Autocrine IL-10 Induces Hallmarks of Alternative Activation in Macrophages and Suppresses Antituberculosis Effector Mechanisms without Compromising T Cell Immunity¹

Tuberculosis (TB)⁵ is one of the most prevalent bacterial infections worldwide and constitutes a leading global health threat (1). Human TB caused by Mycobacterium tuberculosis (Mtb) is responsible for 8 million new cases and ∼2 million deaths annually. Primary TB is thought to depend on a combination of genetic susceptibility of the host and virulence of the infecting Mtb strain, whereas postprimary TB is due to a temporal dysregulation of otherwise intact immune defenses. More detailed elucidation of the balance between protective and inflammatory immune responses to Mtb may accelerate the development of more effective vaccines and therapies targeting both or either form of TB.

The cell-mediated immune response is known to be critical in the host defense against infection with intracellular pathogens such as mycobacteria. Th1 lymphocytes play an important role in granuloma formation by secreting type 1 cytokines, primarily IFN-γ and TNF (2, 3). These cytokines stimulate the antimicrobial activity of infected macrophages, allowing intracellular bacterial killing through reactive nitrogen intermediates (RNI) and LRG-47 (4, 5). IL-12 indirectly promotes an effective cell-mediated immune response by inducing IFN-γ production from T and NK cells. On the other hand, these protective T cell-mediated immune responses also play a causative role in the development of the pathology of this type of chronic disease (6). Hence, down-regulation of the inflammatory response by endogenous factors that limit tissue damage can promote establishment of chronic infection and may also lead to reactivating TB. In fact, increased levels of Th2-type anti-inflammatory cytokines such as IL-4 and IL-10 were detected in patients suffering from postprimary disease (7).

IL-10 is produced by macrophages, T cells, B cells, and a variety of other cell types. The effects of IL-10 on immune responses are mostly inhibitory (8). In macrophages, IL-10 inhibits antimicrobial effector mechanisms, the expression of costimulatory molecules, and the production of proinflammatory cytokines (9, 10). Inhibition of IL-12 production by IL-10 (11) may be one mechanism by which IL-10 blocks the development of protective Th1 immune responses (12). The absence of IL-10 causes overproduction of inflammatory cytokines (13) and the development of chronic inflammatory bowel disease (14). On the other hand, IL-10-deficient (^−/−) mice also show increased resistance to intracellular pathogens such as Listeria monocytogenes (15), Leishmania major (16), and Trypanosoma cruzi (17). Therefore, IL-10 is required to prevent immunopathology, but it can also impair protective responses against several pathogens. The immunoregulatory function of IL-10 and the fact that elevated levels of IL-10 have been detected in TB patients (18, 19) suggest a role for IL-10 in susceptibility to TB. However, one report of Mtb infection in IL-10^−/− mice found neither increased protective immune responses nor exacerbated immunopathology (20).

By contrast, mice lacking IL-10 were shown to clear Mycobacterium bovis bacillus Calmette-Guérin infection more efficiently (21). Taken together, further investigation of Mtb infection in different IL-10 genetic models differing in the level or source of IL-10 production appear warranted. Indeed, mice overexpressing IL-10 in activated T cells (tIL-10^tg mice) were previously reported to be more susceptible to Mtb infection, exhibiting an impaired Th1 development as characterized by decreased numbers of activated T cells in the blood circulation and lung tissue (22). Similarly, another T cell-specific IL-10 transgenic model was also shown to have an increased M. bovis bacillus Calmette-Guérin burden (23). However, because IL-10 is only overexpressed in these mice when T cells become activated, effects of overproduction of IL-10 by innate immune cells could not be studied.

We were interested in analyzing the consequences of a macrophage-specific overexpression of IL-10 during experimental TB for the following reasons: first, macrophages rather than T cells were reported to be the major source of IL-10 in TB patients (18), and second, reactivation disease in patients is typically associated with a strong, rather than diminished, Th1 immune response (24). In this study, infection of mice expressing an epitope-tagged IL-10 under control of the human CD68 promoter specifically in macrophages (macIL-10^tg mice) (25) resulted in a very different immunological phenotype than reported for Mtb-infected tIL-10^tg mice. Even though macIL-10^tg mice were also more susceptible to Mtb infection than nontransgenic littermates, the differentiation, recruitment, and activation of Th1 cells as well as the induction of IFN-γ-dependent effector genes against Mtb were unaffected by macrophage-derived IL-10. Transcriptome analysis revealed that after Mtb infection genes preferentially expressed in macrophages were specifically modulated in the lungs of macIL-10^tg mice. Among these, genes typical for alternative macrophage activation, such as Arg1, were significantly induced in the lung transcriptome of Mtb-infected macIL-10^tg mice. Taken together, our study revealed that macrophage-derived IL-10 induces alternatively activated macrophages and promotes exacerbation of chronic Mtb infection in the face of intact T cell immunity.

Transgenic mice (macIL-10^tg mice) specifically expressing IL-10 in macrophages under control of the human CD68 promoter were bred under specific pathogen-free conditions at the Technical University of Munich (25). macIL-10^tg mice were on an FVB genetic background, and transgene-negative littermates were used as controls (FVB). Myeolid-specific arginase-1-deficient Tie2^cre Arg-1^flox/flox mice were housed at St. Jude Children’s Research Hospital (26). In any given experiment, mice were matched for age and sex. For experiments, mice were maintained under barrier conditions in the BSL 3 facility at the Research Center Borstel in individually ventilated cages. All experiments performed were in accordance with the German Animal Protection Law and were approved by the Animal Research Ethics Board of the Ministry of Environment (Kiel, Germany).

Bone marrow-derived macrophages (BMMφ) were obtained after flushing of femora from Tie2^cre-pos Arg-1^flox/flox mice and cre-negative littermates. To generate BMMφ, bone marrow cells were cultivated in L-929 conditioned medium as source for M-CSF activity for 9 days (27).

