, GM^−/− mice were bred on site. All mice were housed under specific pathogen-free conditions in the Animal Care Facility at the Ann Arbor Veterans Affairs Health System. All studies were conducted according to a protocol approved by the VA Institutional Animal Care and Use Committee. Mice were 6–8 wk of age at the time of infection.

C. neoformans strain 52D was obtained from the American Type Culture Collection (24067; Manassas, VA); this strain displayed smooth colony morphology when grown on Sabouraud dextrose agar. For i.t. inoculation, C. neoformans was grown to a late logarithmic phase (48–72 h) at 37°C in Sabouraud dextrose broth (1% neopeptone and 2% dextrose; DIFCO, Detroit, MI) on a shaker. Cultured yeasts were then washed in nonpyrogenic saline, counted in the presence of Trypan Blue using a hemocytometer, and diluted to 3.3 × 10⁵ CFU/ml in sterile nonpyrogenic saline immediately prior to i.t. inoculation.

Mice were anesthetized by i.p. injection of ketamine (100 mg/kg; Fort Dodge Laboratories, Fort Dodge, IA) and xylazine (6.8 mg/kg; Lloyd Laboratories, Shenandoah, IA). Through a small midline neck incision, the strap muscles were divided and retracted laterally to expose the trachea. The i.t. inoculation was performed under direct vision using a 30-gauge needle attached to a 1-ml syringe mounted on a repetitive pipette (stepper; Tridak, Brookfield, CT). An inoculum of 10⁴ CFU (30 μl) was injected into the trachea. Skin was closed using cyanoacrylate adhesive.

Lungs were perfused in situ via the right heart using PBS containing 0.5 mmol EDTA until pulmonary vessels became grossly clear (∼5 ml). Lungs were then excised, minced, and enzymatically digested to a single-cell suspension as previously described (25). After erythrocyte lysis, cells were washed, filtered over 70-μm mesh, and resuspended in complete medium. Dead cells were removed by centrifugation over a Percoll gradient. Total numbers of viable lung leukocytes were assessed in the presence of Trypan blue using a hemocytometer. PBMC were isolated from peripheral blood obtained from the retro-orbital vein of deeply anesthetized mice and processed as previously described (23). All cell preparations were washed twice in sterile PBS before use for culture or Ab staining.

Lung leukocytes obtained from mice infected for 14 d were cultured at 5 × 10⁶ cells/ml in 24-well plates in 2 ml complete RPMI 1640 media at 37°C and 5% CO₂ for 24 h without additional stimulation. At the end of culture, supernatants were harvested, and cytokine concentration was assessed in triplicates by Luminex assay (Luminex, Austin, TX) per the manufacturer’s instructions.

Lungs were perfused with PBS via the right ventricle of the heart and inflated with 50% OCT compound (Tissue Tek; Sakura Finetek, Torrance, CA) in PBS via the trachea. Lung lobes were then harvested and frozen in OCT. Twenty-micrometer sections were stained using H&E and viewed by light microscopy. Microphotographs were taken using the Digital Microphotography system DFX1200 with ACT-1 software (Nikon, Tokyo, Japan). For immunohistochemistry, sections were fixed in cold acetone, blocked using purified anti-mouse CD16/32 Ab (clone 93; 10 μg/ml; BioLegend, San Diego, CA), and stained with Alexa Fluor 488 anti-mouse CD11c Ab (clone N418, diluted 1:100; BioLegend). Negative control sections were stained with an isotype-matched Ab recommended by the manufacturer (Alexa Fluor 488 Armenian Hamster IgG, clone HTK888; at the same dilution; BioLegend). Sections were mounted using ProLong Gold Antifade Mountant with DAPI (Life Technologies, Grand Island, NY). Confocal microscopy was performed using an Olympus BX51WI microscope (Olympus, Center Valley, PA). Images were collected using a Hamamatsu EM-CCD color digital camera C9100 and the Stereo Investigator software (MicroBright Field; MBF Bioscience, Williston, VT).

