Cooperation between Multiple Microbial Pattern Recognition Systems Is Important for Host Protection against the Intracellular Pathogen Legionella pneumophila

Bacterial strains. L. pneumophila serogroup 1 strain JR32 (47), an flaA mutant (JR32 ΔflaA) (45), and serogroup 1 clinical isolate F2111 (13) were used in this study. L. pneumophila strains were cultured on charcoal-yeast extract (CYE) agar (14) for 2 days and then cultured overnight in N-(2-acetamido)-2-aminoethanesulfonic acid (ACES)-buffered yeast extract (AYE) broth (10 g/liter yeast extract, 10 g/liter ACES, 0.4 g/liter l-cysteine HCl-H₂O, 0.135 g/liter ferric nitrate) prior to use in experiments. For enzyme-linked immunosorbent assay (ELISA) studies and in vivo growth assays, bacteria were grown to an optical density at 600 nm of 1 in AYE broth. For ex vivo growth assays, bacteria were grown to an optical density of 3.4 in AYE broth.

Mice.C57BL/6 (stock number 000664) mice were purchased from Jackson Laboratories. MyD88^−/− (1) and Rip2^−/− (33) mice in a C57BL/6 background have been described previously. MyD88^−/− mice were provided by R. Medzhitov, and Rip2^−/− mice were provided by R. Flavell. For experiments using mice with the nonfunctional Naip5 gene, MyD88^−/− and Rip2^−/− mice in a mixed 129/SvJ × C57BL/6 background were mated with A/J mice to generate mice that were homozygous for the A/J Naip5 allele.

Bone marrow-derived macrophages (BMMs).Bone marrow was collected from the femurs and tibiae of mice. Cells were plated on petri dishes and incubated at 37°C in RPMI 1640 medium containing 20% fetal bovine serum (FBS), 30% macrophage colony-stimulating factor (M-CSF)-conditioned medium, and 1% penicillin-streptomycin. At day 6, cells were harvested and resuspended in RPMI 1640 medium containing 10% fetal bovine serum (FBS) and 5% M-CSF. Cells were then plated in 24-well tissue culture-treated plates and incubated at 37°C. M-CSF was obtained from an L-929 fibroblast cell line (ATCC).

Ex vivo macrophage infections and growth assays.BMMs were added to 24-well plates at a concentration of 2 × 10⁵ cells/well. The cells were infected with either the L. pneumophila JR32 wild-type (WT) or JR32 Δfla strain at a multiplicity of infection (MOI) of 5 and incubated at 37°C. BMMs were washed at 1 h postinfection with warm Dulbecco's phosphate-buffered saline (PBS) to remove extracellular bacteria. Then either the BMMs were lysed immediately with sterile H₂O (day 0) or fresh medium was added until the cells were harvested at 24, 48, and 72 h postinfection. Cell lysates were plated on CYE agar to determine the number of bacterial CFU. Each data point is the value for one mouse for which the average bacterial content of three independent wells was determined. The fold differences were determined by dividing the values obtained at 24, 48, and 72 h by the values obtained on day 0. Ex vivo growth assays were repeated at least once, and the results obtained were similar.

In vivo mouse infections.Mice were anesthetized by intraperitoneal injection of a ketamine (100 mg/kg)-xylazine (10 mg/kg)-PBS solution and infected intranasally with wild-type L. pneumophila strain JR32, the isogenic ΔflaA derivative of JR32, or the clinical isolate F2111 in 40 μl of PBS. For in vivo bacterial growth assays, mice were euthanized by CO₂ either at 4 h postinfection (day 0) or at the times indicated in the figures. Lungs, spleens, or livers were harvested and placed in 10 ml of sterile double-distilled H₂O and homogenized using a PowerGen 125 handheld homogenizer (Fisher) for 30 s. Lysates were plated on CYE agar to determine the numbers of bacterial CFU. Each data point in the figures is the CFU count for a single mouse. All experiments were repeated at least one time independently, and similar results were obtained. The lower limit of detection in the assay was 100 CFU of L. pneumophila.

To determine cytokine levels in bronchoalveolar lavage fluid (BALF), mice were intranasally infected with either 1 × 10⁶ or 4 × 10⁴ CFU of L. pneumophila Δfla. Mice were euthanized by intraperitoneal injection of a ketamine (250 mg/kg)-xylazine (25 mg/kg)-PBS solution. The mouse lungs were lavaged once with 500 μl of PBS, and the BALF was stored at −80°C. IL-12 p40, IL-6, and IFN-γ levels in the BALF were determined by ELISA using BD Pharmingen IL-12 (p40/p70), IL-6, and IFN-γ reagents, respectively. Monocyte chemoattractant protein-1 ([MCP-1] BD OptEIA kit; BD Pharmingen) and keratinocyte-derived chemokine [KC] DuoSet Mouse KC; R&D Systems) levels were also determined by ELISA.

