Toll-Like Receptor 9 Regulates the Lung Macrophage Phenotype and Host Immunity in Murine Pneumonia Caused by Legionella pneumophila

Abstract

Experiments were performed to determine the contribution of TLR9 to the generation of protective immunity against the intracellular respiratory bacterial pathogen Legionella pneumophila. In initial studies, we found that the intratracheal (i.t.) administration of L. pneumophila to mice deficient in TLR9 (TLR9^−/−) resulted in significantly increased mortality, which was associated with an approximately 10-fold increase in the number of lung CFU compared to that of wild-type BALB/c mice. Intrapulmonary bacterial challenge in TLR9^−/− mice resulted in the reduced accumulation of myeloid dendritic cells (DC) and activated CD4⁺ T cells. Lung macrophages isolated from Legionella-infected TLR9^−/− mice displayed the impaired internalization of bacteria and evidence of alternative rather than classical activation, as manifested by the markedly reduced expression of nitric oxide and type 1 cytokines, whereas the expression of Fizz-1 and arginase-1 was enhanced. The adoptive transfer of bone marrow-derived DC from syngeneic wild-type, but not TLR9^−/−, mice administered i.t. reconstituted anti-legionella immunity and restored the macrophage phenotype in TLR9^−/− mice. Finally, the i.t., but not intraperitoneal, administration of the TLR9 agonist molecule CpG oligodeoxynucleotide stimulated protective immunity in Legionella-infected mice. In total, our findings indicate that TLR9 is required for effective innate immune responses against the intracellular bacterial pathogen L. pneumophila, and approaches to maximize TLR9-mediated responses may serve as a means to augment antibacterial immunity in pneumonia.

Legionella pneumophila is a gram-negative bacterial pathogen that often causes a life-threatening pneumonia, especially in immunocompromised individuals. This organism is a facultative intracellular bacillus that usually infects humans via the inhalation of contaminated aerosols from waterborne environmental sources. In the lungs, bacteria multiply intracellularly within monocytes and resident lung macrophages (8, 20). Respiratory tract infection due to L. pneumophila also results in substantial alveolar epithelial injury, resulting in high-permeability edema, respiratory failure, and mortality rates reaching as high as 50% (41, 44).

Protective lung immunity against L. pneumophila requires the ingestion of bacteria by alveolar macrophages and the generation of type 1 cytokines required for the sufficient killing of intracellular organisms (8, 10, 12, 42, 43). Immune responses in the lung are initiated by pattern recognition receptors that recognize pathogen-associated molecular patterns. Toll-like receptors (TLRs) serve as pattern recognition receptors for mammals to recognize and respond to various pathogen-associated molecular patterns (1, 2). For example, bacterial lipopolysaccharides (LPS) generally are recognized by TLR4, whereas peptidoglycans, lipoproteins, and lipopeptides from various microbial pathogens activate TLR2. However, previously it has been shown that LPS isolated from L. pneumophila and viable Legionella organisms function to activate TLR2 in dendritic cells (DC) and macrophages, but not TLR4 (4, 14, 29). Moreover, mice deficient in TLR2, but not TLR4, display the impaired clearance of L. pneumophila after intrapulmonary challenge. Importantly, defects in innate and acquired immune responses in MyD88-deficient mice challenged with L. pneumophila are more profound than those observed in mice deficient in TLR2 (4, 39), suggesting that other MyD88-dependent TLRs are required for optimal lung immunity in Legionella pneumonia.

A candidate TLR well positioned to participate in anti-Legionella immunity is TLR9. TLR9 is a toll receptor that is localized intracellularly within endocytic vesicles and is activated by unmethylated CpG motifs that are present at high frequencies in DNA from various microbes, including bacteria, viruses, and certain fungi (24, 28, 45). The activation of TLR9 by synthetic CpG oligodeoxynucleotides (ODN), microbial DNA, or intact organisms occurs intracellularly within endosomes, resulting in the expression of type 1-associated cytokines and chemokines as well as the up-regulation of costimulatory and major histocompatibility complex (MHC) molecules on the cell surface of professional antigen-presenting cells (21, 32). In mice, TLR9 primarily is expressed on DC (plasmacytoid and myeloid), B cells, and, to a lesser extent, macrophages (23, 30, 36, 37). Alveolar macrophages do not express appreciable quantities of TLR9 in the resting state (40). We and others have recently shown that TLR9 is required for the generation of innate lung immunity against extracellular gram-negative and gram-positive bacterial organisms (3, 6). In murine pneumonia due to Klebsiella pneumoniae, TLR9-mediated DC responses appear to be critical to the effective clearance of bacteria from the lung (6). The role of TLR9 in lung infections due to intracellular bacterial organisms has not been investigated. Importantly, TLR9 has been shown to mediate the production of interleukin-12 (IL-12) in DC infected with Legionella in vitro, suggesting a possible role for TLR9 during pulmonary L. pneumophila infection in vivo (34).

In the current study, we examined the role of TLR9 in host defense in a murine model of Legionella pneumonia. Mice deficient in TLR9 displayed an enhanced susceptibility to pulmonary L. pneumophila infection, as manifested by increased mortality and reduced lung bacterial clearance compared to that of infection in wild-type (WT) mice. Impaired lung bacterial clearance in TLR9^−/− mice was associated with the alternative, rather than classical, activation of lung macrophages, and marked impairment in macrophage NO production. Interestingly, the adoptive transfer of WT but not TLR9^−/− bone marrow-derived DC intratracheally (i.t.) reversed the impairment in lung bacterial clearance and the alternatively activated lung macrophage phenotype in infected TLR9^−/− mice. Finally, the i.t., but not intraperitoneal (i.p.), administration of the synthetic TLR9 agonist CpG ODN stimulated protective immunity in Legionella-infected A/J mice. Collectively, our results highlight the importance of TLR9 in host immunity against intracellular bacterial pathogens of the respiratory tract.

