Vitamin D [1,25(OH)₂D₃] Differentially Regulates Human Innate Cytokine Responses to Bacterial versus Viral Pattern Recognition Receptor Stimuli

Much interest centers on the diverse roles played by vitamin D in the regulation of protective and maladaptive immunity. Vitamin D deficiency, initially seen as a seasonal issue in regions at extreme latitudes, is a global dilemma common even in warm climates (1). A decade ago, epidemiologic studies began to identify linkages between vitamin D insufficiency and the frequency of many diseases, with some of the most influential findings revealing notably increased viral respiratory infections among normally healthy children and adults (2–5). A large number of observational studies subsequently linked vitamin D deficiency with a broad range of undesirable metabolic, immune and clinical outcomes (6, 7). These include the risk of developing and/or surviving cancer (8–10), influenza epidemics (11), other respiratory infections (12–14), asthma (15–19), food allergies (20), inflammatory bowel diseases (21), obesity (22) and several chronic inflammatory disorders (6, 23). One consequence of such research has been that a high level of self-medication with exogenous vitamin D now occurs in many countries, much as extensive high-level vitamin C supplementation became popular for its presumed benefits on the common cold and cancer survival between 1970 and the late 1980s.

Vitamin D acts on immune capacity at multiple levels. Much has been learned about how vitamin D status or exposure to vitamin D metabolites can affect immunity, in particular antimicrobial peptide production (6, 24, 25). Our understanding of the impact of vitamin D on immune regulation, specifically cytokine production by the innate immune response that programs subsequent adaptive immune responses, is much more limited. To date, the majority of mechanistic studies examining these responses have focused on antibacterial (i.e., Mycobacterium tuberculosis) or, when using purified pattern recognition receptor (PRR) ligands, TLR4- or TLR2-driven responses (24, 26, 27). Vitamin D metabolites clearly decrease proinflammatory cytokines stimulated by bacteria or bacterial PRR in a wide variety of murine and human systems (28–30). Our understanding of its impact on virus or viral PRR-stimulated responses, stimuli where much of the most impressive epidemiologic evidence has been identified, is very limited and has largely been derived from experimental animals or long-term human cell lines.

In this study, we test the hypothesis that 1,25(OH)₂D₃ acts on innate immune cells of healthy humans to decrease proinflammatory and to enhance protective anti-inflammatory cytokine and chemokine responses after stimulation with both well-defined synthetic and infectious physiologic viral stimuli. The data reveal a marked qualitative difference between the effects of 1,25(OH)₂D₃ on bacterial (using TLR4 as a representative PRR) versus viral [TLR7/8 or respiratory syncytial virus (RSV)] stimulated responses.

After University of Manitoba Health Research Ethics Board approval and written informed consent, healthy, asymptomatic males and females 16–44 y of age were recruited.

A 2 μg/ml solution of 1,25(OH)₂D₃ (Enzo Life Sciences, Plymouth, PA) in ethanol was examined spectrophotometrically (λmax = 264 nm; ε = 19,000 M⁻¹cm⁻¹; 2.2 μg/ml yielding A₂₆₄ = 0.10). Aliquots were plated in a 96-well plate in an unlit biosafety cabinet for the times indicated (Supplemental Fig. 1), then reexamined at A₂₆₄ as recommended by the manufacturer to assess the kinetics and extent of 1,25(OH)₂D₃ degradation.

