The 8- to 10-week-old C57BL/6 WT, S100A9 knockout (S100A9 KO), TLR4 knock out (TLR4 KO), and RAGE knock out (RAGE KO) mice for in vitro and in vivo experiments were obtained from our animal facility. S100A9 KO mice, TLR4 KO mice, RAGE KO mice, all developed on a C57BL/6 background, were generated as described previously (23–25). To generate TLR4 KO mice, the genomic fragment on chromosome 4 encoding the transmembrane and cytoplasmic domains was replaced with a neomycin resistance gene. To generate RAGE KO mice, the genomic fragment on chromosome 6 encoding the extracellular domain of RAGE was flanked with two loxP sites, which was subsequently deleted on exposure to Cre recombinase. Mice heterozygous for Cre recombination were cross-bred to generate homozygous RAGE KO mice. All animal experiments were approved by the local ethics committee and performed in strict accordance with the German regulations of the Society for Laboratory Animal Science (GV-SOLAS) and the European Health Law of the Federation of Laboratory Animal Science Associations. The protocols were approved by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany.

Dispase was from BD Biosciences (Heidelberg, Germany). DMEM was from Biochrome (Berlin, Germany). Ultrapure LPS from Salmonella enterica, Polymyxin B, DNase I, fibronectin from human plasma, and low melting agarose were from Sigma-Aldrich (Munich, Germany). Airway epithelial media (AEM), a serum free media optimized for airway epithelial cells, along with the following supplements added at indicated final concentrations to AEM—bovine pituitary extract (0.004 ml ml⁻¹), epidermal growth factor (10 ng ml⁻¹), insulin (5 µg ml⁻¹), hydrocortisone (0.5 µg ml⁻¹), epinephrine (0.5 µg ml⁻¹), triiodo-l-thyronine (6.7 ng ml⁻¹), transferrin (10 µg ml⁻¹), and retinoic acid (0.1 ng ml⁻¹) were from PromoCell GmbH (Heidelberg, Germany). Nylon mesh (28 µm) was from Ted Pella, Inc. (CA, USA). Biotinylated rat antimouse CD45 and CD16/32 antibodies were from BD PharMingen (Heidelberg, Germany). Streptavidin-coated magnetic particles were from Promega (Mannheim, Germany). Rat CD16-32-FITC antimouse was from BD Biosciences (Heidelberg, Germany). Rat CD45-PE/Cy7 antimouse were from Biolegend (Fell, Germany). Rabbit Caveolin-1 and rat CD16-32-FITC antimouse were from BD Biosciences (Heidelberg, Germany). Rat CD 34-FITC antimouse and rat EpCAM-APC antimouse were from eBioscience (Frankfurt, Germany). Hamster T1α antimouse was from Novus Biologicals (CO, USA). Rabbit prosurfactant protein C (proSP-C) antimouse, rabbit RAGE antimouse, and rat TLR4-MD-2 (MTS510) biotin antimouse were from Abcam (Cambridge, UK). Goat Alexa Fluor 488 antirabbit was from Life technologies (OR, USA). Alexa Fluor 488 Streptavidin was from Invitrogen (MA, USA). Goat FITC antirabbit, goat Texas red antihamster, and goat Texas red antirabbit were from Dianova (Hamburg, Germany). Carbon capsule and HEPA capsule were from Pall Corporation (PA, USA). Nebulizers were from Omron MicroAir (Amazon.de, Germany). Cylindrical chambers of Pyrex glass were manufactured by Feinmechanische Werkstatt (Physikalisches Institut, Muenster, Germany).

