INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary alveolar proteinosis (PAP) is a rare parenchymal lung disease in which the alveoli fill with phospholipid (PL)- rich proteinaceous material that stains for periodic acid-Schiff (1). The alveolar-filling substance comprises surfactant proteins and PL, and is biochemically nearly identical to surfactant (2). The disease has an unpredictable clinical course, ranging from spontaneous remission to progressive fibrosis (2). Most adults with PAP have a "primary" or idiopathic form of the disease, although more rarely, a secondary form occurs in association with malignancies and immune dysfunction (e.g., human immunodeficiency virus infection) (7, 8). Exposures to certain environmental dusts, such as silica and aluminum, have been implicated in some cases of PAP (9, 10). The mainstay of treatment for PAP is symptomatic relief by removal of the proteinaceous material from the alveolar spaces through whole-lung lavage (4, 11, 12).

Although the specific cellular pathogenesis of PAP is unknown, recent observations in genetically altered mice have shed new light on potential mechanisms for PAP. Mice deficient in the gene for granulocyte-macrophage colony stimulating factor (GM-CSF) develop an alveolar process that is histologically identical to PAP (13, 14). The murine disease can be reversed by insertion of the GM-CSF gene into respiratory epithelium (15). Mice deficient in the -subunit of the GM-CSF receptor also develop an alveolar proteinosis-like disease, which can be reversed by bone marrow transplantation (16). These observations have led to speculation that either absolute GM-CSF deficiency or hyporesponsiveness of alveolar cells to GM-CSF is etiologic to the human form of PAP (13). Other genetically altered animals, including mice with severe combined immunodeficiency, develop a form of alveolar proteinosis (17). Further understanding of how these murine diseases pertain to human PAP may provide clues to the pathogenesis of PAP, thereby leading to pharmacologic therapy for this disorder.

Human data regarding the role of GM-CSF in PAP are limited. A partial remission was reported in one patient after systemic treatment with human recombinant GM-CSF (18). The investigators in this case also found that PAP patients have attenuation of the leukocytotic response after administration of GM-CSF (19). Another study of a single patient with PAP reported that ex vivo stimulation of alveolar macrophages (AM) with lipopolysaccharide increased GM-CSF messenger RNA (mRNA) in the patient's cells, but GM-CSF protein release was not detectable (20).

GM-CSF is a small glycoprotein produced by monocyte/ macrophages, endothelial cells, and epithelial cells (21, 22). This cytokine stimulates proliferation and maturation of monocytic and polymorphonuclear leukocytes and increases the effector functions of mature leukocytes (21). Binding of the macrophage receptor by GM-CSF results in enhanced production of immune mediators, including tumor necrosis factor (TNF)- and superoxide anion (23). Macrophages also secrete GM-CSF in response to stimulation with LPS and other immune effectors (24, 25). In humans, GM-CSF concentrations in serum and alveolar lining have been reported to be in the low pg/ml range. The concentration of GM-CSF in bronchoalveolar lavage fluid (BALF) can be increased in some inflammatory lung conditions such as sarcoidosis, asthma, chronic obstructive pulmonary disease, and usual interstitial pneumonitis (26).

Although the biologic role of GM-CSF in pulmonary homeostasis in humans is not clear, animal studies suggest that defects in GM-CSF production or response in humans could result in PAP. We measured serum and BALF GM-CSF levels in 10 adult human PAP patients to test the hypothesis that these patients had an absolute GM-CSF deficiency. We also tested the hypothesis that AM from patients with PAP are deficient in production of immune mediators in response to GM-CSF, possibly because of a defect in the GM-CSF receptor. Because GM-CSF and interleukin (IL)-3 share the -subunit of their receptor, we also measured IL-3 levels in BALF from PAP patients and compared them with values from control subjects. In addition, we determined whether or not AM from PAP patients release GM-CSF in response to TNF-. Our results show that adult PAP patients have detectable GM-CSF levels in plasma and BALF, and that their macrophages produce GM-CSF in response to stimulation. Moreover, AM from most patients with PAP show similar TNF- release in response to GM-CSF as those from control subjects.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Human Subjects

