IFN-γ- and TNF-Independent Vitamin D-Inducible Human Suppression of Mycobacteria: The Role of Cathelicidin LL-37¹

Tuberculosis (TB)³ is a globally important cause of death (1). Although effective chemopreventive and chemotherapeutic regimes exist, their implementation is not straightforward because of the need for prolonged (6–9 mo) treatment courses. Interruption of such treatment also contributes to the growing problem of drug resistant disease (2). Therefore, there is a need for novel interventions that might help prevent and treat this disease.

It is recognized that the majority of people infected with TB do not develop disease. Instead, a lifelong state of concomitant immunity referred to as latent TB infection develops following primary infection. Perturbation of concomitant immunity by immunosuppression can rapidly lead to the development of reactivation or postprimary TB many years after the initial infection. The CD4 T cell depletion that characterizes HIV infection is by far the strongest associate of such reactivation, although only 7–12% of the global TB burden can be attributed to this cause (3). Immunosuppression by corticosteroid therapy and anti-TNF agents and via rare genetic defects in the IL-12- and IFN-γ-driven type 1 cytokine pathway also predispose to active TB (4, 5, 6, 7, 8). However, the overall contribution of these recognized risk factors to global TB burden is most likely small, and the mechanisms that lead to the breakdown of immunity to TB in most humans remain to be discovered.

A factor that associates epidemiologically with the reactivation of TB is migration, best documented in those moving from endemic environments to more developed areas of the world where TB incidence in indigenous people is very low (9, 10, 11). In 2005 in the United States, the risk of TB in persons born in Asia was 19.6 times greater than in the white population (12). There is some evidence that the incidence of TB in such immigrants is greater than in the country of origin (9). Factors that may change as a consequence of migration are diet and sunlight exposure and thereby the level of vitamin D (13). An association between vitamin D deficiency and the risk of TB in foreign borns in London, U.K., exists (14, 15, 16), and other findings corroborate the idea that vitamin D is protective (17, 18, 19). Historically, vitamin D was used in the treatment of TB until the advent of modern chemotherapy (20).

Vitamin D has no direct antimycobacterial action, but its active metabolite, 1α,25 dihydroxyvitamin D₃ (1α,25(OH)₂D₃), modulates the immune response to Mycobacterium tuberculosis (MTB). Expression of macrophage 25-hydroxyvitamin D 1α-hydroxylase (1α-hydroxylase) has been shown recently to be up-regulated by ligation of macrophage TLRs by MTB Ags (21). This enzyme metabolizes 25-hydroxyvitamin D₃ (25(OH)D₃) to 1α,25(OH)₂D₃. IFN-γ secreted by type 1 T cells potentiates this effect by up-regulating 1α-hydroxylase (22) and inhibiting induction of 25(OH)D 24-hydroxylase (23), a key enzyme in 1α,25(OH)₂D₃ inactivation. 1α,25(OH)₂D₃ induces antimycobacterial activity in vitro in both monocytes (MN) (24) and macrophages (17). Several mechanisms of action have been proposed. Exogenous 1α,25(OH)₂D₃ induces a superoxide burst (25) and enhances phagolysosome fusion (26) in MTB-infected macrophages; both phenomena are mediated by PI3K, suggesting that this response is initiated by ligation of a membrane vitamin D receptor (VDR) (27). 1α,25(OH)₂D₃ also modulates immune responses by binding the nuclear VDR, where it up-regulates protective innate host responses, including the induction of NO synthase, NOS2A (28). There is also evidence that LRG-47 (an IFN-γ-inducible 47-kDa vacuolar guanosine triphosphatase) is required to protect mice against TB (29). There has been no investigation of the single human homolog of LRG-47 (IRGC) in human TB.