Murine S100A8 protein was expressed in Escherichia coli BL21 (DE3) cells and purified as described previously (16, 26–28). Briefly, murine S100A8 cDNA was cloned into the pET11/20 expression vector and expressed in E. coli BL21 (DE3) cells. The bacteria were grown at 37°C in 2× YT. After 24 h, bacteria were harvested, lysed, and inclusion bodies were isolated from the lysates. The inclusion bodies were dissolved in 8 M urea buffer and the sample was adjusted to a pH of 2.0–2.5 by adding hydrochloric acid. In order to establish proper refolding of the protein, samples were then incubated for 60 min at RT and dialyzed stepwise to adapt them to a pH of 7.4 to facilitate refolding in the presence of 2 mM DTT. Samples were then centrifuged at 60,000g for 10 min at 4°C to pellet the aggregated material, and further dialyzed before subjecting to anion exchange column and gel filtration chromatography. Protein concentrations were determined by UV absorption at 280 nm using a specific absorption coefficient of 0.40 (mg ml)⁻¹ cm⁻¹. Limulus amebocyte lysate assay (BioWhitaker, Walkersville, MD, USA) was performed to identify possible endotoxin contamination, which was quantified to 3.6 pg LPS/μg S100A8.

A method modified from a previous protocol (29) was used for isolation of primary lung AECs. The 8- to 10-week-old WT, TLR4 KO, and RAGE KO mice were anesthetized by ketamine/xylazine intraperitoneally according to body weight. When asleep, the mice were fixed and ~10 ml of 0.9% sodium chloride solution was injected through the right ventricle of the heart. The skin from rib cage to the neck was cut to make the trachea visible. A 22G catheter was inserted into the trachea and 1 ml Dispase (5,000 U) was then injected followed by 0.4–0.5 ml of 1% low melt agarose (stored in a 45°C water bath). Lungs were covered with ice for 2–3 min and extracted out and incubated in 2 ml Dispase. The time and temperature for Dispase incubation was optimized, as described in the result section. The lungs incubated in Dispase were then dissected gently by forceps in a Petri-dish containing 7 ml of DMEM media which was supplemented with 25 mM HEPES buffer, 1% glutamine, 1% penicillin/streptomycin, and 0.01% DNase I. The single cell suspension was passed subsequently through 100 and 40 µm strainers followed by passage through a 28 µm nylon mesh. The filtrate was then centrifuged at 130g for 8 min at RT and treated by erythrocyte lysis buffer (155 mM ammonium chloride, 10 mM potassium hydrogen carbonate, and 0.012 mM EDTA in deionized water) to get rid of the erythrocytes, followed by magnetic separation and incubation on Petri-dishes as described in the result section.

The day of isolation was considered day 0. Primary AECs were cultured on fibronectin-coated 24-well plates for 7 days at a density of 0.25–0.35 × 10⁶ cells/well to differentiate into type I cells. To maintain the type II phenotype, cells were cultured on 24-well plates at a density of 0.9–0.95 × 10⁶ cells/well for 28–30 h.

The 8- to 12-week-old mice were placed inside a cylindrical Pyrex chamber which was connected to a nebulizer containing a solution of LPS (0.5 mg ml⁻¹) or S100A8 (0.28 mg ml⁻¹) in saline or only saline. Mice were exposed to LPS or S100A8 or saline vapors for 45 min followed by a resting period of 4 h. After 4 h, mice were anesthetized and fixed on dissecting board and their skin was cut from the abdomen to the throat so as to expose the trachea. Blood was collected from heart to obtain serum for S100A8/A9 analysis by enzyme-linked immunosorbent assay (ELISA). Thereafter, a 22 G catheter was inserted into the trachea. BAL was collected by flushing the lungs five times with 0.8 ml PBS through the trachea and pulling out the BAL from the lungs into the syringe attached to the catheter. The collected BAL was centrifuged at 400g for 5 min at RT. The supernatant was collected and stored at −80°C for later use in analyzing cytokine/chemokine expression by ELISA or bead based immunoassays. The cell pellet was counted by Kimura staining (30) to identify the number of polymorphonuclear neutrophils (PMNs) in BAL.

