Autoantibody-Mediated Pulmonary Alveolar Proteinosis in Rasgrp1-Deficient Mice

Pulmonary alveolar proteinosis (PAP) is a rare pulmonary syndrome defined by the accumulation of pulmonary surfactant in the alveoli leading to respiratory insufficiency. Pulmonary surfactant levels are maintained in part by GM-CSF (also called CSF-2) (1–5). GM-CSF is a pleotropic cytokine that is required for the differentiation, maintenance, and function of alveolar macrophages (6, 7). In the absence of GM-CSF, degradation of pulmonary surfactants by alveolar macrophages is impaired, leading to an accumulation in the alveoli. GM-CSF–deprived alveolar macrophages also are less adherent, and they exhibit impaired phagocytosis and pathogen killing (6). Defective phagocytosis and pathogen killing by alveolar macrophages likely contributes to increased susceptibility to infections with opportunistic pathogens such as Nocardia spp., Cryptococcus spp., and Aspergillus spp. in PAP patients (8).

PAP is classified into three clinical forms including congenital, secondary, and autoimmune, with the latter representing the predominant form comprising 90% of patients (9). In autoimmune PAP (aPAP), GM-CSF–specific autoantibodies are elevated in the serum and bronchoalveolar lavage fluid (BALF) (10, 11). Moreover, transfer of GM-CSF–specific autoantibodies from human aPAP patients to healthy nonhuman primates recapitulates the disease, establishing a causal relationship (12). aPAP patients treated with rituximab, a B cell–depleting anti-CD20 mAb, exhibit improvement in lipid homeostasis and overall outcome, illustrating that B cells and Ab are necessary for the pathology (13–15). GM-CSF–specific autoantibodies in aPAP patients are polyclonal and somatically hypermutated IgG Abs (16, 17). Interestingly, GM-CSF–specific IgG autoantibodies are also detectable in healthy individuals, although titers are significantly lower compared with aPAP patients (18). GM-CSF–specific autoantibodies from both healthy individuals and aPAP patients are comparable in regard to IgG subclass distribution (18), raising a key question of how immunological tolerance is regulated in healthy and pathologic settings. However, identifying underlying immune mechanism(s) resulting in the production of pathological GM-CSF–specific autoantibodies is problematic due to a lack of appropriate animal models.

RasGRP1 is a guanine exchange factor expressed by B and T cells that activates Ras downstream of B cell and T cell Ag receptor stimulation (19). Mice deficient in RasGRP1 are lymphopenic early in life with a block at the double-positive stage in T cell development (20). Later in life, Rasgrp1-deficient mice develop a lymphoproliferative disorder with multiple autoimmune features including enlarged secondary lymphoid tissues, spontaneous formation of germinal centers, and autoantibody production including anti-nuclear autoantibodies (21). The autoimmune features in Rasgrp1-deficient mice are T cell–dependent because Rasgrp1-deficient mice crossed to athymic nude mice have attenuated splenomegaly and lack anti-nuclear Abs (22). In addition, studies using Rasgrp1-deficient mice harboring an autoreactive BCR transgene revealed that B cells in Rasgrp1-deficient mice also breach tolerance at multiple checkpoints, implicating both T cell–independent and –dependent mechanisms (23).

In this study, we show that autoimmune-prone, Rasgrp1-deficient mice develop PAP. The disease in Rasgrp1-deficient mice features characteristic accumulation of pulmonary surfactants in alveolar spaces, reduced pulmonary compliance, impaired alveolar macrophage function, and age-dependent increases in GM-CSF–neutralizing autoantibodies in serum and BALF that correspond with disease onset. Lastly, aged mice lacking both RasGRP1 and CD275/ICOSL, a molecule necessary for conventional T follicular helper cell (Tfh) development and the production of T cell–dependent Ab, do not develop PAP. Therefore, from these data, we suggest that Rasgrp1-deficient mice represent, to our knowledge, the first rodent model for aPAP.

Mice were housed in an American Association for the Accreditation of Laboratory Animal Care–certified pathogen-free facility at the University of South Alabama College of Medicine. Mice were >10 generations backcrossed on a C57BL/6 background. Both C57BL/6 mice and mice heterozygous for Rasgrp1 were used as controls as indicated. Importantly, PAP was not observed in either control group. Mice referred to as “young” herein ranged from 2 to 3 mo of age, whereas those referred to as “aged” ranged from 7 to 12 mo of age. All procedures involving mice were performed according to approved protocols by the University of South Alabama Institutional Animal Care and Use Committee.