MATERIALS AND METHODS

Mice.

Female BALB/c (National Cancer Institute-Harlan, Frederick, MD) or A/J mice (Jackson Laboratories) were used at 8 to 12 weeks of age. Breeding pairs of TLR9^−/− mice generated by S. Akira were obtained from Coley Pharmaceutical (Wellesley, MA) and a colony established at the University of Michigan. These mice were generated on a BALB/c background (with more than five backcrosses), are phenotypically normal in the uninfected state, and reproduce without difficulty.

L. pneumophila inoculation.

We used a clinical isolate of L. pneumophila strain Suzuki (serogroup 1) for animal experiments; this isolate was obtained from the sputum of a patient with L. pneumophila pneumonia (41, 43). In preliminary experiments, we observed that the i.t. inoculation of this isolate (10⁵ to 10⁶ CFU per mouse) consistently induced pneumonia in BALB/c or A/J mice, which was characterized by increases in bacterial numbers, cytokine responses, and pathological changes in the lungs (data not shown). N-(2-acetamido)-2-aminoethanesulfonic acid (ACES; Sigma)-buffered yeast extract broth supplemented with l-cysteine (0.4 mg/ml) and ferric nitrate (0.135 mg/ml) was used as a liquid medium (BYE broth). To prepare solid medium, activated charcoal (2 mg/ml) and agar (15 mg/ml) were added to liquid medium (BCYE agar). Bacteria were incubated on BCYE agar for 4 days at 37°C. A single colony was transferred to 3 ml of BYE broth and then incubated overnight at 37°C with constant shaking. The bacterial suspension again was transferred to fresh BYE broth and incubated overnight under the same conditions. After the confirmation of bacterial motility by microscopic observation, the concentration of bacteria in the broth was determined by measuring the absorbance at 600 nm. Post-exponential-phase bacteria (optical density at 600 nm, 1.7 to 1.8; motility, >30%) were used as challenge organisms, because the expression of virulence in L. pneumophila is dependent on the growth phase. According to a standard of absorbancies based on known numbers of CFU, the bacterial suspension was diluted to the desired concentration in saline. Each animal was anesthetized i.p. with approximately 1.8 to 2 mg of pentobarbital. The trachea was exposed, and 30 μl of inoculum or saline was administered via a sterile 26-gauge needle. The skin incision was closed with surgical staples.

To label bacteria with fluorescein isothiocyanate (FITC), L. pneumophila was incubated with FITC (1 mg/ml) for 60 min at 37°C and then administered i.t. as described above.

CpG ODN.

Active and control CpG ODN were synthesized on a phosphodiester backbone by Oligos Etc. Inc. (Wilsonville, OR). The active CpG ODN contained two CpG motifs (underlined) and had the sequence 5′-TCCATGACGTTCCTGACGTT-3, whereas the C and G of the first CpG motif were reversed in the control peptide (underlined) and had the sequence 5′-TCCATGAGCTTCCTGAGTCT-3. Control or CpG ODN (30 μg) were reconstituted in 30 μl of sterile water or saline. Control groups in the experiments described received either vehicle alone or control ODN. All ODN were free of endotoxin contamination.

Lung harvesting for bacterial number and cytokine analyses.

At designated times, the mice were sacrificed by CO₂ asphyxia. Prior to lung removal, the pulmonary vasculature was perfused with 1 ml of phosphate-buffered saline (PBS) containing 5 mM EDTA via the right ventricle. Whole lungs then were harvested for the assessment of the number of bacterial CFU and cytokine protein expression. After removal, whole lungs were homogenized in 1.0 ml of PBS with protease inhibitor (Boehringer Mannheim Biochemicals, Indianapolis, IN) using a tissue homogenizer (Biospec Products, Inc.) under a vented hood. Portions of homogenates (10 μl) were inoculated on BCYE agar after serial 1:10 dilutions with PBS. The homogenates were incubated on ice for 30 min and then centrifuged at 1,100 × g for 10 min. Supernatants were collected, passed through a 0.45-μm-pore-size filter (Gelman Sciences, Ann Arbor, MI), and stored at −20°C for the assessment of cytokine levels. Lung macrophages (consisting of both alveolar and interstitial macrophages) were isolated by adherence purification and were plated at a concentration of 1 × 10⁶ to 2 × 10⁶ cells/ml.

BAL.

Bronchoalveolar lavage (BAL) was performed for the assessment of the in vivo internalization of FITC-labeled L. pneumophila by alveolar macrophages. At 48 h after the i.t. administration of FITC-labeled bacteria, the trachea was exposed and intubated using a 1.7-mm-outer-diameter polyethylene catheter. BAL was performed by instilling PBS containing 5 mM EDTA in 1-ml aliquots for a total of 3 ml. Lavaged cells were counted, and cytospins were stained with a modified Wright stain to determine the internalization of FITC-labeled bacteria and to identify alveolar macrophages by morphology.

Total lung leukocyte preparation.