The following mAbs purchased from BioLegend were used: N418 (anti-murine CD11c); 2.4G2 (Fc block, anti-murine CD16/CD32); 30-F11 (anti-murine CD45); C/23 (anti-murine CD40); 16-10A1 (anti-murine CD80); GL1 (anti-murine CD86); AF6-120.1 (anti-murine I-A^b); 10F.9G2 (anti-murine programmed cell death ligand 1 [PD-L1]); TY25 (anti-murine PD-L2); 6d5 (anti-murine CD19); 1A8 (anti-murine Ly-6G); 145-2C11 (anti-murine CD3ε); BM8 (anti-murine F4/80); NLDC-145 (anti-murine CD205); C068C2 (anti-murine CD206); XMG1.2 (anti–IFN-γ); TC11-18H10.1 (anti–IL-17A); and TRFK4 (anti–IL-5). The mAb AL-21 (anti-murine Ly-6C) and M1/70 (anti-murine CD11b) and ebio13A (anti–IL-13) were purchased from BD Biosciences (San Diego, CA). The mAb 2f8 (anti-murine CD204) was purchased from AbD Serotec. mAbs were primarily conjugated with Alexa Fluor 700, FITC, PE, PE-Cy7, PerCP-Cy5.5, allophycocyanin, allophycocyanin-Cy7, Brilliant Violet 421, or Pacific Blue. Isotype-matched control mAbs (BioLegend) were tested simultaneously in all experiments.

Cell staining, including blockade of FcRs, and sample analysis by flow cytometry were performed as described previously (25). A minimum of 10,000 events were acquired per sample on an LSR II flow cytometer (BD Biosciences) and analyzed using FlowJo (Tree Star, Ashland, OR).

Lung leukocyte subsets of DC, CD4⁺ T cells, Mϕ, and Ly-6C^high monocytes were identified using established gating strategies (23, 24) and as described in additional detail in the legends to Figs. 3, 4, and 5 and the 13Results section. Briefly, initial gates eliminated debris and doublets, identified CD45⁺ leukocytes, and excluded CD3⁺ T cells, CD19⁺ B cells, and Ly-6G⁺ granulocytes as described previously (23, 24). Thereafter, additional gating identified CD11c⁺ DC and Mϕ. Autofluorescence, assessed in the PerCP-Cy5.5 channel using a 488-nm laser, was used to distinguish nonautofluorescent DC from autofluorescent Mϕ. Within the nonautofluorescent population of DC, a CD11c versus CD11b plot was used to identify total CD11c⁺ CD11b⁺ DC (described as “CD11b⁺ DC” in the text). Within the autofluorescent population of Mϕ, a CD11c versus CD11b plot was used to identify CD11b⁻ Mϕ and CD11b⁺ Mϕ. The phenotype of DC and Mϕ is summarized as follows: CD11b⁺ DC (CD45⁺CD3⁻CD19⁻forward light scatter [FSC]^moderateLy-6G⁻ nonautofluorescent CD11c⁺CD11b⁺); CD11b⁻ Mϕ (CD45⁺CD3⁻CD19⁻FSC^{moderate/high}Ly-6G⁻ autofluorescent CD11c⁺CD11b⁻); and CD11b⁺ Mϕ (CD45⁺CD3⁻CD19⁻FSC^{moderate/high}Ly-6G⁻ autofluorescent CD11c⁺CD11b⁺). In some studies, CD11b⁻ and CD11b⁺ Mϕ were further subdivided into CD11c^high and CD11c^low subsets. Ly-6C^high monocytes were identified within the CD11c⁻FSC^low population of myeloid cells as having the following phenotype: (CD45⁺CD3⁻CD19⁻FSC^lowLy-6G⁻CD11c⁻F4/80⁺CD11b⁺Ly-6C^high). Myeloid cell differentiation and activation was further assessed by measuring cell-surface expression of MHC class II (I-A^b), CD40, CD80, CD86, PD-L2, CD204, CD205, CD206, and Ly-6C relative to an isotype-matched control Ab.

Within the CD45⁺ population, CD4⁺ T cells were identified as CD45⁺FSC^lowside scatter^lowCD4⁺CD8⁻. To detect intracellular cytokine production, lung leukocytes were stimulated with PMA (50 ng/ml) and ionomycin (1 μg/ml) (Sigma-Aldrich, St. Louis, MO) in the presence of brefeldin A and monensin (BioLegend) for 6 h, stained with Aqua Live/Dead fixable viability dye and Abs targeting CD45, CD3, CD4, and CD8, and then fixed with 2% buffered formaldehyde. Intracellular cytokines staining was performed with Abs against IFN-γ, IL-17A, IL-5, and IL-13 in permeabilization buffer (eBioscience).