Ten consecutive patients who underwent therapeutic whole-lung lavage for symptomatic PAP were included in the study. The indications for whole-lung lavage were hypoxemia, frequent infection, intractable cough, or oxyhemoglobin desaturation with exercise. The study protocol was approved by the Duke University Institutional Review Board. The whole-lung lavage was done in the Duke Center for Hyperbaric Medicine, allowing sequential washouts of both lungs on the same day. Hyperbaric pressure was used if necessary to maintain arterial oxygen saturation (Sa_O2) during lavage of the second lung. The procedures were performed under general anesthesia, using independent lung ventilation with a dual-lumen endotracheal tube. While the left lung was ventilated with 100% O₂, the right lung was filled with warm 0.9% NaCl at 37° C. After gravity drainage of 500 ml of lavage fluid, serial exchanges of 500 ml each were made, using gravity drainage augmented by vigorous chest physical therapy. This was continued until the fluid was clear, which usually occurred after drainage of 15 Ls. Samples for this aspect of the study were obtained from the first 1.5 L of fluid, and 15 ml of blood was collected from the arterial line. Control samples were obtained from normal, healthy volunteers by bronchoscopic bronchoalveolar lavage (BAL) at the U.S. Environmental Protection Agency in Chapel Hill, NC, under an approved protocol. For this procedure, informed consent was obtained and topical 2% xylocaine was used to anesthetize the upper airway and larynx. A fiberoptic bronchoscope was passed into the lower airway, where 1% lidocaine was used for topical anesthesia. The lingula of the left lung was accessed, where the bronchoscope was wedged into a subsegment. BAL was performed with 120 ml of 0.9% saline injected serially in 20-ml aliquots, and the fluid was aspirated into a suction trap for studies. This procedure was identically repeated in the right middle lobe.

Blood samples from patients and volunteers were placed in heparinized tubes and centrifuged immediately at 1,000 × g to separate plasma. Plasma was frozen at 80° C until used for measurements of protein and cytokine levels. Lavage fluid from PAP patients and normal subjects was centrifuged on the day of collection at 1,000 × g to separate out the cells. The cell pellet was resuspended in RPMI-1640 medium containing antibiotics, and the cell were counted with a hemocytometer. The volume of cells suspension was adjusted to give 1 × 10⁶ AM/ml, and the cells were placed in primary culture under the experimental conditions outlined subsequently.

Materials

All reagents and chemicals were purchased from Sigma Chemicals unless otherwise specified.

Experimental Design

GM-CSF and TNF- levels were measured in BALF and plasma of control subjects and PAP patients with commercially available enzyme-linked immunosorbent assays (ELISAs) (R&D Systems, Minneapolis, MN). Protein levels were measured in BALF and serum through the Bradford reagent (Sigma). IL-3 levels in BALF of control subjects and PAP patients were measured with an ELISA (R&D Systems).

The test the ability of AM to produce GM-CSF, we cultured cells isolated from the lavage fluid of each subject and patient for 24 h in RPMI-1640 with LPS (100 µg/ml). The GM-CSF concentration in the media after 24 h was measured with ELISA. For each experimental condition, GM-CSF measurements were also made with macrophages cultured for 24 h without LPS to estimate basal GM-CSF production.

To study the ability of AM in PAP patients to respond to GM-CSF, cells from each subject and patient were cultured with GM-CSF (50 ng/ml) in the culture medium. After 24 h, the TNF- concentration in the medium was measured with ELISA. For each experimental condition, TNF- measurements were also made with macrophages cultured for 24 h in RPMI-1640 without GM-CSF, to estimate basal TNF- production.

