25-Hydroxycholesterol acts as an amplifier of inflammatory signaling

Ch25h Amplifies Inflammatory Responses in Macrophages.

We have previously demonstrated that expression of Ch25h is up-regulated in bone-marrow–derived macrophages (BMDMs) by the TLR4 agonist lipopolysaccharide (LPS) (7), and others have shown that Ch25h is an IFN-stimulated gene (19). We tested a variety of purified TLR ligands and found that Ch25h is most strongly induced (10-fold) by the TLR3 agonist polyI:C (PIC) in a type I IFN-dependent manner and least strongly by the TLR1/2 agonist PAM3 (Fig. 1A).

Based on these data, we hypothesized that 25HC might participate in inflammatory signaling following PIC stimulation and conducted transcriptional profiling of BMDMs from wild-type (WT) or Ch25h^−/− mice following stimulation with PIC for 18 h. At this time point, the expression levels of 5,315 genes were altered by more than twofold (P < 0.01) in WT macrophages. Of these PIC-responsive genes, 338 were differentially expressed in Ch25h^−/− macrophages relative to WT cells (P < 0.01) and in almost all cases the transcriptional response to PIC stimulation was dampened by the loss of Ch25h, suggesting that Ch25h acts as an amplifier of the response (Fig. 1 B and C). Among the genes affected by loss of Ch25h are several important inflammatory mediators, including interleukin 6 (Il6), colony stimulating factor 1 (Csf1), and nitric oxide synthase 2, inducible (Nos2) (Fig. 1D). Although loss of Ch25h has a significant impact on the transcriptional response to PIC stimulation, the baseline transcriptional profile is almost entirely unperturbed (Fig. S1A). We confirmed that the lipid product of Ch25h is mediating the amplification by showing that, whereas not inducing Il6 expression by itself (Fig. S1B), 25HC rescues the diminished PIC-induced expression of this cytokine in Ch25h^−/− macrophages and amplifies its expression in WT macrophages (Fig. 1E). This amplification is also reflected in the levels of IL6 protein synthesis (Fig. 1F). It has been suggested that 25HC exerts its antiviral role by altering membrane fluidity and thus inhibiting uptake of viral particles. In contrast, WT and Ch25h^−/− macrophages exhibit equal uptake of FITC-labeled PIC, suggesting that the attenuated responses in Ch25h^−/− macrophages do not arise from altered internalization of inflammatory stimuli (Fig. S2).

Viruses activate multiple innate immune receptors in addition to TLR3 (23, 24). We therefore stimulated macrophages with ligands that are known to activate other pattern-recognition receptors, including the Myd88-dependent TLR7 and TLR9 receptors and the intracellular receptor NLRP3, and found that deletion of Ch25h caused similar changes in the inflammatory response of macrophages to these stimuli (Fig. 2A). Although 25HC is known to activate the liver-X receptors (LXRs), its effect on amplification of the PIC response is unaffected in macrophages isolated from mice lacking both LXRs (LXRαβ^−/− mice) (Fig. 2B).

Ch25h Participates in a Feed-Forward Loop Following PIC Stimulation via a Mechanism Involving AP-1.

To gain insight into the mechanism by which 25HC amplifies the transcriptional response to TLR3 activation, we used a transcription factor binding site prediction algorithm to identify potential regulators and prioritized the resulting list of candidates using microarray data to focus on those transcription factors that were both expressed in macrophages and whose expression was altered by treatment with 25HC. Initially, we analyzed the promoters of 113 up-regulated genes that were most differentially expressed in Ch25h^−/− macrophages relative to WT cells following stimulation with PIC (Fig. 1B, red dots) and identified 29 transcription factor binding site motifs that were enriched in this set relative to a control set of expressed genes. We also analyzed the promoters of 113 PIC-induced genes whose expression was unaffected by deletion of Ch25h (Fig. 1B, subset of light dots) and removed from consideration any motif that was enriched in both groups. Eliminating motifs for regulators not expressed in macrophages yielded a set of eight candidate transcription factors and transcription factor families, some of which were represented by more than one enriched motif: AP-1, NFκB, SCAP, SOX, MAF, TBP, p53, and ATF3.