To ensure consistency in data analysis, gate positions were held constant for all samples. To calculate the total number of cells in each population of interest in each sample, the corresponding percentage was multiplied by the total number of CD45⁺ cells in that sample. The latter value was calculated for each sample as the product of the percentage of CD45⁺ cells and the original hemocytometer count of total cells identified within that sample.

Lung Mϕ were obtained by plating lung leukocytes (isolated as described above) in six-well plates (5 × 10⁶/ml 5% complete media, 2 ml/well) and incubating them at 5% CO₂, 37°C for 90 min. At the end of incubation, RNA was extracted from adherent cells using TRIzol (Invitrogen) and DNase treated using the DNA-free Kit (Ambion by Life Technologies). CD11b⁻ and CD11b⁺ lung Mϕ were obtained by flow sorting of lung leukocytes stained using fluorochrome-conjugated Abs as detailed above. One-step quantitative RT-PCR (qRT-PCR) was performed using the QuantiTect Sybr Green Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol with GAPDH serving as an endogenous reference. Reactions were run in triplicates using a StepOne Plus real-time PCR system (Applied Biosystems). Primers were synthesized by Sigma-Aldrich based on the following 5′-3′ sequences: Arg-1 sense, CAGAAGAATGGAAGAGTCAG; Arg-1 antisense, CAGATATGCAGGGAGTCACC; Retnla (resistin-like molecule-α [Relmα]) sense, TTCTTGCCAATCCAGCTAAC; Retnla (Relmα) antisense, GGGTTCTCCACCTCTTCATT; Mrc1 (CD206) sense, CTCTGTTCAGCTATTGGACGC; Mrc1 (CD206) antisense, CGGAATTTCTGGGATTCAGCTTC; inducible NO synthase (iNOS) sense, GTTCTCAGCCCAACAATACAAGA; iNOS antisense, GTGGACGGGTCGATGTCAC; GAPDH sense, TATGTCGTGGAGTCTACTGGT; and GAPDH antisense, GAGTTGTCATATTTCTCGTGG. Data were analyzed using the 2^−ΔΔ threshold cycle method.

All data were expressed as mean ± SEM. Continuous ratio scale data were evaluated by unpaired Student t test (for comparison between two samples). Statistical calculations were performed on a Dell 270 computer using GraphPad Prism version 6.00 for Windows (GraphPad Software, San Diego, CA). Statistical difference was accepted at p < 0.05.

We previously compared fungal lung burden in GM^−/− and wild-type C57BL/6 (GM^+/+) mice infected by the i.t. route with C. neoformans strain 52D and showed that CFUs in GM^−/− mice increased 2-, 3-, and 6-fold relative to GM^+/+ mice at 14, 21, and 35 d postinfection (dpi), respectively (17). In the current study, similar increases in pulmonary CFU were confirmed in GM^−/− mice at 14 and 28 dpi (Supplemental Fig. 1). As GM-CSF improves control of cryptococcal lung infection, we sought to identify cellular and molecular mechanisms through which GM-CSF mediates this beneficial effect.

To determine whether GM-CSF impacted lung leukocyte accumulation, the number of CD45⁺ lung leukocytes was assessed in GM^+/+ and GM^−/− mice using flow cytometric analysis (FCA) at day 0 (uninfected) and days 7, 14, 21, and 28 dpi. In GM^+/+ mice, the number of CD45⁺ lung leukocytes increased by 7 dpi (7-fold relative to uninfected GM^+/+ mice), peaked at 14 dpi (14-fold), and remained elevated at 21 and 28 dpi (Fig. 1A). In contrast, significantly fewer leukocytes accumulated in the lungs of GM^−/− mice at 7, 14, and 21 dpi, whereas CD45⁺ numbers were equivalent to GM^+/+ mice at 28 dpi.

To broadly determine whether GM-CSF impacted T cell polarization profiles, we next assessed lung leukocytes obtained from GM^+/+ and GM^−/− mice at 14 dpi for their production of cytokines associated with Th1 (IFN-γ and TNF-α), Th2 (IL-4, IL-5, IL-13, and IL-10), and Th17 (IL-17) responses. In GM^+/+ mice, we found robust production of IL-4, IL-5, IL-10, and IL-13 as well as modest expression of IFN-γ, IL-17, and TNF-α (Fig. 1B). In contrast, secretion of IL-4, IL-5, IL-10, and IL-13 were significantly decreased in GM^−/− relative to GM^+/+ mice; nonsignificant trends toward reduced IL-17 and TNF-α were also noted. Interestingly, leukocyte production of IFN-γ was increased in GM^−/− versus GM^+/+ mice.