Western Blot Analysis

BALF samples were mixed in one-half volume of cold lysis buffer (150 mM NaCl; 50 mM Tris, pH 7.6; 1% sodium dodecyl sulfate (SDS) (Bio-Rad, Richmond, CA); 3% NP-40; 5 mM ethylenediamine tetraacetic acid; 1 mM MgCl₂; 2 mM 1,3-dichloroisocoumarin; 2 mM 1,10-phenanthroline; and 0.5 mM E-64). This was mixed with an equal volume of double-strength Laemmli sample buffer (250 mM Tris-HCl (Bio-Rad), pH 6.8; 4% SDS; 10% glycerol; 0.006% bromphenol blue; 2% -mercaptoethanol). Electrophoresis was performed on a 13% polyacrylamide gel under both reducing and nonreducing conditions, with a minigel system (Hoefer Scientific Instruments, San Francisco, CA). All lanes were loaded with 10 µl of boiled sample, and electrophoresis was performed over a period of 1 h at 30 mA. The proteins were electrotransferred on a TE Series Transphor unit (Hoefer) to a polyvinylidene fluoride membrane (Millipore Corporation, New Bedford, MA) and blocked overnight at 4° C in Tris-buffered saline with 1% polyoxyethylene sorbitan monolaureate (TBST) containing 5% nonfat dry milk. On the following day, the membranes were washed for 30 min in TBST at room temperature. Western blots were performed with murine monoclonal antibodies directed against human GM-CSF (R&D Systems). Incubation with the primary antibody was performed for 1 h at room temperature in TBST with 5% milk at a dilution of 1:250. After multiple washes in TBST, the membranes were incubated with horseradish peroxidase-conjugated goat antimouse IgG secondary antibody (Jackson Laboratories, Bar Harbor, ME) at a 1:10,000 dilution in TBST with 5% milk. The membranes were then washed in TBST and the signal detected on Biomax film (Eastman Kodak, Rochester, NY), using the enhanced chemiluminescense kit (Amersham Corporation, Arlington Heights, IL). Western blotting was performed under identical protocol for all samples, under nonreducing conditions.

Statistical Analysis

Statistical analysis was performed with Student's t test. Values are reported as mean ± SD. Statistical significance for differences between PAP patients and controls was assumed at p  0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The demographic data for the PAP patients were consistent with previously reported statistics (3, 5). The mean age of the patients was 37 yr. There were nine males and one female patient, a ratio slightly higher than the reported male-to-female ratio of 4:1 (1, 3). Clinical data for individual patients are reported in Table 1. All of our patients had idiopathic acquired PAP without identifiable predisposing factors. The control subjects were demographically similar overall to the patients with a slightly lower mean age of 30 yr.

Plasma and BALF cytokine and protein values are shown in Table 2. Mean plasma GM-CSF levels as measured with ELISA were 38.4 ± 34.1 pg/ml (mean ± SD) in the patients. This value was significantly higher than that of the control subjects, which was (5.5 ± 2.3 pg/ml). The mean GM-CSF level in the BALF of the PAP patients was 7.2 pg/ml, compared with 1.1 pg/ml for the controls (p  0.05). IL-3 levels in the BALF of the controls and PAP patients were 5.7 ± 12.5 pg/ml (range: 0 to 35.5 pg/ml) in the PAP patients and 1.8 ± 4.0 pg/ml (range: 0 to 11.5 pg/ml) in the controls, with no significant difference between the two groups. Most of the PAP patients and normal subjects had IL-3 levels that were below the detection limit of the assay. Lavage protein was significantly increased in the patients, but plasma protein concentrations were similar to those of the control subjects.

Western blots were done with BALF from the controls and eight of the PAP patients. GM-CSF was below the detection limit in the BALF of all control subjects. Figure 1 shows Western blots of BALF from eight PAP samples that were preserved for electrophoresis under reducing conditions. GM-CSF was detected at approximately 28 kD as a broad band in all patients. Under nonreducing conditions, Western blotting revealed two distinct bands in all patients. A high-molecular-weight band was present at 120 to 200 kD and another band was present at 28 kD, as in the reduced gel. In all cases, the low-molecular-weight band was heterogeneous, in accord with glycosylation or other posttranslational modification of the band protein.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Western blot of GM-CSF in BALF of PAP patients (n = 8 patients). Under reducing conditions, Western blotting was performed on BALF from eight patients, using antihuman GM-CSF antibody. Each lane was loaded with a 10-µl sample. GM-CSF is detectable as a broad band at approximately 28 kD in all patients.

