Critical and Independent Role for SOCS3 in Either Myeloid or T Cells in Resistance to Mycobacterium tuberculosis

Suppressor of cytokine signalling 3 (SOCS3) negatively regulates STAT3 activation in response to several cytokines such as those in the gp130-containing IL-6 receptor family. Thus, SOCS3 may play a major role in immune responses to pathogens. In the present study, the role of SOCS3 in M. tuberculosis infection was examined. All Socs3fl/fl LysM cre, Socs3fl/fl lck cre (with SOCS3-deficient myeloid and lymphoid cells, respectively) and gp130F/F mice, with a mutation in gp130 that impedes binding to SOCS3, showed increased susceptibility to infection with M. tuberculosis. SOCS3 binding to gp130 in myeloid cells conveyed resistance to M. tuberculosis infection via the regulation of IL-6/STAT3 signalling. SOCS3 was redundant for mycobacterial control by macrophages in vitro. Instead, SOCS3 expression in infected macrophages and DCs prevented the IL-6-mediated inhibition of TNF and IL-12 secretion and contributed to a timely CD4+ cell-dependent IFN-γ expression in vivo. In T cells, SOCS3 expression was essential for a gp130-independent control of infection with M. tuberculosis, but was neither required for the control of infection with attenuated M. bovis BCG nor for M. tuberculosis in BCG-vaccinated mice. Socs3fl/fl lck cre mice showed an increased frequency of γδ+ T cells in different organs and an enhanced secretion of IL-17 by γδ+ T cells in response to infection. Socs3fl/fl lck cre γδ+ T cells impaired the control of infection with M. tuberculosis. Thus, SOCS3 expression in either lymphoid or myeloid cells is essential for resistance against M. tuberculosis via discrete mechanisms.


Introduction
Tuberculosis (TB), an infectious disease caused by Mycobacterium tuberculosis, remains a leading public health problem worldwide. The global incidence of TB is rising with 8.8 million new cases and 2 million deaths each year [1]. However, while immune responses to TB clearly show their importance in host defence, it is clear that there are still gaps in our knowledge of the host factors determining the outcome of infection.
Host responses of mycobacterial infections are primarily Th1 immune responses involving cellular effector mechanisms such as macrophage activation. IFN-c is known to be an important mediator of mycobacterial control during clinical and experimental infections [2]. IL-12 is crucial for optimal differentiation and maintenance of IFN-c-secreting antigen-specific Th1 cells [3,4], and in controlling mycobacterial infections in mice and man [5,6].
The ''suppressor of cytokine signalling'' (SOCS) proteins are a family of eight members that inhibit STAT activation by different receptors. SOCS proteins bind either the Janus-activated kinases (JAKs) directly inhibiting their kinase activity, or the intracellular domain of cytokine receptors thereby targeting the receptor complex for ubiquitination and subsequent proteasome-mediated degradation [7]. SOCS3 inhibits STAT3-mediated signalling by binding to the IL-6 receptor family subunit gp130, G-CSF, leptin and the IL-12 receptor [8]. Since SOCS3-deficient mice die during embryogenesis [9,10], the role of SOCS3 in vivo has been studied using conditional knockdown mice. Conditional knockdown of SOCS3 in macrophages protects mice from LPS shock by reducing the secretion of IL-12 and TNF due to the enhanced anti-inflammatory effect of STAT3 [11]. However, mice with SOCS3-deficient macrophages and neutrophils succumb to toxoplasmosis, probably due to reduced IL-12 and IFN-c responses [12]. Furthermore, SOCS3 can also inhibit STAT1 activation thereby preventing IFN-c-like responses in cells stimulated with IL-6 [13,14]. SOCS3 also may have several roles in T cell function. SOCS3 expression in T cells can both obstruct the differentiation of inflammatory IL-17-producing Th17 cells [15,16] and inhibit the secretion of anti-inflammatory IL-10 and TGF-b by T cells [17] and mice with SOCS3-deficient T cells are more susceptible to infection with Leishmania major [17]. On the other hand, SOCS3 has also been shown to impair T-cell memory development, T cell-mediated IFN-c secretion and LCMV virus clearance in mice [18].
In the present study, the role of SOCS3 in the outcome of infection with M. tuberculosis was investigated. We report that the expression of SOCS3, in either myeloid or T cells, is independently required for the control of M. tuberculosis infection in mice. SOCS3 expression in myeloid cells allows a proper IL-12 secretion by hampering an IL-6-mediated inhibition of IL-12 expression. SOCS3 expression in T cells reduces the frequency of cd+ T cells in different organs and the secretion of IL-17 by + T cells in response to infection in a gp130-independent manner.

Results
Socs3 fl/fl LysM cre mice are highly susceptible to infection with M. tuberculosis First, the role of Socs3 expression in myeloid cells in the control of infection with M. tuberculosis was examined by using Socs3 fl/fl LysM cre mice [19]. Lungs and spleens from Socs3 fl/fl LysM cre mice showed significantly higher M. tuberculosis levels than Socs3 fl/fl littermates at 16 and 28 days of infection ( Figure 1A, B). A larger area of the lung parenchyma of Socs3 fl/fl LysM cre mice was occupied by granulomas as compared to controls 4 weeks after infection ( Figure 1C, D). Furthermore, M. tuberculosis-infected Socs3 fl/fl LysM cre mice also showed a higher cumulative mortality ( Figure 1E). Socs3 fl/fl LysM cre mice infected with the attenuated M. bovis BCG displayed higher bacterial levels in the lungs and spleen (but not the liver), although the differences in BCG levels with infected Socs3 fl/fl littermates were not as striking as those observed after infection with M. tuberculosis ( Figure 1F).
Since the LysM promoter is active in neutrophils and SOCS3 has been shown to be a negative regulator of granulopoiesis [20,21], we studied whether the increased susceptibility to M. tuberculosis of Socs3 fl/fl LysM cre mice was associated to increased numbers of neutrophils at the site of infection. Comparable numbers of Gr1+/ F4/80-neutrophils and similar mRNA levels of the neutrophil enzyme myeloperoxidase were detected in lungs from M. tuberculosisinfected Socs3 fl/fl LysM cre and control mice ( Figure 1G, H).

