Granulocyte/macrophage-colony-stimulating factor autoantibodies and myeloid cell immune functions in healthy subjects

Abstract

High levels of granulocyte/macrophage-colony-stimulating factor (GM-CSF) autoantibodies are thought to cause pulmonary alveolar proteinosis (PAP), a rare syndrome characterized by myeloid dysfunction resulting in pulmonary surfactant accumulation and respiratory failure. Paradoxically, GM-CSF autoantibodies have been reported to occur rarely in healthy people and routinely in pharmaceutical intravenous immunoglobulin (IVIG) purified from serum pooled from healthy subjects. These findings suggest that either GM-CSF autoantibodies are normally present in healthy people at low levels that are difficult to detect or that serum pooled for IVIG purification may include asymptomatic persons with high levels of GM-CSF autoantibodies. Using several experimental approaches, GM-CSF autoantibodies were detected in all healthy subjects evaluated (n = 72) at low levels sufficient to rheostatically regulate multiple myeloid functions. Serum GM-CSF was more abundant than previously reported, but more than 99% was bound and neutralized by GM-CSF autoantibody. The critical threshold of GM-CSF autoantibodies associated with the development of PAP was determined. Results demonstrate that free serum GM-CSF is tightly maintained at low levels, identify a novel potential mechanism of innate immune regulation, help define the therapeutic window for potential clinical use of GM-CSF autoantibodies to treat inflammatory and autoimmune diseases, and have implications for the pathogenesis of PAP.

Figure 1

Presence of GM-CSF autoantibodies in healthy subjects. (A) Total IgG was isolated from the serum of healthy control subjects (HC) or patients with PAP (PAP) or from pharmaceutical grade IVIG by protein G chromatography and subjected to ultrafiltration under acidic conditions to remove bound GM-CSF. GM-CSF autoantibodies were then isolated by GM-CSF affinity chromatography and evaluated by far-Western analysis probed with ¹²⁵I-GM-CSF. Shown are Coomassie Blue–stained electrophoresis gels (bottom panels) and corresponding far-Western blots (top panels). Each numbered lane represents the corresponding bound (left panels) and unbound (right panels) chromatography fractions from 1 subject or sample. (B) Serum GM-CSF–binding proteins from 6 healthy human subjects were fractionated individually on gels as in panel A, and proteins in the 180-kDa band from each were extracted, subjected to liquid chromatography and tandem mass spectroscopy, and the results evaluated by comparison to Mascot database as described under “Methods.” Only matches with a probability-based Mowse score greater than 64 (indicating a P value < .05) were considered in the analysis. The percentages of immunoglobulin (Ig) and non-Ig peptides among the top 50 peptide fragment matches identified for each sample are shown. (C) GM-CSF autoantibodies were isolated by GM-CSF affinity chromatography from serum of healthy subjects (HC; n = 10), patients with PAP (PAP, n = 4), or from pharmaceutical IVIG (n = 1 clinical grade vial), and the percentage of IgG subtypes was measured by ELISA. (D) Serum GM-CSF autoantibody concentrations in healthy subjects (HC), patients with PAP (PAP), or IVIG (reconstituted at 9.94 mg/mL in PBS). Serum GM-CSF autoantibody levels in healthy subjects (median (interquartile range [IQR]) = 1.04 [0.63-1.7] μg/mL, n = 72) were lower than in patients with PAP (median [IQR] = 59.8 (27.4–116.5) μg/mL; n = 21; P < .001). Median values (HC, PAP) are indicated by a horizontal bar. (E) Serum GM-CSF autoantibody levels in males (n = 15) and females (n = 57). Data are shown as whisker plots indicating the interquartile range (upper and lower borders of box), the 90th and 10th percentile (error bars), the 95th and 5th percentile (upper and lower open symbols), the median (solid horizontal line in box), and mean (dashed line in box) values of GM-CSF autoantibody levels. (F) Serum GM-CSF autoantibody levels in healthy women (△) and men ( □ ) of various ages (n = 72). Regression analysis did not reveal a significant correlation GM-CSF autoantibody levels with age (R² = 0.08).

