Secretion of S100A8, S100A9, and S100A12 by Neutrophils Involves Reactive Oxygen Species and Potassium Efflux

S100A8/A9 (calprotectin) and S100A12 proinflammatory mediators are found at inflammatory sites and in the serum of patients with inflammatory or autoimmune diseases. These cytoplasmic proteins are secreted by neutrophils at sites of inflammation via alternative secretion pathways of which little is known. This study examined the nature of the stimuli leading to S100A8/A9 and S100A12 secretion as well as the mechanism involved in this alternative secretion pathway. Chemotactic agents, cytokines, and particulate molecules were used to stimulate human neutrophils. MSU crystals, PMA, and H2O2 induced the release of S100A8, S100A9, and S100A12 homodimers, as well as S100A8/A9 heterodimer. High concentrations of S100A8/A9 and S100A12 were secreted in response to nanoparticles like MSU, silica, TiO2, fullerene, and single-wall carbon nanotubes as well as in response to microbe-derived molecules, such as zymosan or HKCA. However, neutrophils exposed to the chemotactic factors fMLP failed to secrete S100A8/A9 or S100A12. Secretion of S100A8/A9 was dependent on the production of reactive oxygen species and required K+ exchanges through the ATP-sensitive K+ channel. Altogether, these findings suggest that S100A12 and S100A8/A9 are secreted independently either via distinct mechanisms of secretion or following the activation of different signal transduction pathways.

a solution of the mAbs 1F8, 6B4, 2A10, and 27E10, as well as pAb anti-S100A8 or anti-S100A9 diluted in PBS/0.1% Tween/2%BSA to a concentration of 1 µg/ml (0.1ug/ml for 27E10) were added. After 45 min, the plates were extensively washed and the HRPconjugated goat anti-mouse or goat anti-rabbit IgG antibody (1:10,000) dilution was added for 45 min. The wells were washed three times, 3,3',5,5'-tetramethylbenzidine solution (TMB) substrate was added and the reaction was stopped by the addition of H2SO4 0.18M.
The optical density was read at 450 nm.
Production of reactive oxygen species. Neutrophils (10 7 cells/ml) were incubated with 10 µM 6-carboxy-2',7'-dichlorodihydrofluorescein diacetate for 20 min at 37˚C in presence or absence of 10 µM DPI for the last 5 min. The cells were washed once in HBSS 1X Tardif et al., supplementary material page 3 containing 1.3 mM CaCl2 and 10 mM HEPES (pH7.4), then stimulated with 1.5 mg/ml MSU crystals or 1µM PMA for increasing periods of time. Fluorescence was measured using a fluorescence ELISA reader.

