Innate and adaptive interferons suppress IL-1α and IL-1β production by distinct pulmonary myeloid subsets during Mycobacterium tuberculosis infection

SUMMARY

Interleukin-1 (IL-1) receptor signaling is required for control of Mycobacterium tuberculosis (Mtb) infection, yet the role of its two ligands, IL-1α and IL-1β, and the regulation of their expression in vivo are poorly understood. Here we show that in addition to IL-1β, IL-1α was independently required for host resistance. We identified two multifunctional inflammatory monocyte-macrophage and DC populations that co-expressed both IL-1 species at the single cell level in lungs of Mtb infected mice. Moreover, we demonstrated that interferons (IFNs) played important roles in fine-tuning IL-1 production by these cell populations in vivo. Type I IFNs inhibited IL-1 production by both subsets while CD4⁺ T cell derived IFNγ suppressed IL-1 expression selectively in inflammatory monocytes. These data provide a cellular basis for the anti-inflammatory effects of IFNs as well as pro-bacterial functions of type I IFNs during Mtb infection and reveal differential regulation of IL-1 induction by specialized cellular sources as an additional layer of complexity in the activity of IL-1 in vivo.

INTRODUCTION

Mycobacterium tuberculosis (Mtb) primarily infects mononuclear phagocytes. Host-resistance against this important pathogen is critically dependent on innate immune functions exerted by these cells including the production of inducible nitric oxide synthase (iNOS), TNFα and IL-12,23 p40 (Cooper et al., 2011; North and Jung, 2004) More recently it has become clear that IL-1 is also of critical importance for host control of Mtb infection as mice deficient in IL-1R or its adaptor MyD88 succumb rapidly to low dose aerosol infection with Mtb (Fremond et al., 2007; Mayer-Barber et al., 2010). While we have shown that IL-1R dependent protection requires IL-1β (Mayer-Barber et al., 2010), the contribution of IL-1α, the second major cytokine agonist for this receptor has been unclear.

A major feature of IL-1 is its complex control at the transcriptional, post-transcriptional and signal transduction levels which is highlighted by the wide variety of immunopathologies and auto-inflammatory diseases that occur in the absence of normal IL-1 regulation (Dinarello, 2009, 2010; Garlanda et al., 2007). Surprisingly little is known about the expression and processing of IL-1 in the context of Mtb infection in vivo. Whereas IL-1β production in response to Mtb in vitro is dependent on the NALP3-ASC inflammasome, we and others have recently reported that in lungs of Mtb infected mice, cleaved IL-1β is found even in the absence of the critical inflammasome component caspase-1. Moreover, although IL-1β is absolutely critical for host control of Mtb infection in mice, NLRP3, ASC or caspase-1 play minor roles in host resistance to this pathogen (Mayer-Barber et al., 2010; McElvania Tekippe et al., 2010; Walter et al., 2010). Although it is not clear what factors mediate the processing of IL-1β during Mtb infection in vivo, the inflammasome independence of the IL-1β response may indicate that it is produced by an atypical cellular source with unique IL-1β processing properties. For example, the neutrophils present in arthritic joints have been shown to be both a key producer of the cytokine and to cleave IL-1β in a caspase-1 independent manner (Guma et al., 2009; Joosten et al., 2009). The cell populations that produce host-protective IL-1 during Mtb infection in vivo, however, have not yet been characterized.

Type II interferon, IFNγ is the signature cytokine of T helper 1 (Th1) cells and has well-established protective functions in host resistance to Mtb and other mycobacterial infections in mice and humans. IFNγ is traditionally regarded as a pro-inflammatory cytokine and its protective function during Mtb infection involves the activation of macrophage (Flynn and Chan, 2001). In contrast, the role of type I IFN in the immune response to virulent Mtb is less clear and even though type I and II IFNs share similar STAT1 dependent signaling pathways, type I IFNs appear to promote rather than control infection. Thus, the hypervirulence of certain Mtb strains correlates with enhanced type I IFN synthesis and type I IFN receptor-deficient mice infected with Mtb display lower bacterial loads when compared to WT animals (Manca et al., 2001; Stanley et al., 2007). In addition, Mtb-infected mice intranasally treated with the type I IFN inducer polyinosinic-polycytidylic acid stabilized with poly-L-lysine (pICLC) exhibit exacerbated lung pathology and increased bacterial burden (Antonelli et al., 2010). The relevance of these observations to human tuberculosis is supported by a recent study in which the whole blood transcriptional profile of TB patients was found to be dominated by type I and II IFNs induced genes and this signature closely correlated with disease severity (Berry et al., 2010). Nevertheless, the cellular mechanisms by which type I IFNs mediate their pro-bacterial effects in Mtb infection and whether or not they involve cross talk with other innate cytokine pathways remains unclear.

Here we establish that IL-1α and IL-1β (IL-1α,β) each had critical functions in murine host-resistance against Mtb and identified and characterized the two major myeloid subsets that produce both cytokines in the lungs of Mtb infected mice as inflammatory monocyte-macrophage (iM) and DCs (iDC) with highly polyfunctional properties. Importantly, we demonstrated that endogenous type I IFN was a potent negative regulator of IL-1α,β production by both of these myeloid cell populations. In addition, our data revealed an anti-inflammatory role for CD4⁺ T cell derived IFNγ in dampening IL-1 expression by iM. Together these findings established polyfunctional mononuclear phagocytes as the major myeloid source of IL-1 during Mtb infection and provided a cellular basis for the cross talk between IFNs and IL-1 in the immune response to this bacterial pathogen in vivo.

RESULTS

IL-1α and IL-1β are each critical for host resistance to Mtb

To investigate the respective contributions of the two ligands IL-1α and IL-1β in IL-1R1-dependent host resistance to Mtb, we performed a series of experiments directly comparing the susceptibility of Il1r1^-/-, Il1a,Il1b^-/-, Il1b^-/- and Il1a^-/- mice to low dose aerosol (100-150 CFU) infection. As described previously (Mayer-Barber et al., 2010), IL-1R1 and IL-1β deficient mice rapidly lost weight and succumbed within 40 days (Figure 1A). No further increase in susceptibility was evident in mice doubly deficient in both IL-1α and IL-1β. Importantly, mice deficient in IL-1α alone displayed similar weight loss and mortality as Il1r1^-/-, Il1a,Il1b^-/- or Il1b^-/- mice and showed increased bacterial loads equivalent to that of infected Il1b^-/- animals (Figure 1A-B). Moreover, we observed similar amounts of IL-1β in the lungs of WT and IL-1α single deficient mice and conversely comparable amounts of IL-1α in WT and IL-1β deficient animals (Figure 1C). IL-1α and IL-1β singly deficient animals each exhibited lower bacterial counts than Il1r1^-/- or Il1a,Il1b^-/- mice, highlighting a dual requirement for both IL-1 species and suggesting a possible synergy in the anti-bacterial function of the two cytokines. Indeed, when infected with a lower dose (<50 CFU) of Mtb both Il1a^-/- and Il1b^{-/- -/-} mice survived longer than Il1r1^-/- animals (Figure S1). Taken together these observations argue that IL-1α and IL-1β each have critical non-redundant functions and co-operate in IL-1R1 mediated host resistance to Mtb.

