Role of the chemokine decoy receptor D6 in balancing inflammation, immune activation, and antimicrobial resistance in Mycobacterium tuberculosis infection

Preliminarily, we investigated the expression of the D6 receptor during M. tuberculosis infection. In human lungs and lymph nodes from a patient with pulmonary tuberculosis, D6 expression was observed in lymphatic endothelial cells and only occasionally in scattered leukocytes (Fig. S1). In the mouse, qualitative RT-PCR experiments indicated that D6 transcripts in the liver, spleen, and lungs do not undergo significant changes at 4, 8, and 12 wk after infection with M. tuberculosis (Fig. S2). These data suggest that D6 is mainly expressed in lymphatic endothelial cells, in agreement with previous reports (17), and that its expression does not change appreciably during M. tuberculosis infection.

To determine whether the D6 receptor plays a role in the control of M. tuberculosis infection, WT and D6^−/− mice were infected via the intranasal (i.n.) route with 2,000 CFU. This dose did not cause mortality in C57BL/6 mice in our study (Fig. 1 A). However, M. tuberculosis infection in D6^−/− mice resulted in increased mortality, with ∼20% of mice succumbing by week 8, 50% of mice succumbing by week 12, and 100% of mice succumbing at week 16, at which time the experiments were terminated (Fig. 1 A). The phenotype of D6^−/− was dramatic and was found highly reproducible over a period of 3 yr in two additional experiments performed on a total of 17 mice per group. Despite the exaggerated susceptibility of D6^−/− mice to M. tuberculosis infection, there was no difference between the two groups of mice in bacterial loads, as determined by CFU counts in the lung, liver, and spleen at 2, 4, 8, and 12 wk after infection (Fig. 1 B).

Of note, CFU data at 4, 8, and 12 wk after M. tuberculosis infection were rechecked in two additional independent experiments with three mice per group performed over a period of 2 yr, confirming the results reported in Fig. 1 B. We then examined by histology the effect of D6 deficiency on the inflammatory response in the lungs, liver, and spleen. Although in WT mice a moderate cellular infiltration was seen at 12 wk after infection and most of the lung airspace remained intact, an exaggerated response was evident in surviving D6^−/− mice (see Fig. 5 A). Histological analysis also showed a diffuse liver necrosis in D6^−/− mice when compared with WT mice (Fig. 1 C). In agreement with the histology results, at 12 wk after M. tuberculosis infection, serum transaminases were abnormally increased in D6^−/− compared with WT mice (Fig. 1 D). Areas of focal necrosis and mononuclear cell infiltrates were also evident in the kidneys (Fig. 1 C), and consistent with this, at 12 wk after M. tuberculosis infection renal function was also severely compromised in D6^−/− mice, which had significantly higher blood urea nitrogen (BUN) levels and frequency of severe proteinuria as compared with WT mice (Fig. 1 E). Of note, although M. tuberculosis growth occurred in the liver (Fig. 1 B), we failed to detect M. tuberculosis CFU in kidneys of both WT and D6^−/− mice over a 12-wk period (not depicted). Thus, despite the successful control of the growth of the infecting mycobacteria, D6 deficiency led to increased mortality associated with lung inflammation and combined renal and liver failure.

We examined the cellular infiltrate in the lungs and liver of D6^−/− mice after M. tuberculosis infection. At predetermined time points, organs were removed, the total number of cells was calculated, and flow cytometric analysis was performed to determine cell populations (Fig. 2). Numbers of DCs (F4/80⁻ and CD11c⁺) in the lungs (Fig. 2 A) were not significantly different between D6^−/− and WT mice. The numbers of macrophages and CD4 and CD8 lymphocytes were significantly higher in D6^−/− mice at all time points after infection. Approximately 45% of CD4 and CD8 T cells infiltrating the lungs of M. tuberculosis–infected D6^−/− mice at 8 and 12 wk after infection expressed activation markers (CD25^high, CD44^high, and CD62L^low), whereas only 30% of CD4 and 35% of CD8 T cells expressed the same activation markers in WT mice, although differences were not statistically significant (unpublished data). Similar findings were also detected in the liver, although the differences between D6^−/− and WT mice were lower than in the lungs and generally reached significance at 8 and 12 wk after infection (Fig. 2 B).