Lymphocyte isolation and intracellular cytokine staining.Lymphocytes were isolated from lungs by digesting minced lungs for 1 h in a 37°C shaking incubator in collagenase buffer (RPMI 1640 medium, 100 U/ml collagenase I [Gibco], 5% fetal bovine serum, 0.1% CaCl₂, 0.1% MgCl₂). Lysates were put through a 70-μm-pore-size nylon cell strainer (BD Falcon), and cells were isolated using a Percoll (Sigma) gradient. Red blood cells were lysed with red blood cell lysing buffer (Sigma). Isolated lymphocytes were resuspended in fluorescence-activated cell sorter (FACS) buffer (PBS, 1% FBS, 0.025% sodium azide) and blocked with purified rat anti-mouse CD16/CD32 (BD Pharmingen). Cells were stained with rat anti-mouse CD11b (Caltag Laboratories), rat anti-mouse Ly-6C/G (Gr-1) (Caltag Laboratories), and rat anti-mouse F4/80 (Caltag Laboratories). Neutrophils were gated as CD11b⁺, Gr-1⁺, and F4/80⁻. Inflammatory monocytes were gated as CD11b⁺, Gr-1⁺, and F4/80⁺. Data were collected using a FACSCalibur flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star).

Generation of bone marrow-chimeric mice.Recipient C57BL/6 (expressing the Ly5.1 antigen) mice were lethally irradiated with two doses of 500 rads over a 5-h period using a ¹³⁷Cs γ-source and reconstituted intravenously with 1 × 10⁷ bone marrow cells from donor C57BL/6 (expressing the Ly5.2 gene) wild-type, MyD88^−/−, or Rip2^−/− MyD88^−/− mice. Mice were placed on Sulfatrim (40 mg/ml sulfamethoxazole-8 mg/ml trimethoprim) for 8 weeks following transplantation of bone marrow cells, and chimerism was confirmed by flow cytometry after differential staining of blood lymphocytes for Ly5.1 and Ly5.2. Mice were infected as described above.

Statistical analysis.A two-tailed, Mann-Whitney U test was used to analyze the significance of differences in means between groups. Survival curves were generated using the Kaplan-Meier method, and the significance of differences was calculated by a log rank test. Differences were considered statistically significant if the P value was <0.05.

L. pneumophila activates Rip2-dependent pathways in vivo.The activation of NF-κΒ by a Rip2-dependent pathway was shown following L. pneumophila infection of MyD88-deficient macrophages ex vivo (35, 52), suggesting that Nod1 and Nod2 are able to detect the presence of L. pneumophila products within the host cell cytosol (52). To determine whether Rip2 signaling is important for in vivo responses to L. pneumophila, wild-type (WT) and Rip2-deficient mice in the C57BL/6 background were infected intranasally with a flagellin-deficient (ΔflaA) strain of L. pneumophila. The L. pneumophila ΔflaA strain was used to evade activation of the Naip5 and NLRC4 signaling pathways. Mice were sacrificed at days 2, 3, and 5 postinfection, and inflammatory cytokine levels were measured in the BALF by ELISA. No significant difference was observed in the levels of IL-12 p40, IFN-γ, or IL-6 in the lung of Rip2-deficient mice compared to WT mice (Fig. 1A). A significant decrease was observed consistently in the levels of the chemokine MCP-1 (Fig. 1A), indicating a role for Rip2 in the detection of L. pneumophila in vivo. To determine whether the absence of Rip2 results in enhanced L. pneumophila replication in the lung, bacterial numbers were measured. No significant difference was detected in the number of L. pneumophila cells in the lung of Rip2-deficient mice compared to WT mice, and the rate of clearance from the lungs was also similar (Fig. 1B). To control for the possibility that the nonmotile phenotype exhibited by L. pneumophila deficient in flagellin might influence these results, the Rip2 deletion was introduced into mice that were deficient in Naip5 signaling as a result of being homozygous for the hypomorphic Naip5 allele from the A/J mouse (34). Similar results were obtained using Rip2-deficient mice having a defect in Naip5 signaling following infection with a clinical isolate of L. pneumophila that produces flagellin (see Fig. S1 in the supplemental material). Thus, bacterial motility and the production of flagellin did not affect the Rip2-dependent response.

