Relative contribution of IL-1α, IL-1β and TNF to the host response to Mycobacterium tuberculosis and attenuated M. bovis BCG

Cross-regulation of IL-1α and IL-1β release by macrophages in response to mycobacteria

Since a mutual induction of IL-1α and IL-1β was reported after stimulation with LPS or turpentine [35], and IL-1β has been identified as a necessary “shuttle” for IL-1α secretion after LPS stimulation [37], we first assessed the interdependence of IL-1α and IL-1β expression in response to mycobacteria in vitro. Bone marrow derived macrophages from mice deficient singly for IL-1α or IL-1β, for both IL-1α and IL-1β, or for IL-1R1 were stimulated with M. tuberculosis H37Rv or M. bovis BCG in vitro and their ability to secrete IL-1α and IL-1β, or TNF, determined. As expected, no IL-1α was detected in the IL-1α deficient mice (Fig. 1A) and no IL-1β was detected in the IL-1β deficient mice (Fig. 1B). The level of IL-1α was partially reduced in IL-1β deficient macrophages stimulated with TLR4 agonist LPS, M. tuberculosis H37Rv or M. bovis BCG, although this did not reach statistical significance. Similarly, the level of IL-1β was slightly reduced in the IL-1α deficient macrophages stimulated with LPS, M. tuberculosis H37Rv or M. bovis BCG. A mutual reduction of IL-1α in the absence of IL-1β and, conversely, a reduction of IL-1β in the absence of IL-1α, were also seen with stimuli triggering different inflammasome pathways such a uric acid crystals (Ref. [36] and unpublished data). Such interdependence of IL-1α and IL-1β release could be due to indirect IL-1 amplification loops, or to the requirement for IL-1β shuttle effect. However, the absence of functional IL-1R1 had little effect on the release of either IL-1α or IL-1β (Fig. 1A and B), pointing against an IL-1/IL-1R1 mediated indirect amplification loop. TNF release induced by mycobacteria was independent of the IL-1α/β/IL-1R1 axis in vitro (Fig. 1C), as expected [10–16]. Conversely, the release of IL-1α and IL-1β in response to mycobacteria was not affected in TNF deficient macrophages (data not shown). Thus, there was some interdependence of IL-1α and IL-1β release after macrophage stimulation with TLR agonists or mycobacteria, which did not affect TNF release.

Lethal M. tuberculosis infection in the absence of both IL-1α and IL-1β, or TNF pathways

We showed earlier that IL-1R1 deficient mice are extremely sensitive to virulent M. tuberculosis H37Rv infection, similar to TNF or MyD88 deficient mice [13,44], and we next wanted to assess the relative contribution of IL-1α and IL-1β to the IL-1R1 mediated immune response to acute M. tuberculosis infection. Indeed, among the early studies addressing the role of IL-1 in host response to M. tuberculosis [10–12], a direct comparison of mice deficient in IL-1α or IL-1β, both IL-1α and IL-1β, IL-1R1 or TNF was missing. Mayer-Barber et al. [16] recently reported a rapid mortality of mice singly or double deficient for IL-1α and/or IL-1β, within 30–40 days post M. tuberculosis infection, although after low dose infection IL-1α or IL-1β single deficient mice succumbed 3–5 months after IL-1R1-deficient mice. Here, mice singly deficient for IL-1α or IL-1β survived the acute phase of infection without bodyweight loss, while mice deficient for both IL-1α and IL-1β, similar to IL-1R1 deficient mice, rapidly lost weight and succumbed within 5 weeks of infection with M. tuberculosis H37Rv, that is, 1 week after TNF deficient mice (Fig. 2A,B). However, mice singly deficient for either IL-1α or IL-1β survived for the duration of the acute M. tuberculosis infection study (3 months) without clinical symptoms of infection (Fig. 2A-C). Therefore, presence of either IL-1α or IL-1β allows some control of acute M. tuberculosis infection.