The purpose of this study was to investigate in vitro mechanisms by which 1α,25(OH)₂D₃ might increase the ability of PBMC from sensitized human donors to resist mycobacteria. We found that 1α,25(OH)₂D₃ suppressed both bacillus Calmette Guérin (BCG) and MTB in infected cell cultures. These effects were primarily mediated by the nuclear VDR. 1α,25(OH)₂D₃ negatively regulated the transcription and secretion of IFN-γ, IL-12p40, and TNF in MTB- and BCG-infected PBMC and macrophage cultures, indicating that vitamin D does not mediate protection via these cytokines. Although the NOS2A gene was up-regulated by 1α,25(OH)₂D₃, inhibition of the formation of reactive nitrogen intermediate (RNI) only had a slight effect on the suppressive effect of 1α,25(OH)₂D₃ on BCG in PBMC cultures. By contrast 1α,25(OH)₂D₃ very strongly up-regulated (and MTB down-regulated) the cathelicidin gene, hCAP18. Intracellular hCAP18 protein was also increased by 1α,25(OH)₂D₃ and synthetic LL-37, the antimicrobial peptide derived from hCAP18, reduced the growth of MTB in culture by up to 75.7%. Our findings indicate that vitamin D mediates protection against TB by “nonclassical” mechanisms, including the induction of antimicrobial peptides.

PBMC were isolated from the buffy coats of healthy purified protein derivative-reactive blood donors over Ficoll as described previously (30). PBMC (5 × 10⁵) were plated in triplicate on 48-well plates in RPMI 1640/10% FCS. Infection and subsequent handling of cultures was according to techniques described previously (31). MN preparations were obtained by adherence, and periodic assessment by FACS found them to be 82 ± 4% CD14 and 76 ± 7% CD11b positive. 1α,25(OH)₂D₃ and its analogs were dissolved in ethanol (0.1–0.2% final concentration in culture). Vehicle control experiments indicated that this concentration had no affect on the growth of bacilli or on the gene expression and cytokine secretion of eukaryotic cells. PBMC were preincubated with 1α,25(OH)₂D₃ and its analogs for 72 h, then infected, then culture supernatants for cytokine analysis were harvested at 24 h for TNF and 96 h for IFN-γ and IL-12p40 for subsequent analysis by ELISA. l-NG-monomethylarginine (l-NMMA) (Sigma-Aldrich) was used in some cell cultures at a final concentration of 5 mM (a 4.35 molar excess over the concentration of l-arginine in RPMI 1640). To investigate the stability of cathelicidin LL-37 in cell culture, PBMC were isolated as above and plated in duplicate on 48-well plates in RPMI 1640/10% FCS spiked with 1 μg/ml synthetic LL-37 (PANATecs). Supernatants were aspirated after 0, 24, 48, 72, and 96 h of culture, and the concentration of cathelicidin LL-37 was determined by ELISA.

MTB-BCG and MTB H37Rv transformed with a replicating vector (pSMT1) containing the luciferase gene of Vibrio harveyi under the control of a constitutive promoter (Hsp60, Rv0440c) were prepared as described previously (32). Frozen aliquots of bacilli were grown to mid-log phase in Middlebrook 7H9 supplemented with 10% Albumin Dextrose Catalase (Difco) and 15 μg/ml hygromycin. Periodic relative luciferase units (RLU)-CFU determinations were conducted to ensure plasmid stability. Knowledge of the CFU:RLU ratio was used to infect with equal numbers of log-phase bacilli corresponding to a multiplicity of infection (MOI) mononuclear phagocyte:bacillus of ∼1:1 into cell culture. Following lysis of eukaryotic cells in 1 ml of H₂O, the luminescence of duplicate 100 μl aliquots of suspension of bacilli was determined by measuring the area under the curve decay in luminescence over 20 s in the presence of excess n-decyl aldehyde substrate (Sigma-Aldrich) in a luminometer (Berthold Technologies) as described previously (33). CFU enumeration was performed by plating serial dilutions of bacilli on 7H11 agar in triplicate.