Mtb (H37Rv) and Mycobacterium avium (TMC724; originally obtained from Dr. F. Collins, Trudeau Institute, Saranac Lake, NY) were grown in Middlebrook 7H9 broth (Difco) supplemented with Middlebrook oleic acid-albumin-dextrose-catalase (OADC) enrichment medium (Life Technologies), 0.002% glycerol, and 0.05% Tween 80. Mid-log-phase cultures were harvested, aliquoted, and frozen at −80°C. After thawing, viable cell counts were determined by plating serial dilutions of the cultures on Middlebrook 7H10 agar plates followed by incubation at 37°C. All experiments were performed in the BSL 3 laboratories at the Research Center Borstel. Before infection of experimental animals, stock solutions of mycobacteria were diluted in sterile distilled water, and pulmonary infection was performed using an inhalation exposure system (Glas-Col). To infect mice with 100 CFU of Mtb or 10⁵ M. avium per lung, animals were exposed for 40 min to an aerosol generated by nebulizing ∼5.5 ml of a suspension containing 2 × 10⁵ or 1 × 10⁹ live bacteria/ml, respectively. Inoculum size was checked 24 h after infection by determining the bacterial load in the lung of infected mice. Mice were regularly weighed before and after infection. In accordance with the Animal Research Ethics Board of the Ministry of Environment, mice that lost 25% of their original weight during the course of infection had to be sacrificed.

Bacterial loads in lung, liver, and spleen were evaluated at different time points after infection with mycobacteria to follow the course of infection. Organs from sacrificed animals were removed aseptically, weighed, and homogenized in PBS containing a proteinase inhibitor mixture (Roche Diagnostics) for subsequent quantification of cytokines prepared according to the manufacturer’s instructions. Ten-fold serial dilutions of organ homogenates were plated in duplicates onto Middlebrook 7H10 agar plates containing 10% OADC and incubated at 37°C for 19–21 days. Colonies on plates were enumerated, and results were expressed as log₁₀ CFU per organ.

One lung lobe per mouse was fixed in 4% formalin-PBS, set in paraffin blocks, and sectioned (2–3 μm). For immunohistological detection of NOS2, tissue sections were deparaffinized, placed in 10-mm sodium citrate buffer (pH 6.0), and then pressure-cooked for exactly 1 min. After blocking for 20 min in 1% H₂O₂ solution, slides were incubated with appropriately diluted polyclonal rabbit anti-mouse inducible NO synthase (NOS2) (BIOMOL) in TBS/10% FCS for 30 min in a humid chamber. Appropriately diluted goat anti-rabbit IgG (bridging Ab) (Dianova) and diluted rabbit anti-goat IgG-peroxidase conjugate (tertiary Ab) (Dianova) were used in sequential incubations of 30 min each. For immunohistological detection of B cells, deparaffinized sections were blocked in H₂O₂ and sequentially incubated with a monoclonal rat anti-mouse CD45R/B220 Ab appropriately diluted in 10% FCS/PBS, a biotinylated goat anti-rat Ab and streptavidin-HRP (all from BD Biosciences). Development was performed using diaminobenzidine (Sigma-Aldrich) and urea superoxide (Sigma-Aldrich), and hemalum was used to counterstain the slides.

Before and at different time points after aerosol infection with Mtb, weighed lung samples were homogenized in 5 ml of 4 M guanidinium-isothiocyanate buffer, and total RNA was extracted by acid phenol extraction.

Affymetrix GeneChips were used for genome-wide expression analysis of the impact of IL-10 overexpression during pulmonary Mtb infection. Tissue from individual mice was processed as biological duplicates (untreated controls and day 25 after infection) and triplicates (day 42 after infection). Per sample, 3 μg of total lung RNA was processed using the GeneChip Expression 3′ Amplification One-Cycle Target Labeling Kit according to the manufacturer’s instruction. Biotinylated cRNA was hybridized on MOE430A 2.0 GeneChips, which were stained, washed, and scanned following standard procedures. Microarray data have been deposited in the Gene Expression Omnibus database and are accessible through GEO Series accession number GSE14136 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE14316). CEL files were processed for global normalization and generation of expression values using the robust multiarray average algorithm in the R affy package (www.R-project.org). To identify differentially regulated genes, a combination of filtering for absolute (max-min >50) and relative (max/min >2) change across all conditions as well as statistical testing using ANOVA for the variables “infection” and “genotype” was applied. Correction for multiple testing was done using Benjamini-Hochberg (28). The 4188 probe sets passing the filtering criteria were then subjected to hierarchical clustering and visualization procedures using the Spotfire DecisionSite Functional Genomics software. Data mining of differentially regulated gene lists for enrichment of cocitations with cells and tissues was performed using Genomatix Bibliosphere software (Genomatix).

From extracted RNA, cDNA was obtained using murine moloney leukemia virus reverse transcriptase (Invitrogen) and oligo(dT) (12–18-mer; Sigma-Aldrich) as a primer. Quantitative PCR was performed on a LightCycler (Roche Diagnostics) as described previously (29). Data were analyzed using the “Fit Points” and “Standard Curve Method” using hprt as housekeeping gene to calculate the level of gene expression normalized for hprt expression. The following primer sets were used: Hprt, sense, 5′-GCA GTA CAG CCC CAA AAT GG-3′, and antisense, 5′-AAC AAA GTC TGG CCT GTA TCC AA-3′; Tnf, sense, 5′-TCT CAT CAG TTC TAT GGC CC-3′, and antisense, 5′-GGG AGT AGA CAA GGT ACA AC-3′; IL-12p40 (Il12b), sense, 5′-CTG GCC AGT ACA CCT GCC AC-3′, and antisense, 5′-GTG CTT CCA ACG CCA GTT CA-3′; Nos2, sense, 5′-AGC TCC TCC CAG GAC CAC AC-3′, and antisense, 5′-ACG CTG AGT ACC TCA TTG GC-3′; LRG-47 (Irgm1), sense, 5′-AGC CGC GAA GAC AAT AAC TG-3′, and antisense, 5′-CAT TTC CGA TAA GGC TTG G-3′; Arg1, sense, 5′-CAG AAG AAT GGA AGA GTC AG-3′, and antisense, 5′-CAG ATA TGC AGG GAG TCA CC-3′; macrophage mannose receptor (Mmr), sense 5′-CCA CAG CAT TGA GGA GTT TG-3′, and antisense, 5′-ACA GCT CAT CAT TTG GCT CA-3′; Ym1, sense 5′-GGG CAT ACC TTT ATC CTG AG-3′, and antisense, 5′-CCA CTG AAG TCA TCC ATG TC-3′; Fizz, sense, 5′-TCC CAG TGA ATA CTG ATG AGA-3′, and antisense, 5′-CCA CTC TGG ATC TCC CAA GA-3′; and Il4 receptor α (Il4ra), sense, 5′-TGT GAC CTA CAA GGA ACC CA-3′, and antisense, 5′-GCA AAA CAA CGG GAT GCA GA-3′.