Figure 1. Mtb infected Il1a^-/- mice display acute mortality and elevated bacterial loads comparable to Il1b^-/- animals.

(A) Body weight and survival of WT, Il1r1^-/-, Il1a,Il1b^-/-, Il1b^-/-and Il1a ^-/- mice after aerosol exposure to Mtb (H37Rv).

(B) Bacterial loads measured 25 days p.i. in lungs of the above mouse strains.

(C) Indicated cytokines were measured by ELISA in BAL fluid of WT (white bars), Il1r1^-/- (black bars), Il1a,Il1b^-/- (dark grey bars), Il1b^-/- (light grey bars) and Il1a^-/- (grey and white striped bars) mice and the means ± SD depicted. Dotted lines indicate the limits of detection of the respective assays. Data are representative of a minimum of 2 independent experiments each involving 5-10 mice per group. * denotes significant (p≤0.05) differences compared to WT controls.

IL-1α and IL-1β are co-expressed by CD11b^pos cells in the lungs of Mtb infected mice

Studies in which we measured IL-1α,β proteins in lung homogenates of Mtb infected mice indicated that the cytokines are induced during the acute phase beginning at 2 wks and peak at 3-4 wks post infection (data not shown). For this reason we chose d25-d28 as the time point for cellular analysis of IL-1 production in our studies. Initial experiments utilizing Il1a,Il1b^-/-→WT BM chimeras confirmed a requirement for hematopoietic cell-derived IL-1 in host resistance (data not shown). Ex vivo cell sorting experiments from Mtb infected lungs were then performed 4 wks p.i. to assess IL-1 production by hematopoietic cells of which mononuclear phagocytes and neutrophils are the major candidates for production of the cytokine in vivo. We generated lung single cell suspensions and FACS sorted non-T and -B cells into CD11b^neg cells, Ly6G^hiCD11b^hi neutrophils, and Ly6G^negCD11b^pos cells and stimulated the cells over night in the presence or absence of live Mtb. IL-1α,β production was found to be largely confined to the Ly6G^neg CD11b^pos sorted fraction, while neutrophils generated little and Ly6G^negCD11b^neg cells undetectable amounts of IL-1 (Figure 2A).

Figure 2. IL-1α and IL-1β are co-expressed by two pulmonary CD11b^pos populations, distinguished by Ly6C, CD11c, CD13 and CD282 expression during Mtb infection.

(A) Pulmonary single cell suspensions from 4 wks infected mice were FACS sorted based on Ly6G and CD11b expression into three populations and subsequently stimulated for 18hrs with H37Rv (Mtb) or left unstimulated (unst.). Data shown are means ± SD of IL-1α and IL-1β in culture supernatants.

(B) Lung cells from naïve or 4 wk infected mice were stimulated for 5hrs with (5hr Mtb) or without (unst.) H37Rv and then stained intracellularly for IL-1α and IL-1β. Numbers indicate mean percentage (± SD) of IL-1 producing cells in depicted gate.

(C) Data depict the mean number of IL-1βproducing (± SD) cells per lung at various time points after infection.

(D) CD68 staining of IL-1 producing cells and use of Ly6C and CD11c for further subsetting (numbers indicate percentage of respective population in depicted gate ± SD).

(E) Proportion of IL-1α,β co-producing cells 4 wks p.i. in each subset after re-stimulation for 5hrs with (Mtb, dark circles) or without (unst., white circles) H37Rv. Each connecting line depicts an individual animal.

(F) CD282 and CD13 expression in correlation with CD11c and IL-1α. Numbers indicate percentage of IL-1 producing cells in depicted gate after re-stimulation for 5hrs with Mtb. Data in A-F are representative of a minimum of 2 experiments with 3-10 mice each.

To analyze IL-1 production at the single cell level we performed intracellular cytokine staining (ICS) of cells from digested lung. Importantly, we observed strong IL-1α,β staining in cells derived from 4 wk infected but not naïve lungs after 5hr ex vivo re-stimulation with irradiated (or live) H37Rv (Figure 2B and data not shown). The latter observation argues that the potential for IL-1 expression by pulmonary CD11b^pos cells is acquired in the context of Mtb infection in vivo likely via changes in cellular composition, recruitment and activation-status, rather than through the 5hr re-stimulation per se. Consistent with the above findings on FACS isolated cell populations, Ly6G^neg CD11b^pos cells expressed the most IL-1α,β by intracellular staining, while neutrophils from infected lungs stained weakly for IL-1α but not IL-1β and CD11b^neg cells were negative for both cytokines (Figure 2B and data not shown). Importantly, the ICS analysis revealed that in the Ly6G^neg CD11b^pos population IL-1α and IL-1β are co-expressed at the single cell level. Moreover, in the early stages of infection the kinetics of the appearance of pulmonary IL-1α,β double-positive (IL-1α,βDP) Ly6G^neg CD11b^pos cells closely mirrored the cytokine proteins measured by ELISA in lung homogenates of infected mice (Figures 2C and data not shown). Together these data show that Ly6G^neg CD11b^pos cells and not neutrophils are the major hematopoietic source of IL-1 in the lungs of infected mice and that this population coordinately expresses IL-1αβ at the cellular level.