Lungs were removed from euthanized animals, and leukocytes were prepared as previously described (6, 11). Briefly, lungs were minced with scissors to a fine slurry in 15 ml of digestion buffer (RPMI medium-10% fetal calf serum-1 mg/ml collagenase [Boehringer Mannheim Biochemical]-30 μg/ml DNase [Sigma]) per lung. Lung slurries were enzymatically digested for 30 min at 37°C. Any undigested fragments were further dispersed by drawing the solution up and down through the bore of a 10-ml syringe. The total lung cell suspension was pelleted, resuspended, and spun through a 40% Percoll gradient to enrich for leukocytes. Cell counts and viability were determined using trypan blue exclusion counting on a hemacytometer. Cytospin slides were prepared and stained with a modified Wright-Giemsa stain.

Multiparameter flow cytometric analyses.

Cells were isolated from lung digests as described above (6, 11). For analyses of T-cell subsets, isolated leukocytes were stained with the following fluorescein isothiocyanate- or phycoerythrin-labeled antibodies: anti-αβ-T-cell receptor (TCR), anti-γδ-TCR, anti-DX5, anti-CD11c, anti-MHC class II, anti-Gr1, and anti-CD69 (unless otherwise noted, all reagents were from PharMingen, San Diego, CA). In addition, cells were stained with anti-CD45-tricolor (Caltag Laboratories, San Francisco, CA), allowing for the discrimination of leukocytes from nonleukocytes and thus eliminating any nonspecific binding of T-cell surface markers on nonleukocytes. T- and NK-cell subsets were analyzed by first being gated on CD45-positive lymphocyte-sized leukocytes and then being examined for FL1 and FL2 fluorescence expression using four-color flow cytometry. Cells were collected on a FACScan or FACScalibur cytometer (Becton Dickinson, San Jose, CA) by using CellQuest software (Becton Dickinson). Analyses of data were performed using the CellQuest software package.

Isolation and culture of bone marrow-derived DC.

Bone marrow was harvested from the long bones of mice by using a previously described technique (6, 38). Recovered marrow cells were seeded in tissue culture flasks in RPMI 1640-based complete medium with murine granulocyte-macrophage colony-stimulating factor (GM-CSF) (10 ng/ml). Medium and cytokines were replaced after 3 days, loosely adherent cells collected after 6 to 7 days, and cells positively selected for CD11c by magnetic bead separation. CD11c⁺ DC were plated overnight and resuspended in fresh medium the following day. The flow cytometry of cells verified >90% purity for DC.

Expression of iNOS and NO by lung macrophages.

To assess spontaneous inducible NO synthase (iNOS) expression in lung macrophages, cells were isolated from lung digest cells by Percoll gradient enrichment and adherence purification at a concentration of 1 × 10⁶ to 2 × 10⁶ cells/well. Cells were washed three times, and then RNA was immediately isolated. Spontaneous NO production was determined using a standard Griess reaction after 16 h in culture.

Murine cytokine ELISAs.

Murine IL-13, TNF-α, IL-12, gamma interferon (IFN-γ), IFN-inducible protein 10 (IP-10), CCL2/monocyte chemoattractant protein (MCP-1), and Fizz-1 were quantitated using a modification of a double-ligand method as previously described (6, 11). The enzyme-linked immunosorbent assay (ELISA) method used consistently detected murine cytokine concentrations of greater than 20 to 50 pg/ml. The ELISAs did not cross-react with other cytokines tested.

Real-time quantitative reverse transcription-PCR.

The measurement of gene expression was performed with the ABI Prism 7000 sequence detection system (Applied Biosystems, Foster City, CA) as previously described (6). Primers and probe nucleotide sequences were the following: miNOS forward, 5′-CCC TCC TGA TCT TGT GTT GGA-3′; reverse, 5′-CAA CCC GAG CTC CTG GAA-3′; and probe, 5′-TGA CCA TGG AGC ATC CCA AGT ACG AGT-3′; mβ-actin forward, 5′-CCG-TGA-AAA-GAT-GAC-CCA-GAT-C-3′; reverse, 5′-CAC-AGC-CTG-GAT-GGC-TAC-GT-3′; and probe, 5′-TTT-GAG-ACC-TTC-AAC-ACC-CCA-GCC-A-3′; arginase-1 forward, 5′-CAG TCT GGC AGT TGG AAG CA-3′; reverse, 5′-GCA TCC ACC CAA ATG ACA CAT-3′; and probe, 5′-/56-carboxyfluorescein/CTG GCC ACG CCA GGG TCC AV/36-TAMSp/-3′; Fizz-1 forward, 5′-CCC TGC TGG GAT GAC TGC TA-3′; reverse, 5′-TCC ACT CTG GAT CTC CCA AGA-3′; and probe, 5′-/56-carboxyfluorescein/TGG GTG TGC TTG TGG CTT TGC C/3-black hole quencher-1/-3′. Specific thermal cycling parameters used with the TaqMan one-step reverse transcription PCR master mix reagents kit included 30 min at 48°C, 10 min at 95°C, and 40 cycles of denaturation at 95°C for 15 s and annealing/extension at 60°C for 1 min. The relative quantitation of cytokine mRNA levels was plotted as the level of change (n-fold) compared to the cytokine mRNA levels in untreated control lung. All experiments were performed in duplicate.

Statistical analysis.

Statistical significance was determined using the unpaired two-tailed alternate Welsh t test and the nonparametric Mann-Whitney test. Calculations were performed using InStat for Macintosh (GraphPad Software, San Diego, CA).