For Ag-specific restimulation and flow cytometric analysis, single-cell suspensions of mediastinal lymph nodes and lungs were prepared from Mtb-infected mice at different time points. Lymph node cells were isolated by straining through a metal sieve. After depletion of erythrocytes, cells were resuspended in complete IMDM (Invitrogen) supplemented with 10% FCS (Life Technologies), 0.05 mM 2-ME (Sigma-Aldrich), and penicillin and streptomycin (100 U/ml and 100 μg/ml; Invitrogen), counted, and used for additional experiments. For preparation of single-cell suspensions from lungs, mice were anesthetized and injected i.p. with 150 U of heparin (Ratiopharm). Lungs were perfused through the right ventricle with warm PBS. Once lungs appeared white, they were removed and sectioned. Dissected lung tissue was then incubated in collagenase A (0.7 mg/ml; Roche Diagnostics) and DNase (30 μg/ml; Sigma-Aldrich) at 37°C for 2 h. Digested lung tissue was gently disrupted by subsequent passage through a 100-μm pore size nylon cell strainer. Recovered lung cells were counted, diluted in IMDM, and used for additional experiments.

For flow cytometric analysis of surface markers, cells were washed and incubated with a mixture containing anti-FcγRIII/II mAb (clone 2.4G2), mouse and rat serum to block nonspecific binding to FcR. Cells were then incubated in consecutive steps for 20 min with optimal concentrations of the following Abs, CD4-allophycocyanin, CD3-PerCP, CD44-FITC, CD62L-PE, IA^q-PE, CD124-PE (IL-4Rα) (all from BD Biosciences), F4/80-AlexaFluor647 (Caltag Laboratories), and CD206 (MMR) (Serotec). For detection of intracellular IFN-γ, an intracellular cytokine staining kit was used (BD Biosciences). Briefly, single-cell suspensions were prepared at 19 days after infection, and 2 × 10⁶ cells were stimulated with plate-bound anti-CD3/CD28 mAb (clone 2C11 and clone 37/51 at 10 μg/ml, respectively) for 4 h in the presence of GolgiPlug (BD Biosciences). Nonspecific Ab binding was blocked by incubation with a mixture containing anti-FcγRIII/II mAb (clone 2.4G2), mouse and rat serum. Cells were washed and incubated with optimal concentrations of anti-CD4-FITC (BD Biosciences). After staining, cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences), and intracellularly accumulated IFN-γ was stained with a PE-labeled anti-IFN-γ mAb (BD Biosciences).

Fluorescence intensity was analyzed on a FACSCalibur (BD Biosciences) gating on propidium iodide-negative macrophages or lymphocytes identified by forward light scatter–side scatter profile, respectively.

To detect IL-10 release in lymph nodes and lungs, 3 × 10⁶ cells from single-cell suspensions of mediastinal lymph nodes and perfused lungs were cultured for 24 h in complete RPMI 1640 (Life Technologies) supplemented with 10% FCS (Life Technologies), 0.05 mM 2-ME (Sigma-Aldrich), and penicillin and streptomycin (100 U/ml and 100 μg/ml; Life Technologies). Supernatants were collected and frozen at −80°C until production of IL-10 was quantified by ELISA. For measuring Ag-specific production of IFN-γ by CD4⁺ T cells in lymph nodes, single-cell suspension of mediastinal lymph nodes were prepared in IMDM from mice 19 days after infection with Mtb. To enrich CD4⁺ T cells, single-cell suspensions from lymph nodes were incubated with magnetic CD4 microbeads (Miltenyi Biotec) and separated from other cells on a MACS separation unit (Miltenyi Biotec). Separated CD4⁺ T cells were collected in IMDM, counted, and diluted. Purity of enriched CD4⁺ T cells was >95% as determined by flow cytometry. For Ag-specific restimulation, 4 × 10⁵ enriched CD4⁺ lymph node cells were incubated with 2 × 10⁵ peritoneal macrophages that were pulsed with 25 μg/ml short-term culture filtrate from Mtb (a gift from P. Andersen, Statens Serum Institute, Copenhagen, Denmark) in IMDM. Resident peritoneal macrophages were obtained 1 day before the experiment by peritoneal lavage of uninfected C57BL/6 mice and incubated overnight in 96-well flat-bottom microplates (Nunc) in complete IMDM. After 72 h of restimulation, supernatants were collected and frozen at −80°C until production of IFN-γ was quantified by ELISA.

To determine cytokine production after pulmonary infection of experimental animals, supernatants were analyzed in 3-fold serial dilutions using the sandwich ELISA system OptEia (BD Biosciences). After incubation with HRP coupled to avidin and developing with tetramethylbenzidine substrate reagent, the absorbance was read on a microplate reader (Sunrise; Tecan). Using a test wavelength of 450 nm and a reference wavelength of 630 nm, samples were compared with appropriate recombinant cytokine standards. The detection limits for all cytokines were 5 pg/ml.