The IL-1α,β producing cells in the lungs of Mtb infected mice consist of two major populations distinguishable by their expression of CD11c and Ly6C

To discriminate between myeloid vs. lymphoid CD11b^pos cells we included the intracellular macrophage-myeloid marker CD68 (Figures 2D and S2A) and found that 4 wks post infection about 30-40% (31±8%) of pulmonary Ly6G^negCD11b^pos cells were of non-myeloid origin, such as NK and T cells (Figure 2D). This analysis revealed that within the CD11b^pos population IL-1α,β expression is restricted to CD68^pos cells (Figure 2D and data not shown). Moreover, CD68^pos, CD11b^pos myeloid cells segregated into two populations based on differential CD11c and Ly6C expression, with the CD11c^pos subset expressing varying degrees of Ly6C (designated hereafter as iDC) while the CD11c^neg cells displayed uniformly high expression of Ly6C (designated hereafter as iM) (Figures 2D-F). Importantly, the IL-1α,β co-producing cells also dissociated into the same two populations based on CD11c and Ly6C expression (Figures 2D, E and data not shown). In addition to CD11c itself, we observed that IL-1α,βDP̣pulmonary iDC cells can also be distinguished from iM cells by their high expression of CD13 (aminopeptidase N) and CD282 (TLR2) (Figure 2F and data not shown).

Further phenotypic characterization revealed that iM cells display striking homogeneity with unimodal expression of all markers screened and thus likely represent a single population (Figure S3). In contrast, expression of most markers by the iDC cells was heterogeneous suggesting the presence of multiple sub-populations. iDC cells, when compared to iM cells also displayed higher expression of markers associated with antigen presentation suggesting that the iDC subsets consists of multiple DC populations (Figure S2B, C). iDC IL-1α,β expressing cells represent one such sub-population displaying a unique phenotype characterized by selective up-regulation of CD14, CD206, CD210, CD64, CD30L, CD70 and CD38 among others (Figures 2B, C and D). Of note, we found that both CD11b^pos subsets expressed the ligand binding chain of the IFNγ receptor (CD119, IFNγR1) as well as IFN inducible markers such as Ly6C, PD-L1, MHC class I+II, Sca-1, Fcγ receptors (Figure S2B-D) and STAT1 phosphorylated at tyrosine 701 (Figure S2E) arguing that both cell populations are subject to IFN mediated signals in vivo during Mtb infection.

Taken together, we identified two distinct pulmonary CD11b^posCD68^pos myeloid subsets based on Ly6C and CD11c expression, that are capable to co-produce both IL-1α and IL-1β at the single cell level after Mtb infection.

Pulmonary IL-1α,β expressing myeloid cells are multifunctional but distinct from IL-12,23p40 producing cells during Mtb infection

We next functionally characterized the two IL-1α,βDP CD11b^pos myeloid populations present in the Mtb infected lung for their ability to produce TNFα, iNOS, IL-10 and IL-12,23p40, mediators previously shown to play important roles in host resistance to the pathogen (Cooper et al., 2011; Redford et al., 2011). The iM cells were found to produce TNFα, iNOS and IL-10 but not IL-12,23p40. In contrast, robust IL-12,23p40 as well as TNFα, iNOS and IL-10 production was observed in iDC cells (Figure 3A-D). To determine whether expression of these mediators correlates with IL-1 production we co-stained the same populations for IL-1α,β. In both populations iNOS, TNFα and IL-10 expression was largely restricted to IL-1α,βDP cells (Figure 3B,C, D and data not shown). Importantly, within the iDC subset IL-1α,βDP cells failed to make IL-12,23p40, which was instead produced by a separate functionally distinct DC sub-population (Figure 3B,C, D and data not shown). Thus, IL-1 producing myeloid cells in the lungs of Mtb infected mice display a high degree of functional heterogeneity at both the population and single cell level and are key producers of the anti-mycobacterial effector molecules iNOS and TNFα. Taken together the above phenotypic and functional analyses provided a platform for investigating the regulation of IL-1α,β cytokine production by myeloid cells at the cellular level during Mtb infection in vivo (Figure S5A).

Figure 3. IL-1α,β co-expressing cells in the lungs of Mtb infected mice are highly poly-functional but distinct from IL-12,23 p40 producing cells.

(A-D) Quantitative and qualitative analysis of cytokine production in indicated subsets 28 days after Mtb infection of WT mice.

(A) Frequencies of iNOS, TNFα, IL-10 and IL-12,23p40 expressing cells within iM (top) and iDC (bottom) subsets after re-stimulation for 5hrs with (Mtb, dark circles) or without (unst., white circles) H37Rv determined by ICS. Each connecting line depicts an individual animal.

(B) Co-staining of IL-1α with indicated cytokines and iNOS.

(C) Percentage of cytokine producing cells within IL-1α expressing cells of the iM (top) or iDC (bottom) subset.

(D) Simultaneous analysis of the functional profile of pulmonary iM (top) and iDC (bottom) subsets after Mtb infection on the basis of IL-1a, IL-1β, iNOS, and TNFα expression. All combinations of the possible cytokine expression patterns are marked on the x-axis, whereas the percentages (mean ± SD) of the distinct cytokine-producing subsets within iM or iDC cells are shown on the y-axis. The data is summarized in pie charts and each slice corresponds to the proportion of iM or iDC cells expressing a given combination of cytokines indicated by the colored boxes at the bottom of the x-axis.

Data in A-D are representative of a minimum of 3 independent experiments with 3-6 animals each.

IFNs negatively regulate the IL-1 pathway in Mtb infected mouse and human mononuclear phagocytes in vitro

Type I and type II IFNs have been implicated in the regulation of the IL-1 pathway in vitro as well as in autoimmune and infectious diseases (Hu et al., 2005; Schindler et al., 1989; Thacker et al., 2010; Tilg et al., 1993). As noted above the major IL-1α,β producing subsets in Mtb infected mice exhibited up-regulated expression of IFN-inducible markers and displayed STAT1 phosporylation. These observations suggested the possible involvement of Type I IFN and IFN-γ signaling in regulating IL-1 expression in the response of myeloid cells to Mtb infection.

To study the effects of type I and type II IFNs on IL-1α,β production in the context of Mtb, we first compared the IL-1α,β response of murine and human mononuclear phagocytes exposed in vitro to Mtb in the presence or absence of IFNs (Figure 4). We found that pICLC-induced Type I IFN and recombinant IFNγ potently inhibited IL-1α,β cytokine production by murine bone marrow derived macrophages (BMMΦ) after Mtb exposure (Figure 4A, B). In contrast, in bone marrow derived dendritic cells (BMDC), IL-1α,β were suppressed by IFNγ but not pICLC. As expected, pICLC significantly inhibited IL-1 production by Ifngr1^-/- but not Ifnar1^-/- BMMΦ and in addition IFNs did not universally down-regulate pro-inflammatory cytokines since TNFα protein was upregulated in the presence of pICLC and IFNγ in both cell types (data not shown). IL-27, which also utilizes STAT1 for signal transduction, was unable to inhibit IL-1 expression arguing that the suppressive effects of IFNs reflect specific functions of the latter cytokines rather than STAT1 activation per se (Figure 4A).