Supplementary data
Characterization of novel antibodies and ELISAs for S100A8, S100A9 and S100A12.
Specific monoclonal and polyclonal antibodies recognising human S100A8, S100A9 or S100A12 were raised by immunising mice and rabbits. mAbs clones 1F8 (anti-S100A8), 6B4 (anti-S100A9) and 2A10 (anti-S100A12) were selected for specificity for their respective targets by direct ELISA and western blot. As shown in Supplementary Figure   1A, mAb 1F8 (anti-S100A8) and pAb anti-S100A8 did not recognize human or mouse S100A9, S100A12, or mouse S100A8, but bound to recombinant and native human S100A8. Similarly, the anti-S100A9 mAb 6B4 detected only human S100A9 and purified calprotectin. Binding to human S100A9, and to a lower extent mouse S100A9 was observed with the pAb against human S100A9. Each of these antibodies bound to a single band of the expected molecular weight. In contrast, bands of approximately 12, 21, 60 and 100 kDa (not shown) were detected by mAb 2A10 (anti-S100A12) in neutrophil's crude extract, presumed to be the monomeric, dimeric and hexameric forms of S100A12.
Specificity of the monoclonal antibodies was then confirmed by direct ELISA using increasing concentrations of recombinant mouse and human S100 proteins (Supp. Fig 1B-G). While all mAbs were specific to human S100A8, S100A9 and S100A12 (1F8, 6B4 and 2A10 respectively) and did not recognize the murine proteins, the pAbs anti-S100A8 and anti-S100A9 bound to both human and mouse proteins.
Binding of the mAbs anti-S100A8 and anti-S100A9 to the heterodimer S100A8/A9 (calprotectin) was next investigated. The mAbs 1F8 and 6B4 bound to recombinant human S100A8 and S100A9 respectively, but had low affinity to purified human calprotectin (Suppl. Fig. 1H-J). Both 1F8 and 6B4 had lower affinity for calprotectin than the anticalprotectin mAb 27E10.
Three sandwich ELISAs (for S100A8, S100A9 and S100A12, respectively) were then developed using these specific monoclonal and polyclonal antibodies. (Suppl. Fig.   1K-N). The ELISAs detecting S100A8 or S100A9 had lower limit of detection of 3.125 ng/ml, lower limit of quantification (LLOQ) of 6.25 ng/ml and an upper limit of quantification (ULOQ) of 100 ng/ml. Sensitivity of S100A12 ELISA was 10 times higher with a lower limit of detection of 0.39 ng/m, a LLOQ of 1.56 ng/ml and an ULOQ of 25 ng/ml. A sandwich ELISA for calprotectin using the polyclonal anti-S100A9 as capture antibody and the commercially available mAb 27E10 (which recognize an epitope formed by the association of S100A8 and S100A9) as detecting antibody was also developed (calprotectin purified from neutrophils was used as standard curve). This ELISA proved highly sensitive with a lower limit of detection of 0.39 ng/ml. Specificity of the sandwich ELISAs detecting S100A8 or S100A9 was confirmed by adding increasing concentrations of purified human calprotectin (7.8 to 4 000 ng/ml). Both ELISAs showed high specificity, with almost no cross-reaction with purified calprotectin below 250 ng/ml (Suppl. Fig. 1O and P). This specificity probably result from the fact that the polyclonal and monoclonal antibodies bind to a common epitope on the C-terminal end of S100A8 or S100A9 (data not shown), restricting the detection to the homodimers. Higher concentrations of calprotectin led to a minor detection of S100A8 and S100A9 (0.04% and 0.1% respectively) likely due to a contamination of the calprotectin preparation with S100A8 and S100A9 homodimers. Therefore, these indirect ELISAs detect specifically the respective proteins in monomer/homocomplex forms and can be used to measure specific forms present in biological fluids like serum, supernatants of stimulates cells or subcellular compartments of cells. Figure S1. Specificity of monoclonal and polyclonal antibodies against human S100A8, S100A9, and S100A12 proteins. A) mAbs and pAbs against S100A8, S100A9, and S100A12 specifically bind to their respective target antigens. Recombinant human S100A8, S100A9, S100A12, mS100A8 (murine S100A8), mS100A9 (murine S100A9), purified human calprotectin, and neutrophils crude extract were loaded onto SDS-PAGE and detected by western blotting using our monoclonal (1F8 (anti-S100A8), 6B4 (anti-S100A9), 2A10 (anti-S100A12)) and polyclonal antibodies as noted. B to G) Recombinant human and murine S100A proteins (1 to 1000 ng in 100 µl) and purified human calprotectin were loaded in high-binding 96-well plates to perform standard ELISA with B) 1F8 mAb, C) rabbit pAb against S100A8, D) 6B4 mAb, E) rabbit polyclonal against S100A9, F) 2A10 mAb, and G rabbit pAb against S100A12 as described in materials and methods. Figure S2. Comparison of binding activity of the different antibodies on wells coated with A) S100A8, B) S100A9, and C) S100A8/A9. Figure S3. Calibration curves of ELISAs specific for calganulins. Monoclonal or polyclonal Abs were coated in high-binding 96-well plates to perform sandwiches ELISA for A) S100A8, B) S100A9, C) S100A8/A9, and D) S100A12. E and F) Absence of crossreactivity of S100A8 and S100A9 ELISAs. Purified human calprotectin (7.8 to 4000 ng/ml) were loaded into the well precoated with E) 1F8 or F) 6B4 to perform S100A8 and S100A9 sandwich ELISA and verify the absence of detection by the polyclonal Abs. Figure S4. Limited secretion of S100A8/A9 and S100A12 induced by chemoattractants.
Neutrophils were stimulated with C5a, IL-8, LTB4, PAF and fMLP, or their diluants for 1 hour. A) S100A8/A9 and B) S100A12 were then quantified in the supernatants. Data are from 5 experiments performed on different blood donors. Figure S5. DPI inhibits the production of reactive oxygen species by neutrophils induced by MSU crystals and PMA. Neutrophils were preincubated with 10 µM DPI for 5 min before being stimulated with A) 1.5 mg/ml MSU crystals or their diluent (HBSS), or B) 1 µM PMA or its diluent (DMSO). Data is from one experiment representative of 3 experiments performed on cells from different blood donors. Figure S6. High extracellular K + concentrations inhibit the secretion of S100A8/A9 induced by MSU crystals. Neutrophils were stimulated with MSU crystals or its diluent (HBSS) in presence or absence of 130 mM K + for 1h at 37˚C. S100A8/A9 in the supernatants was quantified by ELISA. Data are the mean ± sem of 3 experiments performed on cells from different donors. Figure S7. Cytochalasin B does not increase the secretion of S100A8/A9. Neutrophils were stimulated with 10 -7 M fMLP or its diluent (DMSO) in presence or absence of 10 µM cytochalasin B for 1h at 37˚C. S100A8/A9 in the supernatants was quantified by ELISA.
Data are the mean ± sem of 3 experiments performed on cells from different donors. mS100A8 mS100A9 S100A8 S100A9 S100A12 pAb anti-S100A8