SOCS3-deficient macrophages are not impaired in bacterial control and can respond to IFN-c
Next, we studied the expression and role of SOCS3 in mycobacteria-infected macrophages. Bone marrow-derived macrophages (BMM) from wild type (WT) mice showed increased accumulation of Socs3 mRNA after infection with either M. tuberculosis or BCG (Figure 2A-C, S1A, B). Recognition by innate immune receptors was required for SOCS3 expression, since Socs3 mRNA levels after infection were reduced in the Toll-like receptor adaptor molecule MyD88 2/2 BMM and BMM incubated with a NF-kB inhibitor but not in Irf3 2/2 BMM (Figure 2A, B and S1A, B). IRF3 has been shown to detrimentally affect M. tuberculosis infection [22]. As expected, Socs3 mRNA levels were reduced in M. tuberculosis-infected Socs3 fl/fl LysM cre BMM when compared to controls ( Figure 2C), and in vitro infection of BMM with M. tuberculosis stimulated STAT3 phosphorylation that was prolonged in Socs3 fl/fl LysM cre BMM ( Figure 2D).
Whether a defect of macrophages to control intracellular mycobacterial growth could account for the enhanced susceptibility of Socs3 fl/fl LysM cre mice to mycobacteria was then studied.  Figure 2H). Macrophages have been shown to kill mycobacteria through the generation of nitric oxide (NO) by the IFN-c-regulated inducible NO synthase (iNOS) [23]. M. tuberculosis-infected Socs3 fl/fl LysM cre BMM contained higher iNos mRNA and nitrite levels than Socs3 fl/fl BMM ( Figure 2I, J). Similarly, infection of Socs3 fl/fl LysM cre BMM with BCG or stimulation with Pam3CSK4, an agonist for TLR2, a receptor that plays a prominent role in the initiation of responses against M. tuberculosis [24], led to higher NO and iNos mRNA levels compared to controls (Figure S1C, D).
Cells derived from gp130 F/F mice, harbouring a gp130 Y757F mutation to ablate SOCS3 binding to gp130, show an exaggerated gp130-mediated STAT3 signalling as a consequence of an impaired negative feedback loop by SOCS3 to down-modulate gp130/STAT3 signalling [25]. Similar to Socs3 fl/fl LysM cre BMM, gp130 F/F BMM showed increased iNos mRNA and nitrite levels after infection with either M. tuberculosis or BCG, or stimulation with Pam3CSK4 ( Figure 2K and S1E, F). Thus, the increased iNOS response of SOCS3-deficient macrophages was dependent on signalling via gp130. Similarly, the mRNA expression levels of the IFN-c-induced chemokine CXCL10 was also increased in either M. tuberculosisor BCG-infected Socs3 fl/fl LysM cre BMM ( Figure 2L and S1G).
Altogether, these data demonstrated that the higher susceptibility to M. tuberculosis of Socs3 fl/fl LysM cre mice was not associated with a defect of BMM or pulmonary macrophages in controlling intracellular bacterial growth in vitro.

SOCS3-deficient macrophages show decreased TNF and IL-12 responses during mycobacterial infection
After infection with M. tuberculosis, Il-6 mRNA levels were strikingly increased in lungs from Socs3 fl/fl LysM cre mice when