GM-CSF concentrations in the media of AM cultured for 24 h with or without LPS are shown in Figure 2. CM-CSF levels are plotted from macrophages from each PAP patient and normal control subject cultured without and with LPS. The values for the control subjects are shown in Figure 2A and those for the PAP patients are shown in Figure 2B. The AM of PAP patients showed higher basal GM-CSF release after culture without LPS than did the AM of control subjects. AM from both groups showed similar GM-CSF release after culture with LPS. In each group, AM from three individuals did not show increased GM-CSF release after culture with LPS. The mean GM-CSF concentration in media of unstimulated AM after 24 h was 1.4 ± 0.1 pg/ml for the controls and 9.0 ± 4.1 for the PAP patients (p  0.05), correlating with higher basal GM-CSF production by the patients' AM. After 24 h in culture with LPS (100 mg/ml), mean GM-CSF levels in the culture media were similar, at 34.4 ± 9.8 pg/ml for the control subjects and 34.6 ± 9.4 pg/ml for the PAP patients (p = NS).

View larger version (14K): [in this window] [in a new window]   Figure 2.   GM-CSF concentrations in the media of AM cultured for 24 h with or without LPS. (A) Individual GM-CSF levels in control subjects without and with LPS. (B) GM-CSF levels in PAP patients. The unstimulated AM of PAP patients showed higher GM-CSF production after 24 h than did control AM. Mean GM-CSF production in unstimulated cells was 1.4 ± 0.09 pg/ml for control subjects and 9.0 ± 4.093 pg/ml for PAP patients (p  0.05). AM from both groups showed similar GM-CSF production after culture with LPS. After 24 h in culture with LPS (100 ng/ml), mean GM-CSF levels in culture media were similar, at 34.4 ± 9.8 pg/ml for control subjects and 34.6 ± 9.4 pg/ml for PAP patients (n = 10 subjects per group).

As an indicator of the ability of the GM-CSF receptor of AM to respond to GM-CSF, TNF- production of macrophages from each PAP patient and normal subject was measured after 24 h of culture with and without CM-CSF (50 ng/ ml). Figure 3 shows individual data from control subjects and PAP patients, along with mean values. Cells from both control and PAP patients cultured without GM-CSF showed no detectable TNF- after 24 h in culture. AM of all control subjects and PAP patients responded to GM-CSF by producing TNF-, but the overall response of AM from PAP patients tended to be lower than that of controls. The AM of one patient produced minimal TNF- after culture with GM-CSF. The mean TNF- level for AM of PAP patients after culture with GM-CSF was significantly lower that than of AM from the controls' (279 ± 24 pg/ml versus 332 ± 5 pg/ml, p  0.05).

View larger version (10K): [in this window] [in a new window]   Figure 3.   TNF- production of macrophages from control subjects and PAP patients after culture with and without GM-CSF (50 ng/ml). Individual data are shown for control subjects (open squares) and PAP patients (open circles) (solid bar = mean values). Cells from control subjects and PAP patients cultured for 24 h without GM-CSF showed no detectable TNF-. AM from control subjects and PAP patients responded to GM-CSF by producing TNF-, but the response of macrophages of the PAP patients was variable, and cell from one PAP patient failed to increase TNF- production. After culture of macrophages for 24 h with GM-CSF, mean TNF- levels in the media were 278.9 ± 23.9 pg/ ml for PAP patients and 332.1 ± 4.5 pg/ml for controls (p  0.05). Some individuals in each group failed to show increased TNF- after GM-CSF stimulation (n = 10 subjects per group).