Pulmonary compliance was determined by removing the chest wall and injecting 2–20 ml/kg air at 50-μl increments and recording pressure at each increment after a 2- to 3-s pause for the pressure to plateau. Compliance was calculated from the slope of the linear portion of the inflation curve.

Lungs of mice were inflation fixed in situ with 10% phosphate-buffered formalin at a pressure of 15 cm H₂O, embedded in paraffin, sectioned, and stained with H&E or periodic acid–Schiff (PAS). For quantification of proteinaceous material, nonoverlapping lung sections imaged at low power (40× total) encompassing the entire lung section were analyzed using ImageJ software. Percent proteinaceous accumulation was defined as binary area of proteinaceous material divided by binary area of the total lung section analyzed. Total percent proteinaceous accumulation for each mouse was calculated using the mean of three lung sections including an upper, middle, and lower portion of the left lung lobe for each mouse.

To identify surfactant proteins (SP), we deparaffinized and immunostained sections with rabbit anti-mouse SP-A, SP-B (EMD Millipore Corporation, Temecula, CA), SP-C (Santa Cruz Biotechnology, Dallas, TX), SP-D (Bioss, Woburn, MA), or a rabbit IgG isotope control (Cell Signaling Technology, Danvers, MA). Biotinylated horse anti-rabbit IgG (Vector Laboratories, Burlingame, CA) was used as a secondary Ab, which was detected using HRP-conjugated streptavidin (BD Biosciences, San Jose, CA) and using 3-amino-9-ethylcarbazole (Thermo Fisher Scientific, Waltham, MA) as substrate.

Bronchoalveolar lavage (BAL) was collected in situ by washing three times with 0.8 ml PBS (pH 7.4) intratracheally at a pressure of 15 cm H₂O. To determine alveolar macrophage diameter, we cytocentrifuged and stained BAL cells with PAS. Alveolar macrophage diameter was determined using ImageJ. To determine light scatter of alveolar macrophages, we stained BAL cells for flow cytometry using the following Abs: BV510-conjugated anti-CD11c (N418; BioLegend, San Diego, CA), PE-conjugated anti–Siglec F (E50-2440; BD Biosciences, Franklin Lakes, NJ). For the surfactant degradation assay, BAL cells were evenly divided into three samples and incubated at 37°C 5% CO₂ for 30 min and washed to remove nonadherent cells. A 1:1 mixture of rhodamine-labeled 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (Avanti Polar Lipids, Alabaster, AL) and bovine surfactant (Infasurf; Forest Pharmaceuticals, St. Louis, MO), or surfactant mixed with a vehicle control was added to the cells and incubated for 30 min at 37°C 5% CO₂. Surfactant uptake was determined after 30 min by washing cells three times to remove excess surfactant, trypsinizing, and measuring rhodamine fluorescence by flow cytometry. After an additional 4 h, a separate sample was measured to detect degradation. The degradation of labeled surfactant was quantified using the following equation: (30 min MFI − 4 h MFI)/(30 min MFI − control MFI), where MFI represents median fluorescent intensity.

mRNA was prepared from lungs or alveolar macrophages using the RNeasy mini kit (Qiagen, Valencia, CA) and reverse transcribed to cDNA using an iScript cDNA preparation kit (Bio-Rad Laboratories, Hercules, CA). Quantitative PCR was performed by SYBR Green incorporation using a CFX96 (Bio-Rad) thermocycler with the following primers: β₂-microglobulin (5′-CCCGCCTCACATTGAAATCC/GCGTATGTATCAGTCTCAGTGG-3′) SP-A (5′-GTGCACCTGGAGAACATGGA/TGGATCCTTGCAAGCTGAGG-3′), SP-B (5′-CCTCACAAAGATGACCAAGGA/CTGGCTCTGGAAGTAGTCAATAA-3′), SP-C (5′-AGAGGTCCTGATGGAGAGTCC/CATGAGCAGAGCCCCTACAA-3′), SP-D (5′-GGGCCAACAAGGAAGCAATC/GGCCATTCTCTGTTGGGCTA-3′), PU.1 (5′-CCATCGGATGACTTGGTTACTT/GTTCTCAAACTCGTTGTTGTGG-3′), PPARγ (5′-GCCCTTTACCACAGTTGATTTC/GTGGAGATGCAGGTTCTACTT-3′), ABCG1 (5′CAGACGAGAGATGGTCAAAGAG/CTGGTGGGCTCATCAAAGAA-3′).