Exacerbated pulmonary inflammation after M. tuberculosis infection in the absence of IL-1α, IL-1β or IL-1R1, alike in the absence of TNF

We and others reported earlier a dramatic increase in lung inflammation in IL-1R1, MyD88 or TNF deficient mice early after M. tuberculosis infection [10,11,13,43,44]. Since there seems to be little involvement of the major NLRP3 inflammasome pathway leading to IL-1β maturation [31,32], we wished to address further the relative contribution of IL-1α versus IL-1β in the lung inflammation induced during acute M. tuberculosis infection. IL-1α and IL-1β double deficient mice exhibited strongly increased lung weights, an indicator of lung inflammation, 5 weeks post-infection, as did IL-1R1 deficient mice (Fig. 2E) and TNF KO mice which succumbed at 4 weeks (Fig. 2D). Mice deficient only for IL-1α or IL-1β had less pronounced pulmonary inflammation on day 35 post-infection (Fig. 2E). However, IL-1α or IL-1β deficient mice survived to a more chronic infection and lung inflammation was gradually increased in these animals 2 or 3 months post-infection (Fig. 2F and G). Therefore it seemed that expression of either IL-1α or IL-1β helped controlling acute lung inflammation induced by M. tuberculosis infection.

As this may be a consequence of a better contained infection, we next determined the bacterial loads in these animals. Mice double deficient for IL-1α plus IL-1β could not control the infection, similar to IL-1R1 deficient mice on day 35 post-infection (Fig. 2I), although bacterial burdens were not as high as those reached in TNF KO mice which succumbed at 4 weeks (Fig. 2H). Deficiency in either IL-1α or IL-1β allowed some bacterial growth restriction on day 35 post-infection (Fig. 2I), and there was a limited increase in bacterial load at 2–3 months post-infection (Fig. 2J and K). Therefore, mice double deficient in IL-1α plus IL-1β develop a rapid, exacerbated lung inflammation which may be the result of uncontrolled bacterial infection, as seen in IL-1R1 or TNF deficient mice, but expression of either IL-1α or IL-1β was sufficient to contribute to the host control of bacterial growth and lung inflammation during acute M. tuberculosis infection.

Presence of either IL-1β or IL-1α prevents the acute necrotic pneumonia seen after M. tuberculosis infection in IL-1 pathway defective mice, but do not compensate for lack of TNF

The establishment of well-defined granuloma, result of a structured cell mediated immune response, is thought to be crucial for inhibiting mycobacteria growth. In view of the increased lung weight indicative of a strong local inflammation in mice lacking IL-1α and/or IL-1β, and of the defective granuloma formation reported in IL-1R1 deficient mice [13], we next addressed the relative contribution of IL-1α and IL-1β in granuloma formation upon M. tuberculosis infection. Macroscopically, the lungs of mice deficient for both IL-1α and IL-1β displayed pleural adhesions, large confluent nodules, similar to IL-1R1 or TNF deficient mice (Fig. 3A). Microscopic investigation of the lungs of IL-1α plus IL-1β deficient mice revealed severe inflammation with important reduction of ventilated alveolar spaces, massive mononuclear cell and neutrophil infiltrations with extensive confluent necrosis and oedema, in the absence of proper granuloma formation at 35 days (Fig. 3B) and abundant mycobacteria within macrophages and in the extra-cellular space (Fig. 3C). The lung lesions observed in the absence of both IL-1α and IL-1β were similar to those seen in mice deficient for IL-1R1 or TNF, although confluent necrosis was more pronounced in the absence of TNF 4 weeks post-infection (Fig. 3B). A strong iNOS staining confined to macrophages residing in well defined granuloma was seen in wild-type mice on day 35 post-infection (Fig. S1). Pulmonary iNOS expression was attenuated in IL-1α or IL-1β single deficient mice, while in IL-1α and IL-1β double deficient mice and IL-1R1 deficient mice there was a diffuse expression through the inflamed lung at that stage, without strong cell-associated staining as in wild-type mice. Interestingly, mice singly deficient for IL-1α or for IL-1β developed some granuloma with limited oedema, no necrosis and free alveolar space, thus a much less severe lung pathology than that seen in the absence of both IL-1α and IL-1β or of IL-1R1 at 5 weeks post-infection. At 2–3 months post-infection lung pathology in mice deficient for IL-1α or for IL-1β progressed with reduced free alveolar space, increased inflammatory cell infiltration, acid-fast bacilli abundance and oedema, as compared to wild-type mice, but still little necrosis (Fig. 4A and B). Thus, absence of both IL-1α and IL-1β recapitulates the acute necrotic pneumonia resulting from defective IL-1R1 or TNF pathways, while the sole presence of IL-1α or IL-1β limits but does not fully prevent lung infiltration of inflammatory cells, oedema and necrosis after acute M. tuberculosis infection.