RESULTS

Increased mortality and impaired lung bacterial clearance in TLR9^−/− mice after i.t. L. pneumophila administration.

To determine if mice deficient in TLR9 displayed altered susceptibility to intrapulmonary challenge with L. pneumophila, age- and sex-matched WT and TLR9^−/− mice were administered 3 × 10⁷ to 4 × 10⁷ CFU L. pneumophila. This dose resulted in 100% mortality in TLR9^−/− mice, whereas greater early and long-term survival (40%) was observed for WT mice (Fig. 1A). We next evaluated the mechanism of increased mortality in infected TLR9^−/− mice. Legionella pneumophila (1 × 10⁶ CFU) was administered i.t. to WT and TLR9^−/− mice, and lung bacterial clearance was assessed. The number of bacterial CFU was approximately ninefold higher in the lungs of TLR9-deficient mice 72 h after challenge with L. pneumophila than that detected in similarly treated WT mice (Fig. 1B). No bacteremia was detected in WT or TLR9^−/− mice at any time after L. pneumophila administration (data not shown).

FIG. 1.

Survival (A) and lung CFU (B) in WT and TLR9^−/− mice after i.t. Legionella challenge. (A) Survival was assessed in WT and TLR9^−/− mice after the i.t. administration of L. pneumophila at a dose of 1 × 10⁶ (n = 10 animals per group). *, P < 0.05 compared to results for L. pneumophila-infected WT mice. (B) Lung L. pneumophila CFU was assessed in WT and TLR9^−/− mice. They were i.t. administered 1 × 10⁶ CFU L. pneumophila, and then lungs were harvested 72 h later. Results are for 8 to 10 mice per group and are expressed in the log₁₀ scale. *, P < 0.05 compared to results for infected WT mice.

Impaired accumulation and activation of DC and CD4⁺ T cells in TLR9^−/− mice compared to that of WT mice after i.t. Legionella administration.

To determine the role of TLR9 in the accumulation and/or activation of specific leukocyte subsets, WT and TLR9^−/− mice were administered L. pneumophila i.t., and then lungs were harvested 72 h later and cell digests were prepared. This time point was chosen because the maximum recruitment of inflammatory cells is observed at 72 h after i.t. L. pneumophila infection (data not shown). Baseline numbers and proportions of cell types were not different between uninfected WT and TLR9^−/− animals (Table 1). Moreover, TLR9^−/− mice infected with L. pneumophila had numbers of total leukocytes, polymorphonuclear leukocytes, and monocyte/macrophages that were similar to those of infected WT mice.

TABLE 1.

Leukocyte population in lung digest from BALB/c and TLR9-deficient mice 48 h after i.t. L. pneumophila infection^a

Mouse type and infection status	Total no. of cells (107)	No. of PMN (106)	No. of monocytes/macrophages (106)	No. DX5+ (105)	No. CD11c+ and MHC II+ (105)	No. CD4+ (105)	No. CD8+ (105)	No. CD4+ and CD69+ (104)	No. CD8+ and CD69+ (105)
WT
Uninfected	1.3 ± 0.3	1.3 ± 0.1	12 ± 0.1	2.7 ± 0.2	1.3 ± 0.1	8.4 ± 0.1	3.7 ± 0.5	2.0 ± 0.7	0.7 ± 0.2
Infected	2.2 ± 0.3	14 ± 0.9	12 ± 0.1	8.7 ± 2.1	3.4 ± 0.7	19 ± 2.2	6.0 ± 2.5	8.3 ± 2.2	1.3 ± 0.4
TLR9−/−
Uninfected	1.4 ± 0.4	1.5 ± 0.1	12 ± 0.1	2.9 ± 0.3	1.1 ± 0.1	6.2 ± 0.1	4.4 ± 0.2	1.2 ± 0.04	0.5 ± 0.1
Infected	2.7 ± 0.2	13 ± 0.9	9 ± 0.3	9.2 ± 5.2	2.0 ± 0.3	13 ± 1.0	6.1 ± 0.3	5.1 ± 0.4	1.1 ± 0.8

^aResults are from four uninfected mice and six Legionella-infected mice and were combined from two separate experiments. P < 0.05 compared to results for the L. pneumophilla -infected BALB/c group.

To assess DC accumulation, T cells, B cells, and autofluorescent macrophages were eliminated by forward- and side-scatter characteristics, and the remaining cells were assessed for the expression of MHC class II, CD11c, and GR-1 (6). We found no difference in the number of myeloid DC (MHC class II⁺, CD11c⁺) in the lungs of WT and TLR9^−/− mice in the uninfected state. However, the administration of L. pneumophila resulted in a 2.5-fold increase in the number of myeloid DC in lungs of WT mice 72 h after challenge, whereas there was a more modest increase in the total number of these cells in infected TLR9^−/− mice (P < 0.05), indicating the impaired accumulation of myeloid DC after i.t. bacterial challenge. We did not observe appreciable differences in the number of plasmacytoid DC (CD11c⁺, GR-1^moderate) in the lungs of uninfected or Legionella-infected TLR9^−/− animals compared to that of WT mice (data not shown). We next assessed the accumulation of other cell types involved in type 1 immune response generation. Compared to that of the L. pneumophila-challenged WT mice, we observed a significant decrease in the accumulation of CD4⁺ cells and CD4⁺ cells expressing the activational marker CD69 in infected TLR9^−/− mice. In contrast, we observed no change in the total number of DX5⁺ cells (both NK and NKT cells combined), CD8⁺ T cells, or DX5⁺ or CD8⁺ cells coexpressing the activational marker CD69. Collectively, these studies suggest that mice deficient in TLR9 display the reduced accumulation of myeloid DC and CD4⁺ T cells.