The correlation between BALF GM-CSF levels and GM-CSF released over 24 h by unstimulated AM from PAP patients is shown in Figure 4. There was a very high positive correlation in the PAP patients (r² = 0.97) between GM-CSF levels in BALF and in macrophage media after culture with LPS. There was no correlation between plasma and BALF GM-CSF levels in PAP patients, whereas there was a strong correlation between BALF and plasma TNF- levels (r² = 0.85). In PAP patients, correlation between BALF GM-CSF and TNF- production by cultured AM was weak. There was also no correlation in PAP patients between TNF- levels in BALF and TNF- production by AM cultured with GM-CSF.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Plot of BALF GM-CSF levels against GM-CSF produced by cultured AM of PAP patients. There was a high correlation (r2 = 0.97) between GM-CSF in BALF and GM-CSF in media of macrophages cultured for 24 h without cytokines. There was no correlation between serum and BALF GM-CSF levels (n = 10 subjects per group).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In our group of adult patients with active PAP, GM-CSF levels in plasma and BALF at the time of therapeutic lavage were higher than those in control subjects. GM-CSF was detectable by Western blotting in the unconcentrated lavage fluid of PAP patients and not in lavage fluid of control subjects. In addition, AM from the PAP patients showed higher basal GM-CSF release in culture than did control AM. The AM from most patients with PAP also had similar responses to GM-CSF as those of control subjects as measured through TNF- release, although the overall response was decreased in the patient group. The study data demonstrate that GM-CSF protein deficiency is not etiologic in primary PAP, and indicate the presence of a functional GM-CSF receptor in most adult patients with the disease. The possibility of loss of biologic activity of GM-CSF, however, has not been excluded.

Normally, surfactant proteins and PLs are produced and secreted by alveolar type II cells. The clearance of surfactant material from alveolar spaces is a function of both alveolar type II cells and AM (27). Several mechanisms for the pathogenesis of PAP have previously been proposed, including overproduction of surfactant protein by type II pneumocytes and impaired clearance of surfactant by macrophages or type II pneumocytes. Because metabolic studies of PAP patients have indicated normal surfactant formation and clearance by alveolar type II cells, a defect in surfactant clearance by AM appears more likely (2). Several studies have reported defects in phagocytic mechanisms in AM isolated from PAP patients (28). Macrophages isolated from patients with PAP have lysosome dysfunction (29); also, phagocytic function and response to zymogen stimulation decrease in macrophages from normal volunteers after incubation with cell-free fluid from PAP patients (28, 30). Potential causes of phagocytic dysfunction in these macrophages include mechanical interference with lysosome-phagosome fusion, due to "bloating" of the cells with surplus surfactant; intrinsic macrophage defects; or the presence in alveolar material of local mediators that interfere with normal macrophage function (28).

The similarity of the pathologic changes in the lungs of GM-CSF-deficient mice and human PAP patients has generated interest in GM-CSF deficiency as an etiology for PAP. Our studies appear to exclude absolute GM-CSF deficiency as the mechanism for PAP in a typical group of adult patients. Interpretation of case reports of apparent responses of the human disease to systemic recombinant GM-CSF treatment (18) is difficult because of the high spontaneous remission rate that occurs in PAP. Also, a response to pharmacologic doses of GM-CSF does not prove its functional deficiency in PAP, because GM-CSF is clearly important in surfactant homeostasis, and difficulty in clearing surfactant from the lung might be met by an augmented GM-CSF response.

We also investigated the hypothesis that PAP patients have a defect in the receptor for GM-CSF on the AM. Seymour and colleagues (19) reported that patients with PAP did not develop peripheral leukocytosis in response to GM-CSF, suggesting altered responsiveness of hematopoietic progenitor cells to this cytokine. By functional assessment, the majority of our patients appear to have had intact GM-CSF receptors on their AM, as indicated by the ability of these cells to respond to GM-CSF by producing TNF-. Of interest, however, was significant attenuation of the response of AM to GM-CSF in three patients. The patients whose AM did not respond strongly to GM-CSF could represent a subset of patients with a deficient or defective GM-CSF receptor. This variability in macrophage responses is consistent with the clinical heterogeneity of alveolar proteinosis. A decreased effector response by macrophages could also explain the predisposition of PAP patients to infection with Mycobacterium avium complex and other atypical organisms (31). It must also be noted that we measured a single cellular effector function in response to GM-CSF, and that it therefore remains possible that other responses related to GM-CSF signal-transduction pathways are deficient in PAP. Moreover, since the TNF- response is not exclusive for GM-CSF, more complex defects in TNF- production and release cannot be excluded in the etiology of PAP.