• Open in new tab
• Download powerpoint

FIG. 1.

L. pneumophila activates Rip2-dependent pathways in vivo. WT and Rip2-deficient (Rip2^−/−) mice in a C57BL/6 background were given a high intranasal dose (1 × 10⁶ CFU) of L. pneumophila ΔflaA. For each group, five mice were sacrificed at days 2, 3, and 5 postinfection as indicated below each graph. (A) BALF from infected mice was assayed for the indicated cytokines by ELISA. Each circle represents data obtained from a single mouse. The lines indicate the mean values calculated from the data for the two groups of mice. *, P < 0.05. (B) Bacterial numbers were determined from lung lysates at the indicated time points. There was no statistical significance between the two groups of mice (P > 0.05).

Rip2 is important for suppression of L. pneumophila replication in MyD88-deficient mice.Because MyD88-dependent signaling contributes significantly to the innate immune response directed against L. pneumophila, host responses mediated by Rip2 may be masked by MyD88-dependent signaling pathways. To investigate this possibility, mice deficient in both Rip2 and MyD88 were generated in the C57BL/6 background and infected with L. pneumophila ΔflaA. At day 2 postinfection, there was no significant difference between the number of L. pneumophila ΔflaA CFU in the Rip2- and MyD88-deficient (Rip2/MyD88-deficient) mice and in the MyD88-deficient mice (Fig. 2A). At day 5, however, the Rip2/MyD88-deficient mice had slightly higher CFU counts in the lungs, liver, spleen, and blood than mice deficient in only MyD88 (Fig. 2A). This difference was statistically significant. Rip2/MyD88-deficient mice infected with motile wild-type L. pneumophila also had higher CFU counts than MyD88-deficient mice 5 days after infection, indicating that the contribution of Rip2 was independent of flagellin sensing by the Naip5/NLRC4 pathway (see Fig. S2 in the supplemental material). These data suggest that Rip2 signaling contributes to restricting L. pneumophila replication in vivo when MyD88-dependent responses are absent.

• Open in new tab
• Download powerpoint

FIG. 2.

Rip2 is important for suppression of L. pneumophila replication in MyD88-deficient mice. (A) MyD88-deficient (MyD88^−/−) and Rip2/MyD88-deficient (Rip2^−/−MyD88^−/−) mice in a C57BL/6 background were given a low intranasal dose (4 × 10⁴ CFU) of L. pneumophila ΔflaA. At day 2 and day 5 postinfection, seven mice from each group were sacrificed, and bacterial numbers were determined from the lung, blood, liver, and spleen. Each point represents data from a single mouse. The lines indicate the mean values calculated from the data for the two groups of mice. The difference between the two groups of mice was statistically significant at day 5. *, P < 0.005. (B) BMMs from WT (filled circles), Rip2^−/− (open circles), MyD88^−/− (filled squares), and Rip2^−/− MyD88^−/− (open squares) mice were infected with L. pneumophila ΔflaA. The numbers of bacterial CFU were determined at 1 h and 48 h postinfection. Intracellular growth is expressed as the fold increase in the number of CFU detected over this period. Each point represents data for BMMs derived from a single mouse. All data points represent the average increase in the number of bacterial CFU determined from three wells infected independently. Each line indicates the mean calculated from the data for the two different mice.

To determine whether higher levels of replication in Rip2/MyD88-deficient mice result from an intrinsic defect in the ability of macrophages to restrict L. pneumophila multiplication, bone marrow-derived macrophages (BMMs) from WT, Rip2-deficient, MyD88-deficient, and Rip2/MyD88-deficient mice were infected with L. pneumophila ΔflaA, and intracellular growth was monitored for 3 days. L. pneumophila replication in macrophages lacking Rip2 was comparable to that in MyD88-deficient and WT cells (Fig. 2B). Similar results were observed using macrophages deficient in Naip5 signaling and infected with motile L. pneumophila producing flagellin (see Fig. S3 in the supplemental material). Thus, the enhanced replication of L. pneumophila observed in the Rip2/MyD88-deficient mice was not due to intrinsic differences in the ability of naive macrophages derived from these mice to support intracellular replication.