Lung macrophages isolated ex vivo from TLR9^−/− mice after i.t. Legionella challenge have reduced spontaneous iNOS expression and NO production.

Having observed no difference in the influx of innate phagocytic cells in Legionella-infected WT and TLR9^−/− mice to explain differences in bacterial clearance and survival, we performed experiments to assess the effector function of lung macrophages recovered from infected WT and mutant mice. We first evaluated the ability of alveolar macrophages to internalize L. pneumophila in vivo. In these experiments, FITC-labeled L. pneumophila (10⁶ CFU) were administered i.t. to WT and TLR9^−/− mice, and then BAL was performed 48 h later and the numbers of internalized FITC-labeled bacteria were quantitated by fluorescence microscopy from cell cytospins. As shown in Fig. 2A and B, there was a robust ingestion of FITC-labeled L. pneumophila cells within alveolar macrophages from WT mice. In contrast, there was a substantial reduction in the internalization of bacteria by alveolar macrophages from TLR9^−/− mice. Importantly, no differences in the baseline phagocytosis of L. pneumophila by alveolar macrophages from WT and TLR9^−/− mice was noted in the uninfected state (data not shown).

FIG. 2.

Ex vivo phagocytosis, iNOS expression, and NO production by lung macrophages in WT and TLR9^−/− mice after i.t. Legionella challenge. For the assessment of phagocytosis, alveolar macrophages obtained by BAL were isolated from WT and TLR9^−/− mice 48 h after the i.t. administration of FITC-labeled L. pneumophila (10⁶ CFU), and the number of internalized FITC-labeled bacteria per alveolar macrophage was quantitated. (A) A representative fluorescent field. Magnification, ×40. (B) Mean (± standard errors of the means) number of bacteria per alveolar macrophage from three separate animals, with 10 fields being counted from each. **, P < 0.005 compared to the levels of alveolar macrophages from WT mice. For the assessment of spontaneous iNOS and NO production, lung digest macrophages were isolated from WT and TLR9^−/− mice 48 h after i.t. L. pneumophila. iNOS levels were quantitated by real-time PCR (C) and NO levels were measured by Greiss reaction (D). Each value represents the mean results for 6 to 7 animals. *, P < 0.05 compared to results for L. pneumophila-infected WT mice.

To assess for differences of the activation of lung macrophages, WT and TLR9^−/− mice were challenged i.t. with Legionella (10⁶ CFU), and then lungs were harvested 48 h later. Macrophages were isolated from collagenase-treated lung digest and adherence purified. In these studies, we assessed the expression of markers used to define either the classical or alternative activation of lung macrophages. Several studies have identified the importance of NO production by lung macrophages in immunity against a number of intracellular pathogens, including L. pneumophila. Importantly, we found a significantly blunted expression of iNOS mRNA by lung macrophages isolated from TLR9^−/− mice at 48 h after L. pneumophila administration compared to that of the WT macrophages (Fig. 2C). Similarly, spontaneous NO production by lung macrophages isolated from L. pneumophila-infected TLR9-deficient mice was markedly reduced compared to that produced by macrophages isolated from WT mice (Fig. 2D).

Enhanced expression of markers of alternative activation in lung macrophages and whole lung of TLR9^−/− mice after i.t. Legionella infection.

Whereas lung macrophages isolated from TLR9^−/− mice displayed evidence of impaired classical activation, these cells demonstrated a phenotype indicative of alternative activation (15). Specifically, lung macrophages isolated from TLR9^−/− mice at 24, 48, and 72 h after L. pneumophila administration demonstrated a striking induction of Fizz-1 (Fig. 3A) and arginase-1 (Fig. 3B) mRNA in a time-dependent manner, with maximal expression noted at 48 h after bacterial challenge. In contrast, little to no induction of Fizz-1 and arginase-1 mRNA was observed in WT lung macrophages at any time point during Legionella infection (Fig. 4A and B). Consistently with the mRNA data, the spontaneous ex vivo production of Fizz-1 protein by lung macrophages isolated from TLR9^−/− mice 48 h after i.t. L. pneumophila administration was significantly greater than that of macrophages from WT mice (Fig. 4C). Moreover, the levels of Fizz-1 in whole lung were higher in infected TLR9^−/− mice than that observed in infected WT mice (Fig. 4D).

FIG. 3.

Ex vivo spontaneous mRNA expression of Fizz-1 and arginase-1 from lung macrophages isolated from WT and TLR9^−/− mice 48 h after the i.t. administration of L. pneumophila (LP). Lung macrophages were isolated from Legionella-infected WT or TLR9^−/− mice by adherence purification from lung leukocytes, and then they were incubated for 18 h at a concentration of 0.5 × 10⁶ cells/ml medium. Data shown represent mean cytokine mRNA levels quantitated by real-time PCR. Each value represents the mean results from six mice per value. Untreated, uninfected control. *, P < 0.05 compared to results for L. pneumophila-infected WT mice.

FIG. 4.