Author Summary
Tuberculosis is a severe disease caused by infection with the intracellular bacteria Mycobacterium tuberculosis. The protein ''suppressor of cytokine signalling 3'' (SOCS3) inhibits the responses of cells to several cytokines and growth factors that signal via the STAT3 transcription factor. Since STAT3 is a major controller of immune and inflammatory responses, we studied the role of SOCS3 in the control of infection with M. tuberculosis. Mice deficient in the expression of SOCS3 either in myeloid or lymphoid cells were extremely susceptible to infection with M. tuberculosis as measured by elevated bacterial levels, worsened pathology and reduced survival. In myeloid cells, SOCS3 hindered a detrimental role of IL-6. In absence of SOCS3, IL-6 hampered the release of IL-12 by antigenpresenting cells. In T cells, SOCS3-mediated protection was independent of IL-6 signals, and of adequate IFN-c secretion by antigen-specific T cells. Instead, SOCS3 inhibited the in vivo accumulation of, and the IL-17 secretion by cd+ T cells. cd+ T cells accounted in part for the susceptibility to M. tuberculosis infection of mice with SOCS3-deficient T cells. Thus, SOCS3 controls diverse immune mechanisms of myeloid and lymphoid cells that are required for containment of M. tuberculosis. . The mean % lung area with granulomas or free of lesions 6 SEM is displayed. Differences with controls are significant (n = 8 per group, *p,0.05 Student t test). The cumulative mortality of Socs3 fl/fl and Socs3 fl/fl LysM cre mice (n = 10) after aerosol infection with M. tuberculosis is depicted (E). Survival curves are different (Log-rank test p,0.005). CFU per lung, spleen and liver in Socs3 fl/fl and Socs3 fl/fl LysM cre mice (n$5 per group) were assessed 6 weeks after infection with 10 6 BCG i.v. (F). The median CFU and interquartile range per group are depicted. Differences in CFU are significant (**p,0.01 Mann Whitney U test). The mean percentage of Gr1+F4/80-neutrophils in the lung of Socs3 fl/fl and Socs3 fl/fl LysM cre mice (n = 5 per group) 3 weeks after infection with M. tuberculosis 6 SEM was compared to littermates ( Figure 3A). Therefore, we evaluated whether IL-6 secretion was also elevated in mycobacteria-infected Socs3 fl/fl LysM cre BMM. Unexpectedly, we found diminished Il-6 mRNA and protein levels in M. tuberculosisor BCG-infected or in Pam3CSK4-stimulated Socs3 fl/fl LysM cre BMM, peritoneal and pulmonary macrophages as compared to controls ( Figure 3B, C and S2A-C). Similarly, gp130 F/F BMM expressed lower IL-6 protein and mRNA levels after infection with either M. tuberculosis or BCG, or stimulation with Pam3CSK4 ( Figure 3D and S2D). In conclusion, although IL-6 levels are increased in lungs of infected Socs3 fl/fl LysM cre mice, SOCS3 does not impair IL-6 secretion by mycobacteria-infected BMM.
The LPS-induced production of TNF and IL-12 is reduced in SOCS3-deficient macrophages if IL-6 is added [11]. Consistent with this observation, the Tnf mRNA or protein accumulation was reduced in Socs3 fl/fl LysM cre BMM incubated with either M. tuberculosis, BCG-or Pam3CSK4 as compared to controls ( Figure 3E and S2E).
Next, we studied whether IL-6 signalling accounted for SOCS3/gp130-mediated regulation of TNF levels. Tnf mRNA and protein levels were reduced in M. tuberculosisor BCG-infected gp130 F/F BMM and were partially restored in gp130 F/F Il-6 2/2 BMM (Figure 3G and S2F, G), indicating that SOCS3 allows proper infection-induced TNF secretion in macrophages by hampering gp130/IL-6 receptor-mediated signalling. Moreover, the co-incubation with recombinant IL-6 (rIL-6) further diminished TNF levels in M. tuberculosis-infected Socs3 fl/fl LysM cre but not in control BMM ( Figure 3E).
Similar to the results observed for TNF, the Il-12 p40 mRNA and protein levels were reduced in Socs3 fl/fl LysM cre BMM after stimulation with Pam3CSK4, BCG or M. tuberculosis when compared to controls ( Figure 4A-C). IL-12 levels were further decreased in cultures of infected mutant macrophages incubated with rIL-6 ( Figure 4B, C). Accordingly, IL-12 p40 mRNA and protein levels were reduced in M. tuberculosisor BCG-infected gp130 F/F BMM, and such defect was restored in infected gp130 F/F Il-6 2/2 BMM ( Figure 4D and S3A).
Since IL-12 secretion by dendritic cells (DCs) is required for Th1 differentiation, we investigated whether DCs from Socs3 fl/fl LysM cre mice also showed an impaired secretion of IL-12. Socs3 mRNA levels were reduced in Socs3 fl/fl LysM cre bone marrowderived dendritic cells (BMDC) indicating the expression of the LysM cre recombinase ( Figure 4E). The Il-12 p40 mRNA and IL-12 protein expression by Socs3 fl/fl LysM cre BMDC after infection with M. tuberculosis was reduced as compared to controls ( Figure 4F, G). As shown for BMM, incubation with exogenous IL-6 further diminished IL-12 secretion by M. tuberculosis-infected Socs3 fl/fl LysM cre BMDC ( Figure 4G). In order to exclude that the diminished IL-12 secretion in BMDC cultures was due to the response of contaminant macrophages in the culture, the expression of Il-12 p40 mRNA was tested in CD11c+ sorted cells. Both, CD11c+ and CD11c-cells showed diminished IL-12 p40 mRNA accumulation after infection with M. tuberculosis compared with controls ( Figure 4H). Moreover, Socs3 fl/fl LysM cre splenic DCs displayed a diminished secretion of IL-12 after M. tuberculosis ( Figure 4I). Thus, SOCS3 expression promotes IL-12 secretion in M. tuberculosis-stimulated DCs.
Since IL-12 is required for IFN-c secretion by NK cells, we tested the effect of SOCS3 expression by M. tuberculosis-infected DCs in the regulation of IFN-c secretion by NK cells. Co-incubation with M.
tuberculosis-infected splenic CD11c+ DCs induced IFN-c secretion by NK cells. IFN-c expression by NK cells was reduced when these cells were incubated with Socs3 fl/fl LysM cre DCs ( Figure S3B, C).
Lower Il-12 p40 mRNA levels were also found in lungs of M. tuberculosis-infected Socs3 fl/fl LysM cre mice compared to controls ( Figure 4J) and similarly, lower Ifn-c mRNA accumulation in lungs from Socs3 fl/fl LysM cre mice was detected 2.5 weeks after M. tuberculosis infection ( Figure 4K). In order to examine whether an effect of SOCS3-deficient myeloid cells on the cytokine production by CD4+ T cells accounted for the elevated numbers of bacteria in Socs3 fl/fl LysM cre mice, we depleted CD4+ cells during M. tuberculosis infection by administration of anti-CD4 neutralizing antibodies ( Figure S4A). CD4+ cell depletion decreased the Ifn-c mRNA accumulation in lungs from Socs3 fl/fl mice. The Ifn-c mRNA levels in infected Socs3 fl/fl LysM cre mice were similar to those measured in CD4+ cell-depleted mice ( Figure 4K). Moreover, lungs from Socs3 fl/fl LysM cre and control mice depleted of CD4+ cells showed similar bacterial levels ( Figure 4L). In contrast to the decreased Ifn-c mRNA expression, the frequency of CD44+ and CD62L+ CD4+ activated T cells in the lungs of M. tuberculosisinfected Socs3 fl/fl LysM cre and Socs3 fl/fl mice was similar, and higher than in uninfected animals ( Figure S4B, C). When we compared cytokine levels in Socs3 fl/fl LysM cre and Socs3 fl/fl littermates at later time points after M. tuberculosis infection. Higher Ifn-c and iNos, but similar Tnf-mRNA levels were measured in lungs from Socs3 fl/fl LysM cre compared to control mice 8 weeks after infection ( Figure S4D-F).
Regulatory FoxP3+ T-cells have been shown to expand in mice with SOCS3-deficient DCs [26]. However, comparable levels of FoxP3+ CD4+ T cells were found in lungs and pulmonary lymph nodes of infected mutant and control mice ( Figure S4G), suggesting that the susceptibility of Socs3 fl/fl LysM cre mice to M. tuberculosis is not due to higher frequencies of regulatory T cells.
Altogether, the enhanced susceptibility of Socs3 fl/fl LysM cre mice to M. tuberculosis could be associated to a reduced IL-12 secretion resulting in a delayed CD4+ -cell dependent IFN-c-expression.
Mice with SOCS3-deficient T cells are dramatically susceptible to infection with M. tuberculosis Next, the role of SOCS3 expression by T cells in the control of infection with M. tuberculosis was studied. Lungs and spleens from Socs3 fl/fl lck cre mice showed higher numbers of M. tuberculosis bacteria 4 weeks after aerosol infection ( Figure 5A and data not shown), with 500-fold higher bacterial levels in lungs compared to Socs3 fl/fl littermate controls. In contrast, no differences in bacterial load were registered 2 weeks after infection ( Figure 5A). Infected Socs3 fl/fl lck cre mice had a median survival of 38 days after infection while controls survived more than 200 days ( Figure 5B). Four weeks after infection, Socs3 fl/fl lck cre mice displayed an increased severity of pulmonary pathology ( Figure 5C-E) with granulomas containing large necrotic areas ( Figure 5F) and elevated levels of neutrophil myeloperoxidase transcripts ( Figure 5G). The frequency of Foxp3+ CD4+ regulatory T cells in pulmonary lymph nodes was similar ( Figure 5H).
Whether the susceptibility of Socs3 fl/fl lck cre mice was associated to an altered frequency of T cell populations was then evaluated. While the percentage of CD4+ and CD8+ T cells in lungs and spleens from Socs3 fl/fl lck cre and Socs3 fl/fl mice before or after M.   Figure 6D).
The outcome of infection with M. tuberculosis of Rag1 2/2 mice reconstituted with control or Socs3 fl/fl lck cre T cells was then compared. Lungs and spleens from Rag1 2/2 mice transferred with Socs3 fl/fl total T cells (CD90+) showed lower bacterial levels than non-transferred mice while Socs3 fl/fl lck cre T cells failed to transfer protection ( Figure 6F). Moreover, the transfer of a 1:1 mixture of Socs3 fl/fl lck cre and control T cells conferred no protection to Rag1 2/2 mice ( Figure 6F), suggesting that SOCS3-deficient T cells can hamper M. tuberculosis control by wild type T cells. Rag1 2/2 mice transferred with CD4+ Socs3 fl/fl lck cre cells unlike those transferred with total T cells from the same mice, showed reduced M. tuberculosis levels compared to non-transferred controls indicating that in Socs3 fl/fl lck cre mice CD3+CD42 T cells hamper the protective ability of CD4+ cells ( Figure 6G). Therefore, we examined whether cd+ T cells could account for the suppressive activity of CD3+CD42 cells. Indeed, Rag1 2/2 mice transferred with cd+ cell-depleted CD90+ Socs3 fl/fl lck cre T cells showed lower bacterial levels than those transferred with total Socs3 fl/fl lck cre T cells (Figure 6 H).
Since SOCS3 expression in T cells has been shown to impair IL-17 production [27], we speculated that a differential release of IL-17 could be related with the susceptibility of Socs3 fl/fl lck cre mice to M. tuberculosis. Lungs from M. tuberculosis-infected Socs3 fl/fl lck cre mice showed higher levels of Il-17 mRNA than controls ( Figure 7A). The levels of IL-17 in supernatants from lung cells of Socs3 fl/fl lck cre mice stimulated or not with mycobacterial Purified protein derivate (PPD) were higher than controls 2.5 weeks after M. tuberculosis infection, when no differences in bacterial load in lungs were detected ( Figure 7B). Thus, the impaired control of M. tuberculosis-infection in Socs3 fl/fl lck cre mice was associated with increased IL-17 levels. cd+ T cells have been shown to dominate IL-17 secretion during infection with M. tuberculosis [28]. In line with this observation, the incubation of naïve spleen T cells with mycobacteria-infected BMDCs or their supernatants resulted in the secretion of IL-17 ( Figure 7C). Furthermore, IL-17 was secreted by cd+ and total (CD90+) but not by CD4+ T cells after incubation supernatants from mycobacteria-infected BMDCs. The levels of IL-17 secreted by cd+ T cells were higher than those by similar numbers of total T cells. The IL-17 content in supernatants from cd+ and total Socs3 fl/ fl lck cre T cells was higher compared to Socs3 fl/fl controls ( Figure 7D).
IL-17-secreting cells were enumerated by intracellular cytokine staining in PMA/ionomycin-stimulated lung cell suspensions from Socs3 fl/fl lck cre and Socs3 fl/fl mice 16 days after M. tuberculosis infection. The majority of IL-17-secreting lung T cells in infected mice were cd+ rather than CD4+ or CD8+ cells ( Figure 7E, G). Furthermore, M. tuberculosis infection stimulated the IL-17 secretion capability of cd+ T cells ( Figure 7F). However, the frequency of IL-17-secreting cells among cd+ T cells from infected Socs3 fl/fl lck cre and Socs3 fl/fl was similar, suggesting that the lack of SOCS3 does not alter the differentiation of cd+ T cells into IL-17secreting cells.
IL-17 was measured in supernatants from lung cell suspensions from CD4+ or CD90+ Socs3 fl/fl lck cre T cell-transferred Rag1 2/2 mice 4 weeks after M. tuberculosis infection. While IL-17 levels were strikingly higher in supernatants from mice transferred with total T cells compared with non-transferred controls, the IL-17 concentration in cultures from mice inoculated with CD4+ cells was not increased ( Figure 7H). Thus, the inhibition of CD4+ cellmediated protection in Rag1 2/2 mice transferred with CD90+ cells ( Figure 6G) was associated to an increased IL-17 secretion.
Lung cells from M. tuberculosis-infected Socs3 fl/fl lck cre mice contained higher levels of Ifn-c mRNA ( Figure 8A) and secreted higher levels of IFN-c when stimulated with PPD than Socs3 fl/fl controls ( Figure 8B). Thus, the susceptibility of Socs3 fl/fl lck cre mice was not associated to an impaired secretion of IFN-c.
The accumulation of Il-6 mRNA was increased in lungs of Socs3 fl/fl lck cre M. tuberculosis-infected mice at 4 but not at 2.5 weeks after infection compared to controls ( Figure S6A). The IL-6 levels in supernatants from Socs3 fl/fl lck cre and Socs3 fl/fl lung cells obtained 2.5 weeks after infection, stimulated or not with PPD, were similar ( Figure S6B). SOCS3 has been shown to regulate IL-10 secretion by T cells [17]. However, Socs3 fl/fl lck cre and Socs3 fl/fl lung cells from M. tuberculosis-infected mice, stimulated or not with PPD, secreted similar levels of IL-10 ( Figure S6C).
In contrast to results from M. tuberculosis-infected mice, Il-6 and Il-17a mRNA levels in lungs from BCG-infected Socs3 fl/fl lck cre and Socs3 fl/fl mice were similar ( Figure 8C and S6D). Spleen cells from BCG-immunized Socs3 fl/fl lck cre and control mice showed comparable IL-17, IFN-c or IL-6 secretion in response to PPD stimulation when compared with controls ( Figure 8D, E and S6E). The organs from Socs3 fl/fl lck cre and Socs3 fl/fl mice infected with M. bovis BCG contained similar bacterial levels ( Figure 8F). Moreover, bacterial levels in Socs3 fl/fl lck cre and Socs3 fl/fl mice challenged with M. tuberculosis after BCG immunization were comparable ( Figure 8G).
Next, we studied whether a differing stimulation of SOCS3 expression could explain the divergent susceptibility of Socs3 fl/fl lck cre mice to M. tuberculosis and BCG infection. Lungs from mice infected with either M. tuberculosis or BCG showed higher Socs3 mRNA levels compared to uninfected mice. Socs3 transcript levels were higher in M. tuberculosisthan in BCG-infected animals ( Figure 8H, I). However, Socs3 mRNA levels in pulmonary T cells before or after infection with BCG or M. tuberculosis were similar ( Figure 8J). Thus, different expression levels of SOCS3 in T cells do not explain the distinct susceptibility of Socs3 fl/fl lck cre mice to M. tuberculosis and BCG infection.