ELISA plates were coated with 1 μg/ml mouse rGM-CSF (Shenandoah Biotechnology, Warwick, PA) overnight and then blocked with StabilGuard (SurModics, Eden Prairie, MN) for 1 h at room temperature. Biotinylated goat anti-mouse IgG Fc (Southern Biotechnology, Birmingham, AL) was used to detect GM-CSF autoantibodies in serial dilutions of serum followed by HRP-conjugated streptavidin (BD Biosciences, San Jose, CA). Absorbance of serial dilutions was compared with a standard curve generated from nonspecific IgG captured with a goat anti-mouse Ig Ab (Southern Biotechnology).

Neutralization of GM-CSF was determined using the GM-CSF–dependent FDC-P1 cell line (American Type Culture Collection, Manassas, VA). IgG was isolated from pooled serum from 15 mice using protein G–coupled magnetic beads following manufacturer’s instructions (Promega, Madison, WI). Serially diluted serum IgG from mice was added to FDC-P1 cell cultures grown in the presence of 0.01 ng/ml GM-CSF, and proliferation was measured after 3 d by addition of 10 μl 5 mg/ml MTT (Amresco, Solon, OH). Rat GM-CSF–specific IgG mAb (MP1-31G6; BioLegend) was used a positive control. Absorbance was measured at 540 nm after addition of DMSO.

Statistics were determined using GraphPad Prism 6 software. Kaplan–Meier survival curves were compared using a Mantel–Cox log-rank test. Unpaired Student t test was used to compare two means with equal variances, and Welch’s correction was applied when variances were unequal as determined by an F test. One-way ANOVA was used to compare more than two means with the exception of body mass data that were compared using a two-way ANOVA.

Mice deficient in RasGRP1 are autoimmune prone, with breaks in lymphocyte tolerance in both B and T cell compartments (23, 24). Mice were aged to understand the effects of autoimmunity. As controls, Rasgrp1-sufficient littermate mice achieved an average body mass of 35.5 and 42.0 g for females and males, respectively, and >90% of the mice survived past 1 y of age (Fig. 1A, 1B). By comparison, aged Rasgrp1-deficient mice exhibited significantly reduced body mass (25.1 and 30.5 g for females and males, respectively), and survival decreased beginning at 21 wk of age, with only 38% of the mice surviving beyond 1 y of age. Aged Rasgrp1-deficient mice also exhibited labored breathing, prompting us to evaluate lung function. Pulmonary compliance of aged control and Rasgrp1-deficient mice was assessed by generating pressure volume curves. Although aged control mice displayed pressure-volume curves with typical hysteresis within the normal volume range for mice, aged Rasgrp1-deficient mice had comparatively right-shifted pressure-volume curves characteristic of restrictive lung disease (Fig. 1C). Calculation of pulmonary compliance showed that, relative to aged control mice, aged Rasgrp1-deficient mice had reduced compliance (Fig. 1D; control versus Rasgrp1-deficient: 0.032 ± 0.011 versus 0.025 ± 0.004). Therefore, lung function was impaired in aged Rasgrp1-deficient mice.

To identify the nature of restrictive lung disease in aged Rasgrp1-deficient mice, we performed histological examination of lungs. Formalin-fixed, paraffin-embedded sections of lungs from young (2–3 mo) and aged (7–12 mo) control mice (both wild-type and Rasgrp1 heterozygous) demonstrated normal histology, with open alveolar spaces and thin septal walls (Fig. 2A). Lungs from young Rasgrp1-deficient mice also appeared normal. In contrast, analysis of lung sections from aged Rasgrp1-deficient mice indicated an eosinophilic proteinaceous accumulation in the alveolar space characteristic of PAP (Fig. 2A). Perivascular and peribronchial mononuclear cell infiltration was also apparent in the lungs of aged Rasgrp1-deficient mice (data not shown). To assess the extent of pathology, we quantified the amount of proteinaceous accumulation in the lungs (described in 2Materials and Methods). Although lung sections from control and young Rasgrp1-deficient mice had no measurable accumulation, analysis of lung sections from aged Rasgrp1-deficient mice with pathology showed variable accumulation ranging from 5.6 to 38.5% of the lung section (Fig. 2B). Notably, 8 of 11 aged Rasgrp1-deficient mice showed measureable protein accumulation by histology (Fig. 2B), suggesting that ∼70% of aged Rasgrp1-deficient mice develop PAP-like disease. Moreover, because no pathology was observed in young Rasgrp1-deficient mice, development of PAP-like disease is age dependent.