Because of the finding that the lungs of D6^−/− mice contained more macrophages and T lymphocytes than the lungs of WT mice, we asked whether a greater number of these cells migrated from the lungs to the draining lymph nodes. To investigate this possibility, flow cytometric analysis was performed on the mediastinal lymph nodes of D6^−/− and WT mice after M. tuberculosis infection. As shown in Fig. 2 C, the number of macrophages showed a significant increase in the lymph nodes of D6^−/− mice at 4 wk after infection. At 4 and 8 but not 12 wk after infection, there were also statistically significant differences in the total number of DCs in the lymph nodes of D6^−/− mice. The numbers of CD4 and CD8 T lymphocytes were significantly higher in the lymph nodes of D6^−/− mice at 4 and 8 wk after infection, but although the latter constantly increased, the former peaked at week 4 and then decreased. The increased migration of DCs and T lymphocytes to the lymph nodes was expected to result in increased numbers of primed T lymphocytes, followed by migration to the lungs. ELISPOT analysis confirmed this hypothesis, showing that significantly more CD4 and CD8 T lymphocytes from the lungs of D6^−/− mice were primed to produce IFN-γ when stimulated by bacille Calmette-Guérin (BCG)–infected DCs (Fig. 3).

Previous studies have shown that exacerbated inflammation in mice lacking the D6 receptor is associated with higher levels of chemokines in blood and tissues (15, 16, 18). We therefore examined chemokine levels in the bronchoalveolar lavage (BAL) and serum of M. tuberculosis–infected mice. As shown in Fig. 4 A, the concentrations of all tested chemokines were significantly higher in the BAL of D6^−/− than of WT mice at 12 wk after infection. Similarly, concentrations of CCL2/monocyte chemotactic protein 1 (MCP-1), CCL3/macrophage inflammatory protein 1α (MIP-1α), CCL4/MIP-1β, and CCL5/regulated on activation, normal T cell expressed and secreted (RANTES) were also significantly increased in the serum of D6^−/− mice at 12 wk after infection (Fig. 4 B). Control of M. tuberculosis infection depends on the activation of macrophages by IFN-γ and TNF-α, leading to the induction of inducible NO synthase and the production of reactive nitrogen intermediates such as NO. TNF-α is reported to be essential for the early expression of mRNA encoding CC and CXC chemokines, the initial recruitment of CD4 T lymphocytes, and the formation and maintenance of the granulomas (12, 19, 20). To determine whether these components of the antimycobacterial immune response were affected by D6 deficiency, ELISA was performed on the BAL and serum of M. tuberculosis–infected mice at 8 (unpublished data) and 12 wk after infection. Concentrations of TNF-α, IL-1β, and IFN-γ were significantly higher in the BAL (Fig. 4 C) and serum (Fig. 4 D) of M. tuberculosis–infected D6^−/− mice at all tested time points, as compared with infected WT mice. Thus, M. tuberculosis infection in the presence of D6 deficiency causes a local and systemic inflammatory response characterized by increased levels of proinflammatory chemokines and cytokines.

The results in the previous paragraph clearly show that M. tuberculosis infection in the absence of the D6 receptor is characterized by an exaggerated inflammatory response and increased mortality, despite the successful control of bacterial growth. To investigate whether the highly increased levels of chemokines observed in D6^−/− mice were involved in the pathogenesis, we attempted to block inflammatory chemokines in vivo. For this purpose, mice were treated with a mixture of antibodies to the CC chemokines CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES, as previously described (15, 18), or with antibodies to each individual chemokine. On the same days, control mice received i.p. injections of an equivalent amount of irrelevant antibodies. Mice were injected weekly with M. tuberculosis, starting from the third week after infection. Following this treatment schedule, CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES production in the serum of D6^−/− mice was consistently reduced (Table S1), even below levels detected in WT mice, thus demonstrating the efficacy of the blocking. Conversely, D6^−/− mice treated with control antibody had very high levels of serum chemokines, as did untreated D6^−/− mice. TNF-α, IL-1β, and IFN-γ levels in serum were also consistently reduced by treatment with antibodies to CC chemokines (Table S1), and cellular influx to the infected tissues significantly decreased (unpublished data). Additionally, both renal and liver function appeared improved by this treatment, as suggested by the low levels of serum transaminases and BUN, and by the low proteinuria detected in these mice (Table S1). Overall, the most impressive consequence of the neutralization of CC chemokines in D6^−/− mice was a significantly prolonged survival, as demonstrated by the only 10% mortality at 12 wk and 30% mortality at 16 wk after infection with M. tuberculosis (Table I). However, in spite of the ability of chemokine-blocking antibodies to reduce the overall inflammatory response (Fig. 5 A), the ability of D6^−/− mice treated in this manner to control the growth of M. tuberculosis in the lungs was impaired, with higher CFU counts found at 8 and 12 wk (Fig. 5 B). This result therefore indicates that the D6 receptor plays a crucial role in M. tuberculosis infection, and in the balance between immunopathology and protective immune responses.