Gp130 F/F mice display dramatic susceptibility to infection with M. tuberculosis
To further characterize the function of SOCS3 in myeloid and T cells in the control of infection with M. tuberculosis, gp130 F/F knock-in mice were used. gp130 F/F mice displayed a dramatically enhanced susceptibility to M. tuberculosis as measured by their CFU were determined in triplicate cell cultures (H). One out of two independent experiments is shown. Differences with Socs3 fl/fl BMM are significant (*p,0.05 Student t test). Total RNA was extracted from Socs3 fl/fl LysM cre, Socs3 fl/fl (I, L), WT and gp130 F/F (K) BMM at the indicated times after infection with M. tuberculosis at a MOI of 5. The relative accumulation of iNos, Cxcl10 and Hprt mRNA was measured by real time PCR. The mean fold increase of iNos (I, K), or Cxcl10 (L) mRNA 6 SEM in triplicate samples for each genotype and time point in one out of two independent experiments is depicted. Differences with control BMM are significant (*p,0.05 and **p,0.01Student t test). Nitrite concentrations in supernatants of Socs3 fl/fl LysM cre and Socs3 fl/fl BMM at different times after infection with M. tuberculosis. The mean NO 2 2 concentration 6 SEM in triplicate cultures per condition from one of two independent experiments is depicted (***p,0.001 Student t test)(J). doi:10.1371/journal.ppat.1003442.g002   Figure 9A, B), and increased cumulative mortality (gp130 F/F mice died before 48 days after infection whereas all WT controls survived for more than 100 days). Since IL-6 mediated, at least in part, the inhibition of TNF and IL-12 secretion by SOCS3deficient BMM and BMDC, the role of IL-6 in the susceptibility of gp130 F/F mice to infection with M. tuberculosis was studied. We found that gp130 F/F Il-6 2/2 as well as gp130 F/F Stat3 +/2 mutant mice displayed lower levels of M. tuberculosis bacteria in lungs and diminished severity of pulmonary pathology when compared to gp130 F/F mice, indicating that the increased susceptibility of gp130 F/F mice is in part mediated by IL-6 and STAT3 activation ( Figure 9A, B). Lungs from gp130 F/F mice displayed higher Il-6 and lower Il-12 p40 mRNA accumulation than infected control mice, while Il-12 p40 mRNA accumulation in lungs from gp130 F/F Il-6 2/2 was comparable to WT mice ( Figure 9C, D).
Surprisingly, lungs and spleens from M. tuberculosis-infected Rag1 2/2 mice transferred prior to infection with either gp130 F/F or WT T cells contained similar bacterial levels, indicating that T cells play, if any, a redundant role in the gp130-mediated control of M. tuberculosis. Bacterial levels in mice transferred with WT or gp130 F/F cells were lower than those of non-transferred controls ( Figure 9E). Moreover, frequencies of cd+ T cells in organs from WT or gp130 F/F were similar (data not shown). Thus, the susceptibility of Socs3 fl/fl lck cre mice to infection with M. tuberculosis is mediated by receptors other than gp130.
Since T cells did not account for the increased susceptibility of gp130 F/F mice and gp130 F/F mice were significantly more susceptible to M. tuberculosis than Socs3 fl/fl LysM cre mice, we studied the relative contributions of hematopoietic and nonhematopoietic cell lineages to the susceptibility of gp130 F/F mice to infection with M. tuberculosis. Reciprocal bone marrow (BM) radiation chimeras between WT and gp130 F/F mice were generated by inoculation of BM cells into irradiated recipients. WT mice reconstituted with gp130 F/F BM contained higher titers of M. tuberculosis in the lungs than those reconstituted with WT BM (sham chimeric mice), although differences in bacterial levels were notably lower than those observed in the gp130 F/F mice ( Figure 9F). Although gp130 F/F recipients showed significant mortality after irradiation, the few survivors inoculated with WT BM showed very high bacterial levels, similar to those from non-irradiated gp130 F/F mice (data not shown).
Thus, these data suggest that non-lymphoid, hematopoietic cells only partially account for the susceptibility of gp130 F/F mice to M. tuberculosis and suggest a relevant role for non-hematopoietic cells in the high sensitivity of these mice to infection.