Freshly obtained PBMCs were isolated from 20–30 ml peripheral blood via density gradient centrifugation and used immediately for primary cell culture. Cells were cultured in RPMI 1640 (HyClone, Logan, UT), 10% FBS (PAA Laboratories), 1% 200 mM l-glutamine (Lonza, Walkerville, MD) and 0.22% 2-ME (Invitrogen, Grand Island NY) in triplicate at 350,000 cells/well in 96-well U-bottom plates. Medium alone contained 1.7 nM vitamin D [total 25(OH) D₂+D₃] derived from the FBS used for culture. Cultures were stimulated with: 1) medium alone, 2) TLR4 ligand (3.6 ng/ml LPS; Invivogen, San Diego CA), 3) TLR7/8 ligand (1.75 μg/ml CL075/3M002, C₁₃H₁₃N₃S, a thiazoloquinolone derivative containing <0.001 endotoxin unit/μg; Invivogen), or 4) intact infectious RSV [at 10^4.9 median tissue infectious dose/ml, as described previously (31)]. Stimulations were carried out with and without addition of 1,25(OH)₂D₃ to a final concentration of 1, 10, or 100 nM (Enzo Life Sciences, Plymouth, PA) at the time culture was set up. Where used, 25(OH)D₃ (Enzo Life Sciences) was present at 10, 100, or 1000 nM. In selected experiments, monocyte-enriched (CD14⁺) and -depleted (CD14⁻) populations were isolated using CD14⁺ selection kits (Stem Cell Technologies, Vancouver, BC), with cells cultured at 150,000 cells/well, also in triplicate. Flow cytometry demonstrated >90% purity among CD14-enriched and <1% CD14⁺ cell contamination in the CD14-depleted population. Both exhibited >95% viability by flow cytometry and trypan blue exclusion. In all experiments, cells were cultured at 37°C for 24 h after which supernatants were harvested for cytokine analyses and cell pellets stored in 100 μl RNAlater (Ambion) for quantitative PCR (QPCR) analysis.

Primary culture supernatant cytokine and chemokine levels were quantified via ELISA (32). In brief, titrations of four 2-fold dilutions of each sample were assessed with anti-cytokine/anti-chemokine capture and biotinylated Abs purchased from Biolegend (San Diego, CA), Peprotech (Rocky Hill, NJ), and R&D Systems (Minneapolis, MN). Cytokine standards were serially diluted on each plate. All samples were evaluated in triplicate, with concentrations calculated from serial titrations. Interassay ELISA variability was 5–10%. Vitamin D receptor (VDR) protein levels were examined in ELISA (Cloud Clone via Cedarlane, Burlington, ON, Canada) of cell lysates from 6 million PBMCs/donor harvested 24 h after stimulation with medium alone, LPS, or CL075 as cultured as described earlier. The manufacturer’s protocol was followed, except that cell extracts were obtained not with lysis buffer, found to interfere with the standard curve and spiked samples, but via two cycles of snap freeze in lN₂ using PBS, Tween 20 (0.05%), BSA (0.085%), and protease inhibitor (Sigma, St. Louis, MO) followed by quick spin centrifugation of the supernatant. Values are expressed using the vendor’s VDR protein standard.

RNA was extracted from 1 million cells stored at −20°C in RNAlater using RNeasy Plus Mini Kits (Qiagen), and reverse transcribed into cDNA using QuantiTect reverse transcription kits following the manufacturer’s protocols. Quantitative real-time PCR was performed using Roche Light Cycler 480. For enhanced precision, standards were made for each primer set using Human Universal Reference Total RNA (Clontech, Mountain View, CA), hence data are expressed as copy number/ng cDNA rather than fold induction. All data were normalized using 18s as the housekeeping gene. Replicate assays of the same sample on independent days typically resulted in 5–10% coefficient of variation in copy number values obtained.

The normality of data sets was determined with the D’Agostino and Pearson omnibus normality test using Prism 5 software (GraphPad, San Diego, CA). In most instances, nonparametric distributions were observed, so validated two-tailed Wilcoxon matched pairs tests were used. Correlations were examined using Spearman’s rank correlation coefficient. Two-tailed p values are reported, with differences considered significant if p < 0.05.

Unlike 25(OH)D3, 1,25(OH)₂D₃ is extremely light sensitive. We first examined the extent to which low levels of ambient laboratory light destroy 1,25(OH)₂D₃. Supplemental Fig. 1 demonstrates that transfer from light-shielded containers to foil-shielded culture plates within 10 min results in retention of virtually all of its integrity, whereas if exposed to low-level light, it rapidly degrades thereafter. These conditions were used for all experiments described later.