Regulation of pulmonary cytokine expression by IL-1α/IL-1β pathway after M. tuberculosis infection

We next assessed the expression of cytokines involved in innate and adaptive immune responses in these mice. Indeed, the uncontrolled bacilli growth and pulmonary inflammation in mice deficient for both IL-1α and IL-1β was reminiscent of the high infectious burden in M. tuberculosis infected IL-1R1 or MyD88 deficient mice, which was accompanied by increased macrophage and neutrophil recruitment, but a normal adaptive response in terms of recruitment, activation and priming of effector T cells [13,44]. Further, in view of the persistent lung inflammation in single IL-1α or IL-1β deficient mice during chronic M. tuberculosis infection, we asked how the absence of IL-1α and/or IL-1β modulated the immune response in the lung after 1–3 months of infection (Fig. 5). Lung IFNγ levels were highly elevated in mice deficient for both IL-1α and IL-1β or for IL-1R1 at 5 weeks after M. tuberculosis infection (Fig. 5A), reminiscent of TNF deficient mice at 4 weeks, when they succumb to infection. Interestingly, IFNγ levels were also highly elevated in IL-1β deficient mice, but less so in IL-1α deficient mice, indicating that presence of IL-1β sufficed to control IFNγ over-expression. While IL-12/23 p40 levels were reduced in mice deficient for both IL-1α and IL-1β (Fig. 5B), they were essentially normal in IL-1R1 deficient mice or TNF deficient mice and slightly elevated in mice deficient for either IL-1α or IL-1β, although this did not reach statistical significance. IL-23 p19 pulmonary levels were reduced in the highly susceptible, TNF-deficient mice at 4 weeks of M. tuberculosis infection and IL-23 p19 levels were also clearly lower than wild-type controls in mice deficient for both IL-1α and IL-1β (Fig. 5C), and partially decreased in IL-1R1 deficient mice or mice deficient for either IL-1α or IL-1β, 5 weeks after M. tuberculosis infection. The levels of IL-17A were also reduced in mice deficient for both IL-1α and/or IL-1β or IL-1R1 at 5 weeks post-infection and in IL-1α or IL-1β single deficient mice at 8 weeks post-infection (Fig. S2). IL-1α and/or IL-1β gene deficient mice did not express IL-1α and/or IL-1β as expected, while IL-1α and IL-1β pulmonary levels were elevated in TNF-deficient mice at 4 weeks of M. tuberculosis infection, and were also elevated in IL-1R1 deficient mice as compared to wild-type control 5 weeks post infection. IL-1α levels were slightly increased in IL-1β deficient mice, and conversely, IL-1β levels were slightly increased in IL-1α deficient mice during the first 2 months of infection (Fig. 5D and E). TNF pulmonary levels were rather modest, as shown previously [44] and were partially reduced in the absence of IL-1α and/or IL-1β, or IL-1R1 (Fig. 5F). Therefore, complete absence of IL-1α plus IL-1β led to increased IFNγ pulmonary levels, but decreased IL-12/23 p40 and IL-23 p19 expression, which were partially corrected by the presence of either IL-1α or IL-1β. In line with TNF being downstream of IL-1β for restricting M. tuberculosis growth in macrophages in vitro [38], absence of IL-1 pathway impacted on partial reduction of TNF pulmonary levels after in vivo M. tuberculosis infection, while they was no reduction of IL-1α or IL-1β lung content in the absence of TNF.