Figure 4. Type I and II IFNs negatively regulate IL-1α and IL-1β secretion by murine and human myeloid cell subsets infected with Mtb.

(A) Murine cytokines measured by ELISA in supernatants of BMMΦ and BMDC from WT mice after exposure to live Mtb (MOI:1) for 48hrs in the presence or absence of recombinant murine IL-27, IFNγ or pICLC.

(B) IL-1β protein in supernatants of BMMΦ and BMDC from WT, Ifnar1^-/- or Ifngr1^-/- mice incubated with IFNγ or pICLC as indicated after exposure to live Mtb (MOI:1).

(C) Human IL-1α $◻◻\underset{.}{◻}◻\widetilde{◻}1\beta$ measured by ELISA in culture supernatants of human monocyte-derived macrophages (MΦ) or monocyte-derived DCs from 21 healthy donors after 24hrs exposure to live Mtb (MOI:5) in the presence or absence of pICLC or the recombinant human cytokines IFNβ and IFNγ. Horizontal lines indicate the median values. *** denotes significant (p < 0.0001) differences compared to Mtb exposure alone.

(D) IL-1Ra protein in culture supernatants of human MΦ or DCs (top panels) and murine BMMΦ and BMDC (bottom panels) incubated as noted in A and B.

(E) Cytokines in supernatants of BMMΦ derived from WT or Ifnar1^-/- mice after exposure to live Mtb (MOI:1) for 24hrs.

(F) IL-10 protein in supernatants of WT BMMΦ incubated with increasing amounts of pICLC for 40hrs in the presence or absence of Mtb infection.

(G) IL-1β protein in supernatants of WT BMMΦ after exposure to live Mtb (MOI:1) for 24hrs incubated with recombinant murine IFNβ in the presence or absence of neutralizing IL-10 mAb.Murine data presented are the means ± SD and representative of 2-5 independent experiments. * denotes significant (p≤0.05) differences compared to Mtb exposure alone or as indicated with connecting lines.

In order to test the relevance of these observations to the Mtb response of humans, we infected monocyte-derived macrophages (MΦ) or monocyte-derived DCs from 21 healthy blood donors with Mtb and assessed their ability to generate IL-1α,β in the presence or absence of IFNs. In agreement with a recent study from our group focusing on differences in IL-1β regulation in virulent vs. avirulent mycobacteria in human monocytes/macrophages (Novikov et al., 2011), we found that IFNβ inhibits production of the former cytokine in macrophages. Here we show that in Mtb infected DCs as well as MΦ pICLC and IFNβ suppress both IL-1α and IL-1β and that IFNγ inhibits IL-1β but not IL-1α production (Figures 4C). Despite the statistically significant suppression in IL-1β by IFNγ, there was a substantial fraction of donors that responded to Mtb infection with lower amounts of IL-1β (52% ≤ 4ng/ml (MΦ), 8ng/ml (DC):low responders; LR) and in these individuals little to no additional reduction in IL-1β protein was observed after IFNγ treatment (MΦ: 29.6±23.4% reduction in LR; 66.9±13.4% reduction in high responders; HR; DCs: 38.5±30.1% reduction in LR;78.7±0.2% reduction in HR) while pICLC or IFNβ treatment potently suppressed IL-1β by both LR and HR (>70±5%; Figure 4C and data not shown). This donor variation may explain the divergent results in our previous study (Novikov et al., 2011) where we failed to observe suppression of IL-1β by IFNγ in human macrophage.

In the murine as well as human systems, we observed up-regulation of IL-1R antagonist (IL-1Ra) in a dose dependent manner by type I and II IFNs in each cell type with the exception of murine DCs (Figure 4D and data not shown). Thus, in both murine and human mononuclear phagocytes IFNs antagonize the IL-1 pathway by dual mechanisms: first by directly suppressing IL-1α,β cytokine production and secondly by up-regulation of the soluble endogenous IL-1Ra, which opposes IL-1 by competing for IL-1R1 binding.

When we investigated possible sources of type I IFN we found that both murine BMMΦ and BMDC generated detectable amounts of IFNβ in response Mtb infection (Figure 4E). Type I IFNs are potent inducers of IL-10 (Chang et al., 2007) and IL-10 has recently been reported to type I IFN's suppressive effects on IL-1 (Guarda et al., 2011). In the context of Mtb infection in vitro we observed that IL-10 production is critically dependent on IFNαR signaling and that IL-10 is induced by IFNβ and pICLC in a dose dependent manner in BMMΦ and BMDCs (Figure 4E, F and data not shown). To investigate the contribution of IL-10 in type I IFN mediated suppression of IL-1 during Mtb infection in vitro we added exogenous IL-10 to Mtb infected BMDM and observed a significant inhibition of IL-1β production (Figure 4G). More importantly, the Mtb induced IFNβ mediated suppression of IL-1β was partially reversed when endogenous IL-10 was neutralized (Figure 4H). Thus in vitro, besides direct suppressive effects, type I IFNs can also indirectly inhibit IL-1 production during Mtb infection via IFNαR dependent induction of IL-10.

Endogenous type I IFN suppresses IL-1α,β production by both iM and iDC cells in vivo

A pro-bacterial role for endogenous type I IFNs has been demonstrated previously in experiments in which Ifnar1^-/- mice were shown to be less susceptible to Mtb infection (Manca et al., 2005; Stanley et al., 2007). As shown in Figure 5A and 5B this decrease in bacterial burden was closely associated with an increase in IL-1α,β production by pulmonary myeloid cells.

Figure 5. Endogenous type I IFNs suppress IL-1α,β expression by both iM and iDC pulmonary subsets during Mtb infection.

(A) Pulmonary bacterial loads measured 4 wks p.i. in WT or Ifnar1^-/- mice.

(B) ICS for IL-1α,β by pulmonary myeloid cells in WT or Ifnar1^-/- mice 4 wks p.i. Data are representative of 2 independent experiments each involving 3-5 mice per group. * denotes significant (p≤0.05) difference compared to WT control. Numbers indicate mean percentage (± SD) of IL-1 producing cells in depicted gate.

(C) WT CD45.1,1 mice were lethally irradiated and reconstituted with equal ratios of WT (CD45.1,2) and Ifnar1^-/- (CD45.2,2) BM cells and infected with Mtb.