To determine arginase activity in murine tissue, weighed pieces of organs were homogenized in 100 μl of 0.1% Triton X-100 (Sigma-Aldrich) containing a protease inhibitor mixture (Roche). Fifty microliters of 10 mM MnCl₂ (Merck) and 50 mM Tris-HCl (Merck) was added to all samples, and the enzyme was activated by heating for 10 min at 55°C. Arginine hydrolysis was conducted by incubating 25 μl of the activated lysate with 25 μl of 0.5 M l-arginine (Merck) at 37°C for 60 min. The reaction was stopped with 400 μl of H₂SO₄ (96%)/H₃PO₄ (85%)/H₂O (1/3/7, v/v/v). As a degree of arginase activity, the urea concentration was measured at 540 nm after addition of 25 μl of α-isonitrosopropiophenone (Sigma-Aldrich; dissolved in 100% ethanol) followed by heating at 95°C for 45 min. One unit of arginase activity is defined as the amount of enzyme that catalyzes the formation of 1 μmol urea/min.

To detect RNI in uninfected and infected mice, blood was collected at different time points. Using a commercial nitrate reductase kit (Cayman Chemical) NO₃ was converted into NO₂. After adding Griess reagents, the content of NO₂ was determined by photometric measurement reading the absorbance at 540 nm on a microplate reader (Sunrise; Tecan) as described previously (27).

To analyze the antibacterial activity of macrophages, BMMφ were cultured 48 h before infection with Mtb in DMEM supplemented with 10% FCS (Biochrom), 2 mM l-glutamine, 1 mM sodium pyruvate, and 10 mM HEPES (all from PAA) in the presence of medium or incubated with 10 ng/ml IL-4 (BD Biosciences) and 10 ng/ml IL-10 (BD Biosciences). One day before infection, cells were incubated with medium or stimulated with 100 U/ml IFN-γ (PeproTech). Cells were inoculated with Mtb H37Rv at a multiplicity of infection of 0.5 for 2 h. To determine bacterial uptake at different time points, macrophages were lysed by addition of 0.5% Triton X-100 (Serva). Lysates were serially diluted in 0.05% Tween 80 (Merck), plated on Middlebook 7H10 agar containing 10% OADC, and incubated at 37°C for 19–21 days. Colonies on plates were enumerated, and results were expressed as log₁₀ CFU per culture.

Quantifiable data are expressed as the means of individual determinations and SDs. Statistical analysis was performed using the unpaired Student’s t test defining different error probabilities (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001). ANOVA was performed using the Dunnett multiple comparison test, correcting for multiple testing, which defines different error probabilities between untreated cells and macrophages that have been incubated with IFN-γ (#, p ≤ 0.05; ##, p ≤ 0.01) or IL-4/IL-10/IFN-γ (*, p ≤ 0.05; **, p ≤ 0.01).

To determine the influence of increased levels of macrophage-derived IL-10 on the outcome of experimental TB, mice overexpressing IL-10 under control of the macrophage-specific CD68 promoter were infected with 100 CFU of Mtb via the aerosol route. Whereas FVB control mice were able to control mycobacterial growth in infected lungs at a plateau level of ∼6.9 log₁₀ CFU/lung, bacterial loads in IL-10 transgenic littermates steadily increased reaching significantly higher CFU counts 78 days postinfection (Fig. 1,a). Mtb-infected FVB wild-type mice survived for more than 250 days (Fig. 1 b). In contrast, macIL-10^tg animals died significantly earlier, with a mean survival time of ∼100 days.

Priming in lymphoid organs and migration of T lymphocytes to the site of infection is indispensable for the induction of defense mechanisms against Mtb. To evaluate whether T cell activation in macIL-10^tg mice was impaired in draining lymph nodes and lung tissue, activated CD4⁺ T cells were quantified by flow cytometry. After infection with Mtb, comparably increasing relative and absolute numbers of activated CD44^highCD62L^lowCD4⁺ T cells were observed in lymph nodes of both FVB and macIL-10^tg mice (Fig. 2, a and b). Ag-specific restimulation of CD4⁺ T cells purified from the draining lymph nodes revealed that T cells of macIL-10^tg mice were as capable of producing IFN-γ as wild-type littermates (Fig. 2,c). In addition, the relative proportion of activated CD44^highCD62L^lowCD4⁺ T cells in the lung was also not affected during the course of Mtb infection in macIL-10^tg mice (Fig. 2,d). However, the absolute amount of activated CD4⁺ T cells in lungs from macIL-10^tg mice was found to be enhanced at day 84 after Mtb infection (Fig. 2,e). After polyclonal restimulation, IFN-γ production by CD4⁺ T cells was comparable in Mtb-infected FVB and macIL-10^tg mice (Fig. 2 f). Taken together, macrophage-derived IL-10 did not dramatically affect the induction, migration, and function of T effector cells after Mtb infection. The increased infiltration of activated CD4⁺ T cells into lungs from macIL-10^tg mice may reflect an increased inflammatory reaction to the elevated bacterial load at late time points of Mtb infection.

IFN-γ-dependent activation of macrophages at the site of infection is key to the induction of the effector molecules NOS2 and LRG-47 (4, 5), and IL-10 has been previously described to affect antimicrobial effector mechanisms (30, 31). Since the expression of T cell-derived IFN-γ was not affected in Mtb-infected macIL-10^tg mice, downstream effector mechanisms could have been directly impaired by the macrophage-specific overexpression of IL-10. Quantitative real-time PCR revealed that during Mtb infection transcripts for Nos2 were increasingly induced in lungs from both FVB and macIL-10^tg mice to the same extent (Fig. 3,a). In fact, at day 90 postinfection, Nos2 gene expression was found to be even higher in transgenic animals. It should be noted that IL-10 has been shown to induce Nos2 expression in some settings, suggesting that Nos2 is not a canonical target of the IL-10 anti-inflammatory response (32). The kinetics of gene expression of Lrg-47 (Irgm1) was comparable in the lungs of both FVB and macIL-10^tg mice peaking on day 21 of Mtb infection (Fig. 3,b). Six weeks after infection, immunohistochemical detection of NOS2 in lung sections verified these findings on the protein level, showing no differences in NOS2 expression between FVB and macIL-10^tg mice (Fig. 3 c).