PAS-positive staining of the accumulated material in the alveoli is a key feature in PAP. Consistent with PAP, the accumulated material in lung sections from aged Rasgrp1-deficient mice were PAS-positive (Fig. 3A). Further, because impaired surfactant homeostasis in PAP patients leads to buildup of surfactants in the alveoli, lung sections were stained for SP-A, -B, -C, and -D by immunohistochemistry. Consistent with impaired surfactant homeostasis, the accumulated material in lung sections from aged Rasgrp1-deficient mice stained positively for all SPs (Fig. 3B).

Accumulation of pulmonary surfactants in PAP is reportedly caused by impaired degradation rather than increased SP expression (5, 25). To investigate whether increased SP expression contributes to accumulation of surfactants in aged Rasgrp1-deficient lungs, we measured SP gene expression using quantitative RT-PCR. No measurable differences in control and Rasgrp1-deficient mice in SP-A, -B, -C, or -D mRNA were observed, suggesting that the accumulation of pulmonary surfactants is not due to increased gene expression (Fig. 3C). Taken together with histological analysis of lung sections, these data support that Rasgrp1-deficient mice develop PAP in an age-dependent manner.

An important function of alveolar macrophages is to maintain surfactant homeostasis. During PAP, alveolar macrophage function is impaired, rendering them unable to degrade pulmonary surfactants (6). Consistent with impaired function, histological analysis of lung sections from aged Rasgrp1-deficient mice revealed enlarged PAS-positive alveolar macrophages in the alveoli of Rasgrp1-deficient mice in comparison with control mice (Fig. 3A). To quantify alveolar macrophage size, we isolated BAL cells from aged control and Rasgrp1-deficient mice and measured cell diameters. Whereas the mean diameter of alveolar macrophages from littermate control mice was ∼14 μm, those from Rasgrp1-deficient mice had a significantly larger mean diameter (control versus Rasgrp1-deficient: 13.9 ± 0.94 versus 17.2 ± 2.25 μm; Fig. 4A, 4B). Moreover, alveolar macrophages from Rasgrp1-deficient mice were larger and more granular than those from control mice using light scatter by flow cytometry (Fig. 4C). Therefore, alveolar macrophages from aged Rasgrp1-deficient mice are enlarged, consistent with an inability to degrade surfactants.

To measure directly whether alveolar macrophages from mice with PAP have impaired surfactant degradation, we incubated BAL cells with rhodamine-labeled surfactant. At indicated times, the amount of rhodamine label in alveolar macrophages was measured by flow cytometry (Fig. 4D). Surfactant degradation was quantified as the percent reduction in rhodamine signal intensity after normalizing to unstained cells (Fig. 4E). As expected, alveolar macrophages from all mice showed maximal uptake of surfactant after incubating for 30 min, consistent with previous reports (26). At 4-h incubation, alveolar macrophages from all control mice had reduced rhodamine fluorescence, indicating that they had degraded the labeled surfactants. By comparison, alveolar macrophages from aged Rasgrp1-deficient mice, although showing surfactant uptake comparable with control cells, were less able to degrade surfactant. Importantly, alveolar macrophages from young Rasgrp1-deficient mice were able to both take up and degrade surfactants similar to alveolar macrophages from control mice. These data demonstrate that alveolar macrophage function is impaired in Rasgrp1-deficient mice with PAP. Moreover, surfactant degradation by alveolar macrophages from young Rasgrp1-deficient mice is not impaired, indicating alveolar macrophage dysfunction is progressive in Rasgrp1-deficient mice.