To establish whether the observed differences in total lung leukocyte cytokine production reflected specific differences in T cell polarization profiles, we enumerated CD4⁺ T cells and assessed their cytokine production using intracellular FCA of leukocytes obtained from the lungs of GM^+/+ and GM^−/− mice at 14 and 28 dpi. Total numbers of CD4⁺ T cells were increased in GM^−/− mice at both time points (Fig. 2A). As expected based on our lung leukocyte cytokine data (Fig. 1B), the percent and total numbers of CD4⁺ T cells expressing IL-5 and IL-13 (at 14 and 28 dpi) was high in GM^+/+ mice and significantly decreased in GM^−/− mice (Fig. 2B, 2C, top panels). In contrast, the percentage (at 14 dpi) and total numbers (at 14 and 28 dpi) of CD4⁺ T cells expressing IFN-γ was increased in GM^−/− mice, whereas the percentage (at 14 and 28 dpi) and total numbers (at 28 dpi) of CD4⁺ T cells expressing IL-17 was significantly decreased (in GM^−/− mice; Fig. 2B, 2C, bottom panels). No difference in the total numbers of CD8⁺ T cells, or the subset of CD8⁺ T cells expressing IFN-γ was observed between GM^+/+ and GM^−/− mice (data not shown). Taken together, these findings show that GM-CSF enhances Th2 and Th17 responses and diminishes IFN-γ responses in mice that develop persistent cryptococcal lung infection.

The differential effects of GM-CSF on Th cell polarization focused our attention on CD11b⁺ DC, as prior studies have shown that they play a crucial role in orchestrating the adaptive immune response to C. neoformans (23, 25). We used established gating strategies (23, 27) (2Materials and Methods; Fig. 3A) to identify and enumerate CD11b⁺ lung DC by FCA at baseline and throughout the infection (Fig. 3A, 3B). We found similar numbers of CD11b⁺ DC in the lungs of uninfected GM^+/+ and GM^−/− mice (day 0). Following infection in GM^+/+ mice, the numbers of CD11b⁺ lung DC increased rapidly and substantially (21-fold), peaking at 7 dpi, and then slowly declining. In contrast, the numbers of CD11b⁺ lung DC in GM^−/− mice were significantly reduced (relative to their numbers in GM^+/+ mice) at 7, 14, and 21 dpi. Furthermore, their total expansion was low (4-fold) and delayed, peaking at 28 dpi.

We also evaluated the expression of MHC class II and costimulatory molecules CD40, CD80, and CD86 by CD11b⁺ DC in GM^+/+ and GM^−/− mice, as prior studies have demonstrated a positive association between their expression and clearance or containment of cryptococcal lung infection (27, 29–32). No differences in the expression of these molecules were observed on CD11b⁺ DC in the lungs of uninfected GM^+/+ and GM^−/− mice (data not shown). At 14 dpi, expression of CD80 was reduced in GM^−/− mice; a nonsignificant trend toward decreased expression of MHC class II was also observed (relative to GM^+/+ mice; Fig. 3C). Expression of CD40 and CD86 did not differ. We also analyzed the expression of the immunomodulatory receptor PD-L2 by CD11b⁺ lung DC. PD-L2 was not expressed by CD11b⁺ DC in the lungs of uninfected GM^+/+ or GM^−/− mice (data not shown). We observed modest PD-L2 expression by CD11b⁺ DC in the infected GM^+/+ mice and minimal expression on CD11b⁺ lung DC obtained from GM^−/− mice (Fig. 3C). In addition to assessing DC activation markers, we also examined the expression level of Ly-6C on CD11b⁺ DC because we have previously shown that expression of this molecule decreases as DC differentiate from Ly-6C^high monocytes (23). Ly-6C was significantly higher on CD11b⁺ DC in the lungs of GM^−/− relative to GM^+/+ mice (data not shown), suggesting they were less differentiated. Collectively, these studies indicate that GM-CSF promotes the accumulation and activation of CD11b⁺ DC in mice with cryptococcal lung infection.