Figure 2

Concentration of free and IgG-bound GM-CSF in human serum. (A) Total IgG was isolated individually from the sera of healthy subjects (HC) or patients with PAP (PAP) using protein G and evaluated by Western blotting to detect GM-CSF (top panels) or IgGκ (κ) and -λ (λ) light chains (as a loading control, bottom blots). Each lane represents one subject. (B) Detection of free GM-CSF and autoantibody-bound GM-CSF in serum. A set of “standard” samples composed of recombinant human GM-CSF (Leukine) at various concentrations ranging from 0 to 30 ng/mL were prepared in mouse serum in the absence (○) or presence (▽) of purified human GM-CSF autoantibodies (30 μg/mL). Standard samples were diluted 1/30 with 1% BSA in PBS and then GM-CSF was measured using a commercial human ELISA kit (R&D Systems) as directed by the manufacturer. (C) Use of a novel ELISA (SDS-HD ELISA, see “Methods”) to quantify GM-CSF in PBS in the absence or presence of GM-CSF autoantibody (1 μg/mL) and in the absence or presence of a pretreatment with SDS and heat denaturation. Each bar represents the mean of duplicate determinations for 1 of 4 separate experiments with similar results. (D) GM-CSF level evaluated using a novel human GM-CSF ELISA as described in “Methods.” Symbols represent the same samples and conditions as described in the legend to panel B above. GM-CSF was detectable in the absence of GM-CSF autoantibody (○), undetectable in the presence of GM-CSF autoantibody in the absence of SDS-HD pretreatment (▽), and detection was restored in the presence of GM-CSF autoantibody by SDS-HD pretreatment (△). (E) Free GM-CSF (▨) or total GM-CSF (free and autoantibody-bound; ■) were measured in sera of healthy subjects (HC) or patients with PAP (PAP) using a commercially available ELISA or the SDS-HD ELISA, respectively, as described in “Methods.” Total serum GM-CSF levels in HC and PAP were not different (3048 ± 484, n = 11; PAP 2360 ± 668, n = 5; respectively, P = .43).

Figure 3

Regulation of GM-CSF signaling by GM-CSF autoantibodies in healthy subjects and patients with PAP. (A) The serum GM-CSF–neutralizing capacity of GM-CSF autoantibodies was measured using the TF-1 cell-proliferation assay. Equal volumes (30 μL) of serum from healthy subjects (△) or a PAP patient (■) (as a positive control), IVIG reconstituted at physiologic concentration (◇), or culture media (●) (as a negative control) were evaluated. The neutralizing capacity of purified GM-CSF autoantibodies from HC serum or IVIG (dashed line) was intermediate between that of autoantibodies isolated from serum the patient with PAP, which contains high concentrations of GM-CSF autoantibody, and control media, which contains none. (B) Neutrophils isolated from healthy subjects were incubated with various concentrations of GM-CSF affinity-purified autoantibodies isolated from IVIG or with control antibody (1 μg/mL) and stimulated with 10 ng/mL GM-CSF for 15 minutes and total and phosphorylated STAT5 (pSTAT5) was measured by immunoblotting. (C) The signaling activity of free and autoantibody-bound GM-CSF was measured by quantifying the level of STAT5 phosphorylation in isolated neutrophils by immunoblotting (shown as the ratio of phosphorylated STAT5 to total STAT5) as described in “Methods.” The signaling activity of GM-CSF complexed to autoantibody was markedly lower than free GM-CSF (0.142 ± 0.1 vs 2.192 ± 0.2 pg/mL, respectively; n = 3 each; *P < .001). (D) The typical pattern of GM-CSF–stimulated increase in CD11b levels on neutrophils in whole blood (CD11b stimulation index) is shown for a healthy subject (HC) and a patient with PAP (PAP). The amount of exogenous GM-CSF (dashed lines) required to stimulate an increase in neutrophil CD11b levels to the threshold value (dotted line) was lower in HCs than in patients with PAP. The median (IQR) GM-CSF concentration required to reach this stimulation threshold (inset) was significantly higher in patients with PAP than in healthy subjects (120 [80-347] ng/mL, n = 5; and 3.96 [1.07-4.86] ng/mL; n = 12; respectively; P < .002, Mann-Whitney). (E) Concentration-dependent reduction in the CD11b stimulation index by GM-CSF autoantibody purified from IVIG () or patients with PAP (■) and incubated with fresh whole blood at various concentrations. Each bar represents the results of 3 separate determinations. *Significant decrease (P < .001) from baseline determined in the absence of GM-CSF autoantibody. (F) Specificity of purified GM-CSF autoantibody. Neutrophils were incubated in the presence of GM-CSF (10 ng/mL) or IL-8 (100 ng/mL) and in the absence or presence of 1 μg GM-CSF autoantibody or control IgG as indicated. Data represent the level of CD11b in stimulated cells—the level in unstimulated cells. GM-CSF autoantibodies markedly inhibited the GM-CSF–stimulated (■), but resulted in levels of inhibition by IL-8 () that were significantly lower and similar to control (IgG, □). *A significant difference (P < .001) compared with inhibition of the GM-CSF–stimulated increase by GM-CSF autoantibody (■). (G) The CD11b stimulation index was measured in fresh blood from healthy control subjects (○), patients with PAP with active disease (◇), or patients with PAP in clinical remission of the lung disease (♦). The range of serum GM-CSF autoantibody levels separating healthy subjects and patients with PAP with active disease evaluated with this assay is indicated (3.2-24 μg/mL, ▨). Each symbol represents the results of triplicate determinations for one subject. The median (IQR) free serum GM-CSF level in healthy subjects was 0.00 (0.00-0.390) pg/mL and did not correlate with the CD11b stimulation index (P > .05), whereas the median (IQR) GM-CSF autoantibody level (0.90 [0.58-1.19] μg/mL) correlated with CD11b stimulation index (R² = 0.46, P = .03) (Spearman rank order correlation).