Flow cytometric staining was done by collecting ~4–5 × 10⁵ cells in FACS tubes followed by treatment with the Fc blocking reagent, anti-CD16-32 antibody, for 10 min at 4°C. Primary antibody (rat anti-EpCAM or rabbit anti-proSP-C) was then added in 1% FBS/PBS for 30 min at 4°C, followed by secondary antibody (anti-ratIgG-APC or antirabbitIgG-FITC) incubation for 30 min at 4°C. For proSP-C staining, intracellular staining was performed wherein cells were permeabilized using BD Cytofix/Cytoperm Plus Fixation/Permeabilization Kit from BD Biosciences (Heidelberg, Germany) before primary antibody addition, according to the manufacturer’s instructions.

Alveolar epithelial cells were seeded on fibronectin-coated 8-well LabTek chambers at a density of ~3 × 10⁵ (AECI) or ~7 × 10⁵ cells (AECII)/well for 7 or 1 day, respectively. Immunofluorescent staining was performed by fixing the cells with 4% paraformaldehyde for 10 min at RT, followed by permeabilization by 0.5% TritonX-100 for 10 min at RT. Cells were then incubated in 10% normal goat serum blocking buffer for 1 h at RT, followed by incubation in primary antibody (rat anti-TLR4-MD-2-biotin or rabbit anti-RAGE) diluted in 0.2% BSA in PBS-TritonX-100, at 4°C overnight. Secondary antibody (streptavidin Dylight 488 or antirabbit Alexa Fluor 488) was then added in 0.2% BSA in PBS-TritonX-100 to the cells. DAPI was then used to stain the nucleus, followed by addition of fluorescent mounting medium (Dako) and visualization by an inverted fluorescent microscope (Axio Observer, Zeiss, Germany).

Supernatant collected from stimulated AECs was analyzed for IL-6 by murine IL-6 ELISA kit from eBiosciences (Frankfurt, Germany) following manufacturer’s instructions.

BAL from WT and S100A9 KO mice, and serum from WT mice were used to analyze S100A8/A9 by ELISA as described in Ref. (31). Also, BAL of WT and S100A9 KO mice were analyzed for IgM by murine IgM ELISA kit from Thermo Fisher Scientific (Dreieich, Germany) following manufacturer’s instructions.

Several cytokines/chemokines, namely—MCP-1, GM-CSF, IFN-β, IFN-γ, IL-1α, IL-1β, IL-10, IL-12 (p70), IL-17A, IL-23, IL-27, TNF-α, KC, RANTES, IL-5 were measured by a bead based immunoassay called LEGENDplex, according to manufacturer’s instructions, using the principles of sandwich immunoassays. The mouse inflammation panel, proinflammatory chemokine mix and match panel and custom mouse panel from LEGENDplex (Biolegend, Germany) were used to detect various analytes. Analyte measurements were made using Navios flow cytometer.

Statistical significance was calculated of three or more experiments by Prism 5 software using either unpaired two-tailed Student’s t-test compared to the control or two-way ANOVA (with Bonferroni’s or Tukey’s test) or one-way ANOVA (with Bonferroni’s test) (32, 33). Results are presented as means ± SEM. p < 0.05 was considered significant.

A previously described protocol for AEC culture and differentiation (29) was modified and optimized to achieve the desired purification and to prevent pre-activation of AECs. The original protocol was first optimized regarding Dispase digestion, i.e., lungs were incubated in 2 ml dispase for 20 min at 37°C, with constant rotation. For the first purification step, cells were subjected to magnetic bead separation, using biotinylated CD45 and CD16/32 antibodies for 30 min at 37°C, followed by streptavidin-coated magnetic beads for 30 min at RT, as described previously (29). However, the amount of DMEM for suspending the cells in antibodies and beads was optimized to 500 µl. Further purification was achieved by plating cells on Petri-dishes to remove adherent mesenchymal cells. Time of plating was optimized to 4 h, whereas a range of 4–16 h was recommended previously (29). Cells were then stained for Epithelial Cell Adhesion Molecule, EpCAM [epithelial cell-specific marker (34)] by flow cytometry, showing a purity of ~90% (Figure 1A). Contaminating leukocytes (CD45/CD16-32/CD34 positive cells) were completely undetectable (Figure S1A in Supplementary Material).

