Idiopathic Pulmonary Alveolar Proteinosis as an Autoimmune Disease with Neutralizing Antibody against Granulocyte/Macrophage Colony-Stimulating Factor

We obtained BALF from 11 patients with I-PAP, 2 patients with S-PAP, 14 patients with other lung diseases (3 cases each of sarcoidosis and collagen vascular lung disease; 2 cases each of interstitial pneumonitis, hypersensitive pneumonitis, and eosinophilic pneumonia; and 1 case each of bronchiolitis obliterans organizing pneumonia and lung cancer), and 53 normal subjects. Diagnosis of PAP was based on biochemical analysis of BALF and histopathological findings of lung biopsy. The I-PAP patients had no history of hematological disorders, infectious diseases, or toxic inhalation. Two S-PAP patients were in the blast phase of chronic myelogenous leukemia. 5 of 11 I-PAP patients, 1 of 2 S-PAP patients, and 30 normal subjects underwent venipuncture to obtain sera at various periods after bronchoalveolar lavage. Written informed consent to participate in this study was obtained from all subjects.

Recombinant human (rh)GM-CSF was provided by Kirin Brewery Co., Ltd. (Takasaki, Japan), and ¹²⁵I–Bolton Hunter–labeled rhGM-CSF was purchased from NEN Life Science Products. rhIL-3 was purchased from R & D Systems, Inc. Biotinylated or nonlabeled rat anti–human GM-CSF mAb (BVD2-21C11 and BVD2-23B6, respectively) was purchased from PharMingen. Peroxidase-labeled rabbit anti–human IgA, -D, -E, -G, and -M polyclonal antibodies were purchased from DAKO Corp. A GM-CSF– or IL-3–dependent cell line, TF-1 12, was provided by Dr. Kitamura (The Institute of Medical Science, The University of Tokyo, Tokyo, Japan).

Proteins in BALF or sera were subjected to SDS-PAGE using gradient gel (2–15%) under nonreducing conditions at 30 mA constant current for 150 min. Separated proteins were transferred electrophoretically to a polyvinylidene fluoride (PVDF) membrane at 12 V constant voltage for 75 min. The membrane was fixed with 10% (vol/vol) acetic acid and 50% (vol/vol) methanol, stained with Coomassie brilliant blue solution, washed with methanol, and treated with a blocking reagent (PBS supplemented with 1% [wt/vol] BSA and 0.1% [vol/vol] Tween 20) overnight at 4°C. The membrane was then incubated with 0.16 nM ¹²⁵I–GM-CSF for 1 h at room temperature. After washing, the membrane was exposed to x-ray film for 4 d.

The method of this assay was described previously 11. In brief, after coating a micro-ELISA plate with anti–human GM-CSF mAb (BVD2-23B6), rhGM-CSF and BALF were incubated in the wells. rhGM-CSF bound to BVD2-23B6 antibody was detected using biotinylated anti–human GM-CSF mAb (BVD2-21C11) and streptavidin conjugated to horseradish peroxidase. Color was developed using tetramethylbenzidine, and the absorbance was measured at 450 nm using a microplate spectrometer (model 3550; Bio-Rad Labs.). Binding activity was calculated with the following equation: binding activity = 1 − detected GM-CSF (ng/ml)/applied GM-CSF (ng/ml).

BALF was centrifuged at 1,000 g for 5 min. To remove lipoid materials, the supernatant was mixed vigorously with an equal volume of 1-butanol, and the mixture was centrifuged at 1,000 g for 5 min. After removing the butanol layer, the procedure was repeated. The aqueous layer was dialyzed against 10 mM ammonium acetate, pH 7.0, and lyophilized. Delipidated BALF was purified by using HiTrap SP column, a cation exchange column (equilibrated with 20 mM ammonium acetate, pH 6.0, and eluted with a linear sodium chloride gradient); HiTrap Q column, an anion exchange column (equilibrated with 20 mM Tris-HCl, pH 9.0, and eluted with a linear sodium chloride gradient); Superose 12 column, a gel filtration column (equilibrated with PBS containing 0.1% [vol/vol] NP-40 and eluted with the same buffer); RESOURCE Q column, an anion exchange column (equilibrated with 20 mM Tris-HCl, pH 9.0, and eluted with a linear sodium chloride gradient); and RESOURCE S column, a cation exchange column (equilibrated with 20 mM ammonium acetate, pH 6.0, and eluted with a linear sodium chloride gradient). All of these columns were from Pharmacia Biotech.