IL-1 pathway is dispensable for controlling attenuated M. bovis BCG infection, while TNF is essential

Mice deficient for TNF are highly susceptible not only to virulent M. tuberculosis, but also to attenuated M. bovis BCG (Refs. [39–41,45–47] and unpublished data). In contrast, mice deficient for MyD88, an essential adaptor of IL-1R1 pathway, were reported as highly susceptible to virulent M. tuberculosis infection by us and others [13,48,49], while they controlled chronic M. bovis BCG infection [50] and were found less susceptible to M. tuberculosis by other authors [51]. Our data indicated that although IL-1R1 pathway is essential for controlling acute infection by virulent M. tuberculosis, presence of either IL-1α or IL-1β allowed to survive acute M. tuberculosis infection. In previous reports, deficiency in IL-1α or IL-1β led to either early death after M. tuberculosis infection or survival up to 3–6 months at lower dose infection [16]. We next asked how the complete absence of IL-1 pathway affects the control of less virulent mycobacteria.

Mice deficient for IL-1α plus IL-1β, IL-1R1 or TNF were infected with M. bovis BCG (Fig. 6). TNF deficient mice rapidly lost weight and succumbed by 6 weeks of infection (Fig. 6A and B) with highly inflammed lung, spleen and liver, while IL-1R1 and IL-1α plus IL-1β deficient mice survived systemic M. bovis BCG infection with lung, spleen and liver weight similar to wild-type mice at 6 and 11 weeks (Fig. 6C and D). Histologically, liver and lung inflammation were exacerbated in TNF deficient mice, with numerous and large, confluent granulomatous structures containing massive cell infiltrates, while the pathology was much less severe in IL-1R1 or IL-1α plus IL-1β deficient mice (Fig. 6E and F). Therefore, the IL-1 pathway which is essential for controlling virulent M. tuberculosis infection is dispensable for the early control of attenuated M. bovis BCG infection, in sharp contrast with TNF which is essential for both responses.

Mice

Mice deficient for IL-1α or IL-1β [35], or both IL-1α and IL-1β [12], IL1-R1 [59], or TNF [60] were bred in the Transgenose Institute animal facility (CNRS UPS44, Orleans). All mice were backcrossed at least 7–10 times on C57BL/6 genetic background. For experiments, adult (8–12 weeks old) animals were kept in isolators in a biohazard animal unit. The infected mice were monitored regularly for clinical status and weighed twice weekly. All animal experiments complied with the French Government's animal experiment regulations and were approved by the “Ethics Committee for Animal Experimentation of CNRS Campus Orleans” (CCO; N°CLE CCO 2011-026).

Bacteria and infection

M. tuberculosis H37Rv (Pasteur) aliquots kept frozen at −80°C were thawed, diluted in sterile saline containing 0.05% Tween 20 and clumping was disrupted by 30 repeated aspirations through a 26 gauge needle (Omnican, Braun, Germany). Pulmonary infection with M. tuberculosis H37Rv was performed by delivering 1600 ± 300 CFU/lung into the nasal cavities (20 µl each) under xylazine–ketamine anaesthesia, and the inoculum size was verified 24 h after infection by determining bacterial load in the lungs. Alternatively, M. bovis BCG (Pasteur strain 1173P2) or fluorescent GFP-expressing M. bovis BCG (gift from Dr. V. Snewin, Wellcome Trust London, UK) was grown to mid-log phase in Middlebrook 7H9 liquid medium (Difco Laboratories, Detroit, MI) supplemented with 10% oleic acid/albumin/dextrose/catalase (OADC, Difco Laboratories) and 0.05% Hygromycin (Invivogen, San Diego, CA) at 37°C, stored at −80°C in 10% glycerol (Sigma, St Louis, MO), and injected intravenously at 10⁶ CFU/mouse.