A better understanding of the immunological mechanisms of pathogenesis and protection is of essential importance for the design of novel vaccines and immunotherapies against tuberculosis. It has been proposed that the protective response to M. tuberculosis infection requires CD4 and CD8 lymphocytes, the Th1-type cytokines IFN-γ and TNF-α, and activated macrophages (for review see reference 21). The cooperation between the cells and cytokines requires close interaction, which is achieved after migration and granuloma formation in the lungs. The hallmark of infection in the lung is granuloma formation, consisting of clusters of macrophages, lymphocytes, and DCs, which physically contains the mycobacteria and creates a microenvironment for immune cell interaction, limiting M. tuberculosis growth and dissemination. Chemokines are potent leukocyte activators and chemoattractants aiding granuloma formation and thought to be critical for the immune response to M. tuberculosis (14). In vitro experiments demonstrated that M. tuberculosis infection of bone marrow–derived mouse macrophages results in the expression of TNF-α, as well as several chemokines, including ligands for the chemokine receptors CXCR3, CCR5, and CCR2. Neutralization of TNF-α by using antibody or TNFR1-deficient macrophages demonstrated that expression of certain chemokines (CXCL9/monokine induced by IFN-γ, CXCL10/IFN-inducible protein 10, CXCL11/IFN-inducible T cell α chemoattractant, CCL5/RANTES, and CCL2/MCP-1) after M. tuberculosis infection was dependent, at least in part, on TNF-α. However, the lack of TNF-α did not completely abrogate chemokine expression, indicating that there are other factors, induced as a result of infection, that stimulate chemokine production (22).

In mouse models, gene expression of CXC and CC chemokines has been detected in the lungs after M. tuberculosis infection (22–24). CXCR3-deficient mice (25) have an impaired granuloma formation after aerosol infection with M. tuberculosis, although this effect is transient, occurring at the early stages of infection. CCR2-deficient mice are extraordinarily susceptible to moderate- or high-dose M. tuberculosis administered i.v. (26), and susceptibility is dose-dependent (27). Conversely, CCR5 (28) may not be essential to the development of a protective response to M. tuberculosis infection. CCL2/MCP-1–deficient mice did not demonstrate an increased susceptibility to M. tuberculosis infection, but whether cell infiltration or histology was affected in these mice was not reported (29, 30). However, transgenic mice overexpressing CCL2/MCP-1 were more susceptible to tuberculosis (31).