Delipidated BALF was applied on recombinant protein A affinity column (Pharmacia Biotech) equilibrated with 20 mM sodium phosphate, pH 7.0. Ig bound to the column was eluted by pH gradient (pH 3.0–7.0).

NH₂-terminal sequencing of protein was performed by the phenyl isothiocyanate method using the HP G1005A NH₂-terminal protein sequencing system (Hewlett-Packard Bioscience Products).

Various concentrations of Ig purified from BALF of an I-PAP patient (39–5,000 ng/ml) were transferred to micro-ELISA plates coated with 1 μg/ml rhGM-CSF, and the plate was kept at room temperature for 1 h. After washing, 0.3 μg/ml of peroxidase-labeled anti–human IgA, -D, -E, -G, or -M polyclonal antibody was added to each well and incubated at room temperature for 1 h. Color was developed using tetramethylbenzidine, and the absorbance was measured at 450 nm.

The method of this assay was described previously 13. In brief, TF-1 cells (2 × 10⁴ cells/well) were incubated for 3 d with 1 ng/ml of rhGM-CSF or rhIL-3 and 1 μg/ml of Igs purified from BALF of an I-PAP patient. To the culture, 5 μg/ml of 3-[4,5-dimethylthiazol-2yl]-2,5-diphenyltetrazolium bromide (MTT; Sigma Chemical Co.) was added and incubated. After formation of formazan crystals, isopropanol/HCl was added to dissolve the crystal, and the absorbance was measured at 595 nm.

Occurrence of the GM-CSF binding factor in BALF supernatant was studied from 80 donors, including 11 I-PAP patients. As shown in Fig. 1, blot assay with ¹²⁵I–GM-CSF gave a single band with a molecular mass of 180 kD in all I-PAP cases examined. In contrast, no band was detected in S-PAP patients, normal subjects, or patients with other lung diseases such as sarcoidosis, collagen vascular lung disease, interstitial pneumonitis, hypersensitive pneumonitis, and eosinophilic pneumonia.

The binding factor in BALF was purified by cation- and anion-exchange chromatography and gel filtration chromatography (Fig. 2 A). For evaluation of binding activity, a competition assay of GM-CSF binding to the mAb (BVD2-23B6) in ELISA was used. The purified protein showed a single band of 180 kD on SDS-PAGE under nonreducing conditions and two bands of 28 and 57 kD under reducing conditions (Fig. 2 B). ¹²⁵I–GM-CSF binding to the purified 180-kD protein was confirmed by blot assay (Fig. 2 C), and bound ¹²⁵I–GM-CSF was released when treated with citrate buffer, pH 2.0 (data not shown).

The 57-kD band was electroblotted onto a PVDF membrane and sequenced directly. The NH₂-terminal sequence of the 57-kD protein was EVQLVESGGGLVQPGGSLRL, identical to the NH₂-terminal sequence of human Ig H chain. The data suggest that the GM-CSF binding protein in the BALF is an antibody. To show that the binding factor is actually an antibody, Ig in BALF from an I-PAP patient was isolated using a recombinant protein A column. Ig eluted from the column by changing pH gradient showed ¹²⁵I–GM-CSF binding activity (Fig. 2 D). Specificity of binding to GM-CSF was demonstrated by effective competition of ¹²⁵I–GM-CSF binding with nonradioactive GM-CSF (Fig. 2 E). ¹²⁵I–GM-CSF binding activity was not affected by the presence of nonspecific human Ig (data not shown).

To confirm the bioactivity of isolated Ig, its inhibitory effect on growth of the TF-1 cell line was examined using the MTT assay. Growth of TF-1 cells is dependent on either GM-CSF or IL-3. The Ig purified from BALF inhibited GM-CSF–dependent growth of TF-1 cells but not IL-3–dependent growth (Fig. 3). The results indicate that the antibody inhibits growth of this cell line by neutralizing the bioactivity of GM-CSF.