(D) Analysis of donor BM derived CD11b^pos myeloid cells 4 wks p.i. in isolated lung cells marked by CD45.1 and CD45.2 expression (percentage ± SD) and frequency of IL-1α,β expression by WT (white circles) or Ifnar1^-/- (KO, dark circles) total CD11b^pos mononuclear myeloid cells after re-stimulation for 5hrs with (Mtb) or without (unst.) Mtb. Each connecting line depicts an individual animal.

(E) Frequency of IL-1α,β expression by iM and iDC subsets in mixed Ifnar1^-/- BM chimeric mice.

(F-G) Frequencies of iNOS, TNFα and IL-10 expressing cells within pulmonary WT (white circles) or Ifnar1^-/- (KO, dark circles) iM cells (F) and within iDC (G) cells.

Data in A-F are representative of three independent experiments with 3-5 mice each.

* denotes significant (p≤0.05) differences compared to WT controls (ns= not significant).

To test whether endogenously induced type I IFNs can act directly on IL-1α,β producing cells during Mtb infection in vivo we generated mixed bone marrow chimeras allowing us to compare Ifnar1^-/- and WT myeloid cells in the lungs of the same animal. This approach controls internally for potential differences in bacterial load, inflammatory milieu and/or cellular microenvironment. WT CD45.1,1 mice were lethally irradiated and reconstituted with BM from WT (CD45.1,2) donors mixed at a 1:1 ratio with Ifnar1^-/- CD45.2,2 donors (Figure 5C). The mixed Ifnar1^-/-,WT BM chimeras were then infected with Mtb and 4 wks later the frequency of IL-1α,βDP cells within pulmonary WT or Ifnar1^-/- cells was analyzed (Figure 5D). Importantly, in both the iM and the iDC subsets the inability of Ifnar1^-/- cells to receive type I IFN signals resulted in a marked increase in the proportion of IL-1α,βDP, even without Mtb re-stimulation (Figure 5D, E). In addition, we found that type I IFN signaling suppressed iNOS production by iM and iDC cells while in accordance with the in vitro data presented above it also induced IL-10 expression by myeloid cells in vivo after Mtb infection (Figure 5F, G, S5). In contrast, TNFα expression was not subject to type I IFN regulation. Furthermore, phenotypic analysis revealed that type I IFN signaling influenced expression of only a few IFN inducible and surface markers, most notably CD206 and PD-L1 (Figure S3A). Together these data argued that ligation of the type I IFN receptor on each of the two myeloid populations directly suppresses IL-1α,β production in vivo.

Host-protective IFNγ inhibits IL-1α,β production by iM cells

Because in addition to type I IFNs, IFNγ also displayed potent suppressive effects on IL-1 expression in vitro, we next used the same mixed BM chimeric approach to ask whether IFNγ signaling also regulates IL-1α,β production by myeloid cells during Mtb infection in vivo. In these experiments we constructed mixed Ifngr1^-/-,WT BM chimeric mice and functionally and phenotypically analyzed gene deficient or WT CD11b^pos subsets in the same animal for IFN inducible surface marker expression and their capacity to generate innate cytokines (Figure 6, S3B). Importantly, in the lungs of Mtb infected chimeric mice IL-1 production in the iM subset was markedly increased in the absence of IFNγR while in iDC cells no difference in IL-1α,β expression was observed between cells of WT or Ifngr1^-/-origin (Figure 6A, B). In addition, we found that expression of a number of IFN inducible and surface markers were strongly affected by the absence of IFNγR signaling (Figure S3B). Interestingly, in the iDC cells expression of CD206 (macrophage mannose receptor), a surface molecule recently shown to be a marker of TLR induced monocyte-derived DCs (Cheong et al., 2010), and IL-10R (CD210) were both dependent on IFNγR signaling (Figure S3B). As expected, iNOS expression in both iM and iDC subsets was completely abolished in IFNγR deficient cells compared to WT cells in the same animal. Moreover, while TNFα expression in both myeloid subset was unaffected by the absence of IFNγR, IFNα potently suppressed IL-10 production in iM but not iDC cells (Figure 6C, D, S5). Thus, while clearly a host-protective cytokine in Mtb infection, IFNα in common with type I IFN can also be anti-inflammatory suppressing IL-1α,β production by pulmonary myeloid cells.

Figure 6. IFNγ specifically inhibits IL-1α,β expression in the iM but not iDC subset.

WT CD45.1,1 mice were lethally irradiated and reconstituted with equal parts WT (CD45.1,2) and Ifngr1^-/- (CD45.2,2) BM cells and infected with Mtb.

(A-B) Distribution of donor BM derived pulmonary CD11b^pos myeloid cells 4 wks p.i. marked by CD45.1 and CD45.2 expression (percentage ± SD) and analysis of IL-1α,β expression by iM (A) and iDC (B) subsets gated on WT (WT, white circles) or Ifngr1^-/- (KO, dark circles) after stimulation for 5hrs with (Mtb,) or without (unst.) Mtb. Each connecting line depicts an individual animal.

(C-D) Frequencies of iNOS, TNFα and IL-10 expressing cells within pulmonary WT (white circles) or Ifngr1^-/- (KO, dark circles) iM (C) and iDC (D) subsets.

Data in A-E are representative of three independent experiments with 3-5 mice each.

* denotes significant (p≤0.05) differences compared to WT controls (ns= not significant).

CD4⁺ T cells regulate IL-1 expression during Mtb infection through their production of IFNγ

Th1 cells are a major source of IFNγ during Mtb infection in mice and therefore could be a key mediator of the suppression of IL-1 production observed in Ifngr1^-/-,WT BM chimeric mice. In support of this hypothesis, co-culture of Mtb infected BMMΦ for three days with naïve TCR transgenic CD4 T cells recognizing an epitope in the Mtb Ag85b protein (P25), resulted in reduced IL-1β protein when compared to Mtb infected cells alone (Figure S4A). This reduction occurred in an IFNγ-dependent manner and was antigen-specific, since addition of CD4 T cells (OT-II) with an irrelevant specificity failed to suppress IL-1β production. Interestingly, while addition of exogenous IFNγ potently suppressed IL-1β expression by BMDC (Figure 4A), co-culture with P25 cells did not (Figure S4A). Moreover, when IFNγ signaling was disrupted on the BMDCs, P25 cells were able to enhance, rather than suppress IL-1β production (Figure S4A). This finding suggests that CD4 T cells exert both IFNγ-dependent suppressive as well as IFNγ-independent inductive effects on IL-1 production by DCs.