So far, the increased susceptibility of macIL-10^tg mice to experimental TB could neither be attributed to an impaired function of IFN-γ-producing T cells nor to defective IFN-γ-dependent downstream mechanisms in macrophages. IL-10 specifically suppresses the production of proinflammatory cytokines in activated macrophages and dendritic cells (9, 33). To evaluate the suppressive effect of macrophage-derived IL-10 on anti-TB effector cytokines during the course of infection with Mtb, we compared the expression of TNF (produced by T cells and macrophages/dendritic cells) and IL-12p40 (produced by macrophages/dendritic cells) in the lungs of FVB wild-type and IL-10^tg mice. Until day 90 after infection, increasing amounts of Tnf transcripts were found in the lungs of FVB mice (Fig. 4,a). In the lungs of macIL-10^tg mice, however, gene expression was found to be even higher on days 49 and 90 postinfection. In contrast, 49 and 90 days after Mtb infection, gene expression of IL-12p40 (Il12b) was found to be significantly reduced in lungs from macIL-10^tg mice when compared with FVB mice (Fig. 4 b).

After Mtb infection, IL-12 is essential for driving protective Th1 immune responses (34). In lymphoid organs, IL-10 inhibits IL-12-driven Th1 development, whereas it has no suppressive effect on already differentiated Th1 cells in infected tissue (35). Because Mtb-infected macIL-10^tg mice displayed a normal Th1 development, we speculated that the magnitude of transgenic IL-10 expression would be different in draining lymph nodes and lung tissue. In contrast to dendritic cells, only a few macrophages are present in lymph nodes in experimental TB (36). Accordingly, we found only low levels of IL-10 in lymph nodes from macIL-10^tg mice (Fig. 4,c). In contrast, a great amount of IL-10 was secreted in lung tissue from macIL-10^tg animals. In FVB mice, IL-10 was below the detection limit in either organ (Fig. 4,c). In line with the low expression of transgenic IL-10 in the lymph nodes of Mtb-infected macIL-10^tg mice, IL-12p40 expression was not affected. In contrast, elevated levels of transgenic IL-10 in lung tissue led to reduced expression of IL-12p40 (Fig. 4 d). Taken together, IL-10 secretion by macrophages does not seem to have an overt effect on Th1 differentiation and the expression of Th1 cell effector mechanisms. Macrophage-derived IL-10 rather appears to regulate distinct macrophage functions such as cytokine secretion at the site of infection.

To investigate how macrophage-specific overexpression of IL-10 shapes the host response to Mtb at the transcriptome level, genome-wide expression analysis was performed on tissue RNA isolated from the lungs of FVB and macIL-10^tg mice at baseline as well as 25 and 42 days after infection. Excluding variability for the infection itself, only 219 probe sets were different between FVB and macIL-10^tg, indicating a comparatively small effect of IL-10 on the transcriptional changes. The effect of IL-10 on the number of up- and down-regulated genes is summarized in Fig. 5 a. Before infection, only 36 probe sets were up-regulated in lungs from macIL-10^tg mice compared with FVB animals (data not shown). These included some known IL-10-induced genes (Msr1, Ccr2, Ccr5, Fcgr2b, and Saa3). Only four probe sets, all detecting Igg, gave significantly higher signals in lungs from FVB mice. After infection, at the early time point day 25, of all probe sets induced >3-fold in either lung RNA from wild-type or macIL-10^tg mice, only one probe set (C3ar1) had significantly increased expression in macIL-10^tg mice, whereas Igh-VJ588 was higher in FVB control animals. Of the 970 probe sets induced by Mtb 42 days after infection, 158 showed higher expression in macIL-10^tg, whereas 68 were significantly higher in lungs from wild-type mice. Therefore, the majority of IL-10 effects was apparent at the later stages of infection.

Hierarchical clustering of the probe sets regulated by IL-10 (Fig. 5 b; data are listed in supplemental Table I)⁶ revealed groups of genes overexpressed in the lungs of macIL-10^tg mice already under basal conditions (cluster A) or postinfection (cluster B). These clusters contained known target genes of IL-10 in macrophages like Socs3, Bcl3, Saa3, and Il1rn, consistent with previous observations (32). However, many chemokines (e.g., Ccl2 and Ccl7) and cytokines (Il1b and Il6) were also expressed at higher levels in the lungs of macIL-10^tg mice compared with wild-type animals, indicating that the inhibitory effect of IL-10 on proinflammatory gene expression is likely affected by the increased bacterial load in macIL-10^tg mice.

The presence of various transcripts encoding cell type-specific surface proteins, such as the dendritic cell marker Cd207 in cluster C and the monocyte/macrophage marker Msr1 in cluster B, suggested an impact of IL-10 on the infection-induced changes in the cellular composition of the lung tissue. We used a literature-based data mining tool, Genomatix Bibliosphere, to search for overrepresentation of anatomical terms in lists of genes up-regulated in macIL-10^tg or wild-type mice after Mtb infection (Fig. 6 a). Although transcriptional evidence for NK cells, dendritic cells, and T lymphocytes was similar in both genotypes, in the lungs of macIL-10^tg mice, a strong enrichment for probe sets associated with macrophages was evident. In contrast, probe sets up-regulated to higher levels in FVB lungs at day 42 were connected to B lymphocytes. Immunohistological analysis also revealed a moderately reduced infiltration of B cells in lungs from macIL-10^tg mice (supplemental Fig. 1).⁶ Taken together, IL-10 appears to regulate leukocyte subset migration to the lung, or local proliferation of the cells, during Mtb infection.