Because we investigated the GM-CSF response in only one specific resident cell type in the lung, the possibility arises that GM-CSF receptor function is abnormal in other cell types (e.g., type II epithelial cells). This could account for the high GM-CSF levels in BALF of PAP patients. The study by Seymour and colleagues showing an attenuated hematopoietic response to GM-CSF in PAP patients also suggests this possibility (19). A defective receptor for GM-CSF is an attractive hypothesis in PAP for several reasons. Macrophages express a GM-CSF receptor that consists of two glycoprotein subunits, designated  and . The  subunit is shared with the receptors for IL-3 and IL-5 (34). Although only small numbers of this high-affinity receptor are necessary to activate GM-CSF signal transduction (34), small changes in receptor-subunit availability could unbalance these cytokine pathways. However, we were unable to detect high levels of IL-3 in the BALF from our PAP patients, which would have supported this hypothesis.

A potential consequence of GM-CSF receptor hyporesponsiveness could be feedback hypersecretion of GM-CSF. Such a phenomenon could explain increased GM-CSF levels in PAP patients. Lungs from mice that overexpress GM-CSF have alveolar epithelial type II cell hyperplasia and pulmonary fibrosis (35, 36). If such data are relevant to humans, persistent increases in GM-CSF in some PAP patients could explain the pulmonary fibrosis that develops in a subset of patients with chronic disease.

Biologic inactivation of GM-CSF protein is another explanation that could reconcile elevated GM-CSF levels in PAP patients with the pulmonary abnormalities in GM-CSF-deficient mice. For instance, differences in glycosylation are known to affect release of GM-CSF from the cell, and its biologic activity (37, 38). Variable glycosylation of the 15-kD gene product produces a glycoprotein with a final molecular weight of approximately 22 to 26 kD. The deglycosylated GM-CSF protein has increased biologic activity as compared with glycosylated forms (37, 38). In the lavage fluid of PAP patients, we detected a broad band of GM-CSF, consistent with a glycosylated form of this protein. There was no detectable band in the lavage fluid of normal controls for comparison. Examination of the biologic activity of this protein in PAP patients would be of future interest, with attention to differences in glycosylation between patients with PAP and normal controls.

Recently, a GM-CSF-neutralizing factor was reported in the BALF of PAP patients, and was postulated to be an autoantibody (39, 40). This factor in BALF suppressed the cellular response to human recombinant GM-CSF by about 50% (39). The investigators who made this finding subsequently identified a binding factor in BALF and sera of PAP patients as an immunoglobulin that was not present in normal individuals or patients with other lung diseases (40). Through Western blotting, we also detected a high-molecular-weight band in the BALF of PAP patients under nonreducing conditions, which could represent GM-CSF complexed with another protein. However, unbound GM-CSF in the lavage fluid was readily detected under both reducing and nonreducing conditions. This suggests that free GM-CSF is present in the BALF of PAP patients, and argues against substantial inactivation of GM-CSF by a binding factor.

In conclusion, we report that a symptomatic group of adult PAP patients had readily detectable GM-CSF levels in their BALF and blood plasma. In a subset of these patients, a defective GM-CSF receptor on AM has not been excluded. In addition, the biologic activity of the glycoprotein present in the lung in PAP remains to be studied. These data suggest that administration of GM-CSF to PAP patients should be studied carefully, particularly because the long-term effects of pharmacologic doses of GM-CSF on the lung are unknown.