Adoptive transfer experiments were next designed to test the hypothesis that during Mtb infection CD4 derived IFNγ is sufficient to mediate the previously observed suppression of IL-1 by iM in vivo. The approach utilized allowed us to study APC-CD4 T cell interactions through targeted manipulation of the genotype of the CD4⁺ T cells transferred into Tcra^-/- mice (Figure S4B). WT or Ifng^-/- CD4⁺ T cells were adoptively transferred into Tcra^-/- mice prior to aerosol infection with Mtb and bacterial loads were measured 4 wks later (Figure S4B, 7A). Re-stimulated lung cells from Tcra^-/- mice reconstituted with Ifng^-/- CD4⁺ T cells (Tcra^-/-+Ifng^-/- CD4) secreted more IL-1β when compared to lung cells from Tcra^-/- mice that received WT CD4⁺ T cells (Tcra^-/-+WT CD4) despite similar pulmonary bacterial loads in the experimental groups at this early time point (Figure 7A, B). Single cell analysis of the IL-1 producing subsets revealed that iM cells from Tcra^-/-+Ifng^-/- CD4 mice produced significantly more IL-1α,β when compared to cells from Tcra^-/-+WT CD4, even without re-stimulation ex vivo (Figure 7C). Moreover, consistent with our findings in mixed BM chimeric mice, IL-1α,β production by iDC cells was unaffected in the absence of CD4⁺ T cell derived IFNγ while in contrast, iNOS expression was critically dependent on IFNγ from this source (Figures 7D and S4C, S5). These findings thus reveal an anti-inflammatory function for Th1 cell derived IFNγ in modulating IL-1 production by myeloid cells in vivo during Mtb infection.

Figure 7. CD4⁺ T cell derived IFNγ is sufficient to suppress IL-1 production by iM cells in vivo.

(A) Bacterial loads were measured in lungs of WT, □□□□^-/- and □□□□^-/- mice reconstituted with WT or Ifng^-/- CD4⁺ T cells. * denotes significant (p≤0.05) differences compared to WT controls (ns= not significant).

(B) Pulmonary single cell suspensions from animals in the indicated experimental groups were incubated with Mtb and IL-1β was measured by ELISA in the culture supernatant after 8 hrs. Data shown are derived from a pool of 3-5 mice per group.

(C-D) ICS for IL-1α and IL-1β expression by iM (C) and iDC (D) subsets in lungs of □□□□^-/- mice reconstituted with WT (left plot, white squares) or Ifng^-/- (right plot, dark squares) CD4⁺ T cells and frequencies of IL-1α,β expression after stimulation for 5hrs with (Mtb) or without (unst.) Mtb.

Data in A-E are representative of two independent experiments with 3-5 mice each.

* denotes significant (p≤0.05) differences compared to WT controls (ns= not significant).

DISCUSSION

Whereas myeloid cells are the primary targets of mycobacterial infection (Wolf et al., 2007) they are also pivotal effector cells that mediate control of intracellular bacterial growth and orchestrate the inflammatory response to the pathogen. Previous studies have emphasized the roles of myeloid derived iNOS and TNFα as important effector molecules for bacterial control of Mtb in the mouse model (Saunders et al., 2004; Skold and Behar, 2008). We found that IL-1α and IL-1β each have critical non-redundant functions in preventing host mortality but appear to co-operate in mediating IL-1R dependent bacterial control. Indeed, a role for IL-1α in host resistance to Mtb is supported by a recent study in which the induction of autoantibodies against the cytokine resulted in increased mortality during chronic Mtb infection (Guler et al., 2011). IL-1α and IL-1β are also traditionally regarded as myeloid cell derived pro-inflammatory cytokines. Their in vivo source during Mtb infection had not been examined, and therefore it was unclear whether IL-1 derives from the same cell populations that produce the major iNOS and TNFα effector molecules. Similarly, the mechanisms that regulate IL-1 expression during Mtb infection in vivo are poorly understood, particularly the possible role of other cytokine pathways in controlling IL-1 production.

In characterizing the innate immune sources of IL-1α and IL-1β we identified three myeloid cell types (iM, iDC and neutrophils) differentially producing the cytokines in the lungs of Mtb infected mice. Surprisingly, neutrophils did not produce significant amounts of IL-1β. Given our previous findings on the caspase-1 independent processing of IL-1β in vivo during Mtb infection (Mayer-Barber et al., 2010) the neutrophil represented a logical candidate for the cellular source of IL-1β during Mtb infection. Although not producing pro-IL-1β, neutrophils could still play a role in the IL-1 response to Mtb in mice by secreting proteases that would cleave extracellular pro-IL-1β (Dinarello, 2010).

A key finding of the present study is that in lungs of Mtb infected mice, two major CD11b^pos iM and iDC subpopulations co-produce IL-1α and IL-1β at the single cell level. These cells also produce large amounts of the host-protective mediators iNOS and TNFα. The iM (Ly6C^hi, Cd11c^neg, CD282^int, CD13^neg) cells represent a single mononuclear phagocyte population and produce predominantly IL-1α, IL-1β and TNFα They display phenotypic markers characteristic of inflammatory monocytes/macrophages previously described in non-lymphoid tissue (Dunay et al., 2008; Geissmann et al., 2010; Varol et al., 2009). In addition they share similar functional properties, e.g. IL-10 and iNOS production, with myeloid derived suppressor cells that have been extensively studied in long-term unresolved pathological conditions such as chronic infections, inflammation and cancer (Biswas and Mantovani, 2010).

In contrast, the iDC (CD11c^pos, Ly6C^neg-int, CD13^pos, CD282^int,hi) cells express high levels of MHCII, CD80 and CD86 as well as additional markers associated with antigen presentation and APC-T cell interactions and likely reflect pulmonary DCs. Based on their cytokine expression profiles these cells contains at least two functionally distinct DC subsets, one producing IL-1α,β (CD282^hi, Ly6C^pos) in concert with iNOS, TNFα and IL-10 and the second producing IL-12,23p40 (CD282^int, Ly6C^neg). The IL-1α,β producing iDC cells closely resemble monocyte derived inflammatory DCs, an iNOS and TNFα producing DC subset that has been previously implicated in murine resistance to intracellular bacteria, parasites or viruses and in human psoriasis (De Trez et al., 2009; Lin et al., 2008; Lowes et al., 2005; Serbina et al., 2003). In contrast, the IL-12,23 p40 expressing iDC cells are similar to a subset of conventional DCs previously characterized in lung (GeurtsvanKessel and Lambrecht, 2008). Moreover, both IL-1α,β and IL-12,23 p40 producing DC subsets were detected in the lung draining lymph node (data not shown) and it is possible that these two functionally separate DC populations play important yet distinct roles in T helper cell differentiation during Mtb infection. Thus, IL-12 is required for the generation of IFNγ producing Th1 cells (Cooper et al., 2011), while as described here IL-17 protein and Th17 responses (data not shown) were diminished in lungs of IL-1 deficient animals infected with Mtb. Such a functional division of labor between DC populations could play a key factor in determining Th lineage fate decisions in lymph nodes and peripheral tissue sites during infection.