Among the group of genes induced to high levels in infected macIL-10^tg mice was Arg1 (Fig. 6,b), encoding the enzyme arginase-1, whose extrahepatic expression is typically associated with IL-4/IL-13-induced alternative macrophage activation, indicating that after Mtb infection IL-10 may promote the development of alternatively activated macrophages. Although Il4 and Il13 mRNA was not detectably expressed in the lung data set, small amounts of IL-4 or IL-13 could promote alternative activation of lung macrophages in macIL-10^tg mice. IL-10 may enhance the sensitivity to IL-4 because it increases Il4ra expression on macrophages in vitro (32), an effect that was also observed in the lung data set after Mtb infection (Fig. 6,b). In addition to arginase-1, many other markers have been described to define IL-4Rα-mediated alternatively activated macrophages, especially during pulmonary inflammation such as Fizz1, various chitinases, and MMR (29). Among these genes, only Fizz1 (encoded by Retlna) and the chitinase Chi3l1 were clearly expressed at higher levels in macIL-10^tg lungs (Fig. 6 b), whereas the chitinases Ym1 (= Chi3l3) and Ym2 (= Chi3l4) as well as the Mmr (encoded by Mrc1) were down-regulated after infection to similar levels in both genotypes. The differential induction of markers for alternatively activated macrophages in Mtb-infected macIL-10^tg mice reflects also some opposite biological effects of IL-10 to IL-4 and IL-13, e.g., the regulation of MMR expression and function (37). In contrast to this rather inhibitory effect, IL-10 has been clearly shown to contribute to alternative macrophage activation during TLR ligation by a strong synergistic induction of arginase-1 (32). Taken together, our genome-wide expression analysis revealed that after infection with Mtb, macrophage-derived IL-10 induces several features of alternatively activated macrophages.

To verify our findings revealed by genome-wide expression analysis, we next performed quantitative RT-PCR analysis to test the model that macrophage-derived IL-10 promotes alternative macrophage activation after Mtb infection (Fig. 7,a). Six weeks after infection, we found significantly elevated gene expression of Fizz1 and Arg1. In contrast, transcripts for Mmr and the chitinase Ym1 were not elevated. In whole tissue, Il4ra gene expression in macIL-10^tg mice was elevated but not significantly different from gene expression in lungs from FVB wild-type mice. To determine the activation state of macrophages in Mtb-infected macIL-10^tg mice, we analyzed macrophages from perfused lungs by flow cytometry. Six weeks after infection, IA^q, CD80, CD86, and the MMR were expressed to the same extent on propidium iodide-negative F4/80-positive cells from FVB and macIL-10^tg mice (Fig. 7 b). Whereas only a subpopulation of macrophages from infected FVB mice expressed the IL-4Rα, all macrophages from macIL-10^tg mice were found to be positive for this receptor subunit. Taken together, we found that after Mtb infection IL-10 secretion by macrophages induces an alternatively activated program characterized by an enhanced expression of Fizz1, Arg1, and the Il4ra.

Classical macrophage activation plays an important role in combating infection with Mtb through IFN-γ-induced expression of effector molecules such as NOS2-dependent RNI (4). This induction may be counterregulated in alternatively activated macrophages. Because in alternatively activated macrophages arginase-1 hydrolyzes l-arginine to urea and l-ornithine, this pathway is also discussed to regulate RNI production in macrophages through depletion of l-arginine as the substrate for NOS2 (39, 40). Therefore, we speculated that in macIL-10^tg mice an increased arginase-1 activity modulates the production of RNI, thereby affecting effector mechanisms against Mtb usually exerted by classically activated macrophages. A kinetic quantification of gene expression revealed that Arg1 is not induced in lungs from FVB mice during the course of Mtb infection (Fig. 8,a). In contrast, Arg1 expression in macIL-10^tg mice was found to be significantly induced during the late phase of Mtb infection. Accordingly, arginase activity in lungs from macIL-10^tg mice was found to be elevated already 49 days postinfection (Fig. 8,b). In contrast, arginase activity was detectable in the lungs of FVB mice but did not increase after infection (Fig. 8,b). Concomitant with the increased arginase-1 activity in infected macIL-10^tg mice, RNI were only moderately produced in these animals (Fig. 8,c). In distinction, significantly increasing amounts of RNI were found in sera from Mtb-infected FVB mice (Fig. 8,c). Hence, the increased susceptibility in macIL-10t^g mice after infection with Mtb appeared to be caused by an elevated induction of arginase-1 in alternatively activated macrophages, which in turn led to an inefficient production of RNI. To examine whether IL-10-induced arginase-1 expression in macrophages diminishes antimycobacterial effector mechanisms in macrophages, we performed a Mtb growth inhibition assay in arginase-1-deficient BMMφ (Fig. 8 d). Incubation of BMMφ from Tie2^cre-neg Arg-1^flox/flox arginase-1-competent and Tie2^cre-pos Arg-1^flox/flox arginase-1-deficient mice lead to a significant reduction of mycobacterial growth 48 and 72 h after infection with Mtb when compared with infected macrophages that have not been activated with IFN-γ. Preincubation with IL-4 and IL-10, however, diminished IFN-γ-dependent mycobacterial growth inhibition in arginase-1-competent BMMφ. In contrast, IL-4/IL-10 treatment of IFN-γ-activated arginase-1-deficient macrophages did not lead to reduced antimycobacterial effector mechanisms, indicating that induction of arginase-1 in macrophages reduces the capability of controlling Mtb growth.

To test our hypothesis that macrophage-derived IL-10 mostly subverts NOS2-dependent effector mechanisms, we infected macIL-10^tg mice aerogenically with M. avium, a pathogen that is insensitive to NOS2-derived RNI production (6, 41), and determined bacterial loads in different organs 105 days after infection (Fig. 9). Bacterial loads in macIL-10^tg mice were found to be only moderately increased in lung and liver and comparable to FVB mice in spleens. However, infection of NOS2^−/− mice with M. avium revealed that bacterial loads rather decrease in the absence of NOS2 (40), corroborating the contention that after infection with RNI-insensitive mycobacteria the suppressive effect of NO on T cells appears (6). Therefore, the inability of macIL-10^tg mice to efficiently control bacterial growth after infection with RNI-sensitive and, to some extent, also -insensitive mycobacteria indicates that autocrine IL-10 operates partially by reducing RNI production via arginase-1 induction.