The action of chemokines is tightly controlled by decoy receptors (9), which internalize and target chemokines for intracellular degradation, thus avoiding excessive inflammatory responses and the consequent tissue damage. D6 is a decoy receptor that recognizes most inflammatory CC chemokines, implying that it may play a key role in resolving CC chemokine–driven inflammatory responses. The results reported in this paper confirm this possibility in a mouse model of M. tuberculosis infection through the mucosal (i.n.) route and show that D6^−/− mice have an exaggerated inflammatory response leading to death, despite their ability to efficiently control bacterial load. Upon i.n. delivery of low-dose M. tuberculosis, 20% of mice succumbed at week 8, 50% of mice succumbed at week 12, and 100% of mice succumbed at week 16, despite the successful control of the growth of M. tuberculosis bacilli. The phenotype of D6^−/− was dramatic and was found highly reproducible over a period of 3 yr in two additional experiments performed on a total of 17 mice per group. Moreover, CFU data at 4, 8, and 12 wk after M. tuberculosis infection were found highly reproducible in two additional independent experiments performed over a period of 2 yr. Mortality in D6^−/− mice coincided with an overwhelming local and systemic inflammatory response that mainly compromised liver and kidney functions in the host, characterized by increased and sustained numbers of macrophages, DCs, and CD4 and CD8 T lymphocytes, and increased production of inflammatory CC chemokines and cytokines (TNF-α, IL-1β, and IFN-γ). However, we failed to detect M. tuberculosis CFU in kidneys of both WT and D6^−/− mice over a 12-wk period. This might reflect either M. tuberculosis growth below the CFU detection limit or the development of kidney damage in the absence of kidney infection. A similar increase in macrophage, DC, and T cell numbers was detected in the mediastinal lymph nodes, suggesting the possibility that more T lymphocytes were being primed and migrating to the lungs. Such increased priming likely accounts for the substantially increased numbers of T lymphocytes migrating to the lungs of D6^−/−mice after M. tuberculosis infection, as confirmed by ELISPOT analysis showing that significantly more CD4 and CD8 T lymphocytes from the lungs of D6^−/− mice were primed to produce IFN-γ when stimulated in vitro by BCG-infected DCs. The finding that D6^−/− mice had more inflammation is intriguing and convincingly supports an immunoregulatory role for the D6 receptor. Recent data in other systems have shown that D6^−/− mice display exaggerated responses to inflammatory stimuli, resulting in an aggressive cutaneous inflammatory pathology after the application of phorbol esters (15). In another study (16), D6^−/− mice displayed an exaggerated response to the subcutaneous injection of complete Freund's adjuvant. Unfortunately, none of these studies has been able to define a mechanism for the heightened immune response in the D6^−/− mice. Although there are several different possibilities for the increased inflammatory response in M. tuberculosis–infected D6^−/− mice, the one we favor is that lack of the D6 receptor fails to remove CC chemokines from many different organs and tissues, determining their increased concentration. Such increased expression/circulation of CC chemokines may, in turn, lead to signaling through their receptors and could result in the influx of immune cells and abundant, uncontrolled production of proinflammatory cytokines, which ultimately mediate tissue damage. However, the increased expression may also have been a reaction to the numbers of cells in infected organs. Whatever the case, the concentrations of CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES were found to be highly increased in the BAL and serum of M. tuberculosis–infected D6^−/− mice and sustained over time; moreover, blocking all four of them in D6^−/− mice in vivo (CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES) by repeated injection of a cocktail of neutralizing antibodies significantly reduced cell recruitment, production of proinflammatory CC chemokines and cytokines, liver and kidney damage, and, most importantly, the mortality of D6^−/− mice. However, despite controlling inflammatory responses, these mice became more susceptible to infection with M. tuberculosis, as demonstrated by higher CFU counts in the lungs at 8 and 12 wk after infection. The data reported in Table I clearly show that none of the neutralized chemokines alone accounts for the reduced inflammatory response and mortality, and for the increased susceptibility to infection.

CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES are all ligands of CCR5, which can attract and activate macrophages and Th1 lymphocytes. However studies in CCR5^−/− mice have found that although CCR5 plays a role in the migration of DCs to and from lymph nodes (28, 32), it is not indispensable for granuloma formation and immune protection against M. tuberculosis infection (32). CCL5/RANTES has been associated with the generation of type 1 cytokine–producing granulomas (33). CCR2-deficient mice are extraordinarily susceptible to moderate- or high-dose M. tuberculosis administered i.v. but not to low aerosol doses (27), and CCL2/MCP-1–deficient mice do not demonstrate an increased susceptibility to M. tuberculosis infection (29). However, there is redundancy in the chemokine system. Specific to this study, the chemokines CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, and CCL5/RANTES are ligands of CCR5, but CCL3/MIP-1α and CCL5/RANTES can also signal through CCR1 and CCR3. This finding makes it difficult to dissect the precise roles of individual chemokines and their receptors. In this study, enhancing the type 1 T cell response or the ability of T lymphocytes to enter the lungs over the WT level does not have an obvious beneficial effect on the control of infection and inflammation, indicating that there are additional factors that must be induced or enhanced to increase the ability of the host to eliminate M. tuberculosis infection.

In summary, our study suggests an important role for the D6 receptor in immune regulation. D6 controls chemokine accumulation and activity, migration, and possibly, the maintenance of DCs and T cells in infected organs, as well as the production of inflammatory cytokines. Therefore D6, through its ability to avoid an excess of chemokines in the inflamed tissue, may favor resolution of chronic inflammatory responses and overall provide a fine mechanism for the control of the balance between protective immune responses and immunopathology.

