Interleukin-23 dependent IL-17 drives Th1 responses following Mycobacterium bovis BCG vaccination

Animals

C57BL/6 (B6), OT-II αβ TCR Transgenic (Tg) mice (OT-II) which are MHC class II I-A^b restricted and specific for OVA_323–339 and il-10^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME). il23p19^−/− [20], il17ra^−/−[11] on the B6 background were maintained at the University of Pittsburgh animal facility. All experimental mice were age-and sex-matched and were used between the ages of 6 to 8 wks according to University of Pittsburgh IACUC guidelines.

Vaccinations

BCG Pasteur was grown in Proskauer Beck (PB) medium containing 0.05% Tween 80 to mid-log phase and then frozen in 1-ml aliquots at −80°C. Bacterial stocks were plated on 7H11 agar plates to calculate colony forming units (CFU). Mice were vaccinated subcutaneously with 1×10⁶ CFU of BCG in PBS. BCG-vaccinated mice received COX2 inhibitor (NS-398; Sigma 10 mg/kg of body weight), isotype control antibody (Clone 54447, R and D Biosystems) and IL-17 neutralizing antibody (Clone 50104, R and D Biosystems) every 48 hours following vaccination.

Experimental Infection

The H37Rv strain of M.tuberculosis was grown as previously described [21]. For aerosol infections, mice were infected with 100 CFU of bacteria using a Glas-Col airborne infection system as described before [21]. Lung bacterial burden was estimated by plating the lung homogenates on 7H11 agar plates.

Draining Lymph nodes (DLN) cell preparation

DLNs were collected in ice cold DMEM and dispersed through a 70-µM pore size nylon tissue strainer (Falcon; BD Biosciences). Cells suspensions were treated with Gey’s solution, washed and counted (Beckman Coulter). Single cells were used for ELISpot, flow cytometric analyses or for sorting purified populations.

Detection of cytokine producing cells by ELISpot assay

Detection of antigen-specific IFNγ- and IL-17-producing cells was carried out using an ELISpot assay as described before [22]. Cells from unvaccinated and vaccinated mice were seeded at an initial concentration of 2–5×10⁶ cells/well and doubling dilutions made. Irradiated B6 splenocytes were used as APCs, while Ag85B_240–254 was used as antigen in assays from BCG-vaccinated mice to detect responding CD4⁺ cells [23] ; mouse rIL-2 (Sigma-Aldrich; 10U/ml) was added to all wells. Spots were enumerated by using CTL-Immuno Spot analyzer and frequency of responding cells determined and applied to the number of cells per sample to generate the total number of responding cells per organ. Wells without antigen were included as controls and did not yield cytokine-producing spots.

Generation and stimulation of BMDCs

BMDCs (DCs) were generated by culturing bone marrow cells in cDMEM containing GM-CSF (PeproTech) [21]. On day 7, nonadherent cells were collected and stimulated with BCG at a multiplicity of infection (MOI) of 5. Culture supernatants were analyzed by luminex assays.

Naive CD4⁺ T cell isolation and in vitro effector generation

Naïve CD4⁺ T cells were isolated from OT-II TCRαβ Tg mice using magnetic CD4⁺ beads (L3T4) (Miltenyi Biotec). Naive OT-II CD4⁺ T cells (1×10⁶ cells/ml) were cultured with BCG-stimulated DCs (MOI 5) or unstimulated DCs (1×10⁶ cells/ml) and OVA_323–339 peptide (5 µM) for 5 days. In some wells, DCs were treated with COX2 inhibitor (Celecoxib, 10µM), anti-IL-10 (10 µg/ml; Clone JES 052A5, R and D Biosystems) [24], isotype control (10 µg/ml; Clone 43414, R and D Biosystems) or IL-17A (100 ng/ml, R and D Biosystems) was added. Protein levels in the supernatants were assayed by ELISA.

Determination of protein amounts

IL-17, IFNγ, IL-10 protein amounts were measured in cell culture supernatants using a mouse Luminex assay (Linco/Millipore). ELISA antibody pairs were used to detect IL-12p70 levels (R and D Biosystems) and IL-23 levels (R and D Biosystems) in supernatants. PGE2 quantitative ELISA was performed in DC culture supernatants according to manufacturer's protocol (R&D Biosystems) and serum PGE2 levels determined using DetectX PGE2 kit validated for mouse serum (Arbor Assays).

Real Time PCR (RT-PCR)

RNA was extracted as previously described, reverse transcribed and amplified using FAM-labeled probe and primers on the ABI Prism 7900 detection system [21]. Fold increase in signal over that derived from control samples was determined using the ∆∆ct calculation. In some cases, the levels of mRNA relative to house keeping gene (GAPDH) was calculated. The primer and probes sequences have been previously published [21] or were commercially purchased (ABI Biosystems).