Initially, AECs were cultured in DMEM + 10%FBS as described earlier (29, 35), but they were spontaneously activated when cultured in FBS containing medium. Therefore, AECs were subsequently cultured in a special AEM along with supplements that supported cell growth without FBS addition. AECs cultured in this media showed a significantly reduced spontaneous activation (Figure S1B in Supplementary Material).

Freshly isolated primary AECs were mainly type II cells, as confirmed by analysis for proSP-C (Figure S1C in Supplementary Material), which is the precursor protein of type II-specific marker SPC (36). AECs were cultured on fibronectin-coated culture plates for 7 days to obtain a type I phenotype by differentiation of AECII (6). A type II monolayer was obtained by culturing AECs at a high density for 28–30 h after isolation. The type I cells (AECI) and type II cells (AECII) were characterized based on cell-specific markers (Figures 1B–D). Most AECs were positive for proSP-C on day 1. By day 7, the cells lost proSP-C indicating that they had differentiated, now expressing AECI-specific markers T1α and Caveolin-1, confirming differentiation of AECII to AECI during 7 days in culture.

Alveolar epithelial type I cell (AECI) and type II cell (AECII) were stimulated for 6 or 24 h with increasing concentrations of murine S100A8 as indicated (Figures 2A–D) and IL-6 in the supernatant was measured by ELISA after each time point. For both AECI and AECII, IL-6 release increased proportionally with increasing concentrations of S100A8 at each time point of stimulation. For both AECI and AECII, IL-6 release from stimulated and non-stimulated cells increased over time from 6 to 24 h. Generally, AECI showed a stronger secretion of IL-6 than AECII (6 and 24 h) (Figure 2). The observed effects were not influenced by possible endotoxin contamination due to the low content of 3.6 pg LPS/μg S100A8 in the preparation. Up to 100 pg ml⁻¹ LPS did not activate human monocytes (Figure S2A in Supplementary Material) and addition of Polymyxin B (an LPS inhibitor) did not affect the release of TNF-α from S100A8 stimulated monocytes. Human monocytes were isolated and stimulated as described earlier (26). Also, up to 50 pg ml⁻¹ LPS did not activate AECI to release IL-6 and addition of Polymyxin B to LPS (10 ng ml⁻¹) completely blocked its IL-6 release capacity from AECI (Figure S2B in Supplementary Material). In contrast, addition of Polymyxin B to S100A8 (5 µg ml⁻¹) did not affect its IL-6 release capacity from AECI.

In order to understand the interplay of DAMPs (S100A8) and PAMPs (LPS) in activating epithelial cells layers, confluent monolayers of AECI and AECII were stimulated for 6 and 24 h by murine S100A8 or LPS or a combination of both. Supernatant collected after 6 and 24 h was analyzed by either ELISA or bead based immunoassay for several cytokines/chemokines, namely—GM-CSF, IFN-β, IFN-γ, IL-1α, IL-1β, IL-6, IL-10, IL-12 (p70), IL-17A, IL-23, IL-27, MCP-1, RANTES, TNF-α, and KC. It was observed that AECs released only IL-6, MCP-1, and KC in response to S100A8 or LPS or their combination (Figures 3A–L).

S100A8 caused a significant release of IL-6, MCP-1, and KC from AECI (Figures 3A–C) and AECII (Figures 3G–I) as compared to non-stimulated controls at 6 h. LPS caused a significant release of IL-6 and KC from AECI (Figures 3A,C) and AECII (Figures 3G,I) as compared to non-stimulated controls at 6 h. Release of IL-6/MCP-1/KC induced by combination of S100A8 and LPS was significantly higher than controls and for most conditions, we could observe mainly additive effects after 6 h of stimulation. Although, for some conditions, the cytokine/chemokine release by the combination of S100A8 and LPS did not reach the summary of individual stimulations (S100A8 or LPS), it is clear that there are definitely no competitive effects between S100A8 and LPS stimulation. The release of IL-6/MCP-1/KC after 24 h was similar irrespective of the stimulant used in case of both AECI and AECII (Figures 3D–F and 3J–L).