Inflammasome activation is not required for host-resistance and IL-1β processing during Mtb infection in vivo (Mayer-Barber et al., 2010), so we focused on mechanisms that could regulate the induction and/or function of both IL-1α and IL-1β. Previous in vitro studies had found that type I and type II IFNs modulate pro-IL-1β expression (Guarda et al., 2011; Guarda et al., 2009; Masters et al., 2010; Novikov et al., 2011). To investigate the role of IFNs in regulating IL-1α,β production in vivo at the cellular level we utilized mixed BM chimeras that allowed the side-by-side comparison of WT and gene deficient populations in the same animal.

Type I IFNs had a major suppressive effect on both IL-1α and IL-1β production by the iM and iDC cells in the lungs of Mtb infected mice. The precise mechanism by which IFNγR signaling mediates suppression of IL-1α,β expression by these myeloid subsets is unclear. In the recent study by Guarda et al. it was shown that type I IFN inhibits LPS induced pro-IL-1β expression via IL-10 (Guarda et al., 2011). Moreover, IL-10 has been shown to promote susceptibility to Mtb in mice (Redford et al., 2010; Redford et al., 2011; Turner et al., 2002). Here, we implicate IL-10 in type I IFN mediated IL-1 suppression in vitro during Mtb infection and extend these observation in vivo by showing that IL-10 production by pulmonary iM and iDC cells during Mtb infection is type I IFN dependent. Furthermore, we show that this IL-10 production is limited primarily to IL-1α,β expressing cells. While IL-10 may be involved indirectly in the type I IFN mediated suppression of IL-1 in vivo, our mixed BM chimera experiments clearly argue that IFNαR-signals act directly on the cytokine producing cells to inhibit IL-1α,β expression in the lungs of Mtb infected mice.

Although inhibition of IFNγR (CD119) expression has been implicated in the pro-bacterial effects of type I IFNs (Rayamajhi et al., 2010b), we failed to observe IFNαR mediated down-regulation of CD119 during Mtb infection. Rather our data reveal that type I IFNs act on multiple innate immune cell types to coordinate a multifaceted anti-inflammatory response to Mtb infection, simultaneously suppressing IL-1α,β and iNOS, while promoting expression of immunosuppressive IL-10 and IL-1Ra. There is now growing evidence that induction of type I IFN directly serves the pathogen as a host evasion mechanism to promote bacterial survival (Rayamajhi et al., 2010a; Trinchieri, 2010) and our findings argue that this may stem from combined repression of a collection of critical host protective pathways in which IL-1 should now be included as a member.

In addition to type I IFNs, we demonstrated that type II IFN (IFNγ) also modulates IL-1 expression in vivo during Mtb infection. However, we found that IFNγ suppressed IL-1α,β production in the iM but not the iDC population. This observation was unexpected since both subsets not only expressed IFNγR1 but were also highly responsive to IFNγ during Mtb infection as evidenced by the ablation of iNOS expression in the Ifngr1^-/- cells. There are several possible explanations for this differential regulation. Firstly, the iM subset represents a single cellular population, while iDC cells are comprised of multiple subpopulations. In the absence of IFNγR expression the iDC subpopulations underwent major phenotypic changes raising the possibility of selective outgrowth and/or recruitment of receptor deficient cells under these conditions. Secondly, the two subsets may integrate IFNγR mediated signals in a qualitatively different fashion (Hu et al., 2006; Masters et al., 2010). Lastly, the differential regulation of IL-1 production by inflammatory monocytes/macrophages and DCs could reflect opposing activities of Th1 cells on DCs. CD4 T cells have been shown to be capable of both suppressing and inducing IL-1β expression in myeloid cells (Guarda et al., 2009; Jayaraman et al., 2010). Indeed, our in vitro studies with BMDCs support the hypothesis that CD4 T cells suppress IL-1β via IFNγ while simultaneously enhancing its expression through an IFNγ-independent mechanism, as previously proposed with Tim-3 (Jayaraman et al., 2010). These two counter-regulatory effects mediated by Th1 cells may cancel each other out thereby explaining the apparent lack of IFNγ mediated suppression of IL-1α,β production in the iDC subset in vivo.

An intriguing paradox raised by our findings concerns the apparently opposing functions of IFNγ in mediating host-resistance to this pathogen and in suppressing the host protective cytokines IL-1α and IL-1β. On the one hand, IFNγ promotes inflammation by inducing up-regulation of MHC class II and iNOS in iM and iDC cells while suppressing IL-10 expression in iM cells. On the other hand, IFNγ mediates its anti-inflammatory effects by suppressing IL-1α,β production in iM cells, up-regulating IL-10R expression on iDC cells, and acting as the principal inducer of the inhibitory ligand PD-L1. Our findings provide a context for the role of IL-1 in the complex pro-inflammatory and suppressive pathways simultaneously induced by IFNγ: in contrast to other host-protective factors induced by IFNγ, IL-1 is actively suppressed.

Collectively, our data demonstrate that the host detrimental type I IFN and host protective IFNγ both down-modulate IL-1α,β production. This situation may have evolved to mutually benefit the bacteria and the host by simultaneously inhibiting IL-1 dependent control of infection while limiting immunopathology. Although the in vivo findings presented here derive exclusively from studies in the murine model, their relevance to human Mtb infection is supported by the recent work of Berry et al. (2010) in which a transcriptional profile marked by type I and II IFN induced genes was found to correlate with disease severity in patients. Our data linking IFN signaling with IL-1 suppression provide a testable hypothesis for explaining this correlation between IFN expression and human disease.