Flow cytometry and cell sorting

Single cell suspensions were stained with fluorochrome-labeled antibodies specific for CD3 (17A2), CD4 (RM4-5), CD8 (53-6.7). Intracellular staining was performed by using anti-IFNγ (XMG1.2) on cells stimulated with PMA and Ionomycin as per the method described [12]. To sort for purified DLN cell population, stained cells were sorted on BD FACS Aria flow cytometer as CD3⁺ CD4⁺ (>94% purity). For analysis, FlowJo (Tree Star Inc, CA) was used.

Statistical analysis

Differences between the means of groups were analyzed using the two tailed Student’s t-test in GraphPad Prism 5 (La Jolla, CA). Inherently, logarithmic data from bacterial growth and RT-PCR were transformed for statistical analyses.

Vaccine-induced IL-17 is required to generate Th1 responses following BCG vaccination in mice

IL-17 is required for generation of Th1 responses and host immunity against F.tularensis LVS and C.muridarum [12–13]. Therefore, we determined if IL-17 was involved in generation of Th1 responses following vaccination with BCG. We subcutaneously vaccinated wild type C57BL/6J (B6) mice or IL-17 Receptor A deficient mice (il17ra^−/−) with live BCG and evaluated IFNγ responses in the draining lymph nodes (DLN) of vaccinated mice. BCG-vaccinated B6 mice induced both CD4⁺ and CD8⁺ IFNγ-producing cells in DLNs, with higher numbers of CD8⁺ IFNγ-producing cells, when compared to unvaccinated mice (Figure 1A,B). Interestingly, il17ra^−/− mice had significantly decreased CD4⁺ and CD8⁺ IFNγ-producing cells compared to B6 BCG-vaccinated mice (Figure 1A,B; Supplementary Figure 1A, B). In order to detect antigen-specific responses in CD4⁺ T cells, Ag85B_240–254 peptide containing the motif that is conserved for Class II I-A^b and requires processing by APCs to prime Ag85B-specific CD4⁺ T cells was used to stimulate lymph node cells [23]. Ag85B-specific Th1 cells were significantly lower in il17ra^−/− BCG-vaccinated DLNs (Figure 1C) and correlated with a decreased expression of IFNγ mRNA in il17ra^−/− DLN cells when compared to B6-vaccinated mice (Figure 1D). However, no Ag85B-specific cytokine responses were detected in the lungs of BCG-vaccinated mice at any of the time points tested (data not shown).

IL-17 induces production of IL-12 in DCs [12–13] and subsequent Th1 cell differentiation [12]. Therefore, we hypothesized that reduced Th1 responses in the il17ra^−/− BCG-vaccinated mice was due to decreased induction of IL-17-dependent IL-12 production. Consistent with this hypothesis, significantly reduced IL-12p35 and IL-12p40 mRNA levels were detected in DLN cells of BCG-vaccinated il17ra^−/− mice (Figure 1D) and correlated with decreased mRNA expression of the Th1 transcription factor, Tbet [25]. As expected [12], there was increased induction of IL-17 mRNA in the il17ra^−/− BCG-vaccinated DLN cells compared to DLN cells isolated from B6 BCG-vaccinated mice (Figure 1D). Also, the increased levels of IL-17 mRNA correlated with increased expression of the Th17 transcription factor, RORγt [26] in DLN cells from il17ra^−/− BCG-vaccinated mice (Figure 1D). These data suggest that IL-17 is required for induction of vaccine-induced Th1 responses following BCG vaccination.

IL-23-dependent Th17 response precedes and is required for BCG vaccine-induced Th1 immunity following M.tuberculosis challenge

IL-23 is critical for Th17 responses in vivo following mycobacterial exposure [21–22, 27] and therefore, we vaccinated B6 mice and IL-23 gene deficient mice (il23p19^−/−) and evaluated generation of Ag85B-specific Th17 and Th1 responses in the DLNs. The generation of Ag85B-specific Th17 responses (Figure 2A) and Th1 responses (Figure 2B) were significantly decreased in il23p19^−/− mice when compared to B6 BCG-vaccinated mice. Induction of an effective Th1 vaccine response is crucial for vaccine-induced protection against M.tuberculosis challenge [22]. Therefore, we tested whether reduced Th17 and Th1 vaccine-induced responses resulted in decreased protection in the M.tuberculosis-challenged il23p19^−/− BCG-vaccinated mice. il23p19^−/− mice were vaccinated with BCG, rested for 30 days, following which they were challenged with aerosolized M.tuberculosis and the lung bacterial burdens determined in BCG-vaccinated and unvaccinated mice. As previously described, no differences in bacterial burden in lungs of B6 and il23p19^−/− unvaccinated mice were detected [21] (Figure 2C). However, we found significantly higher lung bacterial burden in il23p19^−/− M.tuberculosis challenged BCG-vaccinated mice when compared to B6 BCG-vaccinated mice (Figure 2C). These data demonstrate the importance of the IL-23/Th17 pathway in mediating Th1 responses and protective BCG vaccine-induced immunity in response to pulmonary M.tuberculosis challenge.