The impact on TLR4-dependent cytokine and chemokine responses among healthy adults was determined. Fresh PBMCs were cultured plus/minus LPS stimulation with 1,25(OH)₂D₃ over the range of 1 to 100 nM, broadly corresponding to low physiologic up to pharmacologic levels. A common panel of characteristic proinflammatory and anti-inflammatory biomarkers that are expressed at substantial levels upon both TLR4 and TLR7/8 stimulation were selected to allow direct comparisons. As anticipated, exogenous 1,25(OH)₂D₃ markedly inhibited production of each proinflammatory cytokine and chemokine examined including CCL2 (∼65%), CXCL8 (∼45%), and IL-6 (∼40%) (Fig. 1). Importantly, a broad panel of anti-inflammatory cytokine responses were also inhibited, with median decreases of 25% for IL-10, 42% for IL-1Ra, and 35% for sTNF-RII production. In each experiment, 1,25(OH)₂D₃ alone had minimal or no detectable effects on the low-level background cytokine responses produced in the absence of PRR ligands (Fig. 1). These findings extend our understanding of how vitamin D modulates human innate immune responses, indicating that global inhibition of both proinflammatory and anti-inflammatory cytokine production occurs after activation with this representative bacterial PRR ligand.

To determine whether and how 1,25(OH)₂D₃ modulates cytokine responses elicited by a viral PRR stimulus, we cultured PBMCs directly ex vivo with synthetic TLR7/8 ligand CL075. This viral mimic stimulates strong proinflammatory and anti-inflammatory responses among the same cytokines examined earlier (Fig. 2). In contrast with its strongly inhibitory effects on TLR4-driven responses, 1,25(OH)₂D₃ failed to reduce proinflammatory chemokine production. This finding was consistent despite its use over a concentration range spanning three orders of magnitude. Indeed, a statistically significant but very minor increase (∼12%) was found for median CCL2 production after TLR7/8 stimulation at the highest level_. The panel of anti-inflammatory responses examined, IL-10, IL-1Ra, and sTNF-RII, were also unaffected.

IL-27/35 (EBI3), innate cytokines which play important roles in regulating both innate immunity and CD4 T cell differentiation, were markedly induced after TLR4 or TLR7/8 ligation (Fig. 3). As described earlier, 1,25(OH)₂D₃ inhibited responses elicited by LPS (4-fold decrease, p = 0.0001). EBI3 responses in cells cultured with medium alone were decreased by ∼40% in the presence of 1,25(OH)₂D₃, with no incremental effect seen on production of this immunoregulatory molecule upon TLR7/8 stimulation.

As anticipated, 1,25(OH)₂D₃ alone or in combination with TLR4 or TLR7/8 stimulation had no impact on apoptosis or cell viability under any of the conditions tested (data not shown). Collectively, these data indicate that although TLR4 PRR-mediated activation is extremely sensitive to such regulation, a broad panel of TLR7/8-driven cytokine and chemokine responses are essentially unaffected.

The synthetic ligands used earlier were chosen for their ability to target PRRs with high selectivity. We therefore next examined whether the differences seen in responsiveness between these representative bacterial and viral ligand-driven human cytokine responses were evident when fresh PBMCs were stimulated with a common infectious respiratory virus. RSV interacts with multiple PRRs (33) and has been extensively implicated in asthma. The effects on both proinflammatory and anti-inflammatory cytokine production were examined after virus exposure. RSV stimulated strong innate proinflammatory (CCL2, CCL8, CCL5) and anti-inflammatory (IL-10) cytokine production (Fig. 4). 1,25(OH)₂D₃ failed to inhibit either proinflammatory or anti-inflammatory responses in 15 independent experiments, despite use of up to 100 nM concentrations. These data reinforce the finding of differential effects on TLR4 versus these virally stimulated innate cytokine responses among healthy human populations.

To control for the limited availability of blood from human donors, and more importantly, the fact that individuals vary widely in the proportion of different cell types represented in their peripheral blood, we chose unselected PBMCs for the earlier analyses. To assess the relative contribution of different cell populations to these effects, we performed experiments using enriched CD14⁺ monocyte and monocyte-depleted populations obtained from fresh PBMCs in eight independent experiments with novel donors. CD14-enriched, CD14-depleted, and unfractionated PBMC populations were stimulated with TLR4 or TLR7/8 ligands as previously described. Fig. 5 demonstrates that the great majority of cytokine production is derived from the CD14 population. As shown earlier, cytokine production was markedly inhibited by 1,25(OH)₂D₃ upon TLR4 but not TLR7/8 stimulation.