Human AECII had been shown to express very low basal amounts of TLR4, which increased upon stimulation by LPS (37). Based on these observations, WT and TLR4 KO murine AECI and AECII monolayers cultured on fibronectin-coated Labtek chambers were stimulated with murine S100A8 or LPS for 6 h, before staining for TLR4-MD-2 (Figure 4A). WT AECI and AECII showed a low expression of TLR4-MD-2 under non-stimulated conditions. AECI showed no increase in TLR4-MD-2 expression from its basal level, after stimulation. However, expression of TLR4-MD-2 increased markedly in WT murine AECII after stimulation by S100A8 and LPS. To confirm that the increase in the expression of TLR4-MD-2 induced in WT AECII by S100A8 was not due to the presence of endotoxin contamination, we compared the TLR4-MD-2 expression of WT AECII stimulated by S100A8 and LPS to that of WT AECII stimulated by a combination of Polymyxin B and S100A8 or LPS (Figure S3 in Supplementary Material). Increase in TLR4-MD-2 expression in AECII by stimulation with S100A8 + Polymyxin B was identical to S100A8 stimulation alone. However, when Polymyxin B was added to LPS stimulation, the TLR4-MD-2 expression markedly decreased as compared to LPS stimulation alone, and reached basal levels. RAGE expression was examined in AECs by stimulating WT and RAGE KO AECI and AECII with S100A8 or LPS for 6 h and subsequent staining for RAGE (Figure 4B). WT AECI showed a high basal expression of RAGE which did not increase further upon S100A8/LPS stimulation. AECII showed no RAGE expression.

As mentioned earlier, S100A8 binds to TLR4 and RAGE in vitro. To explore the role of TLR4 and RAGE in AEC activation, AECs were isolated from WT, TLR4 KO and RAGE KO mice and cultured for 7 days (AECI) and 1 day or 28 h (AECII). Thereafter, AECs were stimulated with murine S100A8 or LPS for 6 and 24 h and supernatants were analyzed for IL-6 by ELISA (Figures 5A–D). At both time points, IL-6 release from S100A8/LPS stimulated RAGE KO AECI was not reduced compared to that from S100A8/LPS stimulated WT AECI, in case of LPS it was even higher in RAGE KO AECI. In contrast, IL-6 release from stimulated TLR4 KO AECI was significantly repressed as compared to that from stimulated WT AECI. IL-6 release from TLR4 KO AECI was similar to that released from non-stimulated controls indicating that in absence of TLR4, AECI were not responding to S100A8 or LPS stimulation. Thus, although AECI expresses RAGE in high amounts as compared to low amounts of TLR4-MD-2 expression, IL-6 release from S100A8 stimulated AECI is TLR4 dependent and independent of RAGE. Accordingly, IL-6 release induced by S100A8, from TLR4 KO AECII was significantly repressed as compared to that from WT AECII at each time point but RAGE KO AECII showed similar levels of IL-6 release compared to WT AECII. LPS stimulated TLR4 KO AECII showed already a lower IL-6 release after 6 h compared to WT AECII which, however, did not reach statistical significance. This trend at 6 h became significant after 24 h stimulation.

In order to prove that the unresponsiveness of TLR4 KO AECs to S100A8 was due to the lack of TLR4 and not due to secondary mechanisms, AECI and AECII from WT and TLR4 KO mice were stimulated with the TLR4 agonist LPS or the TLR1/2 agonist Pam3CSK4 (Figures 5E,F). IL-6 release from Pam3CSK4 stimulated TLR4 KO AECs was similar to that from Pam3CSK4 stimulated WT AECs. Therefore, the lack of TLR4 in AECs has no general effect on the inflammatory response of AECs.

