25-Hydroxycholesterol secreted by macrophages in response to Toll-like receptor activation suppresses immunoglobulin A production

Synthesis of 25-Hydroxycholesterol in Response to Macrophage TLR Activation.

Sterol responses to macrophage TLR activation were monitored by mass spectrometry. Levels of the membrane-permeable oxysterol 25-hydroxycholesterol increased linearly between 2 and 12 h in both cells and medium of thioglycolate-elicited intraperitoneal macrophages stimulated with the TLR4-selective agonist Kdo₂-Lipid A (KDO) (Fig. 1A). By 24 h, levels of 25-hydroxycholesterol in the medium reached a concentration of ≈65 nM. Consistent with these changes were large increases in the levels of cholesterol 25-hydroxylase (CH25H) mRNA (Fig. 1B) and protein (Fig. 1C). Treatment of bone marrow–derived macrophages with KDO produced even larger elevations in 25-hydroxycholesterol and CH25H mRNA and protein (Fig. 1 D–F), indicating that induction of the enzyme was a generalized response of macrophages to activation of TLR4. Similar results were reported recently by others using bone marrow–derived macrophages (11).

To establish whether the induction of CH25H was unique to TLR4 activation, intraperitoneal macrophages were stimulated with agonists of different TLRs and CH25H mRNA levels measured. Incubation with LPS derived from Escherichia coli or Salmonella enterica induced CH25H mRNA (Fig. 2A), as did selective agonists for TLR2 (peptidoglycan from Staphylococcus aureus), TLR3 (poly I:C), and TLR2/6 (lipoteichoic acid) (Fig. 2B). Coincubation of macrophages with KDO and increasing concentrations of the MAPK inhibitors Sp600125 (Fig. 2C), U0126, or SB203580 progressively attenuated the increase in CH25H mRNA. The NF-κB inhibitors curcumin (Fig. 2D) and resveratrol also blocked CH25H induction. Similar MAPK and NF-κB inhibitor responses were measured in bone marrow–derived macrophages.

In Vivo Induction of CH25H.

To determine whether CH25H induction in response to TLR activation occurred in vivo, wild-type C57BL/6J mice were injected intraperitoneally with KDO and levels of CH25H mRNA, protein, and 25-hydroxycholesterol were measured in various tissues. Induction of CH25H mRNA was observed in all tissues examined from wild-type mice (Fig. 3A), with the largest fold increases detected in the liver (≈250-fold at 8 h) and heart (≈50-fold at 8 h). In some tissues, CH25H mRNA levels peaked at 8 h after KDO injection (liver, heart), whereas in others maximum stimulation was observed after 16 h treatment (lung, colon, brain, kidney, thymus, skin, and muscle). Consistent with these increases in mRNA, CH25H protein levels were elevated in the lungs and livers of wild-type but not Ch25h^−/− mice (Fig. 3B), and 25-hydroxycholesterol levels were elevated in the lungs (Fig. 3C) and sera (Fig. 3D) of wild-type mice 16 and 24 h after KDO treatment. Taken together, these data indicated that the response of CH25H to TLR4 activation in vivo was similar to that observed in cultured macrophages. Furthermore, the maximum induction of CH25H was observed in tissues with resident macrophage populations (e.g., Kupffer cells in the liver, alveolar macrophages in the lung, and microglia in the brain).

Altered IgA Levels in Ch25h^−/− and Cyp7b1^−/− Mice.

We next determined the effects of CH25H inactivation on the RNA expression patterns of different tissues, using microarray analysis. Ch25h^−/− mice had higher levels of mRNAs encoding the heavy chain of IgA (IgA_H) and the Ig joining chain (IgJ). The latter protein assembles monomeric IgA into a polymeric form and acts as a ligand for the Ig receptor that transports IgA across the mucosal epithelium (12). To confirm the microarray results, quantitative PCR (qPCR) was performed in the wild-type and mutant mice (Fig. 4A). Marked increases (5–25-fold) in the levels of IgA_H and IgJ mRNAs were observed in all Ch25h^−/− tissues, including the small intestine, lymph nodes, circulating cells, and liver. Levels of IgJ mRNA were also elevated in these tissues, but to a lesser extent (2–8-fold), whereas IgM mRNA levels did not vary between mice of different Ch25h genotypes.

In agreement with the mRNA results, IgA levels as determined by a qualitative isotyping assay were significantly higher in sera (n = 24), bronchial alveolar lavages (BAL) (n = 24), and intestinal mucosa (n = 12) of 2–4-month old Ch25h^−/− mice (Fig. 4B). In these assays, levels of IgG1 and IgM were unchanged in Ch25h^−/− mice, whereas IgG2a levels were significantly elevated in all tissues, and IgG2b and IgG3 levels were elevated only in the mucosa [supporting information (SI) Fig. S1]. Quantitative measurements using an ELISA in pooled samples from mice of each Ch25h genotype (n = 24) revealed 3–5-fold elevations in IgA levels in the sera, lungs, and mucosa of mutant mice (Fig. 4C).