We next examined why cytokine responses after stimulation with bacterial PRR ligand are inhibited, but not those stimulated by these viral PRRs. We hypothesized that the failure of 1,25(OH)₂D₃ to inhibit TLR7/8 cytokine responses was linked to excessive induction of CYP24A1, the enzyme hydroxylating 1,25(OH)₂D₃. Fig. 6A demonstrates that 1,25(OH)₂D₃ alone results in intense induction of CYP24A1 as anticipated. TLR4 or TLR7/8 stimulation alone does not lead to CYP24A1 induction. After TLR4 plus 1,25(OH)₂D₃ stimulation, CYP24A1 expression was induced at levels ∼75% less than those seen with active vitamin D alone (p < 0.01). TLR7/8 plus 1,25(OH)₂D₃ elicited extremely low levels of CYP24A1 expression. Thus, median CYP24A1 copy numbers of ∼40,000 versus 10,000 versus 1,400 were found for medium control, TLR4-stimulated, and TLR7/8-stimulated cultures, respectively, over 15 experiments (p < 0.002 to 0.0001). These data, demonstrating a 97% decrease in the ability of TLR7/8-stimulated cells to generate CYP24A1, argue strongly against excessive induction of this enzyme as a mechanism for the unresponsiveness of TLR7/8-stimulated cells to 1,25(OH)₂D₃.

The hypothesis that 1,25(OH)₂D₃ inhibited a characteristic bacterial but not viral PRR-driven response is due to selective downregulation of TLR4 expression, thereby selectively limiting the capacity to sense infection was examined. However, Fig. 6B and 6C reveal that PRR expression is inhibited to a similar extent for both PRRs, with TLR4 levels reduced by median 50% and TLR8 by 75%. These data argue against preferential reductions in TLR4 versus TLR8 TLR expression as a cause of the different sensitivities of these prototypical antiviral versus bacterial innate responses to modulation.

We next examined the converse hypothesis. We speculated that a differential impact of TLR4 versus TLR7/8 activation on VDR expression underlies the capacity to differentially modulate these responses. LPS stimulation alone had no detectable impact on the intensity of VDR expression levels (p = 0.40; Fig. 7A). In contrast, TLR7/8 stimulation reduced VDR expression levels by 60% (p < 0.0001). Independent experiments in five individuals where PBMCs were stimulated with LPS or CL075 for 24 h and then whole-cell lysates were examined for VDR protein levels by ELISA reinforced this conclusion (Fig. 7B).

Subsequent experiments with highly enriched monocytes versus CD14⁻ cells in a population of eight additional volunteers (Fig. 7B) demonstrated that vitamin D receptor mRNA expression was predominately evident in CD14⁺ cells rather than the CD14-depleted population. As seen for unfractionated PBMCs (Fig. 7A), TLR7/8 stimulation of CD14⁺ cells results in 50–60% reductions in VDR relative to TLR4 stimulation (Fig. 7C). Attempts to “rescue” VDR expression by adding exogenous 1,25(OH)₂D₃ did not result in increases under any condition tested (p > 0.05, data not shown). Thus, TLR7/8, but not TLR4, stimulation severely restricts VDR expression and the capacity of primary cells to respond to 1,25(OH)₂D₃.

A very large number of genes contain vitamin D response elements. Evidence was also sought for differential consequences on cellular proteins that act to inhibit NF-κB, a common link in TLR4- and TLR7/8-driven cytokine production. Fig. 8 demonstrates that IkBα, a potent negative regulator of NF-κB (34), is enhanced by 1,25(OH)₂D₃ after TLR4 stimulation of CD14⁺ monocytes, with a mean increase of 247%. Conversely, the impact on TLR7/8-stimulated cells was a minor trend (18% decrease, not significant) downward. MKP-1 (a negative regulator of MAPK) and, hence, NF-κB (28, 35) were not differentially regulated after TLR4 or TLR7/8 stimulation.

Several conclusions emerge from this study. First, active vitamin D [1,25(OH)₂D₃] exerts a qualitatively different impact on TLR4- versus TLR7/8-stimulated innate immunity in humans. Although both proinflammatory and anti-inflammatory cytokine and chemokine responses to TLR4 were strongly inhibited in the presence of 1,25(OH)₂D₃, those stimulated via TLR7/8 were unaffected throughout a titration range from physiologic up to pharmacologic. This resistance of a panel of viral PRR-triggered innate cytokine responses to modulation was recapitulated using an intact infectious virus that has been extensively liked to the pathogenesis of asthma. We are not aware of prior reports identifying differential susceptibility of bacterial versus viral PRR responses to such modulation.