To understand the in vivo role of endogenous S100A8/A9 in neutrophil recruitment during ALI, a murine model of ALI was created by exposure to LPS or S100A8 vapors. Saline exposed mice served as controls. For these studies, WT and S100A9 KO C57BL/6 mice were used. S100A9 KO mice lack both S100A8 and S100A9 on the protein level (23, 38), as observed in the western blot of bone marrow cells from WT and S100A9 KO mice (Figure S4 in Supplementary Material). Murine bone marrow cells were isolated as described earlier (16) and western blot was performed in a manner similar to that described in a previous study (26).

BAL from LPS exposed WT and S100A9 KO mice was analyzed after 4 h, for PMN counts, which were significantly higher in the BAL of WT mice than in the BAL of S100A9 KO mice (Figure 6A). The reason for higher PMN counts in WT compared to S100A9 KO mice could be because LPS exposure caused a high release of S100A8/A9 (~600 ng ml⁻¹) in BAL of WT mice (Figure 6B), which then acted as a chemoattractant for PMNs. LPS exposed S100A9 KO mice, as expected, showed non-detectable S100A8/A9 levels. The strong induction of S100A8/A9 release by LPS could be further confirmed by comparing the high S100A8/A9 levels in LPS-treated WT mice to non-detectable S100A8/A9 levels in untreated (saline treated) WT mice. To confirm that LPS exposure together with the S100A8/A9 release caused PMN infiltration in BAL, we examined the effect of exogenous S100A8 application on PMN counts in BAL of WT mice. We observed that PMN counts of WT mice exposed to S100A8 were higher than saline exposed WT mice (Figure 6C). S100A9 KO mice exposed to S100A8 showed basal levels of PMNs in BAL which was similar to that shown by saline-treated (WT or S100A9 KO) mice and was significantly lower than WT mice exposed to S100A8, confirming that S100A8/A9 is required for PMN infiltration in BAL in response to DAMP/PAMP exposure. Systemic release of S100A8/A9 in LPS or S100A8 exposed WT mice was comparable to saline exposed WT mice (Figure 6D) indicating no systemic involvement of S100A8/A9 in PMN recruitment to BAL at this early time point. PMNs in BAL were also counted 24 h after LPS exposure (data not shown), but the differences between WT and S100A9 KO PMN counts were not significant after 24 h. We also studied the release of IgM in BAL, which is a biomarker of lung injury (Figure 6E). We compared the release of IgM in BAL of WT mice exposed to LPS or S100A8 to that of untreated (saline treated) WT mice. The IgM response in BAL was significantly higher after 4 h of LPS exposure. Although not significant, the IgM response in BAL of S100A8-treated mice was higher than that of untreated mice. Thus, it can be concluded that after 4 h, LPS exposure leads to a strong injury response in BAL of mice and S100A8 exposure leads to a mild injury response in BAL of mice. We also compared the IgM response between LPS/S100A8 exposed WT and S100A9 KO mice and found that IgM response of S100A9 KO mice, although non-significant, was lower than the corresponding LPS/S100A8 exposed WT mice. The role of TLR4 and RAGE receptors in neutrophil recruitment to BAL was also studied, by analyzing BAL from LPS/S100A8 exposed WT, TLR4 KO and RAGE KO mice for PMN counts (Figure 6F). For S100A8 treatment, there was no difference in the PMN counts between the WT, TLR4 KO and RAGE KO mice, indicating that TLR4 and RAGE do not influence neutrophil recruitment in vivo during S100A8 exposure. However, for LPS treatment, PMN counts in BAL of TLR4 KO mice were significantly lower than that of WT mice. Although not significant, the PMN counts in BAL of LPS-treated RAGE KO mice were lower than that of WT mice, and thus RAGE could play a role in PMN recruitment to BAL during LPS exposure, independent of S100A8/A9. The mechanism behind this finding is not known and could be analyzed by further investigation.