Discussion
In the present study, we demonstrated that SOCS3 expression in lymphoid and myeloid cell populations is essential for the resistance against M. tuberculosis in mice via distinct mechanisms. M. tuberculosis and BCG infections were potent stimuli for Socs3 expression in vivo and in myeloid cell populations in vitro.
In line with previous studies, we found that SOCS3 expression in macrophages was mediated by MyD88 and NF-kB [29]. By using Socs3 fl/fl LysM cre mice, our data suggest that myeloid SOCS3 expression contributes to a timely CD4+ cell-dependent IFN-csecretion rather than to improved innate effector immune mechanisms by macrophages. The observation that CD4+ celldepleted Socs3 fl/fl LysM cre and control mice had similar bacterial loads supports this hypothesis. Moreover, SOCS3-deficient and control BMM as well as pulmonary macrophages showed comparable intracellular mycobacterial growth, and IFN-c diminished bacterial numbers with equal efficiency in Socs3 fl/fl LysM cre and WT macrophages, as also recently demonstrated for T. gondii [12]. However, whether a defective IFN-c secretion either by antigen-specific CD4+ T cells, or alternatively by NKT or CD8+ T cells underlies the susceptibility of Socs3 fl/fl LysM cre mice to infection with M. tuberculosis remains to be determined.
SOCS3-deficient BMM showed increased STAT3 activation and diminished secretion of TNF and IL-12 after infection with either M. tuberculosis or BCG. This confirms previous data showing reduced TNF and IL-12 release by Socs3 fl/fl LysM cre and gp130 F/F macrophages in response to LPS, when co-incubated with IL-6 [11,30]. Moreover, constitutively active STAT3 has been found to inhibit Il-12 p40 mRNA accumulation in LPS-stimulated BMDC [31]. Even though the LysM promoter is primarily active in neutrophils and macrophages, LysM promoter activity in DCs has previously been shown [26]. Socs3 fl/fl LysM cre BMDC showed limited IL-12 production in response to mycobacterial stimulation and Il-12 p40 levels were also reduced in the lungs of infected Socs3 fl/fl LysM cre and gp130 F/F mice. Importantly, we found that IFN-c levels were diminished in lungs of Socs3 fl/fl LysM cre at 16 days but not at later time points after infection with M. tuberculosis. This delay in the establishment of immune protective responses might underlie the increased susceptibility of Socs3 fl/fl LysM cre mice. In support of this notion, the resistance of different mouse strains to M. tuberculosis is associated with the timing of IFN-c responses [32]. Although we observed a reduced secretion of IFNc by NK-cells during in vitro co-culture with M. tuberculosis-infected Socs3 fl/fl LysM cre and control splenic DCs, the NK cell involvement in the enhanced susceptibility of Socs3 fl/fl LysM cre mice is unlikely since NK cells were not required for controlling mycobacterial infections [33,34].
co-incubated with 50 ng/ml IL-6 was determined by ELISA (C). The mean IL-12 concentration 6 SEM from triplicate cultures per condition in one of two independent experiments is depicted, (*p,0.05, **p,0.01, ***p,0.001 Student t test). The levels of Il-12 p40 mRNA were measured in triplicate cultures of gp130 F/F , gp130 F/F Il-6 2/2 or WT BMM infected with M. tuberculosis (D), (**p,0.05 and ***p,0.001 Student t test). The levels of Socs3 (E) and Il-12 p40 mRNA (F) in triplicate cultures of M. tuberculosis-infected Socs3 fl/fl LysM cre and Socs3 fl/fl BMDC were determined by real time PCR. One of two independent experiments is shown, (*p,0.05 Student t test). The concentration of IL-12 was determined by ELISA in supernatants from triplicate cultures of M. tuberculosis-infected Socs3 fl/fl LysM cre and Socs3 fl/fl BMDC co-incubated or not with 50 ng/ml IL-6 (G), (*p,0.05, Student t test). Total RNA was extracted from M. tuberculosis-infected CD11c+ and CD11c-Socs3 fl/fl and Socs3 fl/fl LysM cre BMDC cultures 24 h after M. tuberculosis infection. The mean Il-12p40 mRNA levels 6 SEM levels measured by real time PCR are depicted (H). The concentration of IL-12p40 in supernatants from M. tuberculosis-infected Socs3 fl/fl and Socs3 fl/fl LysM cre splenic DCs was determined by ELISA (I). The mean IL-12p40 6 SEM pg/ml from triplicate cultures is depicted (*p,0.05, **p,0.01, ***p,0.001 Student t test). The fold increase of Il-12 p40 mRNA in the lungs of M. tuberculosis-infected Socs3 fl/fl LysM cre and Socs3 fl/fl mice relative to uninfected mice is displayed (J). The data is pooled from 2 independent experiments with n$5 animals per group in each one (*p,0.05, Student t test). Total RNA was isolated from the lungs of TNF is of major importance in the control of M. tuberculosis [35]. Although TNF secretion by infected SOCS3-deficient macrophages is reduced, TNF expression in the lungs of M. tuberculosisinfected Socs3 fl/fl LysM cre mice was not diminished, suggesting that a role for TNF in the susceptibility to infection of Socs3 fl/fl LysM cre mice is unlikely.
The reduction of TNF and IL-12 levels observed in gp130 F/F BMM was reversed when using gp130 F/F Il-6 2/2 cells, and addition of rIL-6 further diminished the release of IL-12 and TNF by either mycobacteria-infected, or Pam3CSK4-stimulated Socs3 fl/ fl LysM cre BMM, suggesting that in macrophages, SOCS3 allows proper TNF and IL-12 secretion by hampering an IL-6-mediated inhibition of the secretion of these cytokines.
Lungs from M. tuberculosis-infected Socs3 fl/fl LysM cre and lck cre mice contained higher Il-6 mRNA levels than controls. However, neither macrophages nor T cells are likely to account for the elevated IL-6 levels in M. tuberculosis-infected SOCS3-deficient mice. Epithelial cells, fibroblasts and adipocytes have all been shown to secrete IL-6 in response to inflammatory stimuli [36,37]. Whether non-hematopoietic cells are major IL-6 producers during M. tuberculosis infection remains to be investigated.
Socs3 fl/fl lck cre mice showed a dramatically enhanced susceptibility to M. tuberculosis infection. However, SOCS3 expression in T cells was not required for the development of protective immune responses against M. tuberculosis in BCG-vaccinated mice. Thus, the requirement of SOCS3 in the control of mycobacterial infection depends on the mycobacterial species and on the immune status of the host.
Our results show a hitherto unknown role for SOCS3 controlling the frequency of cd+ T cells in different organs before and during M. tuberculosis infection while frequencies of CD4+ or CD8+ T cells were not regulated by SOCS3. Moreover, cd+ T cells impaired the transfer of protection by Socs3 fl/fl lck cre CD4+T cells, suggesting that SOCS3 inhibits a non-redundant detrimental role of cd+ T cells in the outcome of infection with M. tuberculosis. The detrimental activity of SOCS3-deficient cd+ T cells contrasts with previous reports that have shown a minor role of WT cd+ T cells in resistance to M. tuberculosis [38,39].
SOCS3 can impair the secretion of IL-17 [16,40]. The increased IL-17 mRNA and protein levels in Socs3 fl/fl lck cre mice in M. tuberculosis-, but not in BCG-infected mice suggested that IL-17 levels might be causally associated to the increased susceptibility to M. tuberculosis infection. When cd+ T cells but not CD4+ Socs3 fl/fl lck cre T cells were adoptively transferred in Rag1 2/2 mice, an increased IL-17 secretion by lung cells and impaired transfer of protection against M. tuberculosis was observed. The function of IL-17 during primary mycobacterial infections is controversial since only after high dose intratracheal infection mice deficient in IL-17 were reported to be unable to control M. tuberculosis infection [41][42][43]. On the other hand, IL-17 has been implicated to increase bacterial dissemination, recruitment of neutrophils and morbidity during infection with M. tuberculosis [44][45][46]. Socs3 fl/fl lck cre mice showed elevated levels of neutrophilderived molecules and necrotic granulomas during M. tuberculosis infection. Our results confirmed a previous report indicating that cd+T cells dominate IL-17 production during M. tuberculosis infection [28]. SOCS3 also hampered the secretion of IL-17 by cd+ T cells when incubated with infected DCs or their supernatants. However, SOCS3 did not impair the development of cd+ T cells that are capable of secreting IL-17. Thus, the increased IL-17 levels in Socs3 fl/fl lck cre mice are probably the consequence of both the increased numbers of cd+ T cells and their unrestricted secretion of IL-17 in response to cytokines released by mycobacteria-stimulated DCs.
The increased susceptibility to M. tuberculosis of Socs3 fl/fl lck cre mice was not associated to an impaired IFN-c secretion by antigen-specific T cells, suggesting that SOCS3 is not required for