One potentially important difference between the ionic environment in 7H9 medium and the phagolysosome is free iron concentration: 1.5 × 10⁻⁴ M in 7H9 broth but estimated to be 10⁻⁸ M within the phagolysosome (34). Iron-free 7H9 medium was prepared from its constituent ingredients (omitting ferric ammonium citrate) in bottles washed with 6 M HCl. Iron-depleted stocks of H37Rv lux were prepared by inoculation of iron-replete stock into iron-free medium to yield a final iron concentration of 2 μM, grown to log phase, and frozen after addition of an equal concentration of 30% glycerol (final concentration 1 μM Fe in 15% glycerol). Inocula from these iron-depleted stocks were then added to iron-free medium with or without 5 μg/ml LL-37 at 1/100 dilution to attain 10⁻⁸ M final Fe concentration.

For RNA extraction, 5 × 10⁶ PBMC or MN in 2 ml were set up in 6-well plates. In addition, both mononuclear phagocytes (isolated by adherence as previously described (35)) and PBMC from the same donors were studied. In these cultures, the infecting dose of MTB was normalized to average MN count (10% total PBMC) such that the MOI was 0.1:1 for PBMC and 1:1 for mononuclear phagocytes. Culture supernatant was aspirated and the cell monolayer immediately lysed using the RNEasy extraction kit (Qiagen). RNA was reverse transcribed using the Quantitect reverse transcription kit (Qiagen) that includes a DNase digest step. As the gene for LRG-47 has a single exon, we confirmed the completion of DNase digestion by PCR from partially processed samples that had not undergone the reverse transcriptase step. cDNA was used in quantitative PCR for IFN-γ, TNF, IL-12p40, NOS2A, LRG-47 (IRGC), LL-37, and β-actin on the ABI Prism 7000 platform. Primers and probes were obtained as predeveloped assay reagents (Applied Biosystems) with the exception of IRGC (XM_293893) for which the primers and probe sequences were as follows: forward primer, 5′-TCCCACTTTTCAAATGTGGTGTT-3′; reverse primer 5′-TCAGGTAGTTCTCCAGGGTTGTG-3′; and probe 5′-6-FAM-ACCTGCCTGGCACAGGGTCTGC-3′-TAMRA.

Each reaction was multiplexed by, and normalized to the β-actin content and fold induction over unstimulated samples was calculated by the ΔΔC_T method as described previously (user bulletin no. 2, available for download from www.appliedbiosystems.com).

Supernatants were analyzed for the presence of TNF and IL-12p40 by ELISA using Ab pairs from R&D Systems, and IFN-γ was assayed using an Ab pair from BD Pharmingen (catalog nos. 554548 and 554550). The sensitivity of these assays was between 10 and 77 pg/ml. LL-37 was assayed using a kit from Hycult Biotechnology for LL-37 (HK321) whose sensitivity was 1 ng/ml.

Adherent MN were cocultured in the presence of varying concentrations of 1α,25(OH)₂D₃ in 0.1% ethanol for 72 h. Staining for surface HLA-DR expression was performed according to standard protocols using PE-conjugated anti-HLA-DR and PE-conjugated mouse IgG1 as isotype control (Serotec). Stained cells were analyzed on a FACSCalibur flow cytometer equipped with CellQuest software.

Rabbit hCAP-18 polyclonal antiserum (36) and a PE-conjugated mouse mAb to CD14 (BD Biosciences) were used to detect hCAP-18 in mononuclear phagocytes. The rabbit polyclonal Ab is highly specific for hCAP-18, as demonstrated by immunoblotting of both neutrophil homogenate and plasma (36), and it reacts with both the LL-37 and the cathelin domain of the hCAP-18 molecule (37). PBMC from a healthy laboratory donor were isolated above and cultured in quadruplicate with 10⁻⁶ M 1α,25(OH)₂D₃ or 0.1% ethanol control for 72 h on 12-mm poly-l-lysine-coated coverslips (VWR) in 24-well plates. Cells were fixed in 3.7% paraformaldehyde, washed three times in PBS 10 mM HEPES, and incubated for 15 min at room temperature with permeabilization solution (PS) containing 10% calf serum, 0.05% saponin, 10 mM glycine and 10 mM HEPES. Cells were then incubated at room temperature for 30 min in PS containing 2% MACS FcR-blocking reagent (Miltenyi Biotec) with or without 1% hCAP-18 antiserum. Following three additional washes in PS, all cells were incubated at room temperature for 30 min in PS containing 1% goat anti-rabbit Alexa Fluor 644 (Molecular Probes) 17% CD14 PE Ab (BD Biosciences) and 2% MACS FcR-blocking reagent. After three additional washes in PS, coverslips were mounted onto glass slides and dried overnight at 4°C before confocal microscopy. Experiments substituting preimmune rabbit IgG for anti-hCAP-18 antiserum (to control for nonspecific binding of primary Ab) and staining in the absence of anti-hCAP-18 antiserum (to control for nonspecific binding of secondary Ab) were also performed.