TB is a systemic disease that becomes manifest most prominently in the lung (24). Progressive primary TB occurs when the initial inflammatory focus is unable to contain mycobacterial replication, and the disease spreads to other parts of the lung. Postprimary disease develops when a lesion containing a previously “dormant”, i.e., growth-arrested organism resumes growth after a period of clinical latency. This organism may have been implanted years before during a primary infection that was successfully contained by a protective Th1-type immune response. Reactivation is often a consequence of a perturbed immune status of the host, although the cause of deregulation may be multifactorial.

Elevated levels of the Th2 cytokine IL-10 were detected in individuals with active TB (18, 19). Importantly, macrophages rather than T cells were reported to be the major source of IL-10 in these patients (18). However, whether and how macrophage-derived IL-10 promotes exacerbation of primary infection or reactivation of chronic latent infection is not clear. Our study revealed that in macIL-10^tg mice an increased production of IL-10 by macrophages resulted in enhanced Mtb replication, particularly during the chronic phase of infection, and thus bears some resemblance to reactivation TB in humans.

Humans infected with Mtb typically have a strong DTH response, as measured by the purified protein derivative skin test (24). This holds also true for patients suffering from postprimary TB. Thus, despite an apparently adequate initial differentiation of cell-mediated immune responses, Mtb is capable of persisting and can even resume replication within macrophages. In the draining lymph nodes of Mtb-infected macIL-10^tg mice, T cells efficiently differentiated into Th1 cells and migrated to the site of infection. Despite the presence of macrophage-derived IL-10 in the lungs of infected transgenic mice, Th1 cells efficiently expressed effector functions and secreted sufficient IFN-γ to induce a classical macrophage activation characterized by expression of NOS2 and LRG-47. However, bacteria survived and successfully proliferated within IFN-γ-activated IL-10^tg macrophages. It appears, therefore, that during experimental TB macrophage-derived IL-10 does not affect initiation and maintenance of Th1 immune responses but specifically acts on macrophage functions in a partially dominant manner over IFN-γ at the functional level.

It is important to note that overexpression of IL-10 specifically in T cells resulted in completely different immunosuppressive mechanisms after Mtb infection (22). tIL-10^tg mice failed to efficiently prime naive CD4⁺ T cells in draining lymph nodes, resulting in decreased numbers of activated T cells in the blood circulation and lung tissue; this defective Th1 response eventually led to an increased bacterial burden, due to a failure of classical macrophage activation. In contrast, IL-10 expression in macrophages at the site of infection induces an alternative activation program specifically in macrophages that counteracts concomitant classical macrophage activation. Since T cell-mediated immune responses remain largely unaffected (as is the case in human postprimary TB), the outcome of experimental TB in macIL-10^tg mice may therefore better reflect the situation in patients with reactivating TB.

We could link the increased susceptibility of Mtb-infected macIL-10^tg mice neither to an impaired function of IFN-γ-producing T cells nor to defective IFN-γ-dependent downstream mechanisms in macrophages. Transcriptome analysis revealed a strong enrichment for probe sets associated with macrophage function in the lungs of Mtb-infected macIL-10^tg mice. IL-10 has previously been shown to act primarily on activated macrophages, where IL-10R expression is highest. The IL-10R activates STAT3, and loss of this transcription factor in macrophages mimics loss of IL-10 itself (42). Hence, because STAT3 in myeloid cells is both necessary and sufficient for the effects of IL-10, macrophages are major targets of IL-10 in vivo (42, 43, 44, 45). On a functional level, IL-10 has been shown to negatively regulate macrophage function in a variety of experimental systems (17, 30, 31, 46). In vitro, IL-10 antagonizes the IFN-γ-dependent induction of NOS2 (30, 31). Our finding that macIL-10^tg mice fail to efficiently control bacterial growth especially after infection with RNI-sensitive mycobacteria corroborates our contention that the mode of action of autocrine IL-10 proceeds mainly via arginase-1-mediated reduction of RNI production in macrophages. However, we found NOS2 normally expressed in Mtb-infected macIL-10^tg mice, whereas the production of RNI was reduced during the course of infection. Hence, other pathways must be affected in macrophages of macIl-10^tg mice that lead to an impaired expression of effector molecules such as RNI.

Among the group of genes induced to high levels in Mtb-infected macIL-10^tg mice, some were associated with alternative macrophage activation such as Il4ra, Fizz1, and Arg1 (29). Alternative macrophage activation is typically induced by IL-4Rα ligation. Although Il4 and Il13 mRNA was not detectably expressed in the lung data set, small amounts of IL-4 may still drive alternative activation of lung macrophages in macIL-10^tg mice. IL-10 may enhance the sensitivity to IL-4 because it increases IL-4 receptor expression on macrophages in vitro (32), an effect that was also observed in the lung data set after Mtb infection. The differential induction of markers for alternatively activated macrophages in Mtb-infected macIL-10^tg mice reflects also some opposite biological effects of IL-10 to IL-4 and IL-13, e.g., the regulation of MMR expression and function (37). In contrast to this rather inhibitory effect, IL-10 has been clearly shown to contribute to alternative macrophage activation during TLR ligation by a strong synergistic induction of arginase-1 (32). Hence, during experimental TB, Arg1 expression may be directly induced by IL-10 or indirectly by IL-10-dependent up-regulation of the high-affinity IL-4Rα. Alternative activation of macrophages via the IL-4Rα has been shown to differentially regulate protective and immunopathological immune responses (29). Importantly, analysis of macrophage-specific IL-4Rα-deficient mice clearly demonstrated immunomodulation by macrophages to be a central regulatory mechanism during inflammatory immune responses in vivo (47). Because alternative macrophage activation reduces RNI production (40, 48, 49) and proinflammatory cytokine release (50), protective (39, 51, 52) and pathological (50) immune responses are significantly regulated by these types of macrophages.