The isotype of the antibody was determined by antigen capture assay. The Ig captured by GM-CSF reacted with only anti–human IgG, whereas no other antiisotype antibody reacted, indicating that the isotype of the antibody is IgG (data not shown).

To know whether the I-PAP patients have circulating antibody against GM-CSF, we performed a blot assay with ¹²⁵I–GM-CSF of sera from 36 donors, including 5 I-PAP patients, 1 S-PAP patient, and 30 normal subjects who underwent bronchoalveolar lavage during this study. Serum samples from all I-PAP patients showed a single 180-kD band, whereas no such band was detected in samples from the S-PAP patient and normal subjects (Fig. 4). Interestingly, the band was observed even in those from three patients who had entered remission and whose chest x-rays showed no opacity at the time of study. The isotype of the antibody in sera was also IgG (data not shown). These results indicate that the antibody is not limited to the lung but exists systemically in I-PAP patients.

We have shown in this paper that autoantibody against GM-CSF is present in the lungs and sera of I-PAP patients but not in those of S-PAP patients, normal subjects, or patients with other lung diseases. Because GM-CSF is a key factor for maintaining the differentiation and proliferation of macrophages, dysfunction of AMs due to the neutralization of GM-CSF bioactivity by an autoantibody is a plausible explanation for the pathogenesis of human I-PAP. In fact, AMs from patients with active acquired PAP are known to be functionally impaired 14,15. Furthermore, one type of congenital PAP was known to be associated with a defect of the GM-CSF receptor, and in murine models, knockout of the gene for GM-CSF or its receptor induced PAP-like disease. The disease in these murine models was corrected by transgenic expression of GM-CSF in the pulmonary epithelia or bone marrow transplantation from wild-type mice 16,17.

It is increasingly understood that sera in some healthy and diseased individuals contain autoantibodies against cytokines. Natural antibodies have been reported against rh IL-1β, IL-2, IL-8, and TNF-α 18,19,20,21. Most of these autoantibodies were neutralizing antibodies and interfered with the binding of cytokines to receptors by simple competition 19,20,21. Recently, it was reported that anti–GM-CSF autoantibody was frequently detected in pharmaceutically prepared human IgG 22 and that 0.3–2% of sera had a low titer of anti–GM-CSF antibody 22,23. However, it was not reported that autoantibody against GM-CSF is associated with any disease or symptoms. Our results strongly suggest that I-PAP is an autoimmune disease with neutralizing autoantibody against the cytokine. As mentioned above, previous work in the patient case report and in mice has established that impaired production or action of GM-CSF can cause PAP. It is, therefore, reasonable to ascribe causality to the presence of autoantibody in the cases presented here, frequently and solely in the patients with I-PAP. This is, therefore, the first disease for which it can be argued that the cause is an autoantibody to a cytokine.

It is not clear at the time of this writing why autoantibody is generated in I-PAP patients. One possibility is that the autoantibody was provoked by a cross-reactive antigen, possibly from infectious pathogen(s). Alternatively, the autoantibody was generated to endogenously regulate GM-CSF bioactivity.

Although the functional consequences are predominantly manifest within the lung, our data suggest that the pathophysiologic disorder of I-PAP may originate from the systemic generation of autoantibody against GM-CSF. In this regard, recent clinical trials of GM-CSF administration to acquired PAP are intriguing 24. The hematopoietic response to GM-CSF in the patients was attenuated. The impaired response was not due to altered expression of GM-CSF receptor on PBMCs. It is possible that the administrated GM-CSF was neutralized by anti–GM-CSF antibody before it reached the target organs.

Finally, our observations have therapeutic implications. Whole lung lavage provides temporary remission in most cases, but additional lavage is required in more than 50% of patients 25. Measurement of the autoantibody in the lung after whole lung lavage may identify patients who require further therapy or who have risk for recurrence.

The authors thank T. Tomita, T. Nukiwa, A. Kurashima, T. Sakai, T. Abe, Y. Abe, T. Sugie, M. Niijima, K. Watari, M. Fujisawa, S. Kudoh, and A. Azuma for providing patient samples and clinical information. We are very grateful to W. Rom and M. Weiden for critical review of the manuscript and K. Arai, K. Akagawa, and S. Mori for valuable discussions.