Normally distributed variables were analyzed by Student’s paired or unpaired t test as appropriate. Correlation was performed by calculation of the Pearson or Spearman ρ according to the normality of variables. RNA fold induction values were normalized by log₁₀ transformation. Significance was inferred when p values were <0.05.

The ability of 1α,25(OH)₂D₃ to suppress the growth of BCG-lux in the PBMC of 10 healthy donors was tested. PBMC (5 × 10⁵) were inoculated with a fixed RLU content of BCG lux (t = 0) corresponding to a MOI of 0.1 CFU:cell. The growth of BCG over the following 96 h in the presence of varying amounts of 1α,25(OH)₂D₃ was determined by dividing the luminescence at 96 h by that at t = 0 to give a luminescence ratio (LR) for each donor. 1α,25(OH)₂D₃ was associated with a dose-dependent reduction in LR that became statistically significant at 10⁻⁸ M (Fig. 1 A). The effect was dependent on the presence of cells because 1α,25(OH)₂D₃ had no affect on the LR of BCG-lux that had been cultured in tissue culture medium alone.

The coculture of BCG-lux with PBMC is a convenient assay system, but we wished to ensure our observations were generalizable to virulent MTB. Therefore, we cultured PBMC from 10 donors with BCG-lux and MTB-lux in the presence or absence of 10⁻⁶ M 1α,25(OH)₂D₃. The 96-h BCG and MTB RLU content for each donor were highly correlated (Pearson, r = 0.5941, p = 0.006). 1α,25(OH)₂D₃ induced a fall in BCG-lux growth from 17.4 × 10⁵ to 9.66 × 10⁵ RLU/ml and a fall in MTB-lux growth from 10.5 × 10⁵ to 5.81 × 10⁵ RLU/ml (p = 0.002 and p < 0.0001, respectively; Fig. 1,B). To investigate the relationship between RLU and CFU, we cocultured the PBMC of six donors with MTB-lux at a MOI of 0.1:1 for 96 h in the presence or absence of 10⁻⁶ M 1α,25(OH)₂D₃. Significant suppression of the CFU of MTB was again seen in all donors with the mean CFU/ml decreasing from 8.83 × 10⁴ to 5.51 × 10⁴ (p = 0.031; Fig. 1 C). A similar decrease in luminescence in the same cultures was also observed (p = 0.063). Therefore, we concluded that the effects we had observed when using BCG-lux were generalizable to a pathogenic strain and that there was a relationship between RLU and CFU per milliliter.

1α,25(OH)₂D₃ can ligate a membrane-bound VDR (VDR_mem) to initiate rapid effects or a nuclear receptor (VDR_nuc) to modulate downstream gene transcription (27). We investigated the pathway responsible for the suppression of mycobacteria using conformationally locked analogs of 1α,25(OH)₂D₃ that selectively agonize or antagonize these receptors (Table I). All analogs were dissolved in ethanol, which was present in cell culture at final concentration of 0.1–0.2%. Neither vehicle control nor any analog had an effect on the luminescence of BCG-lux cultured in medium alone (data not shown).