To further narrow the list of possible transcriptional regulators that might mediate the effect of Ch25h on inflammatory signaling, we treated BMDMs with 25HC, performed transcriptional profiling by microarray, and identified 534 genes whose expression levels were altered by at least twofold (P < 0.01). Of the candidate regulators, only Atf3, v-maf musculoaponeurotic fibrosarcoma oncogene family, protein F (avian) (MafF), and, strikingly, seven members of the AP-1 family (Fos, Fosb, Fosl1, Fosl2, Jun, Junb, and Jund) were up-regulated by 25HC treatment (Fig. 3A). To determine which of these proteins was most likely to mediate the Ch25h amplification effect, we performed chromatin immunoprecipitation with real-time PCR quantification (ChIP-qPCR) using antibodies directed against FOS, JUN, MAFF, and ATF3 on chromatin isolated from WT and Ch25h^−/− macrophages following stimulation with PIC for 18 h. PIC stimulation induced recruitment of FOS, JUN, and ATF3 to their predicted binding sites in the Csf1 and Il6 promoters in WT cells but did not alter the binding of MAFF (Fig. 3 B–E). In cells lacking Ch25h, neither of the canonical AP-1 components, FOS and JUN, was recruited to these promoters, whereas recruitment of ATF3 was unaffected (Fig. 3 B–E). Importantly, the differences in the recruitment of FOS and JUN between WT and Ch25h^−/− macrophages cannot be explained by differences in their abundances as their mRNA and protein levels are equivalent in both genotypes (Fig. S3). Taken together these data suggest that 25HC recruits or helps to retain the AP-1 FOS–JUN heterodimer on the promoters of some PIC-responsive genes to augment their expression.

Expression of Ch25h Exacerbates Morbidity Following Influenza Infection.

Next we sought to determine the impact of 25HC-amplified inflammatory signaling on the outcome of an in vivo infection. We found that Ch25h is strongly up-regulated in both mouse lungs and human airway epithelial cells following infection with a variety of influenza viruses including the 2009 H1N1 pandemic strain (Fig. 4A and Fig. S4). This, coupled with recent studies that demonstrated strong antiviral activity of 25HC in vitro and decreased capacity of Ch25h^−/− mice to control herpes virus infection (19), led us to examine the role of Ch25h in control of influenza.

We infected age- and sex-matched WT and Ch25h^−/− mice with the mouse-adapted H1N1 influenza virus PR8 and assessed weight loss and mortality. Surprisingly, Ch25h^−/− mice were relatively protected following infection with influenza (Fig. 4B). We did not detect significant differences in viral load in vivo (Fig. 4C and Fig. S5), despite the observations that immortalized airway epithelial cells (Let1a cells) (25) derived from Ch25h^−/− mice are more susceptible to influenza infection in vitro (Fig. 4D), and that exogenous 25HC has potent antiviral activity on these cells (Fig. 4E). Analysis of the lung pathology showed that, whereas the influenza-induced lesions were similar in both genotypes, they were much less severe and the extent of the pathology was significantly lower in Ch25h^−/− mice (Fig. 4F). Histologic changes included D2–D4 bronchitis and bronchiolitis with respiratory epithelial necrosis and exudate, associated inflammatory infiltrates in regional alveoli spaces and interstitium (alveolitis), edema, and hemorrhage (Fig. 4G). Similar mortality and histologic differences were observed between genotypes following infection with the H3N2 influenza virus X31 (Fig. S6). In concordance with the histopathological evidence for an attenuated inflammatory response to influenza infection in Ch25h^−/− mice, several of the critical cytokines and antiviral factors whose inductions are attenuated in Ch25h^−/− macrophages following PIC stimulation were also induced to lower levels by influenza infection in vivo (Fig. 4 H and I).

Cell Culture, Stimulation, and Infection.

Cells were cultured at 37 °C and in 5% (vol/vol) CO₂ in a humidified incubator. Primary mouse BMDMs were prepared as previously described (7). On day 7, cells were stimulated as indicated in Results. Stimuli used were: PIC, 6 μg/mL (Amersham), 25HC, 5 μM (Sigma), and unmodified cholesterol, 5 μM (Sigma). Transfected PIC, 333 ng/mL, was delivered as previously described (38). LET1a cells were generated as previously described (25). Before infection, LET1a cells were plated at the desired density and adhered overnight. Cells were rinsed with PBS to remove protease inhibitors and infected at the indicated MOI for 1 h in OptiMEM (Gibco) + 0.3% BSA (Sigma; A8412). Infection media was removed and the cells were incubated for an additional 6 h in OptiMEM + BSA. Cells were lifted, fixed in formalin, permeabilized with 0.2% saponin, and stained with anti–influenza-NP-FITC antibody (ViroStat) and analyzed by flow cytometry.

Mice and in Vivo Infections.