Serum levels of 25-hydroxycholesterol are normally low, owing to rapid metabolism in the liver by the CYP7B1 oxysterol 7α-hydroxylase (13). Loss of CYP7B1 in mice (14) or humans (15) leads to large increases in circulating levels of 25-hydroxycholesterol. IgA levels in Cyp7b1^−/− mice were significantly reduced in the sera, lungs, and mucosa as judged using a qualitative assay (Fig. 4B). Thus, genetic manipulation of 25-hydroxycholesterol synthesis and catabolism had opposite effects on serum IgA levels.

Normal Leukocyte Populations and Cytokine Expression in Ch25h^−/− Mice.

Elevated levels of IgA in Ch25h^−/− mice could be due to increases in leukocyte numbers, specific leukocyte subpopulations, or cytokine production. Wild-type and knockout mice had similar numbers of total white blood cells and of neutrophils, lymphocytes, monocytes, eosinophils, and basophils (Fig. S2A). The total number of immune cells present in immunologic tissues [spleen, lung, bone marrow, blood, lymph nodes (inguinal, auxiliary, and mesenteric), Peyer's Patches, and thymus] also were similar between Ch25h^−/− and wild-type mice (Fig. S2B). More refined FACS analyses revealed no significant differences in the percentages of B cells, T cells, macrophages, neutrophils, helper T cells, and cytotoxic T cells in tissues isolated from knockout and wild-type mice (Fig. S2C). Serum levels of 33 cytokines, including TGF-β1, IL-5, and others known to affect IgA levels were quantified in 3 separate experiments (n = 8 mice per experiment per genotype) and no significant differences were observed (Fig. S3). Collectively, these data indicated that the elevated levels of IgA observed in Ch25h^−/− mice were not explained by alterations in leukocyte number, subpopulation distribution, or cytokine production.

Regulation of IgA Secretion by 25-Hydroxycholesterol In Vitro.

To determine whether 25-hydroxycholesterol affected IgA production in vitro, splenic B220⁺ cells were isolated by sorting from wild-type C57BL/6J mice, placed in culture, and treated with E. coli LPS for 16 h. Cytokines known to stimulate IgA CSR and subsequent IgA secretion were then added to the cells and levels of IgA determined by ELISA over a 6-day period. IgA increased in the medium of B220⁺ cells treated with TGF-β1, TGF-β1 plus IL-5, TGF-β1 plus IL-2, and TGF-β1 plus IL-5 plus IL-2 (Fig. 5A). The addition of 250 nM 25-hydroxycholesterol to the cellular medium on either days 0, 1, 2, or 3 of the experiment suppressed secretion of IgA when measured on day 6, but the oxysterol had no effect when added on days 4 and 5 (Fig. 5B). These results suggested that early exposure of naive B cells to 25-hydroxycholesterol caused suppression of CSR but had no effect on IgA expression or secretion once rearrangement of the Ig heavy chain locus had taken place (approximately day 4). The oxysterol mediated similar suppressive effects on IgA production by B220⁺ cells isolated from Peyer's patches. The suppressive effect of the oxysterol was also seen when APRIL was added to the cellular medium (Fig. S4), indicating that 25-hydroxycholesterol can inhibit T cell–dependent (Fig. 5B) and T cell–independent CSR in naïve B cells. In control experiments, 250 nM 25-hydroxycholesterol did not affect IL-4–stimulated IgE or IgG1 CSR in splenic B220⁺ cells (Fig. S5).

Dose–response studies indicated that the IC₅₀ for 25-hydroxycholesterol–mediated suppression of IgA secretion was ≈50 nM (Fig. 5C), a concentration that was similar to that produced in vitro by KDO-treated thioglycolate-elicited intraperitoneal macrophages (≈65 nM) and was significantly lower than that produced by activated bone marrow–derived macrophages (≈500 nM) (Fig. 1). To explore the chemical specificity of the response, B220⁺ cells were incubated with different sterols (each at 250 nM) and IgA secretion measured after cytokine stimulation. Both 25-hydroxycholesterol and 27-hydroxycholesterol suppressed IgA, whereas 22(R)-hydroxycholesterol and 24(R/S)-hydroxycholesterol, 2 ligands for the liver X receptor (16), were inactive in this assay (Fig. 5D). 24-Dihydrolanosterol, a sterol that suppresses cholesterol biosynthesis (17), was similarly inactive. The addition of cholesterol together with 25-hydroxycholesterol failed to reverse the suppressive effects of the oxysterol, suggesting that the oxysterol effect was independent of cellular cholesterol levels.

Mechanism of IgA Suppression by 25-Hydroxycholesterol.