When added to the PBMC of four donors infected with BCG-lux, the nuclear agonist V was associated with dose-dependent suppression of the bacteria that was evident at a concentration as low as 10⁻¹⁰ M and significant at 10⁻⁹ M (p = 0.046), whereas the membrane agonist JN had no affect (Fig. 2,A). As previously observed, a higher but still modest dose (10⁻⁸ M) of 1α,25(OH)₂D₃ significantly suppressed BCG-lux (p < 0.001). Addition of increasing concentrations of the VDR_mem antagonist 1β,25(OH)₂D₃ (HL) to such 1α,25(OH)₂D₃-stimulated cultures did not significantly attenuate this effect (Fig. 2,B). The partial VDR_nuc antagonist 23S-25-dehydro-1 α,25(OH)D₃-26,23-lactone (MK) exerted a moderate although statistically nonsignificant agonist effect at 10⁻⁷ M (Fig. 2,C). Full antagonist activity was revealed by complete reversal of the suppressive effects on BCG-lux of 10⁻⁸ M 1α,25(OH)₂D₃ (p = 0.016 by comparison with cultures that contained 1α,25(OH)₂D₃ alone; Fig. 2 C).

There is clear evidence that containment of MTB requires IL-12-mediated differentiation of type 1 lymphocytes producing IFN-γ that thereby activates mononuclear phagocytes to increase their ability to present Ag via up-regulation of HLA-DR and to produce a variety of mediators, including TNF. We were interested to determine whether the 1α,25(OH)₂D₃-mediated in vitro suppression of BCG-lux that we had observed could be related to these factors. We therefore assayed the BCG-lux-induced secretion of IFN-γ and IL-12p40 at 96 h and TNF at 24 h in the presence of varying concentrations of 1α,25(OH)₂D₃ in culture supernatants from 13 donors. In a subset of 4 donors, we also determined the effect of 1α,25(OH)₂D₃ on HLA-DR expression (in the absence of BCG) by FACS analysis. There was a 1α,25(OH)₂D₃ dose-dependent reduction in IFN-γ, IL-12p40, and TNF secretion that became significant for all three mediators at 10⁻⁹ M 1α,25(OH)₂D₃ (p ≤ 0.008; Fig. 3, A–C). There was also a clear trend toward decreased expression of HLA-DR, although in the smaller numbers of donors tested, this did not attain significance (p = 0.125; Fig. 3 D).

In addition to IL-12, IFN-γ, and TNF, there is also evidence that NO and LRG-47 (an IFN-γ-inducible 47-kDa vacuolar guanosine triphosphatase) are required to protect mice against TB (29, 38). There has been no investigation of the single human homolog of LRG-47 (IRGC) in human TB. To further investigate NO and IRGC and to generalize the cytokine results we had obtained in 1α,25(OH)₂D₃ and BCG-lux-stimulated cultures, we therefore investigated the regulatory effects of both MTB and 1α,25(OH)₂D₃ on these genes.

The PBMC and MN of 10 donors were used in these experiments. An initial sample of RNA was extracted from both cell types to determine the effect of preincubation with 10⁻⁶ M 1α,25(OH)₂D₃ or solvent (0.1% ethanol) for 72 h before infection with MTB. After this 72-h preincubation, time point 0 RNA samples were extracted for both cell types. These samples served as controls for the RNA extracted from 6 to 24 h. MTB-stimulated samples in the continued presence or absence of 1α,25(OH)₂D₃. Two broad patterns of regulation were observed: genes up-regulated strongly by MTB and down-regulated by 1α,25(OH)₂D₃ (IL-12p40, IFN-γ, and TNF; Fig. 4, A–C) and genes whose expression was variably affected by MTB and 1α,25(OH)₂D₃ (IRGC and NOS2A; Fig. 4, D and E).

The detailed fold change in RNA levels were as follows. 1α,25(OH)₂D₃ moderately decreased constitutive IL-12p40 in MN and IFN-γ expression in both MN and PBMC over the initial 72 h (p ≤ 0.03). With the exception of IL-12p40 at 6 h in PBMC, MTB strongly up-regulated (8- to 724-fold) IL-12p40, IFN-γ, and TNF in both cell types at both time points (p ≤ 0.011). Again, with the exception of IL-12p40 at 6 h, this up-regulation was abolished or even reversed (34- to 17,660-fold decrease) by the continued presence of 1α,25(OH)₂D₃ (p ≤ 0.003; Fig. 4, A–C).