Our next objective was to determine whether the reduction in CD11b⁺ DC observed in GM^−/− mice was attributable to impaired accumulation of their Ly-6C^high monocyte precursors. Established gating schemes and FCA (23, 24) (Fig. 4A) were used to identify and enumerate Ly-6C^high monocytes in GM^+/+ and GM^−/− mice at baseline and following infection. In uninfected mice, no differences were identified in the frequency or total numbers of Ly-6C^high monocytes (Fig. 4B). The numbers of Ly-6C^high monocytes in the lungs of GM^+/+ mice increased markedly (23-fold) and rapidly, peaking at 7 dpi (Fig. 4B). Interestingly, the percentage of Ly-6C^high monocytes in the lungs of GM^−/− mice significantly exceeded that observed in GM^+/+ mice at 7, 14, and 21 dpi (Fig. 4B, left panel). Despite the reduction in total lung leukocytes in infected GM^−/− mice relative to GM^+/+ mice (refer to Fig. 1A), total numbers of Ly-6C^high monocytes in GM^−/− and GM^+/+ mice were similar at 7, 14, and 28 dpi and were increased in GM^−/− mice at 21 dpi (Fig. 4B, right panel). Lastly, the percentage of Ly-6C^high monocytes in peripheral blood of GM^+/+ and GM^−/− mice was determined. Results show that the percentage of Ly-6C^high monocytes in peripheral blood of GM^+/+ and GM^−/− mice was similar at baseline and increased with infection. Furthermore, similar to what was observed in the lung, the percentage of Ly-6C^high monocytes in GM^−/− mice was comparable or significantly elevated relative to GM^+/+ mice (Fig. 4C). Collectively, these data demonstrate that GM-CSF deficiency does not impair the accumulation of Ly-6C^high monocytes in the blood or lungs of mice with cryptococcal lung infection.

Increased fungal burdens in GM^−/− mice suggested a Mϕ defect as these are the critical effector cells responsible for cryptococcal killing. To investigate this possibility, we used established gating strategies and FCA (24, 27) (2Materials and Methods; Fig. 5A) to identify and enumerate two subpopulations of lung Mϕ based on their expression of CD11b. In uninfected GM^+/+ mice, we identified a large population of autofluorescent CD11c⁺CD11b⁻ Mϕ and a smaller population of CD11c⁺CD11b⁺ Mϕ (Fig. 5A). Note that in prior studies we have identified CD11b⁻ and CD11b⁺ Mϕ as alveolar and exudate Mϕ, respectively (24, 27, 33). In uninfected GM^−/− mice, the subpopulation of CD11b⁺ Mϕ predominated (Fig. 4A) consistent with other reports demonstrating that few, if any, CD11b⁻ Mϕ are present in the lung of uninfected GM-CSF^−/− mice; those that are present are felt to represent an immature subset (34). In GM^+/+ and GM^−/− mice, both CD11b⁻ and CD11b⁺ Mϕ expressed the F4/80 cell-surface marker, further confirming their myeloid origin (Fig. 5B).

In response to infection in GM^+/+ mice, the numbers of CD11b⁻ Mϕ increased by a total of 5-fold and peaked at 14 dpi (Fig. 5C, left panel), whereas the numbers of CD11b⁺ Mϕ increased by a total of 37-fold and peaked at 21 dpi (Fig. 5C, right panel). In contrast, the numbers of CD11b- Mϕ in infected GM^−/− mice were significantly reduced at 7, 14, and 21 dpi. The numbers of CD11b⁺ Mϕ in infected GM^−/− mice were also reduced at 7, 14, and 21 dpi; however, this reduction reached statistical significance only at 7 dpi.

Prior studies identified decreased CD11c cell-surface expression by lung Mϕ in uninfected GM-CSF^−/− mice (35, 36). Yet, whether both CD11b⁻ and CD11b⁺ Mϕ are similarly affected and whether decreased CD11c expression is observed on Mϕ in the lungs of GM^−/− mice during cryptococcal lung infection is unclear. Using a modified gating scheme in which lung Mϕ were subdivided on the basis of high versus low CD11c expression (Fig. 5D), we observed prominent reductions in the CD11c^high subsets of both CD11b⁻ and CD11b⁺ Mϕ (Fig. 5E).