IFN-c secretion by T cells and that cd+ T cells do not modulate IFN-c secretion by ab+ T cells.
Since SOCS3 regulates signalling via various receptors, we investigated whether signals mediated via the gp130 receptor account for the susceptibility to M. tuberculosis of SOCS3 conditional knockdown animals. We found that gp130 F/F mice are highly susceptible to infection with M. tuberculosis. The susceptibility of gp130 F/F is mediated by both IL-6-dependent as well as IL-6-independent signalling events, since gp130 F/F Il-6 2/2 mice showed lower bacterial load than gp130 F/F mice but higher bacterial levels than controls. The cytokine responses of gp130 F/F and Socs3 fl/fl LysM cre macrophages to mycobacterial infections were similar suggesting that the protective role of SOCS3 in myeloid cells is dependent on gp130.
IL-1b and IL-23 have been shown to stimulate IL-17 production by cd+ T cells [47,48], but SOCS3 impairs IL-23 signalling [27]. Thus, IL-1 b and IL-23 might mediate the elevated IL-17 secretion in Socs3 fl/fl lck cre T cells. Moreover, IL-23 signalling [49] as well as the increased frequency of cd+ T cells in Socs3 fl/fl lck cre mice are both independent of gp130 signalling (data not shown). Accordingly, T cells from highly susceptible gp130 F/F mice transferred resistance to M. tuberculosis as previously shown for T. gondii infection [50]. Since gp130 F/F were more susceptible to M. tuberculosis than Socs3 fl/fl LysM cre mice, and lethally irradiated WT that were reconstituted with gp130 F/F BM were more resistant to infection than gp130 F/F or WT BMR gp130 F/F mice, we also suggest that gp130-dependent SOCS3signalling in non-hematopoietic cells contributes to the control of infection with M. tuberculosis.
Collectively, our data indicate that the expression of SOCS3 either in myeloid or in T cells is essential for control of M. tuberculosis infection (Figure 10). SOCS3 mediates protection through inhibition of IL-6/gp-130 signalling in myeloid cells,

Ethics statement
The animals were housed and handled at the Dept.