D6^−/− mice were generated as previously described (15). Homogeneous populations were established by backcrossing heterozygous mice to C57BL/6 mice for more than eight generations. The resultant heterozygous mice were bred to obtain homozygotes. WT C57BL/6J and D6^−/− mice were bred in a specific pathogen-free/viral antibody–free barrier facility and obtained from Charles River Laboratories. 8–14-wk-old male and female mice were used in accordance with institutional guidelines in compliance with national (D.L. N.116, Gazzetta Ufficiale della Repubblica Italiana, supplement 40, 18-2-1992) and international law and policies (European Economic Community Council, 1987, Directive 86/609, Official Journal of European Communites L 358,1; and Institute of Laboratory Animal Resources, Committee on Life Sciences, National Research Council, 1996, Guide for the Care and Use of Laboratory Animals). Each experimental group consisted of 7–10 mice. All infected mice were maintained under germ-free conditions and were routinely monitored for mouse pathogens. Experiments were performed in specific pathogen-free facilities.

All chemicals were purchased from Sigma-Aldrich, unless otherwise noted. Middlebrook 7H9 liquid medium and 7H10 agar were obtained from Difco Laboratories. Antibodies used in flow cytometry were obtained from BD Bioscience or R&D Systems.

Mice were infected via the i.n. route with M. tuberculosis H37Rv (2 × 10³ CFU in 20 μl). This resulted in reproducible delivery of 50–100 viable CFU M. tuberculosis, as confirmed by CFU determination on the lungs of infected mice 1 d after infection. In fact, in a total of 10 mice per group from four independent experiments, day 1 M. tuberculosis CFU ranged from 50 to 100 (with only one exception in one WT mouse of 110 CFU), and in any case no statistical significant differences between day 1 CFU in WT (72.5 ± 21) and D6^−/− (75 ± 19) mice were detected (P = 0.9658 by the Student's t test). Moreover, using the i.n. infection route, we failed to detect M. tuberculosis CFU in nasal tissues in two different experiments. The tissue bacillary load was quantified by plating serial dilutions of the lung, liver, and spleen homogenates into 7H10 agar, as described previously (34).

To determine cellular infiltrate, the lungs, liver, and mediastinal lymph nodes were removed at different weeks after infection and digested in the presence of 200 U/ml of collagenase, and mononuclear cell suspensions were obtained through Lympholyte M (Cederlane Laboratories) gradient centrifugation. The viability of cells, as determined by Trypan blue exclusion, was >90%. A single-cell suspension was prepared by pushing the tissue through a cell strainer. In some experiments, lung mononuclear cells were enriched in T cells by passage through a nylon wool column, and CD4 and CD8 T cells were then sorted by anti-CD4 or -CD8 immunomagnetic beads (Miltenyi Biotec), according to the manufacturer's instructions (34). The cells were incubated for 24 h at 37°C in complete medium to allow cells and beads to dissociate. Single-cell suspension were counted. The samples were triple stained with fluorochrome-conjugated anti-CD3, -CD4, -CD8, -CD44, -CD62L, -CD25, -CD11c, and -F4/80, in FACS buffer (0.1% Na azide, 0.1% BSA, and 20% mouse serum). Purified or PE-, FITC-, CyChrome-, or allophycocyanin-conjugated isotype control Igs were used as controls. After washes, the cells were fixed in 4% paraformaldehyde for 1 h and collected on a FACSCalibur (BD Biosciences). Analysis was performed on CellQuest software (BD Biosciences).

The ELISPOT method (35) was used to detect IFN-γ secretion by individual CD4 or CD8 T cells from infected mice, after stimulation with BCG-infected DCs in vitro prepared as previously described (36). In brief, ELISPOT plates (BD Biosciences) were coated with IFN-γ capture antibody overnight at 4°C. The capture antibody was discarded, and the plates were washed and blocked with complete media for 2 h at room temperature. 150,000 purified CD4 or CD8 T cells per well were cultured with irradiated BCG-infected WT DCs for 36–40 h at 37°C. The cells were discarded, and plates were washed with deionized water and PBS/Tween 20. Secondary biotinylated antibody was added for 2 h and incubated at room temperature, followed by washing with PBS/Tween 20. Streptavidin-alkaline phosphatase was added to the plates for 1 h, followed by washing and development of a color reaction using the AEC substrate reagent kit (BD Biosciences). The reaction was stopped when spots developed by running the plate under water. The spots were enumerated using an ELISPOT reader (Bioreader 3000; BioSys).