C57BL/6J, Ch25h^−/−, Lxrα^−/−, and Lxrβ^−/−mice were purchased from The Jackson Laboratory and Ifnar^−/− mice were the kind gift of Michael Gale (University of Washington, Seattle). Lxrαβ^−/− mice were bred in house. Mice were housed under specific pathogen-free conditions. Mice (8–12 wk of age) were anesthetized with a ketamine/xylazine mixture and infected intranasally with the indicated amount of influenza in 30 µL of sterile PBS. Mice were euthanized by CO₂ inhalation and lungs were extracted for RNA isolation using TRIzol reagent (Invitrogen) or for histologic analysis. Differences in influenza-induced mortality between genotypes were assessed using the log-rank test on Kaplan–Meier survival curves. All mice in this study were used according to protocols approved by the Institutional Animal Care and Use Committees at the Seattle Biomedical Research Institute and at the Institute for Systems Biology, Seattle.

Histological and Immunohistochemical Analysis.

Dissected mouse lungs (left lungs) were fixed in 10% neutral-buffered formalin, processed routinely into paraffin, and stained with hematoxylin and eosin (H&E). The scoring system has been previously described (39). Formalin-fixed and paraffin-embedded sections were stained with goat anti-influenza A virus (Meridian Life Science; 4–5 mg/mL) followed by secondary antibody rabbit anti-goat IgG (Jackson ImmunoResearch; 1.8 mg/mL) as described (39). H&E and IHC-stained slides were digitized with an Olympus Nanozoomer and images were captured with Nikon Digital Pathology viewing software.

cDNA, Real-Time PCR, and ELISA.

Total RNA was isolated with TRIzol reagent (Invitrogen), DNase treated using the Turbo DNase kit (Ambion), reverse transcribed using SuperScript II (Invitrogen), and analyzed by real-time PCR with TaqMan Gene Expression assays (Applied Biosystems) in a 10-μL reaction volume on an ABI 7900 Fast Real-Time PCR system. Individual measurements were normalized to the expression levels of Ef1a mRNA. Lung RNA was obtained by homogenizing whole lung tissue in 1 mL of TRIzol with a stainless steel bead in a Retsch MM400 for 2 min. Protein expression levels in bronchiolar lavage (BAL) fluid were determined by multiplexed immunoassay per manufacturer instructions (Millipore).

Microarray Analysis.

Transcription Factor Binding Site Motif Analysis.

Promoter sequences (1,500/500 bp before/after the transcription start site) were obtained from an internal database that was populated from BioMart and scanned using the position-weight matrix scanning program Clover using all vertebrate transcription factor binding site matrices from the TRANSFAC Professional 2010 database (BioBase). Promoter sequences from all genes determined to be expressed in the microarray analysis (mean intensity 5% higher than negative control probes) were used as background for P-value calculations.

ChIP Real-Time PCR.

BMDMs were cultured from mice of the indicated genotypes and on day 7, cells were stimulated as described in Results. Immunoprecipitation (IP) was carried out as described in ref. 7. Antibodies used for IP were as follows: Jun (Santa Cruz Sc-1694 antibody for c-Jun with cross-reactivity for JunD) and Fos (ThermoScientific). As a control, IPs were performed with serum from unimmunized rabbits (normal rabbit serum) on chromatin pooled from each condition. For ChIP-qPCR at the Il6 promoter, the primers used were as follows: forward primer, TCCCATCAAGACATGCTCAAG; reverse primer, GTCGTTTAGCATCGAAAGAATCAC; probe, 5′-FAM/AGAGTGCTCATGCTTCTTAGGGCT/3′-TAMSp (IDT). For ChIP-qPCR at the Csf1 promoter, the primers used were as follows: forward primer, CACACACACAAACACACTCAC; reverse primer, ACTTTGAGGAGGCTGCAC; probe, 5′-FAM/AGGCACTGACACATACTACACCC/3′-TAMSp (IDT).

Plaque Assays.

Titers of infectious virus were measured by plaque assay on monolayers of Madin–Darby canine kidney (MDCK) cells as previously described (44).

ELISA.

Levels of IL6 protein were measured using a commercially available ELISA (R&D Systems; DY406) according to the manufacturer’s instructions.

Western Blots.

Protein levels of AP-1 components in BMDMs stimulated with PIC were measured by Western blotting using antibodies specific for Fos (Thermo Scientific; PA1-830) and Jun (Santa Cruz Biotechnology; sc-1694). The intensity of each band was measured from digital scans of each membrane using ImageJ.

Statistical Analysis.

Unless otherwise indicated, error bars in all figures represent the SEM and significance was assessed using a two-tailed t test with P < 0.05 considered statistically significant.