Mice
Socs3 fl/fl mice containing loxP-flanked Socs3 alleles have been described before [51]. For a T cell-specific deletion Socs3 fl/fl mice were bred with transgenic lck cre mice [52] and for a myeloidspecific deletion with transgenic LysM cre mice [19]. Offsprings were genotyped as described [51] and Socs3 fl/fl littermates negative for cre expression were used as controls for all experiments. Gp130 F/F mice possess a homozygous substitution of tyrosine (Y) 757 to phenylalanine (F) within the common IL-6 family receptor gp130 abrogating the SOCS3 binding site. Gp130 F/F mice and their corresponding compound mutant homozygous null for IL-6 (gp130 F/F Il-6 2/2 ) or heterozygous for STAT3 (gp130 F/F Stat3 +/2 ) have been described previously [53,54], all on a mixed C57Bl/66129/Sv background which were used as controls. Rag1 2/2 mice were generated by homologous recombination in embryonic stem cells [55] and crossed to C57Bl/6 background. Bacteria were quantified on Middlebrook 7H11 agar containing 10% enrichment of oleic acid, albumin, dextrose, catalase, 5 mg of amphotericin B per ml and 8 mg/ml polymyxin B grown for 3 weeks at 37uC.

T cell transfer
Briefly, single-cell suspensions from spleens were selected for CD4+ or CD90+ T cells with magnetic beads (Miltenyi Biotech, Cologne, Germany) as specified by the manufacturer. When indicated, CD90+ cells were depleted of cd+ cells by FACS sorting. 1-3610 6 T cells were inoculated i.v. into Rag1 2/2 mice. Two weeks after transfer, mice were infected with M. tuberculosis Harlingen.

T cell depletion
Mice were injected i. p. three consecutive days with 0.5 mg/ mouse of Sepharose G affinity-purified anti-CD4 (GK1.5) antibody one week before infection. CD4-specific depletion was controlled in blood using flow analysis. Two weeks after the first injection, additional 0.5 mg/mouse anti-CD4 antibody was injected to maintain CD4+ cells depleted.

Immunization
Mice were immunized with 5610 6 heat-killed BCG (60 min at 80uC) s. c. and boosted after 2 weeks with 2.5610 6 heat-killed BCG. 4 weeks after the first injection, splenocytes from immunized and non-immunized mice were re-stimulated with 15 mg/ml PPD (Statens Seruminstitut, Copenhagen, Denmark) and supernatants collected after 72 h.

BCG immunization and M. tuberculosis challenge
Mice were immunized with 1610 6 BCG i. v. and kept for 10 weeks before aerosol infection with M. tuberculosis together with non-immunized controls. Mice were sacrificed 4 weeks after M. tuberculosis strain Harlingen infection and bacterial loads were determined in the lungs.

Bone marrow chimera
Recipient C57Bl/66129/Sv and gp130 F/F were irradiated 26 with 550 rad and received 5610 6 BM cell from either C57Bl/ 66129/Sv or gp130 F/F mice. Mice were kept for 3 weeks on antibiotics (Tribrissen in drinking water) and were infected with M. tuberculosis 8 weeks after transfer.

Generation of mouse bone marrow-derived macrophages
Bone marrow was extracted from tibia and femurs of mice and resuspended in Dulbecco's modified Eagle's medium (DMEM) containing glucose and supplemented with 2 mM L-glutamine, 10% FCS, 10 mM Hepes, 100 mg/ml streptomycin, 100 U/ml penicillin (all from Sigma), and 30% L929 cell-conditioned medium (as a source of macrophage-colony stimulating factor). Bone marrow cells were passed through a 70 mm cell strainer, plated and incubated for 6 days at 37uC, 5% CO 2 . Bone marrowderived macrophage (BMM) cultures were then washed vigorously to remove non-adherent cells, trypsinized, counted and cultured for one day at 37uC in 24, 12 or 6 well plates. We have previously Figure 6. cd+ T cell numbers are increased in organs of Socs3 fl/fl lck cre mice. The frequency of cd+ T cells within CD3+ cells in the thymus (A), spleen (B) and lung (C) of Socs3 fl/fl lck cre and Socs3 fl/fl mice obtained before or 2.5 weeks after M. tuberculosis infection were determined by FACS. Mean percentage (A-C) and the total numbers (D) of cd+ within CD3+ T cells in the lungs 6 SEM are depicted. Differences with Socs3 fl/fl mice (n = 4 per group) are significant (*p,0.05, ***p,0.001 Student t test). The gating strategy and representative dot-plots for spleens of infected Socs3 fl/fl lck cre and Socs3 fl/fl mice are shown (E). Two million Socs3 fl/fl lck cre or Socs3 fl/fl CD90+ T cells positively selected from spleens using magnetic beads were inoculated i.v. into Rag1 2/2 mice. A group of animals was also inoculated with both Socs3 fl/fl lck cre and Socs3 fl/fl T cells (F). Rag1 2/2 mice were alternatively transferred with 1.2610 6 CD4+ or 2610 6 CD90+ Socs3 fl/fl lck cre spleen cells (G) or in a different experiment with 2610 6 FACS-sorted CD3+ depleted of cd+ T cells or total CD3+ Socs3 fl/fl lck cre spleen cells (H). Two weeks after transfer, mice were infected with M. tuberculosis via the aerosol route. Mice (n$6 per group) were sacrificed 4 weeks after infection. Box and whisker diagrams showing the median CFU, quartiles and the 99 th percentiles in lungs and spleens are depicted (F-H). Differences in CFU are significant (*p,0.05, **p,0.001, ***p,0.001 Mann Whitney U test). doi:10.1371/journal.ppat.1003442.g006 shown by immunofluorescence staining that these BMM are F4/ 80+, CD14+ and Mac-3+ [57].

Generation of mouse bone marrow-derived dendritic cells
Mouse bone marrow-derived dendritic cells (BMDC) were differentiated as previously described [58]. Briefly, bone marrow was extracted from tibia and femurs and cell suspensions cultured in RPMI-1640 medium containing 10% FCS, 100 U/ml penicillin, 100 mg/ml streptomycin and 2 ng/ml GM-CSF (Peprotech, Rocky Hill, NJ). Fresh medium and cytokine were replaced after 3 days. After six days of culture, loosely adherent cells were harvested and seeded in concentrations for infection. In some cases, harvested cells were further selected for CD11c expression with magnetic beads (Miltenyi Biotech) before seeding.