The effects of both 1α,25(OH)₂D₃ and MTB on IRGC were moderate and variable between donors (Fig. 4,D). The only statistically significant effects on IRGC expression were a moderate (2.1-fold, p = 0.017) increase in constitutive expression in MN and conversely a 2.9-fold 1α,25(OH)₂D₃-associated decrease in MTB-stimulated PBMC at 6 h. 1α,25(OH)₂D₃ increased the constitutive expression of NOS2A 3.9-fold in PBMC (p = 0.0008; Fig. 4 E). NOS2A was not regulated by MTB, but at 24 h, its mean expression in MTB-stimulated PBMC was also 5.25-fold increased (p = 0.005) by 1α,25(OH)₂D₃.

Because 1α,25(OH)₂D₃ increased the expression of NOS2A, we determined whether inhibition of the formation of RNI by l-NMMA would reverse the 1α,25(OH)₂D₃-mediated decrease in luminescence. The PBMC of four donors were therefore set up with BCG-lux in the presence or l-NMMA with or without 10⁻⁸ M 1α,25(OH)₂D₃. l-NMMA has no effect on the luminescence of BCG-lux grown in medium. In PBMC culture, l-NMMA increased (by an average of 58%) the 96-h LR, suggesting that RNI do play a role in controlling BCG-lux under the conditions of this assay (Fig. 5). 1α,25(OH)₂D₃ decreased the luminescence as observed previously. However, the addition of l-NMMA to the 1α,25(OH)₂D₃-stimulated culture resulted only in a modest (25%) increase in RLU per milliliter. Taken together, these results suggest that the major component of 1α,25(OH)₂D₃-mediated suppression of mycobacteria is by a mechanism other than the generation of NO.

1α,25(OH)₂D₃ is known to induce the expression of the hCAP-18 gene that encodes the antimicrobial peptide LL-37 (39). Therefore, we were interested to determine the effect of MTB on the expression of this gene and whether LL-37 had activity against MTB. PBMC and MN of 10 donors were set up, and RNA was extracted for quantitative RT-PCR as described above. 1α,25(OH)₂D₃ increased constitutive hCAP-18 expression in both MN and PBMC over the initial 72 h by between 50- and 206-fold (p ≤ 0.002; Fig. 6,A). In MN, MTB down-regulated (4.59- to 13.7-fold) LL-37 at 6 and 24 h (p ≤ 0.015). 1α,25(OH)₂D₃ (10⁻⁶ M) completely reversed (527- to 774-fold increase) this MTB mediated suppression of hCAP-18 (p ≤ 0.0001). The presence of hCAP-18 peptide in 1α,25(OH)₂D₃-treated mononuclear phagocytes was confirmed by immunohistochemical staining; some peptide colocalized with CD14 (Fig. 6,B). No staining for hCAP-18 was observed in the absence of primary Ab (data not shown) or when preimmune rabbit IgG was substituted for primary Ab (Fig. 6 B), indicating that staining was specific for hCAP-18.

Based on these findings, we determined the level of LL-37 in supernatants from the PBMC of four donors cultured for 72 h with 10⁻⁶ M 1α,25(OH)₂D₃. Interestingly, no detectable LL-37 protein was found in these experiments. Our inability to detect secreted LL-37 was not due to degradation of LL-37 in cell culture because we “spiked” the supernatant of a PBMC culture with synthetic LL-37 and determined detectable concentrations up to 96 h of culture. LL-37 was detectable in all supernatants, with concentrations declining by 57% over 96 h of culture.