Figure 4

Correlation of GM-CSF autoantibody level and basal neutrophil function in vivo, and effects of GM-CSF autoantibody level on neutrophil function in vitro. (A) The phagocytic capacity of unstimulated neutrophils in whole blood was measured in healthy subjects (○), patients with PAP with active disease (◇), and patients with PAP in clinical remission of the lung disease (♦) by quantifying the uptake of IgG-opsonized latex microspheres as described under “Methods.” The range of serum GM-CSF autoantibody levels separating healthy subjects and patients with PAP with active disease evaluated with this assay is indicated (3.2-39 μg /mL, ▨). Each symbol represents the results for triplicate determinations for one subject. The mean (IQR) free serum GM-CSF in healthy subjects was 0.00 (0.00-0.267) pg/mL serum and did not correlate with the neutrophil phagocytic capacity (P > .05), whereas GM-CSF autoantibody levels (1.07 [0.74-1.66] μg/mL) correlated with neutrophil phagocytic capacity (R² = −0.70, P = .001) (Spearman rank order correlation). (B) Neutrophil chemotaxis was measured as described in “Methods.” In brief, neutrophils were placed in the upper chamber of a transwell culture plate and IL-8 (10 ng/mL) was placed in the lower chamber, both the upper and lower chambers (chemokinesis [CK] control) or was omitted (No IL-8) and GM-CSF autoantibody (0.5, or 1.0 μg/mL) or isotype control antibody (0.5 or 1.0 μg/mL) was placed in the upper chamber. Each bar represents the mean (± SE) for results from 3 determinations. Compared with the respective isotype antibody controls, increasing concentrations of GM-CSF autoantibody reduced neutrophil chemotaxis in rheostatic fashion (*P < .05; **P < .005).

Figure 5

Histogram showing the frequency distribution of serum GM-CSF autoantibody levels in healthy subjects (, n = 72), patients with PAP with active lung disease (■, n = 21), and patients with PAP in clinical remission of the lung disease (□, n = 2). Clinical remission was defined as formerly diagnosed patients with PAP who were currently presenting no respiratory insufficiency and had normal chest X-ray images. The range of serum GM-CSF autoantibody levels separating subjects with no or active lung disease from those without active disease is indicated (10.4-19 μg/mL, ▨).

Figure 6

Schematic of the proposed mechanism of innate immune regulation by GM-CSF autoantibodies showing the relationship between endogenous GM-CSF autoantibody level (abscissa), GM-CSF–dependent myeloid cell functions, and GM-CSF bioactivity (ordinate). More than a range of low autoantibody levels present in healthy subjects, myeloid cell functions vary inversely with level of GM-CSF autoantibodies (ordinate, ▨) and increased levels of GM-CSF (eg, present at inflammatory sites or from exogenous administration) increase myeloid cell functions above baseline levels by a mechanism known as “GM-CSF priming” (ordinate, ■). At and above GM-CSF autoantibody levels sufficient to completely neutralize GM-CSF bioactivity (eg, the critical threshold), GM-CSF–stimulated myeloid cell functions are minimal or zero (ordinate, □) and the risk of PAP is increased.