Crystalline silica-induced proinflammatory eicosanoid storm in novel alveolar macrophage model quelled by docosahexaenoic acid

Introduction: Workplace exposure to respirable crystalline silica (cSiO2) is associated with chronic inflammatory and autoimmune diseases. At the mechanistic level, cSiO2 particles are quickly phagocytosed by resident alveolar macrophages (AMs) in the lung, causing a robust cycle of proinflammatory cytokine release, lysosomal rupture, mitochondrial toxicity, and immunogenic cell death if the particle is not efficiently cleared by the lung. We and others have demonstrated in bone marrow-derived and transformed macrophage models that supplementation with the ω-3 polyunsaturated fatty acid (PUFA) docosahexaenoic acid (DHA) contributes to increased membrane phospholipid content of DHA and subsequent suppression of cSiO2-triggered inflammatory responses. However, mechanistic exploration of ω-3 PUFA effects in AMs is challenging due to reliance on short-lived primary AMs derived from lung lavage fluid.

Methods: To address these limitations, we have employed a recently developed novel self-renewing AM model from C57BL/6 mice, fetal liver-derived alveolar-like macrophages (FLAMs), that is phenotypically representative of primary lung AM populations. We found that incubation of FLAMs with 25 µM  DHA as ethanolic suspensions or as complexes with bovine serum albumin were equally effective at increasing ω-3 PUFA content of phospholipids at the expense of the ω-6 PUFA arachidonic acid (ARA) and the ω-9 monounsaturated fatty acid oleic acid. Based on these findings, FLAMs were treated with 25 µM DHA in EtOH or EtOH vehicle (VEH) for 24 h, with or without LPS for 2 h, and with or without cSiO2 for 1.5 or 4 h then proinflammatory cytokine release, lysosomal membrane permeabilization, and mitochondrial depolarization assessed. In addition, oxylipin metabolites were measured using a targeted LC-MS lipidomics panel of 156 metabolites.

Results: Regardless of whether FLAMs were LPS-primed, cSiO2-triggered lysosomal permeability, mitochondrial toxicity, and cell death were not impacted by DHA. LPS+cSiO2 elicited marked IL-1α, IL-1β, and TNF-α release after 1.5 and 4 h of cSiO2 exposure, which was significantly inhibited by DHA. In VEH-treated cells, cSiO2 alone and LPS+cSiO2 induced synthesis of ARA-derived proinflammatory oxylipins including prostaglandins, leukotrienes, and thromboxanes that was suppressed by DHA. In addition, DHA promoted synthesis of pro-resolving DHA-derived oxylipins at the expense of ARA-derived oxylipins.

Discussion:  FLAMs were amenable to lipidome modulation by DHA, which suppressed cSiO2-triggered proinflammatory cytokine responses and ARA-derived oxylipins that potentially contribute to the particle’s toxicity in the lung. FLAMs are a promising in vitro alternative to primary AMs for investigating interventions against toxicant-triggered inflammation and autoimmunity in the lung.

Methods
Experimental design
Fetal liver-derived alveolar-like macrophages (FLAMs) were seeded in 6-well plates at a density of 4.50×105 cells/well in complete FLAM medium (RPMI 1640, 10% fetal bovine serum, 1% penicillin-streptomycin, 30 ng/ml GM-CSF, 20 ng/ml TGF-β). Cells were incubated overnight to achieve 70-90% confluency before beginning treatments. The next day, cells washed once with sterile PBS, fresh complete FLAM medium containing either 25 µM ethanolic DHA or ethanol vehicle (VEH) added, then cells incubated for 24 h. Following DHA or VEH treatment, cells were washed once with sterile PBS, treated with either 20 ng/ml lipopolysaccharide (LPS) or VEH in DPBS+/+ (DPBS containing calcium and magnesium) for 2 h, then exposed to 12.5 µg/cm2 crystalline silica (cSiO2) or VEH for 1.5 or 4 h. Cell culture supernatants and cells were pooled together and collected at t = 2 h (after 2 h LPS priming), 3.5 h (after 1.5 h cSiO2 exposure), and 6 h (after 4 h cSiO2 exposure). LPS and cSiO2 exposures were done in DPBS+/+ to minimize interference from fatty acids present in cell culture medium during oxylipin analyses.

Measurement of intracellular and extracellular oxylipins
Sample preparation
Following treatments, ice-cold methanol was added to each well, resulting in a final sample volume of 3 ml (1 ml cell culture supernatant + 2 ml methanol). To each sample, 60 µl of antioxidant cocktail (0.2 mg/ml butylated hydroxytoluene, 0.2 mg/ml triphenylphosphine, 0.6 mg/ml EDTA) was added to achieve a total cocktail concentration of 5% (v/v). Cells and supernatants within each well were pooled together, then samples were frozen at -80 °C until liquid chromatography-mass spectrometry (LC-MS) analysis.

Liquid chromatography-mass spectrometry (LC-MS) analysis
Targeted LC-MS lipidomics for 156 lipid metabolites was conducted at the Lipidomics Core Facility at Wayne State University. Briefly, 100 µl aliquots of cellular samples were thawed and spiked with a cocktail of deuterated internal standards (5 ng each of PGE1-d4, RvD2-d5, LTB4-d4, and 15[S]-HETE-d8; Cayman Chemical, Ann Arbor, MI) for quantification of oxylipins and recovery. Then, fatty acyl lipid metabolites were extracted by using C18 extraction columns that were washed with 15% (v/v) methanol and subsequently hexane, dried in a vacuum, eluted with methanol containing 0.1% (v/v) formic acid, dried under nitrogen gas, and dissolved in a 1:1 mixture of methanol:25 mM aqueous ammonium acetate. Extracted fatty acyl metabolites were subjected to high-performance liquid chromatography (HPLC) using a Luna C18 (3 µm, 2.1×150 mm) column connected to a Prominence XR system (Shimadzu, Somerset, NJ) then analyzed with a QTrap5500 mass spectrometer (AB Sciex, Singapore) set to negative ion mode. Analyst 1.6 software (AB Sciex) and MultiQuant software (AB Sciex) were used to collect and quantify the data in units of ng, respectively.

Data analysis and statistics
For oxylipin data, MetaboAnalyst Version 5.0 (Xia Lab, Quebec, Canada, www.metaboanalyst.ca/) were used to conduct statistical analyses. First, raw ng values were converted to corresponding pmol values in Microsoft Excel. Then, in MetaboAnalyst, the one factor statistical analysis module was chosen, and data were uploaded as a comma separated values (.csv) file with samples in unpaired columns and features (i.e., metabolites) in rows. Features with >70% missing data were removed from the dataset, and remaining missing values were estimated by replacing with the corresponding limits of detection (LODs; 1/5 of the minimum positive value of each variable). After the data was cleaned, the data was normalized by auto scaling only, then the data editor option was used to select experimental groups of interest to compare. For comparisons between experimental groups, one-way analysis of variance (ANOVA) (FDR = 0.05) followed by Tukey’s honestly significant difference (HSD) post-hoc test was used, with FDR q < 0.05 considered statistically significant.