Second, although extensive murine studies, and several with primary human cells ex vivo, convincingly demonstrate that 1,25(OH)₂D₃ inhibits proinflammatory cytokine production, much less is understood of its impact on PRR-stimulated anti-inflammatory responses. When such responses were examined in previous studies, IL-10 was usually the sole representative assessed. The data in this report reveal that a broad panel of anti-inflammatory cytokines evoked by TLR4 stimulation, including IL-10, sTNF-RII, IL-1R antagonist, and the subunit common to the recently described anti-inflammatory cytokines IL-27/IL-35 (EBI3), are all substantially inhibited. This suggests that the commonly cited conclusion that vitamin D is inherently anti-inflammatory offers an oversimplification of its multiple effects on the human immune system. Indeed, the complexity of vitamin D biology in different species, tissues and stimuli is increasingly evident (36, 37).

No evidence was found to link deficient vitamin D activating enzyme (CYP27B1, data not shown), enhanced CYP24A1, or differentially reduced TLR4 versus TLR7/8 PRR expression to these distinct biological effects in humans. Rather, the data support roles for TLR7/8 (and not TLR4) agonist-dependent reductions in VDR expression and enhanced TLR4 (but not TLR7/8)-driven expression of NF-κB negative regulator, IkBα, as key contributing factors. We recognize that with well >4000 genes directly or indirectly regulated by vitamin D and its metabolites, additional mechanisms of regulation beyond those examined in this article are also likely to contribute to this differential activity. Carlberg et al. (38) recently performed a genome-wide analysis of the VDR showing that monocytes alone had 2340 VDR binding sites of which approximately one third were typical DR3-type response elements. Further investigation into VDR binding sites also indicates poor overlap between monocytes and lymphocytes in terms of 1,25(OH)₂D₃/VDR target genes, underlining how multiple components of vitamin D regulation are likely to be cell type specific.

The benefits of aggressive vitamin D supplementation in vivo have recently become controversial (2, 39–42). A commonly cited epidemiologic rationale for such supplementation is to enhance protective immune capacity and to decrease inflammation. Although human proinflammatory responses to bacteria or purified bacterial PRR ligands are clearly inhibited, our finding that proinflammatory (and anti-inflammatory) innate immunoregulatory responses stimulated by an infectious virus or this viral ligand were unaltered is important. Similarly, whether large-scale vitamin D supplementation assists healthy human populations in preventing or reducing the severity of viral respiratory tract infections, or other clinical disorders rooted in maladaptive immunity, is increasingly inconclusive (40, 41, 43).

This study has several limitations. First, although the data identify a differential impact on panels of these protypical bacterial (TLR4) versus viral (TLR7/8) PRR-dependent cytokine responses, a vast and increasing universe of PRR and of PRR ligands are being discovered. Whether each exhibits this dichotomy remains to be determined. Because most natural environmental and infectious innate stimuli use several PRRs to trigger immunity, and these PRR interact in a complex manner, it will be important to assess sensitivity to a broader range of stimuli. We caution that the data from any individual ligand studied, including those used in this study, are unlikely to apply to all stimuli and tissue types. Second, we recognize that with several thousand direct target genes and likely many more downstream, the effects may differ in different cell types. Finally, we emphasize that this study was conducted in healthy adult populations. Populations exhibiting chronic ongoing inflammatory diseases or those receiving broad-spectrum immunosuppressants may differ in their response to the impact on immune functions. Several interventional trials are under way that are examining the impact of vitamin D supplementation on asthma exacerbation rates, a clinical outcome primarily driven by viral infection.

Collectively, the data argue for more critical assessment of the benefits and associations of linking the presumed protective benefits of vitamin D for resistance to or resolution of viral infections in healthy populations.

We extend sincere appreciation to Rishma Chooniedass and Bill Stefura for technical assistance, to Caroline Graham for editorial assistance, and to all the study participants.

The authors have no financial conflicts of interest.