Levels of chemokines and type 1 cytokine protein from macrophages isolated from the lungs of WT and TLR9^−/− mice after i.t. L. pneumophila challenge. WT and TLR9^−/− mice were i.t. administered 10⁶ CFU L. pneumophila, and then lung macrophages were harvested 48 h later by lung digests. Then they were purified by adherence purification and cultured for 18 h, and protein levels in supernatants were measured by ELISA. Each value represents the mean results from four to five mice per value. No appreciable levels were detected in the untreated group. *, P < 0.05 compared to results for L. pneumophila-infected WT mice.

Impaired ex vivo production of inflammatory and type 1 cytokines by lung macrophages isolated from TLR9^−/− mice after i.t. Legionella infection.

To further explore functional changes in lung macrophages and mechanisms that may contribute to the impaired accumulation and/or activation of immune cells in TLR9^−/− mice, we assessed the spontaneous production of inflammatory/type 1 cytokines and chemokines by lung macrophages isolated from Legionella-infected mice. In these experiments, WT and TLR9^−/− mice were infected with 1 × 10⁶ CFU L. pneumophila i.t. and macrophages were isolated from lungs 48 h after bacterial challenge, and then they were cultured for 18 h and assessed for the production of TNF-α, IL-12, CCL2/MCP-1, and IL-13 by ELISA. As shown in Fig. 4, we observed a significant decrease in the production of IL-12, TNF-α, and MCP-1 by TLR9^−/− macrophages compared to that of lung macrophages isolated from infected WT mice. In contrast, the spontaneous production of IL-13 by macrophages isolated from Legionella-infected TLR9^−/− mice was significantly more robust than that produced by WT macrophages.

Decreased levels of inflammatory and type 1 cytokines/chemokines in BAL fluid from TLR9^−/− mice after i.t. Legionella infection.

To assess the differences in lung cytokine responses in vivo, WT and TLR9^−/− mice were infected with 10⁶ CFU L. pneumophila i.t., BAL was performed at 48 h after bacterial challenge, and then cell-free BAL fluid was assayed for protein levels of TNF-α, IL-12, IFN-γ, IP-10, and CCL2/MCP-1 by ELISA. Compared to the levels for uninfected animals, we observed an increase in the levels of all cytokines in infected WT mice at 48 h after challenge (Fig. 5). The BAL fluid levels of these cytokines were significantly reduced in TLR9^−/− mice. We were unable to detect IL-13 in the BAL fluid of either uninfected or infected mice (data not shown).

FIG. 5.

Levels of chemokines and type 1 cytokines in cell-free BAL fluid recovered from WT and TLR9^−/− mice 48 h after i.t. L. pneumophila challenge (10⁶ CFU). TNF-α, IL-12, IFN-γ, CXCL10/IP-10, and CCL2/MCP-1 levels were determined by ELISA. Each value represents the means ± standard errors of the means of four to five mice per value. Cytokine/chemokine levels were not detected in the BAL of untreated animals. *, P < 0.05 compared to results for L. pneumophila-infected WT mice.

Reconstitution of immunity and lung macrophage phenotype in TLR9^−/− mice after adoptive transfer of syngeneic bone marrow-derived DC.

Previous studies have shown that DC express considerably greater quantities of TLR9 than other leukocyte populations, and they are one of the predominant cell types to respond in a TLR9-dependent fashion. Moreover, our data indicate the impaired accumulation of DC in the lungs of TLR9^−/− mice after the i.t. administration of L. pneumophila. To determine if altered DC responses contributed to the impaired macrophage phenotype and decreased bacterial clearance observed in TLR9-deficient mice, we adoptively transferred bone marrow-derived DC into WT or mutant mice and then assessed their effects on bacterial clearance and markers of alternative activation of lung macrophages. For these experiments, we administered DC i.t., an approach that we and others have previously shown to stimulate intrapulmonary immunity in other model systems in which the disease is localized to the lung (6, 38, 47). Bone marrow cells were harvested from WT and TLR9^−/− mice and cultured in complete medium for 6 days in the presence of murine recombinant GM-CSF (10 ng/ml), loosely adherent cells were collected, and DC were positively selected for CD11c⁺ DC by magnetic bead separation. Bone marrow-derived DC (10⁶ cells) obtained from WT mice or TLR9^−/− mice were administered i.t. concomitantly with the administration of a sublethal dose of L. pneumophila (5 × 10⁵ CFU). Lungs were harvested 48 h later, and the numbers of bacterial CFU were determined. As previously observed, the numbers of lung L. pneumophila CFU were substantially higher in infected TLR9^−/− mice than in WT mice (Fig. 6A). Importantly, the i.t. administration of WT DC into infected TLR9^−/− mice resulted in a significant reduction in the number of bacterial CFU in the lung compared to that observed in infected WT mice, whereas the i.t. transfer of DC isolated from TLR9^−/− mice had no effect on bacterial clearance in TLR9^−/− mice.

FIG. 6.

Effect of the i.t. transfer of DC on bacterial clearance and cytokine production in WT and TLR9^−/− mice with Legionella pneumonia. Bone marrow cells were isolated from WT and TLR9^−/− mice, incubated with GM-CSF (10 ng/ml) for 6 days, and positively selected for CD11c. The DC then were administered i.t. L. pneumophila (1 × 10⁶ CFU) in a volume of 30 μl. L. pneumophila CFU in lung (A) and Fizz-1 levels by ELISA (B) were measured 48 h after i.t. L. pneumophila administration. The numbers of bacterial CFU are expressed on a log₁₀ scale ± standard errors of the means (n = 5 to 7 per group; results are a composite of two separate experiments). *, P < 0.05 compared to results for L. pneumophila-infected WT mice.