In the context of TB and alternatively activated macrophages, one genome-wide expression analysis in IL-4-stimulated murine bone marrow-derived macrophages infected with Mtb has been published so far (51). A specific reanalysis of our lung transcriptome data for differences in the expression of the 15 most highly increased genes in Mtb-infected IL-4-stimulated macrophages revealed that 5 genes were expressed at significantly higher levels in macIL-10^tg lungs than in FVB mice after infection (Fcrg2b, Fn1, Cish, Ccl9, and Mrc1), and Cdh1 was increased under basal conditions in macIL-10^tg lungs (data not shown). Mmp12 (encoding matrix metalloproteinase-12, MMP-12) and Tfr (encoding the transferrin receptor, TfR) were found to be significantly up-regulated in Mtb-infected alternatively activated macrophages in vitro (51). The induction of the TfR appears to be of major importance in the context of experimental TB because intracellular Mtb exploit the transferrin-transferring receptor complex for iron utilization (53). However, neither our genome-wide expression analysis nor quantitative RT-PCR revealed a differential regulation of Tfr in Mtb-infected macIL-10^tg mice in vivo (data not shown). Additionally, Kahnert et al. (51) suggested MMP-12 induced in alternatively activated macrophages to be involved in the pathogenesis of TB. Although our genome-wide expression analysis in Mtb-infected macIL-10^tg mice revealed an increased expression of Mmp12 in macIL-10^tg mice at later time points of infection, the difference was not significant (data not shown). Moreover, quantitative RT-PCR of Mmp12 did not confirm a difference between wild-type and macIL-10^tg mice (data not shown). Taken together, genome-wide expression analysis in Mtb-infected and IL-4-induced alternatively activated macrophages in vitro and in Mtb-infected macIL-10^tg mice in vivo revealed an overlapping set of regulated genes, consistent with our interpretation that IL-10 supports a macrophage phenotype in vivo that is phenotypically related to alternatively activated macrophages.

Until recently, little was known about the impact of alternative macrophage activation and arginase-1 on the outcome of Mtb infection. Classical macrophage activation plays a central role in combating infection with Mtb through IFN-γ-induced expression of effector molecules such as the NOS2-dependent production of RNI (4, 54). Importantly, inhibition of endogenous NOS2 reactivates latent infection in Mtb-infected mice, indicating that NOS2-dependent mechanisms in macrophages are crucial for the control of Mtb persisting in immunocompetent hosts (55). The production of RNI might be counterregulated by IL-4Rα- and/or IL-10-dependent mechanisms, leading to arginase-1 expressing alternatively activated macrophages. In fact, RNI production in Mtb-infected alternatively activated macrophages was found to be reduced (51). In line with these findings, we here show the induction of arginase-1 in macrophages subverts IFN-γ-induced antimycobacterial effector mechanisms in macrophages. Since NOS2 shares l-arginine as a substrate with arginase-1, substrate depletion by either enzyme is a key regulatory mechanism in macrophages, and differential expression of NOS2 and arginase-1 is important for regulating macrophage effector functions (40, 49). Very recently, El Kasmi et al. (26) showed that specific deletion of arginase-1 in macrophages reduced lung bacterial loads during Mtb infection, supporting our contention that arginase-1 induction in macrophages undermines NOS2-dependent effector mechanisms against Mtb by depletion of the common substrate l-arginine. In addition, arginase-dependent production of polyamines was shown to promote elevated growth of L. major in macrophages (52) and may also be beneficial for the proliferation of Mtb. Importantly, in human pulmonary TB, high production of IL-10 also parallels arginase activity in PBMC (56). Hence, elevated IL-10-dependent arginase activity during the chronic phase of infection may promote recrudescence of Mtb growth. We do not exclude that macrophage-derived IL-10 suppresses other antimycobacterial effector mechanisms in addition to regulating RNI synthesis. In contrast to infection with Mtb, macIL-10^tg mice were indeed able to control bacterial growth after infection with M. avium, which has been shown to be insensitive to the effects of RNI (6, 41). However, because bacterial loads have been described to be further reduced after infection with M. avium in the complete absence of RNI in NOS2^−/− mice, it appears that although autocrine IL-10 subverts RNI synthesis in macrophages other suppressive mechanisms may also be induced by IL-10.

We conclude that during Mtb infection macrophage-derived IL-10 drives reactivation during the chronic stage of the disease by acting primarily at the level of the macrophage. Macrophage-derived IL-10 appears to override IFN-γ-dependent classical macrophage activation and effector mechanisms against Mtb by inducing an alternatively activated phenotype. Specifically, Arg1 expression in these macrophages may provide better growth conditions for Mtb. Th2 cytokines that potently drive alternative macrophage activation have been associated with the development of postprimary TB in patients (7, 57). Thus, reactivating TB may, in part, be mediated by a Th2-type immune reaction characterized by an alternative activation state of macrophages that eventually subverts protective immune mechanisms. It is possible that the identification of an increased type 2 cytokine immune response may identify infected individuals at risk to reactivate a latent Mtb infection. Additionally, prophylactic treatment of these individuals with immunomodulating drugs aimed at inhibiting the production of type 2 cytokines or downstream immune mechanisms may prevent or ameliorate reactivation TB.

We thank Alexandra Hölscher, Tanja Sonntag, Susanne Metken, and Manfred Richter for excellent technical assistance; Ilka Monath, Sven Mohr, and Claus Möller for organizing the animal facility and taking care of the mice at the Research Center Borstel; Angela Servatius for management of the macIL-10^tg colony and processing the microarray samples at the Institute of Medical Microbiology, Immunology and Hygiene, Technical University Munich; members of P. Murray’s laboratory for bone marrow isolation; and Bianca Schneider for help with the Mtb growth inhibition assay.

The authors have no financial conflict of interest.