Bacterial load in tissues

Bacterial loads in the lung of infected mice were evaluated at different time points after infection with M. tuberculosis H37Rv as described [46]. Organs were weighed and defined aliquots were homogenised in PBS in a Dispomix homogeniser (Medic Tools, Axonlab, Baden-Daettwil, Switzerland). Tenfold serial dilutions of organ homogenates in 0.05% Tween 20 containing 0.9% NaCl were plated in duplicates onto Middlebrook 7H11 (Difco) agar plates containing 10% OADC and incubated at 37°C. Colonies were enumerated at 3 weeks and results are expressed as log₁₀ CFU per organ.

Pulmonary cytokine determination

Lung homogenates were centrifuged (3 min at 14,500 rpm), the supernatants sterilised by centrifugation through 0.22 μm filter (3 min at 14,500 rpm; Costar-Corning, Badhoevedorp, The Netherlands), immediately frozen on dry ice and stored at −80°C until determination of IL-1α, IL-1β, IL-12/IL-23p40, IL-23p19, TNF and IFNγ levels by ELISA (Duoset R&D Systems, Abingdon, UK).

Histopathological analysis

For histological analysis left lobe of lungs from M. tuberculosis infected mice were fixed in 4% phosphate buffered formalin and paraffin-embedded. Two to 3-μm sections were stained with Hematoxylin and Eosin and a modified Ziehl-Neelsen method. The latter involved staining in a prewarmed (60°C) carbol-fuchsin solution for 10 min followed by destaining in 20% sulphuric acid and 90% ethanol before counterstaining with methylene blue. Free alveolar space, lung cellular infiltration, oedema and necrosis were quantified using a semi-quantitative score with increasing severity of changes (0–5) by two independent observers including a trained pathologist (BR). Liver and lung from M. bovis BCG infected mice were assessed histologically by H&E staining and the slides scored for inflammatory cell infiltration and granuloma lesions as above.

Primary macrophage cultures

Murine bone marrow cells were isolated from femurs and differentiated into macrophages after culturing at 10⁶ cells/ml for 7 days in DMEM (Sigma) supplemented with 10 mM l-glutamine, 25 mM Hepes, 100 U/ml penicillin and 100 U/ml streptomycin, plus 20% horse serum and 30% L929 cell-conditioned medium as a source of M-CSF [43]. Three days after washing and re-culturing in fresh medium, the cell preparation contained a homogenous population of macrophages. Macrophages were plated in 96 well microculture plates (at 10⁵ cells/well in supplemented DMEM as above), and stimulated with LPS (Escherichia coli, serotype O111:B4, Invivogen, Saint Louis, MO at 100 ng/ml), heat-killed M. tuberculosis H37Rv (heat-killed 90 min at 80°C; 2 bacteria per cell), live or heat-killed M. bovis BCG (HKBCG; from Pasteur Institute, Paris, both at a MOI of 2 bacteria per cell). Cell supernatants were harvested after 24 h of stimulation for IL-1α, IL-1β and TNF quantification by ELISA (R&D Duoset).

Statistical analysis

Statistical significance was determined with Graph Pad Prism (version 5.04 for Windows, GraphPad Software, La Jolla, CA). Differences between multiple in vivo groups were analysed by means of one-way non-parametric ANOVA test (Kruskal–Wallis followed by Dunn's multiple comparison test) and values of P ≤ 0.05 were considered significant. Two-tailed, non-parametric Mann–Whitney t-test was used for analyzing in vitro results in Figure 1.