Rasgrp1-deficient mice develop germinal centers that contain autoreactive B cells and have increased levels of serum anti-nuclear autoantibodies beginning at 2–3 mo of age (23). In aPAP patients, GM-CSF–specific IgG autoantibodies are elevated in serum and in BALF, and these Abs neutralize GM-CSF (10). To determine whether GM-CSF–specific autoantibodies were linked to development of PAP in Rasgrp1-deficient mice, we measured the concentration of GM-CSF–specific IgG in serum and BALF using ELISA. Aged control mice had a mean serum GM-CSF–specific IgG concentration of ∼10 ng/ml (10.4 ± 4.9; Fig. 5A). In contrast, aged-matched Rasgrp1-deficient mice had a mean serum GM-CSF–specific IgG concentration of 32.5 ng/ml, representing a 3-fold increase relative to control mice. Consistent with the lack of disease in younger Rasgrp1-deficient mice, 2- to 3-mo-old Rasgrp1-deficient mice exhibited GM-CSF–specific IgG concentration comparable with control mice (3.5 ± 2.6; Fig. 5C). The concentration of GM-CSF–specific IgG in BALF from aged Rasgrp1-deficient mice was also increased in comparison with levels measured from control mice (Fig. 5B). Therefore, GM-CSF–specific IgG autoantibodies are elevated in Rasgrp1-deficient mice in an age-dependent manner.

To test whether GM-CSF–specific autoantibodies from Rasgrp1-deficient mice have the capacity to neutralize GM-CSF, we tested isolated serum IgG for the ability to block proliferation of GM-CSF–dependent FDC-P1 cells (Fig. 5D). As a positive control, a monoclonal rat anti-mouse GM-CSF IgG prevented growth of FDC-P1 cells. Importantly, IgG from wild-type mice failed to inhibit GM-CSF–dependent growth. By comparison, serum IgG isolated from aged Rasgrp1-deficient mice completely inhibited GM-CSF–dependent growth. Therefore, aged Rasgrp1-deficient mice have elevated GM-CSF–neutralizing serum IgG, suggesting that progressive development of PAP in Rasgrp1-deficient mice is autoimmune.

In the absence of GM-CSF, alveolar macrophages have reduced expression of transcription factors PU.1 (6, 27) and PPARγ 28), as well as the ABCG1 transporter (29). Accordingly, if GM-CSF is neutralized by autoantibody in Rasgrp1-deficient mice, then PU.1, PPARγ, and ABCG1 mRNA levels would be reduced. To test the bioactivity of GM-CSF in Rasgrp1-deficient mice, we measured using quantitative RT-PCR the expression of PU.1, PPARγ, and ABCG1 in alveolar macrophages (GR1⁻, CD11c⁺ Siglec F⁺) sorted from the lungs of aged control and Rasgrp1-deficient mice (Fig. 6). Consistent with aPAP patients, the relative expression of PU.1, PPARγ, and ABCG1 was reduced in alveolar macrophages from aged Rasgrp1-deficient mice compared with aged control mice. Collectively, these data suggest that bioactivity of GM-CSF is neutralized in Rasgrp1-deficient mice.

CD278/ICOS–CD275/ICOSL interactions are necessary for conventional T cell–dependent humoral responses (30–32). GM-CSF autoantibodies are IgG, and therefore likely to be dependent on T cell help. To test whether disrupting B cell–T cell interactions via CD278-CD275 is required for the development of aPAP and GM-CSF autoantibodies, we crossed Rasgrp1-deficient mice to mice deficient in CD275. As before, histological lung sections from aged Rasgrp1-deficient mice (n = 4) revealed lesions characteristic of PAP, whereas those from aged control mice (n = 3) showed normal airways. Interestingly, 9 of 11 aged mice lacking both RasGRP1 and CD275 (Rasgrp1^−/− CD275^−/−) had normal lung histology and, therefore, showed no signs of PAP (Fig. 7A). Although lung sections from two aged Rasgrp1^−/− CD275^−/− mice exhibited enlarged alveolar macrophages, there was little evidence of surfactant accumulation compared with the pathology observed in aged Rasgrp1^−/− mice (data not shown). To determine whether the absence of PAP in Rasgrp1^−/− CD275^−/− mice was due to a reduced autoimmunity, the concentration of serum GM-CSF–specific autoantibodies were measured. Similar to earlier data (Fig. 5), the concentration of serum GM-CSF–specific autoantibodies from Rasgrp1-deficient mice was ∼40 ng/ml. In contrast, the concentration of serum GM-CSF–specific autoantibodies from Rasgrp1^−/− CD275^−/− mice was reduced and comparable with those observed in control mice (Fig. 7B). Therefore, disruption of CD278–CD275 interactions abrogated the development of aPAP.