In additional studies, we found that the i.t. transfer of WT DC, but not DC from TLR9^−/− mice, resulted in a reduction in whole lung Fizz-1 levels compared to those observed in infected WT mice. (Fig. 6B).

CpG ODN stimulates protective immunity in murine Legionella pneumonia.

Having shown that mice deficient in TLR9 display impaired immunity against L. pneumophila, we next performed experiments to determine if the pretreatment of mice with the synthetic TLR9 agonist CpG ODN could stimulate protective immunity against L. pneumophila in a mouse strain in which macrophages are inherently permissive for intracellular Legionella replication. A/J mice have mutations in their nucleotide-binding oligomerization domain-leucine-rich repeat naip5 locus, making both the animals and macrophages from these animals susceptible to L. pneumophila infection (13). We chose to use an A-class CpG ODN in these experiments, as A-class CpG ODN primarily exert stimulatory effects on DC and NK cells, and we have previously shown that pretreatment with an A-class CpG ODN can augment innate pulmonary immunity against the extracellular gram-negative bacterial pathogen K. pneumoniae (5, 11, 16). In these experiments, mice were pretreated with control or CpG ODN (30 μg) i.t. or i.p., followed 48 h later by the i.t. administration of L. pneumophila (5 × 10⁵ CFU), and then the numbers of lung bacterial CFU were determined 72 h later. In animals administered CpG alone, we observed a very modest and transient increase in polymorphonuclear leukocytes and no histological evidence of lung injury (11 and data not shown). As shown in Fig. 7, the i.t., but not i.p., pretreatment of A/J mice with CpG ODN resulted in a 7.5-fold reduction in L. pneumophila CFU at 72 h after bacterial administration. Similarly, we found that the i.t. pretreatment with the C-class CpG ODN 10101 (30 μg) reduced the number of L. pneumophila CFU in A/J mice by approximately eightfold by 72 h (data not shown). These studies suggest that compartmental pretreatment with a synthetic TLR9 agonist can stimulate protective innate immunity in murine Legionella pneumonia.

FIG. 7.

Effect of i.t. CpG A-class ODN administration on lung clearance of L. pneumophila. A/J mice were administered A-class CpG (CpG-A) or control ODN (Ctl) (30 μg) i.t. or i.p. and then challenged with 10⁶ CFU of L. pneumophila 48 h later. The numbers of CFU in lungs were quantitated 72 h after bacterial challenge. Values represent the means ± standard errors of the means from results for four to five animals per group. *, P < 0.05 compared to results for Legionella-infected animals pretreated with control ODN i.t. or CpG-A i.p.

DISCUSSION

TLR signaling is a critical component of the innate immune response and is required for the detection of many types of microbial pathogens, including bacteria (1, 2). TLR9 is a predominantly intracellular receptor that recognizes microbial DNA and synthetic CpG ODN (45). Legionella is an intracellular organism that produces several factors that are potential TLR agonists, including LPS, lipoproteins, flagellin, unmethylated CpG DNA, and peptidoglycan. Several TLRs are important for host responses against Legionella infection. Specifically, TLR2 is required for murine macrophage cytokine and prostaglandin responses to purified L. pneumophila lipid A and intact L. pneumophila (33). It has been shown that the L. pneumophila flagellin protein activated TLR5, and polymorphisms of TLR5 in humans have been linked to enhanced susceptibility to Legionnaires' disease (17, 18). In contrast, no defect in antibacterial responses were observed in TLR4 mutant mice with Legionella pneumonia (14, 29). Our data are consistent with the findings of Newton and associates, who demonstrated that TLR9 was essential for IL-12 production by macrophages and DC ex vivo in response to incubation with L. pneumophila (34). Our study is the first to demonstrate an important role for TLR9 in L. pneumophila pneumonia in vivo. Finally, marked impairments in IFN-γ production and bacterial clearance have been observed in the lungs of MyD88^−/− mice challenged with L. pneumophila and are significantly reduced in association with the inability of MyD88-deficient macrophages to produce measurable IL-12 and other proinflammatory cytokines upon L. pneumophila infection (4). The fact that the defect in innate antibacterial responses observed in MyD88-deficient mice challenged with bacterial pathogens is more profound than that observed in mice in which a single TLR is deleted and/or nonfunctional argues strongly for the participation of multiple TLRs.

The mechanism of the impaired clearance of L. pneumophila from the lungs of TLR9^−/− mice has not been completely defined. We found no differences in the recruitment of phagocytic cells (macrophages and neutrophils). However, we found the reduced uptake of bacteria by alveolar macrophages in infected TLR9^−/− mice compared to that of the WT mice. Defects in the phagocytosis of Streptococcus pneumoniae have been demonstrated in alveolar macrophages from TLR9-deficient mice (3). Interestingly, we observed for the first time that the defect in the ingestion of bacteria was observed only in vivo, as we found no defect in resting alveolar macrophages isolated from TLR9^−/− mice at baseline (data not shown). These findings suggest that the cytokine milieu within the lung microenvironment during infection is important to enhance phagocytosis, and that these signals are deficient or inadequate in TLR9^−/− mice.