Since IL-17 appeared to be a prerequisite for effective generation of BCG-induced Th1 responses (Figure 1), we determined the kinetics of Ag85B-specific Th1 and Th17 responses in B6 and il17ra^−/− BCG-vaccinated mice. We found that significant Ag85B-specific Th17 responses occurred between days 4–8 in the DLNs of BCG-vaccinated B6 mice, which was prior to the detection of Ag85B-specific Th1 responses on day 14 post-vaccination (Figure 2D). Furthermore, IL-17 signaling is crucial for the generation of Ag85B-specific Th1 responses, since il17ra^−/− BCG-vaccinated mice, although able to induce an early Ag85B-specific Th17 response, are unable to induce an effective Ag85B-specific Th1 response on day 14 (Figure 2E). To further confirm these findings we purified CD4⁺ T cells from B6 BCG-vaccinated and unvaccinated DLNs at different time points post vaccination and measured cytokine mRNA induction in these cells. Consistent with data shown in Figure 2D, IL-17 mRNA induction occurred in CD4⁺ T cells earlier than the induction of IFNγ mRNA, which was detected on day 14 post vaccination (Figure 2F). Together, our data show that BCG-vaccination induces an early IL-23-dependent Th17 response that precedes the Th1 response, and is required for induction of an effective BCG vaccine-induced Th1 response.

BCG-induced PGE-2 drives IL-10 and inhibits IFNγ production in T cells

Th17 cells are induced early in vivo following BCG vaccination and are important for subsequent generation of vaccine-induced Th1 cells at later time points (Figure 1, 2). Therefore, we then addressed whether the Th1 and Th17 polarizing cytokines namely IL-12 or IL-23 are induced in Dendritic cells (DCs) in response to BCG exposure. We found that following BCG exposure, DCs produced both IL-23 and IL-12 cytokines (Figure 3A,B). Interestingly, BCG also induced high levels of the anti-inflammatory cytokine, IL-10 in BCG-exposed DCs (Figure 3C). IL-10 is an anti-inflammatory cytokine that inhibits IL-12 production and Th1 cell differentiation [28]. Accordingly, IL-10 also inhibits IL-12 production in BCG-infected DCs and generation of IFNγ-producing cells [29]. Based on these data, we hypothesized that absence of early Th1 responses in vivo following BCG vaccination was due to high BCG-induced IL-10 levels (Figure 3C) and that IL-17-dependence to induce Th1 responses (Figure 1) was a host strategy to overcome the IL-10-mediated inhibition. To address this hypothesis, we first treated BCG-stimulated DCs with IL-10 neutralizing antibody and measured IL-12 production in supernatants. As expected [29], neutralization of BCG-induced IL-10 resulted in significantly increased production of IL-12 (Figure 3D). We also determined the effect of IL-10 neutralization on Th1 cell generation by co-culturing naïve OT-II TCR Tg T cells with BCG/OVA_323–339--treated DCs in the presence of IL-10 neutralizing antibody. Consistent with our hypothesis, we report that T-cell derived IFNγ production was inhibited in the presence of BCG and neutralization of IL-10 reversed BCG-mediated inhibition of IFNγ production in T cell supernatants (Figure 3E). These data suggest that despite induction of some IL-12 in BCG-exposed DCs, coincident induction of IL-10 inhibits Th1 responses. Importantly, Ag85B-specific Th1 responses detected in vivo were also increased in BCG-vaccinated il10^−/− mice when compared to B6 BCG-vaccinated mice (Figure 3F). Consistent with a recent published study [30], we did not find any differences between lung bacterial burden in B6 and il-10^−/− unvaccinated mice at early time points following infection with M.tuberculosis (Figure 3G). However, we found that il10^−/− BCG vaccinated mice when challenged with aerosolized M.tuberculosis mediated significantly better bacterial control in the lungs when compared to challenged B6 BCG-vaccinated mice (Figure 3G). These data suggest that IL-10 reduces the efficacy of BCG vaccine-induced immunity against M.tuberculosis challenge.