The immunophenotype of CD11b⁻ and CD11b⁺ Mϕ in the lungs of infected GM^+/+ and GM^−/− mice at 14 dpi was further assessed using an Ab panel that targeted I-A^b (MHC class II), cell-surface pathogen recognition receptors, and costimulatory molecules. Expression of I-A^b, CD204 (scavenger receptor A), CD205 (DEC-205), and CD206 (mannose receptor) was decreased in CD11b⁻ Mϕ in the lungs of infected GM^−/− mice relative to GM^+/+ mice (Fig. 6A). No difference in CD11b⁻ Mϕ expression of CD40, CD80, or CD86 was observed, whereas PD-L2 expression was reduced relative to GM^+/+ mice (data not shown). CD11b⁺ Mϕ expression of I-A^b and CD206 were diminished in GM^−/− mice, whereas CD204 and CD205 expression did not differ (relative to GM^+/+ mice; Fig. 6B). No difference in CD11b⁺ Mϕ expression of CD40 and CD80 was identified, whereas CD86 was increased and PD-L2 decreased in GM^−/− mice (relative to GM^+/+ mice; data not shown).

Macrophage activation was further assessed by gene expression profiling performed on adherence-enriched total lung Mϕ preparations obtained from individual GM^+/+ and GM^−/− mice at 14 and 28 dpi. The reduced expression of Th2 cytokines in GM^−/− mice (Fig. 1B) led us to first evaluate gene expression of three proteins—Arginase 1, Relmα (also referred to as Fizz1), and CD206—associated with alternative Mϕ activation (Fig. 6C). Consistent with this hypothesis, qRT-PCR analysis showed that transcript levels of all three genes were decreased in Mϕ obtained from GM^−/− versus GM^+/+ mice (Fig. 6C). Expression of iNOS, which is associated with classical Mϕ activation, was unchanged at 14 dpi, but significantly increased at 28 dpi (relative to GM^+/+ mice; Fig. 6C). Lastly, we determined whether these patterns of gene expression differed between CD11b⁻ and CD11b⁺ Mϕ by performing qRT-PCR analysis on each individual subpopulation after their isolation by FACS (using gating strategies described in Fig. 5A) from pooled samples of lung leukocytes obtained from GM^+/+ and GM^−/− mice at 14 dpi. Expression of genes encoding for arginase and Relmα was decreased in both CD11b⁻ and CD11b⁺ Mϕ isolated from GM^−/− mice, whereas iNOS was increased (relative to GM^+\+ mice; Fig. 5D). Taken together, our studies of Mϕ numbers, immunophenotype, and gene expression demonstrate that GM-CSF promotes substantial accumulation and alternative activation of CD11b⁻ and CD11b⁺ Mϕ in the lungs of mice with cryptococcal lung infection.

Our findings identified that GM-CSF mediated some immune effects commonly associated with cryptococcal clearance (enhanced Th17 responses and the accumulation of activated CD11b⁺ DC and Mϕ) as well as effects often associated with persistence (enhanced Th2 responses, decreased Th1 responses, and alternative Mϕ activation) (26, 27, 30, 37). To further elucidate the effects of GM-CSF on control of cryptococcal lung infection, we assessed microanatomic features of the infection using lung sections obtained from infected GM^+/+ and GM^−/− mice at 14 and 28 dpi by light microscopy (Fig. 7). In the lungs of GM^+/+ mice obtained at 14 dpi, we observed cryptococci located in alveolar regions encompassed by patchy infiltrates comprised of numerous mononuclear cells, eosinophils, and larger cells with abundant cytoplasm (Fig. 7A, 7B). At 28 dpi, these infiltrates had coalesced into loosely formed foci of alveolar granulomas in which most fungal organisms were contained intracellularly within large foamy Mϕ (Fig. 7E, 7F). In contrast, examination of lung sections obtained from infected GM^−/− mice at 14 dpi revealed alveolar spaces containing numerous cryptococci, but relatively few leukocytes (Fig. 7C, 7D). At 28 dpi, dense bronchovascular infiltrates containing small mononuclear cells were identified adjacent to airways, whereas scarce inflammatory cells and numerous “uncontained” cryptoccoci were observed in the alveolar spaces (Fig. 7G, 7H).