To assay the effect of synthetic LL-37 peptide on the growth of MTB, six replicates of MTB were set up in broth containing LL-37 and cultured for 96 h. There was a dose-dependent reduction in the CFU of MTB that was maximal (75.7% reduction) at 200 μg/ml (p = 0.04; Fig. 7,A). The microbicidal effects of LL-37 depend on the ionic environment (40). One potentially important difference between the ionic environment in 7H9 medium and the phagolysosome is free iron concentration: 1.5 × 10⁻⁴ M in 7H9 broth but estimated to be 10⁻⁸ M within the phagolysosome (34). MTB requires free iron for electron transport and the generation of ATP. Therefore, we investigated the effects of a physiological low-dose (5 μg/ml) synthetic LL-37 on MTB under iron-limiting conditions. At 96 h, the suppressive effects of both iron depletion in the presence or absence of LL-37 were moderate and did not achieve significance. However, when iron-depleted cultures were prolonged to 192 h, the CFU recovered from cultures containing 5 μg/ml LL-37 were 1.8-fold reduced (2.42 × 10⁶ vs 4.42 × 10⁶, p < 0.0001) when compared with cultures without LL-37. By contrast, no effect of 5 μg/ml LL-37 on MTB CFU was observed in the presence of 150 μM iron (Fig. 7 B).

We have demonstrated that 1α,25(OH)₂D₃ reduces both ATP generation by and the growth of BCG and MTB in infected cells. These effects were primarily mediated by the nuclear VDR. 1α,25(OH)₂D₃ strongly down-regulated the transcription and secretion of IFN-γ, IL-12p40, and TNF in MTB- and BCG-infected cells, indicating that these cytokines are not mechanistically implicated in its protective actions. Although the NOS2A gene was moderately up-regulated by 1α,25(OH)₂D₃, inhibition of the formation of RNI only marginally affected the suppressive effect of 1α,25(OH)₂D₃ on BCG in infected PBMC cultures. By contrast, 1α,25(OH)₂D₃ strongly up-regulated (and MTB down-regulated) the hCAP18 gene. Synthetic LL-37 (200 μg/ml) reduced the growth of MTB in culture by 75.7% and 20 μg/ml reduced growth by 52.4%. Although LL-37 protein was undetectable in the supernatant of 1α,25(OH)₂D₃-stimulated PBMC cultures, intracellular hCAP-18 protein was demonstrated by immunostaining. These findings indicate that vitamin D mediates protection against TB by “nonclassical” mechanisms that may include the induction of antimicrobial hCAP-18 but that further work on the regulation of this gene is necessary before concluding vitamin D acts solely in this way.

When modeling the replication of mycobacteria in humans, we and others (30, 35, 41) have classically infected mononuclear phagocytes with washing off of nonphagocytozed bacilli after an interval (the “CFU assay”). In this way, the intracellular replication of bacteria in variously differentiated cells can be assessed. However, we did not in these experiments restrict our consideration of the action of 1α,25(OH)₂D₃ to mononuclear phagocytes. In addition, MTB is able to replicate extracellularly in vivo. Lastly, the CFU assay is cumbersome for the high-throughput analyses that we desired. We therefore adopted coculture of light emitting BCG and MTB with PBMC to study the many of these effects. Although our experience is that tissue culture medium poorly supports the growth of mycobacteria, we acknowledge that, as in tissue, they may have been able to metabolize and replicate extracellularly. However, the presence of cells consistently reduced this replication, and 1α,25(OH)₂D₃ had no affect in the absence of cells (Fig. 1).

IFN-γ and TNF are two of the best-characterized cytokine factors necessary for protection against TB (8, 42). The independence of 1α,25(OH)₂D₃-mediated suppression of mycobacteria from these cytokines is evident in our experiments: both genes (together with the protective IL-12p40 gene) were strongly down-regulated by 1α,25(OH)₂D₃ reflected by decreased cytokine secretion in culture (Figs. 4 and 3, respectively). Humans have only one intact IFN-inducible p47 GTPase (IRGC) whose expression has been reported from testis but not THP-1 cells (43). Although our data indicate that expression is present in primary myeloid cells, the effect of 1α,25(OH)₂D₃ on IRGC expression was small and inconsistent, and so we cannot readily implicate this molecule in the effects we observed. 1α,25(OH)₂D₃ did, however, increase the constitutive expression of NOS2A 3.9-fold in PBMC (Fig. 4,E). NOS2A was not regulated by MTB, but at 24 h, its mean expression in MTB-stimulated PBMC was 5.25-fold increased (p = 0.005) by 1α,25(OH)₂D₃. These effects are complex. NOS2A clearly appears 1α,25(OH)₂D₃ inducible, and this gene is also known to be induced by IFN-γ (38). However, 1α,25(OH)₂D₃ down-regulates IFN-γ (Figs. 3 and 4), and so the effects we observed in the PBMC of sensitized donors will have been composite. Unlike IFN-γ and TNF, we cannot therefore exclude a role for vitamin D-induced RNI in our system. However, the effects of inhibiting RNI formation on 1α,25(OH)₂D₃-mediated suppression of BCG-lux were moderate (Fig. 5).