aPAP is a rare autoimmune disorder caused by neutralizing autoantibodies to GM-CSF. Currently, there are no identified animal models for aPAP that spontaneously produce GM-CSF–specific autoantibodies. In this article, we show that aged Rasgrp1-deficient mice have an increase in morbidity and mortality associated with reduced pulmonary compliance. The histopathology in lungs of aged Rasgrp1-deficient mice is indistinguishable from human PAP with eosinophilic, PAS-positive, and SP-positive material within the alveoli of the lungs. Rasgrp1-deficient mice also have enlarged, foam-like alveolar macrophages that are impaired at degrading surfactants. Lastly, Rasgrp1-deficient mice exhibit an age- and CD275-dependent increase in GM-CSF–specific autoantibodies that are capable of neutralizing GM-CSF bioactivity. Collectively, the present studies strongly support that Rasgpr1-deficient mice develop autoantibody-mediated PAP.

GM-CSF–specific autoantibody is elevated concomitant with disease onset, and IgG from aged Rasgrp1-deficient mice neutralizes GM-CSF, suggesting that Ab-mediated neutralization of GM-CSF is the mechanism of PAP in Rasgrp1-deficient mice. Moreover, the onset of pathology in Rasgrp1-deficient mice presents at 6–8 mo of age, which compares with aPAP patients who typically present from 30 to 46 y of age (8). This contrasts with GM-CSF– and GM-CSFR–deficient mice (1, 2, 4) that develop disease as early as 2 mo of age. Similarly, congenital disease in humans resulting from GM-CSFR mutations also presents early in life (33–35). Importantly, Rasgrp1-deficient mice do not show signs of disease early in life; therefore, we consider it highly unlikely that PAP in aged Rasgrp1-deficient mice is caused by congenital deficiency. Thus, the kinetics of disease in Rasgrp1-deficient mice is most consistent with aPAP rather than congenital disease.

It should be noted that the concentration of GM-CSF–specific autoantibodies observed in Rasgrp1-deficient mice is lower than those established in human aPAP patients (18, 36, 37). This difference may be caused by assay conditions used to quantify GM-CSF–specific Abs in mouse. In our study, the concentration of GM-CSF–specific Abs was determined by comparison with a standard curve generated using total Ig capture Ab (i.e., not GM-CSF–specific capture). Therefore, the concentrations reported in this study are relative to detection of total, nonspecific IgG rather than GM-CSF–specific IgG.

Reduced mRNA levels of GM-CSF–dependent genes are consistent with GM-CSF neutralization as the cause of PAP in Rasgrp1-deficient mice. Alveolar macrophages from aPAP patients and GM-CSF–deficient mice have markedly reduced levels of ABC transporter ABCG1 (29), a molecule involved in lipid transport (38–40). The reduction of ABCG1 is attributed to lower levels of PPARγ as a consequence of GM-CSF deficiency (28, 41). Interestingly, mice deficient in ABCG1 display increased surfactants in the alveoli later in life; however, these mice also have elevated SP-D and reduced SP-B and SP-C mRNA expression consistent with disorders of surfactant production (42, 43). Disorders of surfactant production are considered distinct from PAP, which is defined by an inability to degrade surfactants (44). Notably, although alveolar macrophages from aged Rasgrp1-deficient mice have a reduction in ABCG1 expression, they also have a reduction in PPARγ mRNA. Further, expression of SPs in the lung is normal. Therefore, gene expression in alveolar macrophages from aged Rasgrp1-deficient mice is most consistent with GM-CSF neutralization and aPAP rather than with disorders of surfactant production. Neutralization of GM-CSF in aPAP also causes reduced levels of PU.1, a transcription factor regulated by GM-CSF (6, 12, 27). Enforced expression of PU.1 in alveolar macrophages from GM-CSF–deficient mice restores cell size, adherence, and surfactant degradation (6), demonstrating that PU.1 is sufficient to recoup alveolar macrophage function. Notably, alveolar macrophages from aged Rasgrp1-deficient mice have reduced PU.1 mRNA, again implicating GM-CSF neutralization as a cause of pathology in aged Rasgrp1-deficient mice.