CSR requires progression of B cells through 2 or more cell division cycles (18, 19). To resolve whether 25-hydroxycholesterol acted by suppressing cell division, we determined whether the sterol inhibited the acute proliferation that occurs after treatment of naïve B cells with LPS and cytokines. As shown in Fig. 6A, addition of 25-hydroxycholesterol to the medium did not significantly reduce the incorporation of [³H]thymidine into the DNA of B220⁺ cells stimulated for 56 h with LPS, TGF-β1, IL-2, and IL-5. In long-term (7-day) dose–response experiments, 25-hydroxycholesterol did not affect proliferation of B220⁺ cells treated with LPS and TGF-β1 or TGF-β1 plus IL-5; however, at the 3 highest levels tested (25, 100, and 250 nM), the oxysterol did reduce proliferation when IL-2 was included in the cytokine mixture (Fig. 6B). The effects of 27-hydroxycholesterol on cellular proliferation were similar to those observed with 25-hydroxycholesterol, whereas 22(R)-hydroxycholesterol and 24-dihydrolanosterol did not affect growth over the 7-day period. These results suggested that one mechanism by which 25-hydroxycholesterol reduced IgA levels was by counteracting the proliferative effect of IL-2 on B220⁺ cells.

In contrast, the mechanism by which 25-hydroxycholesterol inhibited IgA production in cells stimulated with LPS, TGF-β1, and IL-5 involved suppression of AID (Fig. 6C). Incubation of B cells with LPS alone resulted in a 14-fold increase in AID mRNA as judged by qPCR, and this induction was not affected by 25-hydroxycholesterol (Fig. 6C). When TGF-β1 and IL-5 were combined with LPS, a similar 8–12-fold increase in AID mRNA was measured; however, this elevation was blocked by 25-hydroxycholesterol. The induction of AID by LPS, TGF-β1, and IL-5 led to a 90-fold increase in IgA_H mRNA, and as expected, 25-hydroxycholesterol reduced this increase. The suppressive effect of 25-hydroxycholesterol was specific for AID and IgA_H mRNAs, given that the induction of RelA, Bcl-XL, NF-κBiα, NF-κBiβ, and NF-κBiε mRNAs in B cells by LPS and cytokines was not affected by the oxysterol, nor was the level of IgM_H mRNA altered (Fig. 6C).

Flow Cytometry.

Single-cell solutions were prepared from spleen, lung, bone marrow, blood, lymph node (inguinal and auxiliary), mesenteric lymph node, Peyer's patches, and thymus from Ch25h^+/+ and Ch25h^−/− mice (C57BL/6J, male, aged 2–4 months) as described in SI Materials and Methods, resuspended in FACS buffer, and live cells counted by trypan blue exclusion (Sigma). Approximately 1 × 10⁶ cells per tissue were pretreated with rat serum (Sigma) and then stained with the following antibodies from BD Bioscience: anti-CD45R (clone RA3–6B2)/B220-phycoerythrin (PE)-Cy7, anti-CD3ε (clone 500A2)-Pacific Blue, anti-CD11b (clone M1/70)-allophycocyanin, anti-CD8α (clone 53–6.7)-FITC, anti-CD4 (clone GK1.5)-PE, anti-Ly-6G, and anti-Ly-6C (clone RB6–8C5)-Alexa700. Dead cells were excluded by propidium iodine staining. The percentages of B cells, T cells, macrophages, neutrophils, T-helper cells, and cytotoxic T-cells were determined for all tissues with at least 3 × 10⁴ viable cells acquired using an LSR II (BD Bioscience) and data analyzed by FlowJo (Tree Star). FACS data were determined from 4 independent experiments with Ch25h^+/+ and Ch25h^−/− mice spleens (n = 19), lungs (n = 23), bone marrow (n = 12), blood (n = 21), lymph nodes (inguinal and auxiliary) (n = 15), mesenteric lymph nodes (n = 10), Peyer's patches (n = 10), and thymus (n = 15).

B Cell Experiments.

Single-cell suspensions prepared from spleens as described above were resuspended in FACS buffer and stained with anti-CD45R/B220-biotin (BD Biosciences). Cells were washed, counterstained with streptavidin microbeads (Miltenyi Biotech), washed again, and B220⁺ cells isolated by positive magnetic separation using an autoMACS separator (Miltenyi Biotech). Isolated cells were >93% B220⁺ as determined by FACS. The cells were centrifuged, counted after staining with trypan blue, plated at 5 × 10⁶ cells per well (24-well plate), and cultured with 8 μg/mL of LPS from E. coli 0111:B4 in B cell medium (RPMI 1640, 10% FCS, 1% penicillin/streptomycin, and 50 μM 2-mercaptoethanol) for 16 h as previously reported (31–33). Treated cells were diluted to 1 × 10⁵ cells per well (96-well plate) into medium containing the indicated cytokines (TGF-β1, IL-5, and IL-2) with or without oxysterols (250 nM) in triplicate. B cells were grown for an additional 6 days and cells and medium harvested. Cellular DNA content and medium IGA levels were quantified by fluorometry and ELISA according to the manufacturer's protocol (Molecular Probes, Bethyl Labs).