BAL was obtained by flushing 2 ml PBS into the lungs of killed mice (37). The concentrations of chemokines (CCL2/MCP-1, CCL3/MIP-1α, CCL4/MIP-1β, CCL5/RANTES, CCL11/eotaxin, and CCL22/MDC) and cytokines (TNF-α and IFN-γ) were quantified by ELISA in BAL and sera using commercially available kits (R&D Systems).

Serum alanine aminotransferase (ALT; EC 2.6.1.2) and aspartate aminotransferase (AST; EC 2.6.1.1) were measured by the standard photometric method using an automatic analyzer (type 7350; Hitachi Chemical Diagnostics) and a commercial kit (Sigma-Aldrich) adapted to small-sample volumes, as previously described (38). The normal ranges of ALT and AST were obtained testing sera from 10 mice, age- and sex-matched with animals used in experimental groups (38). Proteinuria was measured on a 0–4+ scale using a colorimetric assay strip (38) for albumin (Albustix; Bayer): 0, absent; 1+, ≤30 mg/100 ml (mild); 2+, 100 mg/100 ml (moderate); 3+, 300 mg/100 ml (severe); and 4+, >2,000 mg/100 ml (severe). BUN levels were measured by impregnating Azostix strips (Bayer) with a drop of fresh blood (38) and using the following scale: 1+, 5–15 mg/100 ml (normal); 2+, 15–26 mg/100 ml (mild); and 3+, ≥30 mg/100 ml (severe).

To block inflammatory chemokines, mice were treated with a mixture of goat antibodies to the CC chemokines CCL2/MCP-1, CCL3/MIP-1α, and CCL4/MIP-1β), and a monoclonal antibody to CCL5/RANTES purchased lyophilized from R&D Systems, resuspended in sterile PBS, and mixed as previously described (16, 18). Mice were given i.p. injections of 200 μl of the mixture, equivalent to 100 μg of each antibody. Alternatively, mice were treated with 400 μg of individual antibodies specific for a single CC chemokine. On the same days, control mice received i.p. injections equivalent to 400 μg of irrelevant antibodies (normal goat IgG; Sigma-Aldrich) in 200 μl PBS. Mice were injected weekly, starting from the third week after infection with M. tuberculosis.

Tissue samples for histological studies were fixed in 10% normal buffered formalin, followed by paraffin embedment. 5–6-μm sections were stained with hematoxylin and eosin. For histometry, 5-μm-thick sections from each specimen were stained with hematoxilin and eosin and photographed at 6× using a stereoscopic zoom microscope (SMZ800; Nikon) and a digital camera (Coolpix 990; Nikon). Sections of at least eight lung lobes were studied in each case. A sequence of appropriate software programs were used (Scion Image [Scion Corporation]; Photoshop 5.0 [Adobe Systems Inc.]) to determine the area of each single lesion and the total tissue area on photomicrographs at each time point. Measurements were done blindly by two independent pathologists (P.J. Cardona and N. Caceres). Data are percentages calculated by dividing the granuloma-involved area by the global tissue area and multiplying by 100.

The double-tailed Student's t test was used to compare the significance of differences between groups. Kaplan-Meier analysis was used to determine the statistical significance of the differences in survival times of WT and D6^−/− mice. The significance of differences of log₁₀ CFU counts between groups was determined using one-way analysis of variance (ANOVA), with Scheffe tests for post-ANOVA individual comparisons. P < 0.05 was considered significant.

Fig. S1 shows lung and draining lymph nodes sections from M. tuberculosis patients stained for D6 and the macrophage marker CD68. Fig. S2 shows expression levels of the D6 transcript in mouse tissues during M. tuberculosis infection. Table S1 reports the effects of treatment of D6^−/− mice with anti–CC chemokine antibodies.

This study was carried out with financial support from the European Union FP6 contracts LSHP-CT-2003-503240 (MUVAPRED), LSHB-CT-2005-518167 (INNOCHEM), and LSHG-CT-2005-005203 (MUGEN); the CARIPLO Foundation (NOBEL project); the Ministero dell'Istruzione, Università e Ricerca (FIRB project); and the Universities of Palermo and Milan (FIRST projects). This work was conducted with the support of the Fondazione Humanitas per la Ricerca and Italian Association for Cancer Research.

The authors have no conflicting financial interests.