We then further determined the molecular mechanism by which BCG induced IL-10 inhibits Th1 responses. PGE2, is known to induce IL-10 and inhibit IL-12 production in DCs [16]. However, it is not known if BCG can induce PGE2 production in DCs and whether it impacts generation of BCG-induced T-cell responses. We report that BCG induced high levels of PGE2 in DC culture supernatants (Figure 4A). PGE2 synthesis involves the release of endogenous arachidonic acid and conversion to PGE2 via the rate limiting enzyme cyclooxygenase 2 (COX2). Accordingly, co-treatment of BCG-exposed DCs with a COX2 inhibitor (Celecoxib) abrogated PGE2 production (Figure 4A). Consistent with a role for PGE2 in IL-10 production, addition of COX2 inhibitor significantly reduced BCG-induced IL-10 levels (Figure 4B) and increased IL-12 production (Figure 4C). Furthermore, treatment with COX2 inhibitor was also able to reverse BCG-mediated inhibition of IFNγ production in T cells cultured with BCG-exposed DCs (Figure 4D) in DC-T-cell co-cultures. These data show that BCG exposure induces PGE2 and downstream induction of IL-10; however, this pathway also limits early IL-12 production and T cell derived IFNγ responses. These data together show that presence of BCG-induced IL-10 is detrimental to generation of effective Th1 responses and vaccine-induced protection against M.tuberculosis challenge.

BCG-induced PGE2 drives IL-23 and Ag85B-specific Th17 responses

Addition of exogenous PGE2 is a potent inducer of IL-23 in DCs and drives the production of IL-17 in T cells in vitro [18–19]. Since PGE2 drives IL-10 in BCG-exposed DCs (Figure 4B), we then examined whether PGE-2 had dual functions following mycobacterial exposure and can also drive IL-23 production in DCs. Accordingly, we treated BCG-exposed DCs with COX2 inhibitor and determined IL-23 levels in culture supernatants. Our data show that BCG-induced PGE-2 is critical for induction of IL-23 since we detected decreased IL-23 production in response to BCG stimulation in COX2 treated samples (Figure 4E). To further determine if PGE2-induced IL-23 production is required for the generation of BCG-induced Th17 responses, we co-cultured naïve CD4⁺ OT-II TCR Tg T cells with BCG/ OVA_323–339-treated DCs in the presence or absence of COX2 inhibitor. We found BCG/OVA_323–339--treated DCs primed T cells produced IL-17, while addition of COX2 inhibitor significantly reduced the production of IL-17 in T cell cultures (Figure 4F). These data show for the first time that that BCG-induced PGE2 production in DCs serves dual functions- not only does it mediate IL-10 production and limit IFNγ production (Figure 3), but also mediates IL-23 production and IL-17 production in T cells (Figure 4). In support of this, we found that treatment of BCG-vaccinated mice with COX2 inhibitor in vivo significantly reduced Ag85B-specific Th17 responses (Figure 4G), but not Ag85B-specific Th1 responses (Figure 4H) or vaccine-induced protection in the lung following M.tuberculosis challenge (Figure 4I). These data suggest that both in vivo and in vitro, blocking PGE2 results in reduced Th17 responses. Importantly, despite reduced Th17 responses, a competent Th1 response is generated, likely due to coincident loss of IL-10 production that can confer vaccine-induced protection. These data suggest a role for IL-17 specifically to overcome IL-10 inhibitory effects.

IL-17 is not required to drive Th1 responses in the absence of IL-10

Consistent with a role for IL-17 in induction of IL-12 in DCs [12–13], we found that IL-17 treatment of BCG-exposed DCs enhanced IL-12 (Figure 5A). Importantly, treatment of IL-17 significantly decreased BCG-induced IL-10 production in DCs (Figure 5B). These data suggest that BCG-exposure results in induction of PGE2 and high levels of IL-10 that initially inhibits IL-12 production and Th1 responses in vivo (Figure 2). Accordingly, the peak of PGE2 induction in vivo following BCG vaccination was at day 4, with significantly lower levels of PGE2 at later time points (Figure 5C). However, BCG-induced PGE2 also mediates IL-23 induction and drives subsequent Th17 responses. IL-17 then induces IL-12 production and inhibits IL-10 production and mediates IFNγ responses at later time points. Therefore, IFNγ protein expression was not detected early, but at later time points following BCG vaccination in vivo (Figure 5D). In order to confirm that PGE-dependent IL-17 mediates Th1 responses to overcome BCG-mediated IL-10 inhibition, we depleted IL-17 in il10^−/− BCG-vaccinated mice and measured Ag85B-specific Th1 responses in DLNs. Our data show that depletion of IL-17 in B6 mice resulted in decreased Ag85B-specific Th1 response (Figure 5E). However, depletion of IL-17 in il10^−/− mice did not result in decreased Ag85B-specific Th1 responses when compared to il-10^−/− BCG vaccinated mice treated with isotype control antibody (Figure 5E). These data suggest that IL-17 responses are required to drive Th1 responses, specifically to overcome IL-10-dependent Th1 inhibitory effects.