Our assessment by light microscopy of the microanatomic characteristics of cryptococcal lung infection showed a reduction in immune infiltrates at the site of infection in alveolar spaces in GM^−/− mice (relative to GM^+/+ mice; Fig. 7). Total reductions in lung DC and Mϕ numbers were also identified (by FCA; Figs. 3, 5). We next compared the anatomic distribution and relative abundance of lung DC and Mϕ in GM^+/+ versus GM^−/− mice using CD11c Ab staining and fluorescent microscopy (Fig. 8). Numerous CD11c⁺ cells were identified within early alveolar infiltrates in the lungs of GM^+/+ mice at 14 dpi (Fig. 8A and inset). In contrast, few CD11c⁺ cells were identified in the lungs of infected GM^−/− mice at this time point (Fig. 8B). At 28 dpi, numerous large CD11c⁺ cells displaying Mϕ phenotype were identified within established alveolar infiltrates in the lungs of GM^+/+ mice (Fig. 8C and inset), whereas CD11c⁺ cells were still absent in the alveolar regions of GM^−/− mice (Fig. 8D). Rather, lungs from infected GM^−/− mice displayed dense accumulations of small CD11c-negative cells within bronchovascular infiltrates. These data, in conjunction with those obtained by light microscopy (Fig. 7), show that GM-CSF promotes the localization of lung DC and Mϕ to sites of alveolar infection.

Accumulating evidence identifies GM-CSF deficiency as a risk factor for the development of cryptococcal infections in humans (6, 7). Our prior study (17) and current findings show that control of cryptococcal lung infection is impaired in GM^−/− mice. Our thorough temporal and spatial immunophenotyping of this response yielded the following observations in infected GM^−/− mice (relative to GM^+/+ mice): 1) Th2 and Th17 responses are decreased, whereas Th1 responses are increased; 2) accumulation of CD11b⁺ DC and Mϕ are decreased, whereas the numbers of Ly-6C^high monocytes are unchanged or increased; 3) CD11b⁺ DC activation and alternative Mϕ activation are decreased; and 4) localization of DC and Mϕ at sites of alveolar infection are decreased. Thus, the current study provides important evidence that the net positive effect of GM-CSF on control of cryptococcal lung infection results from a complex interplay between local immune processes affecting T cell polarization and the in situ differentiation, accumulation, activation, and alveolar localization of lung DC and Mϕ.

The most striking finding in the lungs of GM^−/− mice with cryptococcal infection was the pronounced reduction of CD11b⁺ DC and Mϕ, especially the subset of cells expressing the highest amounts of CD11c. This observation supports the previous findings of Paine et al. (36), who reported decreased CD11c expression on lung Mϕ obtained from GM^−/− mice and that of Guth et al. (35), who showed that a subset of fluorescently labeled cells obtained from the bone marrow of GM^−/− mice gained expression of CD11c following their i.t. injection and subsequent recovery from the lungs of GM^+/+ mice. In a related study, Guilliams et al. (34) showed that perinatal intrapulmonary GM-CSF administration to uninfected GM^−/− mice restored high CD11c expression and alveolar Mϕ development from fetal monocytes. This group suggested that Mϕ present in the lungs of uninfected GM^−/− mice are not bona fide Mϕ but rather a monocyte–Mϕ intermediate; our data support this conclusion. Furthermore, whereas their studies were performed in naive mice, our studies performed in infected mice suggest that the maturation defect observed in the absence of GM-CSF persists over time and despite a considerable microbial burden.

The consequence of this reduction in fully differentiated CD11c⁺ DC or Mϕ was most notable at the site of infection within the alveolar spaces of GM^−/− mice. Loose granulomatous inflammation was prevalent in GM^+/+ mice, and we observed numerous mature appearing DC and Mϕ surrounding and containing individual cryptococci. In contrast, alveolar inflammation was markedly reduced in infected GM^−/− mice, and numerous “free” Cryptococci were identified. Collectively, these findings demonstrate that GM-CSF promotes the accumulation of lung DC and Mϕ to the site of infection within the alveolar compartment. These reductions in alveolar infiltrates identified in mice with cryptococcal lung infection resemble those observed in GM-CSF deficient mice infected with Mycobacterium tuberculosis (38), suggesting a common mechanism leading to the increased susceptibility to pulmonary pathogens observed in GM^−/− mice and patients with acquired GM-CSF deficiency.