Isolation of pulmonary macrophages
Pulmonary macrophages were isolated as previously described [59]. Briefly, lungs from Socs3 fl/fl and Socs3 fl/fl LysM cre mice were dissected, digested with 1.8 U/ml dispase for 60 min at RT, followed by digestion with DNase (both form Sigma) for 30 min at 37uC. After red blood cell lysis, hematopoietic CD45+ lung cells were positively enriched using magnetic beads (Miltenyi Biotech), and pulmonary macrophages selected by plastic-adherence. Fortyeight hours after seeding to culture plates, CD45+ adherent cells were washed four times with RPMI media.

Isolation of splenic dendritic cells
Splenic DCs were isolated as previously described [60]. Briefly, splenocyte suspensions were positively selected using anti-CD11ccoupled magnetic beads (Miltenyi Biotech). This protocol lead to a purity .95% and an approximate yield of 0.5-1610 6 DCs per spleen.

Real time PCR
Transcripts were quantified by real time PCR as previously described [56]. Hprt was used as a control gene to calculate the DC t values for individual samples. The relative amount of cytokine/Hprt transcripts was calculated using the 2 2(DDCt) method. These values were then used to calculate the relative expression of cytokine mRNA in uninfected and infected cells and tissues.

Flow cytometry and intracellular cytokine staining
Lungs were perfused with PBS through the heart before removal from mice. Following digestion with Collagenase D and DNase I, erythrocytes were lysed and single-cell suspensions prepared by filtering lung tissue through 70-mm nylon cell strainers. Single spleen cell suspensions were obtained by mechanical disruption, lysis of erythrocytes and straining over a 70-mm nylon mesh. Lung cells and splenocytes were stained for CD3, CD4, CD8 and cd TCR (all eBioscience) or F4/80 and Gr1 (BD Biosciences) and fixed before acquisition.
CD11c+ splenic DCs were infected with M. tuberculosis H37Rv MOI5 for 4 hours, washed and cultured overnight with DX5+ NK cells separated from spleens with magnetic beads (Miltenyi Biotech). The next day, cells were treated with brefeldin A (5 mg/ ml) for 4 hours, followed by a FACS stain for DX5 (PE, BD Biosciences) and intracellular IFN-c (eBioscience).
Data were acquired in CyAn ADP flow cytometer (Beckman Coulter) and analyzed using FlowJo software (Tree star Inc., Ashland, OR).

Western blot
M. tuberculosis-infected and uninfected BMM were lysed and separated on 10% separating/5% stacking SDS-polyacrylamide gels. Samples were then transferred onto nitrocellulose membranes (BioRad, Hercules, CA) by electroblotting at 100 V, 250 mA for 80 min. Immunostaining was performed using polyclonal rabbit anti-phosphorylated (Tyr701) STAT3, total STAT3 (Cell signaling technology, Beverly, MA) or anti-actin (Sigma). Membranes were then washed and incubated with horse-radish peroxidase-conjugated polyclonal goat anti-rabbit immunoglobulin (DAKO) and developed using ECL-Plus (Amersham Biosciences, Buckinghamshire, UK) and photographed using a Fuji intelligent dark box II digital camera.

Histopathological analysis
Formalin fixed left lungs of mice experimentally inoculated with M. tuberculosis were blocked on paraffin. From each lung sample 4 sections were obtained, one longitudinal along the long axis of the lobe and 3 across/transversal of the remaining piece of lung.
The blocks were processed and sections were stained with haematoxylin-eosin. All sections were interpreted by the same pathologist (D. G-W.) and scored semi-quantitatively, blinded to the variables of the experiment.
The following features were scored: infected mice is displayed (n = 6, *p,0.05 Mann Whitney U test) (F). The mean frequency of IL-17-secreting CD4+, CD8+ and cd+ within CD3+ cells in lungs of infected mice (5 mice per group) 6 SEM is depicted (G).     peritoneal macrophages (B). The mean IL-6 in Socs3 fl/fl and Socs3 fl/ fl LysM cre (A, B) and gp130 F/F and WT (D) BMM 6 SEM as determined by ELISA is depicted. Differences with control BMM are significant (*p,0.05 and ***p,0.001 Student t test). The mean levels of Il-6 mRNA in Socs3 fl/fl and Socs3 fl/fl LysM cre BMM either infected with BCG or treated with Pam3CSK4 were determined by real time PCR in triplicate independent cultures per condition and compared to non infected cultures (C). One representative of two independent experiments is shown. Differences with control BMM are significant (**p,0.01Student t test). The mean levels of Tnf mRNA in Socs3 fl/fl and Socs3 fl/fl LysM cre (E) and gp130 F/F , gp130 F/F Il6 2/2 and WT (G) BMM either infected with BCG or treated with Pam3CSK4 determined by real time PCR in triplicate independent cultures per condition compared to non infected cultures is depicted. Differences with control BMM are significant (**p,0.01Student t test). The mean TNF concentration in supernatants of BCG-infected gp130 F/F , gp130 F/ F Il6 2/2 and WT BMM 6 SEM as determined by ELISA in triplicate cultures per condition (F). Differences with control BMM are significant (**p,0.01Student t test). Ifn-c (D), Tnf (E) and iNos (F) transcripts was measured by real time PCR. The mean fold cytokine mRNA increase 6 SEM in lungs from infected mice (n = at least 5 per group) was calculated. Differences with controls are significant (* p,0.05 Student t test). The mean frequency of FoxP3+ within CD4 + T cells in the pulmonary lymph nodes and lungs from mice (n = 5 mice per group) 6 weeks after infection with M. tuberculosis was determined by FACS (G). mice were sacrificed at the indicated time points and the total RNA was extracted from lungs. The accumulation of Il-6 (A) transcripts was measured by real time PCR. The mean fold mRNA increase 6 SEM in lungs from infected mice (n$5) was calculated. Differences with controls are significant (*p,0.05 Student t test). Lung cell suspensions from Socs3 fl/fl lck cre and Socs3 fl/fl mice at 2.5 weeks after infection with M. tuberculosis were stimulated with 20 mg/ml PPD. The content of IL-6 (B) and IL-10 (C) in supernatants analysed by a cytokine bead array (CBA) 48 h after stimulation. The mean cytokine concentration 6 SEM (n = 6 animals per group) is depicted. Levels of Il-6 mRNA expression was determined by real time PCR in lungs of Socs3 fl/fl and Socs3 fl/fl lck cre mice, before and 6 weeks after BCG infection (n = 6 per group) (D). Socs3 fl/fl and Socs3 fl/fl lck cre mice were immunized s.c. with 5610 6 heat-killed BCG and boosted after 2 weeks with 2.5610 6 heat-killed BCG. After 2 weeks, mice were sacrificed and splenocytes from immunized or nonimmunized control mice were co-incubated or not with 20 mg/ ml PPD for 72 h. The mean IL-6 concentration 6 SEM (n = 6 animals per group) in lung cell supernatants was determined by ELISA (E). (EPS)