Cathelicidins are a structurally diverse family of antimicrobial peptide precursors with widespread distribution in mammals, characterized by the presence of a highly conserved cathelin domain of ∼100 residues. Humans express only one, hCAP-18, which is found in alveolar macrophages, lymphocytes, neutrophils, and epithelial cells (44). The promoter of the hCAP-18 gene contains a consensus vitamin D response element, and 1α,25(OH)₂D₃ induces hCAP-18 gene expression in human cell lines (39). The protein product of hCAP-18 undergoes extracellular cleavage by the neutrophil azurophil granule proteinase 3 to generate a 37-residue peptide LL-37 (45). LL-37 also exerts immunomodulatory activity, being chemoattractant for MN, T cells, and neutrophils (46) and up-regulates IL-8 and MCP1 in human whole blood (47). It also possesses broad spectrum bactericidal activity: it kills microbes by disruption of the cell membrane (48). Therefore, it seemed plausible that 1α,25(OH)₂D₃-induced antimycobacterial activity might be mediated by LL-37.

LL-37 (200 μg/ml) suppressed CFU to a similar extent as 10⁻⁶ M 1α,25(OH)₂D₃. The antimicrobial activity of LL-37 is known to be dependent on ionic environment (40). Our observation that a moderate concentration of LL-37 (5 μg/ml) induces superior suppression of H37Rv RLU and CFU under iron-limiting conditions supports the argument that LL-37 could be responsible for 1α,25(OH)₂D₃-induced suppression of mycobacteria in MN (21). However, we did not detect LL-37 in the supernatant of 1α,25(OH)₂D₃-stimulated PBMC by ELISA. Because immunostaining demonstrated the presence of intracellular hCAP-18 protein in monocytes (Fig. 6 B), we attribute this finding either to a lack of secretion of hCAP-18 protein by PBMC or to a failure of extracellular cleavage of hCAP-18 protein to release LL-37 detectable by ELISA. It is feasible that neutrophils may fulfil the latter role in vivo.

Our demonstration that 1α,25(OH)₂D₃-induced antimycobacterial activity is primarily mediated by nuclear-initiated signaling focused our investigation on transcriptional events. It is possible that the induction of LL-37 is a class effect and that other 1α,25(OH)₂D₃-induced antimicrobial peptides may be of importance. The observation that antimycobacterial activity is primarily mediated via nuclear-initiated signaling may also have significance for drug development: doses of vitamin D administered to patients with active TB may be limited by induction of hypercalcemia (20). Therefore, there may be a place for the use of noncalcemic vitamin D analogs as novel adjunctive treatments for active TB. Our findings suggest that these analogs should possess activity at the nuclear VDR.

In conclusion, our data tend to support and extend the recent article (21) that vitamin D-inducible LL-37 may play a role in phagocyte defense against MTB. However, the effects of substantial concentrations of LL-37 on MTB are moderate, and the possibility that vitamin D may also regulate other antimicrobial peptides is a current focus of our work. Overall, our findings illuminate at the cellular level mechanisms by which micronutrient status might considerably influence susceptibility to TB. Greater research of these factors in clinical studies and by randomized controlled trials allied to in vitro research might delineate better the extent to which these factors may contribute to population susceptibility and novel routes to prevent and treat this complex and devastating disease.

The funders of this study are public or nonprofit organizations that support science in general. They had no role in gathering, analyzing, or interpreting the data. We thank Bradley February for assistance in CFU enumeration and Drs. Geoff Packe and Mark Nicol for helpful discussions and review of the manuscript.

The authors have no financial conflict of interest.