EXPERIMENTAL PROCEDURES

Mice and Mtb infections

C57BL6 (WT, B6) and Ifngr1^-/- mice were purchased from Taconic Farms (Hudson, NY) and Jackson Laboratories (Bar Harbor, ME) respectively. B6.SJL (CD45.1,1), Tcra^-/-, Ifng^-/-, Ifnar1^-/-, Il1r1^-/-, OT-II mice were obtained through a supply contract between the National Institute of Allergy and Infectious Diseases (NIAID) and Taconic Farms and were backcrossed to B6 for a minimum of 10 generations. Ag85b-specific P25 TCR transgenic mice, originally generated by Takatsu and colleagues were kindly provided by Joel Ernst (NYU). Il1a^-/-, Il1b^-/-, Il1a,Il1b^-/- mice were originally derived by Y. Iwakura (Tokyo University) and generously supplied by Tod Merkel (FDA). All animals were maintained in an AALAC-accredited BSL2 or BSL3 facilities at the NIH and experiments performed in compliance with an animal study proposal approved by the NIAID Animal Care and Use Committee. Aerosol infection with the H37Rv strain of Mtb (100-150 CFU/mouse unless noted otherwise) and determination of lung bacterial growth were performed as previously described (Mayer-Barber et al., 2010).

Macrophage and DC differentiation and in vitro cultures

Murine BM cells were cultured for 7 days in either 15% GM-CSF media to generate BMDC or 30% L929 supernatant media to differentiate BMMΦ and then exposed to H37Rv at a multiplicity of infection of one (MOI:1) in the presence or absence of 30ng/ml IL-27 (Ebioscience, San Diego, CA), 200U/ml IFNγ (Ebioscience), 30ng/ml IL-10 (Ebioscience), 30ng/ml IFNβ (R&D Systems, Minneapolis, MN) and 20μg/ml pICLC (kindly provided by A. Salazar, Oncovir Inc.) for 24-48hrs and supernatants harvested. In some experiments P25 or OTII CD4 T cells were enriched from spleen and lymph nodes of uninfected mice by magnetic bead separation (Miltenyi Biotech, Auburn, CA) and used for 60hr co-culture at a ratio of 2:1 (T cells:BM derived cells).

Human elutriated monocytes were obtained from peripheral blood of 21 healthy CMV negative donors. MΦ were generated by culturing monocytes with media containing M-CSF (60ng/ml ) for 7 days and DCs after 5 days of culture in the presence of GM-CSF (800U/ml) and IL-4 (1000U/ml) followed by maturation for 48hrs with TNFα (20ng/ml). Fresh media with indicated growth factors was added every 48hrs. Cells were exposed to H37Rv (MOI=5) in the presence or absence IFNγ (10ng/ml), IFNβ (10ng/ml) or pICLC (10μg/ml) for 24hrs. All recombinant human cytokines were from Peprotech (Rocky Hill, NJ).

Cytokine and nitric oxide measurements in culture supernatants and biological fluids

Murine and human cytokine concentrations in culture supernatants were quantitated using commercial ELISA kits (R&D Systems, Minneapolis, MN). Bronchoalveolar lavage (BAL) fluid and cell free lung homogenates were obtained and cytokines and nitric oxide measured as described previously (Mayer-Barber et al., 2010).

Flow cytometry

Abs against mouse surface antigens and cytokines were purchased from Ebioscience, Biolegend (San Diego, CA), AbD Serotec (Raleigh, NC) and BD Pharmingen (San Diego, CA) and used in 12-color flow cytometry either biotinylated or directly conjugated. The Abs used were directed against Ly6C (clone AL-21), Ly6G (1A8), CD13 (R3-63), CD11c (HL3 and N418), CD282 (6C2), CD68 (FA-11), CD45.1 (A20), CD45.2 (104), I-A^b (M5/114.15.2), IL-1α PE (ALF-161), IL-1β APC (polyclonal and replacement pro-IL-1β monoclonal NJTEN3), iNOS (polyclonal), IL-12,23 p40 (C15.6), TNFα (MP6-XT22), IL-10 (JES5-16E3), CD11b (M1/70), TCRβ (H57-597), CD19 (6D5), NK1.1 (PK136), 7/4 (7/4), IL10R/CD210 (1B1.3a), CD206 (MR5D3), IFNγRI/CD119 (GR20), CD14 (Sa14-2), PDL-1/CD274 (MIH5), Sca-1(D7), CD16/32 (93). Biotinylated antibodies were detected with streptavidin-conjugated Qdot 605 from Molecular Probes-Invitrogen (Carlsbad, Ca). Ultraviolet fixable live/dead cell stain was purchased from Molecular Probes-Invitrogen and used according to the manufacturer's protocol. All samples were acquired on a LSRII flow cytometer (BD Biosciences, San Jose, CA) and analyzed using FlowJo software (Three Star, Ashland, OR).

Isolation and re-stimulation of myeloid cells from lung tissue

Perfused lungs from infected mice were cut into 1-2mm pieces and subsequently digested with Liberase Cl (0.4mg/ml in PBS; Roche-Diagnostics, Indianapolis, IN). The reaction was stopped after 30-45 min. with an equal volume of fetal calf serum. Digested lung was fully dispersed by passage through a 100μm pore size cell strainer and an aliquot removed for bacterial load measurements. Isolated cells were then washed, counted and re-suspended in media containing monensin (0.1%, Ebioscience) in the presence or absence of 100ug/ml irradiated Mtb H37Rv at 1×10⁶ cells per well in 96 well plates and incubated at 37°C, 5% CO₂. 5 hrs later cells were surface stained, fixed, permeabilized and ICS performed.

Preparation of mixed BM chimeric mice

B6.SJL (CD45.1,1) mice were lethally irradiated (950 rad) and reconstituted with a total of 10⁷ donor BM cells from C57BL/6 CD45.1,2 wild-type (WT) mice mixed at equal parts with BM cells from CD45.2,2 mice deficient (KO) in either IFNγR1 or IFNαR1. Mice were allowed to reconstitute for 8-10 weeks before aerosol infection with H37Rv.

T cell isolation and adoptive transfer

CD4 T cells for adoptive transfer were enriched from the spleen and lymph nodes of uninfected B6 or Ifng^-/- mice by magnetic bead separation to approximately 90%-95% purity, according to the manufacturer's protocol (Miltenyi Biotec) and injected i.v. into Tcra^-/- mice (4×10⁶ cells/animal ) 1-7 days prior to aerosol infection with H37Rv.

Statistical analyses

The statistical significance of differences between data groups was determined using the Mann-Whitney test or the Wilcoxon matched pairs test (mixed BM chimera experiments).