GM-CSF deficiency or neutralization results in an increased susceptibility to pulmonary pathogens (8) including many fungal pathogens such as Cryptococcus neoformans (45), Pneumocystis carinii (46), and Histoplasma capsulatum (47). Consistent with autoantibody-mediated neutralization of GM-CSF, lung sections from aged Rasgrp1-deficient mice display positive staining for Grocott’s methenamine silver (data not shown). Although PAP can develop secondary to infection, particularly Pneumocystis spp. infection, secondary PAP remains unlikely because: 1) aged Rasgrp1-deficient mice with PAP and littermate controls tested negative in a serological test for Pneumocystis spp. (data not shown); 2) aged Rasgrp1-deficient mice that do not develop PAP pathology are frequently caged with aged Rasgrp1-deficient mice that develop PAP pathology; and 3) SP-B levels are decreased in a mouse model of secondary PAP (48), whereas SP-B expression is unchanged in aged Rasgrp1-deficient mice in comparison with control mice. It is more likely that autoantibody neutralization of GM-CSF contributes to impaired clearance by alveolar macrophages (6) and/or inhibition of pulmonary macrophage and dendritic cell maturation (7, 49). Future studies can elucidate whether fungal infections in aged Rasgrp1-deficient mice contribute to mortality because GM-CSF–deficient mice exhibit a normal life span (1).

CD278/ICOS–CD275/ICOSL interactions are required for humoral responses to T cell–dependent Ags (30). This study demonstrates that aged Rasgrp1-deficient mice also lacking CD275 produce normal amounts of GM-CSF–specific IgG Abs and do not develop PAP. Thus, these data support that PAP in Rasgrp1-deficient mice is autoantibody dependent. In MRL/lpr (50), collagen-induced arthritis (51), and diabetes/NOD mouse models (52), disruption of CD278–CD275 interactions mitigates severity of disease. CD278–CD275 interactions are considered key for the generation of Tfhs (53, 54). In recent work, we demonstrated that Rasgrp1-deficient mice have increased frequency and number of IL-17A–producing Tfh cells (55). Interestingly, Rasgrp1-deficient mice also lacking CD275 still produced Tfh17 cells, formed germinal centers, and produced similar titers of anti-nuclear autoantibodies compared with mice lacking only RasGRP1. However, whereas Tfh17 cells from Rasgrp1-deficient mice also produced IL-21, Tfh17 cells from Rasgrp1^−/− CD275^−/− mice did not. Because GM-CSF–specific autoantibodies in mice lacking both RasGRP1 and CD275 are reduced to levels comparable with control mice, we speculate that IL-21 production by Tfh17 cells is critical for GM-CSF–specific but not anti-nuclear autoantibodies. Formal testing of this hypothesis may be significant for human disease as well. To date, no studies have examined the importance of Tfhs, CD275–CD278 interactions, and/or IL-21 in human disease, although GM-CSF–specific autoantibodies from aPAP patients are heavily somatically hypermutated (16), suggesting a role for Tfh cells.

Mutations in BCR and TCR signaling pathways can facilitate breaks in B and T cell tolerance. For example, Rasgrp1 deficiency in mice leads to a break in tolerance in B cells early in development, at the transitional stage, and in the germinal center (23). These studies used an Ig knock-in transgene with broad specificity to ssRNA and nuclear Ags. It is unknown whether GM-CSF–specific B cells are regulated similarly. Notably, anti-nuclear and transgene-derived autoantibodies are elevated at 2–3 mo of age in Rasgrp1-deficient mice, an age at which the concentration of GM-CSF–specific autoantibodies is similar to those observed in control mice. As discussed earlier, elevation of GM-CSF–specific autoantibodies is not evident until 6–8 mo of age. Whether these age-dependent changes in anti–GM-CSF Ab levels are due to escape from discrete tolerance checkpoints is unknown. To our knowledge, no studies are examining whether BCR and TCR signaling mutations or specific tolerance checkpoints associate with aPAP in humans.

In conclusion, our data strongly suggest, to our knowledge, that Rasgrp1-deficient mice are the first rodent model for aPAP. Thus, Rasgrp1-deficient mice develop age-dependent PAP-like pathology in the lung that corresponds with an age-dependent increase in GM-CSF–specific neutralizing autoantibodies. This novel aPAP model may reveal important underlying immune mechanisms and identify new therapeutic targets for human aPAP.

We thank Kristin O’Donnell for assistance with measuring pulmonary compliance and Audrey Vasauskas for assistance with surfactant staining.

The authors have no financial conflicts of interest.