Whereas alveolar inflammation was reduced in infected GM^−/− mice, we identified dense mononuclear infiltrates surrounding bronchovascular structures that appeared anatomically distinct from the infected alveolar compartment. We believe it likely that these bronchovascular infiltrates contain an abundance of recruited monocytes that have failed to fully differentiate into CD11b⁺ DC or Mϕ in the absence of GM-CSF. In support of this hypothesis, data obtained by FCA revealed that the percentage and number of their Ly-6C^high monocyte precursors were either equivalent or increased within the lungs and peripheral blood (percentage data only) of GM^−/− mice (relative to infected GM^+/+ mice), showing that the accumulation of these cells in response to infection is GM-CSF independent. These findings, as well as our prior studies (23, 24) and those of Guth et al. (35, discussed above), suggest that the impaired accumulation of CD11b⁺ DC and Mϕ in GM^−/− mice is attributable to their failure to differentiate from Ly-6C^high monocytes within the infected pulmonary microenvironment.

We identified complex effects of GM-CSF on DC and Mϕ activation that might be viewed as both favorable and unfavorable. In the absence of GM-CSF, CD11b⁺ DC expressed less CD80 and a trend toward less MHC class II relative to GM^+/+ mice. Similarly, expression of MHC class II, CD204, and CD205 were decreased in CD11b⁻ and CD11b⁺ Mϕ when GM-CSF was absent. These findings show that GM-CSF promotes the expression of numerous important markers associated with Ag presentation and phagocytosis that would be expected to improve fungal clearance. However, GM-CSF deficiency was also associated with reduced Mϕ expression of CD206, Arginase 1, and Relmα and increased expression of iNOS. Thus, GM-CSF expression was linked with alternative Mϕ activation, which might be expected to impair fungal clearance. Although the myeloid cells in GM^−/− mice express iNOS (likely in response to the observed increase in IFN-γ), their inability to fully differentiate and mobilize to infected alveolar spaces likely limits their fungicidal capabilities. The degree to which GM-CSF affects these activation profiles directly (by signaling through the GM-CSF receptor; reviewed in Ref. 9) or indirectly (through changes in the cytokine or surfactant milieu) is uncertain and warrants additional investigation in future studies.

Our data also identify similarly complex effects of GM-CSF on T cell polarization responses in this model. In infected GM^−/− mice, production of IL-4, IL-5, IL-10, and IL-13 by lung leukocytes was reduced as were the percentage and number of CD4⁺ T cells staining for intracellular IL-5 and IL-13. These results confirm and extend findings from our initial study (17) and provide clear and convincing evidence that GM-CSF promotes Th2 responses in this model. In contrast, Th1 responses appeared increased in the absence of GM-CSF, and we believe this may be attributable to reduced counter-regulation by IL-10, as we have recently shown that blockade of IL-10 signaling increases IFN-γ production by CD4⁺ T cells in mice with cryptococcal lung infection (27). Note that based on results of prior studies (26, 39), we might have expected that the decreased Th2 response and increased Th1 responses observed in GM^−/− mice (relative to GM^+/+ mice) would have decreased the fungal burden in the lung, which was not the case. Yet, it remains possible that the increased Th1/Th2 balance observed in GM^−/− mice had some favorable effect on fungal clearance, which was nonetheless insufficient to overcome the more pronounced impairments observed in myeloid cells in the absence of GM-CSF. In addition, Th17 responses were not increased in parallel with Th1 responses, but rather were decreased in GM^−/− mice, demonstrating that these responses can be differentially regulated and also offering another potential explanation for the beneficial effect of GM-CSF, as we have shown that IL-17 provides modest protection against progressive cryptococcal infection in this model (40).

In summary, our findings reinforce a net-protective role for GM-CSF in host defenses against cryptococcal lung infection. Despite promoting maladaptive Th2 responses, our results show that GM-CSF exerts a favorable local effect within the infected pulmonary microenvironment to promote the differentiation, accumulation, activation, and alveolar localization of lung DC and Mϕ. Our collective findings show that GM-CSF critically contributes to the formation of loose granulomatous inflammation that controls fungal burdens in this model of persistent cryptococcal lung infection. These findings advance our understanding of the role of GM-CSF in antifungal host defenses and help explain the susceptibility to cryptococcal infections in patients with acquired anti–GM-CSF Abs.

We thank Valerie R. Stolberg for assistance with fluorescence microscopy.

The authors have no financial conflicts of interest.