Additionally, we found convincing evidence of the impaired classical activation of lung macrophages, as manifested by decreased inflammatory cytokine production, a striking impairment in iNOS expression, and spontaneous NO elaboration by lung macrophages isolated from Legionella-infected TLR9^−/− mice. It has been shown previously that NO is a key effecter molecule in the resistance of cultured murine macrophages to intracellular pathogens, including L. pneumophila (9, 19). In the absence of NO production, the macrophage remains permissive for L. pneumophila replication. NO inhibits the growth of L. pneumophila in the lung during the initial stages (i.e., within 3 to 5 days) of infection, suggesting that this mediator likely plays a pivotal role in the host's first line of defense (i.e., nonspecific immune response) to the bacteria, thereby limiting intrapulmonary replication prior to the development of L. pneumophila-specific cellular immunity. Classically activated macrophages (also called M1) are induced by IFN-γ alone or in combination with microbial stimuli (e.g., LPS [31]). Importantly, we found the reduced accumulation of activated CD4⁺ T cells and the expression of type 1 cytokines and chemokines, including IFN-γ and CXCL10/IP-10, in Legionella-infected TLR9 mutant mice. We cannot exclude the impaired internalization of bacteria as a meaningful contributor to the defective classical activation of lung macrophages during infection.

Conversely, macrophages isolated from infected TLR9 mutant mice displayed markers characteristic of an alternatively activated phenotype, including the enhanced expression of arginase-1 and Fizz-1 (15). Alternatively activated macrophages (M2) express high levels of mannose- and galactose-type receptors, and arginine metabolism is shifted toward the production of ornithine and polyamines via upregulated arginase-1. Consequently, alternatively activated macrophages have a reduced expression of NO and impaired inflammatory cytokine expression, responses that are essential for protective innate immunity against lung bacterial pathogens. The M2 phenotype can be observed in the setting of parasitic infection and chronic inflammation and is driven in vitro by cytokines/mediators such as IL-4, IL-13, IL-10, immune complexes, and glucocorticoids (15, 31). Lung macrophages isolated from Legionella-infected TLR9^−/− mice produce greater quantities of IL-13 and reduced quantities of IL-12 relative to those produced by WT macrophages. Whether there is enhanced expression of IL-13 or other cytokines that promote alternative activation by other lung cells in infected mutant mice remains to be determined. It is noteworthy that some phagocytic responses, particularly those mediated by the mannose receptor, are enhanced in alternatively activated macrophages (15). However, the involvement of the mannose receptor in the internalization of Legionella spp. has not been shown previously. Our results indicate that TLR9 is required for classical lung macrophage activation in intracellular bacterial infection, and this is the first study to identify the critical role of TLR9 in the programming of the macrophage phenotype.

In our study, we observed the impaired expression of type 1 cytokines and the reduced accumulation/activation of CD4⁺ T cells in infected TLR9^−/− mice. This defect raises the possibility of altered DC function in the absence of TLR9. TLR9 has been shown previously to be responsible for DC activation and the generation of type 1 responses in a variety of in vitro and in vivo immune responses (26, 27, 34). We observed reduced lung DC accumulation in TLR9-deficient mice after challenge with L. pneumophila. Similarly, we have recently shown impaired DC accumulation and activation in the lungs of TLR9^−/− mice infected with the extracellular gram-negative bacterium K. pneumoniae (6). In the current study, we demonstrated that host immunity could be restored in mutant mice by the adoptive transfer of syngeneic WT, but not TLR9^−/−, bone marrow-derived DC directly into the lungs of TLR9^−/− mice infected with L. pneumophila. Moreover, adoptive transfer resulted in a decrease in the levels of Fizz-1 in infected knockout mice, indicating a partial reversal of the M2 phenotype. These data provide compelling evidence that impaired DC function in TLR9^−/− mice is largely responsible for the immune defects observed, and that the changes in lung macrophage phenotype are attributable, at least in part, to impaired DC function. We cannot exclude a direct TLR9-mediated effect on lung macrophages, but it is noteworthy that murine alveolar macrophages do not express appreciable amounts of TLR9 in the resting state (40), and we have not found a substantial induction of TLR9 message in lung macrophages during murine Legionella pneumonia (data not shown).

We found that i.t, but not systemic (i.p.), treatment with CpG ODN enhanced the intrapulmonary clearance of L. pneumophila. Similar immunostimulatory effects of intrapulmonary CpG have been observed in respiratory tract infections due to other microbial pathogens, including K. pneumoniae, Chlamydia trachomatis, Mycobacterium tuberculosis, and Aspergillus fumigatus (7, 11, 22, 35). Mechanisms of protection conferred by CpG generally are due to the promotion of type 1 cytokine responses, including the enhanced expression of IL-12, IFN-γ, and IP-10 (11, 25, 46, 48). The effect of CpG is largely mediated by the activation of TLR9, but it is noteworthy that the ODN backbone itself has been shown to possess immunostimulatory effects independent of TLR9 agonist activity. A-class and C-class CpG ODN primarily stimulate DC effector function, whereas B-class CpG ODN primarily activate B cells and, to a lesser extent, monocytes/macrophages (5, 16, 24). Finding that both A-class and C-class ODN can augment protective immunity further supports a direct role for DC as the predominant cells responding in a TLR9-specific manner during pneumonia.

In conclusion, our study demonstrates the importance of TLR9 in the generation of protective lung immunity and regulating lung macrophage phenotype during infection with virulent facultative intracellular bacterial pathogens. Approaches to augment TLR9-mediated responses has some promise as a immunoadjuvant therapy in the treatment of severe life-threatening respiratory tract infections such as Legionella pneumonia.