Cellular recruitment defects in Rip2/MyD88-deficient mice after L. pneumophila infection.The cytokine IL-6 and chemokines KC and MCP-1 are detected at low levels after L. pneumophila infection of MyD88-deficient mice (Fig. 3A). To determine whether the higher bacterial numbers observed in the Rip2/MyD88-deficient mice correspond with a defect in production of proinflammatory chemokines or cytokines, MyD88-deficient and Rip2/MyD88-deficient mice were infected intranasally with L. pneumophila ΔflaA. Mice were sacrificed at days 2 and 5 postinfection, and ELISA measurements of BALF were used to assess the host response to infection. The levels of IL-6 at days 2 and 5 postinfection and of KC and MCP-1 at day 2 postinfection were reduced significantly in Rip2/MyD88-deficient mice compared to MyD88-deficient mice (Fig. 3A). Because the KC and MCP-1 attract neutrophils and inflammatory monocytes to infected tissues, respectively (9, 49), we examined whether the reduction of these chemoattractants in Rip2/MyD88-deficient mice early upon infection correlate with differences in cellular recruitment to the lung of infected mice. Rip2/MyD88-deficient mice had a significant reduction in the proportion of neutrophils (CD11b⁺, Gr-1⁺, and F4/80⁻) in the lung at day 2 postinfection and fewer inflammatory monocytes (CD11b⁺, Gr-1⁺, and F4/80⁺) at days 2 and 5 postinfection than MyD88-deficient mice (Fig. 3B and C). Thus, Rip2 signaling contributes to cytokine production and cellular recruitment to the lung of L. pneumophila-infected mice.

• Open in new tab
• Download powerpoint

FIG. 3.

Cellular recruitment defects in Rip2/MyD88-deficient mice after L. pneumophila infection. MyD88^−/− and Rip2^−/− MyD88^−/− mice in a C57BL/6 background were given a low intranasal dose (4 × 10⁴ CFU) of L. pneumophila ΔflaA. Seven mice from each group were sacrificed at days 2 and 5 postinfection. (A) BALF from infected mice was assayed for the indicated cytokines by ELISA. Each point represents data from a single mouse. The horizontal lines indicate the mean values calculated from the data for the two groups of mice. (B and C) Cell suspensions from isolated lungs were stained for CD11b, Gr-1, and F4/80 and examined by flow cytometry. (B) Representative plots indicating the frequency of neutrophils (CD11b⁺, Gr-1⁺, and F4/80⁻; upper-left quadrant) and inflammatory monocytes (CD11b⁺, Gr-1⁺, and F4/80⁺; upper-right quadrant) from an individual MyD88^−/− or Rip2^−/− MyD88^−/− mouse sacrificed at day 2 or day 5. (C) Average percentage of neutrophils or inflammatory monocytes from individual MyD88^−/− or Rip2^−/− MyD88^−/− mice sacrificed at day 2 or day 5. Plots for mouse groups are arranged as in panel A. *, P < 0.05; **, P < 0.005.

Mice deficient for both Rip2 and MyD88 are highly susceptible to lethal infection by L. pneumophila.To investigate whether Rip2 signaling contributes to host protection upon infection by L. pneumophila, MyD88-deficient and Rip2/MyD88-deficient mice were infected with L. pneumophila ΔflaA and monitored for survival. Most of the infected Rip2/MyD88-deficient mice died within the first 7 days, and all of the mice died by day 11 (Fig. 4A). In contrast, all of the infected MyD88-deficient mice were alive at day 9, and 50% of the MyD88-deficient mice survived infection and appeared healthy at day 30. The surviving MyD88-deficient mice were sacrificed, and the numbers of bacteria in the lungs, liver, spleen, and blood were determined. The surviving MyD88-deficient mice had high levels of L. pneumophila in the lung, but bacteria were not detected in the liver, spleen, or blood (Fig. 4B). Similar findings were observed for Rip2/MyD88-deficient mice having a defect in Naip5 signaling and infected with motile L. pneumophila producing flagellin (see Fig. S4 in the supplemental material). These data indicate that Rip2 signaling in response to L. pneumophila activates pathways that provide host protection against infection and that these pathways play a critical role in host survival when MyD88 signaling is disrupted.

• Open in new tab
• Download powerpoint

FIG. 4.

Mice deficient for both Rip2 and MyD88 are highly susceptible to lethal infection by L. pneumophila. MyD88^−/− and Rip2^−/− MyD88^−/− mice in a C57BL/6 background were given a low intranasal dose (4 × 10⁴ CFU) of L. pneumophila ΔflaA. (A) The graph represents the percentage of MyD88^−/− (n = 19) and Rip2^−/− MyD88^−/− (n = 18) mice surviving at the time points indicated. The difference between the two groups of mice was significant (P < 0.0001). (B) Surviving MyD88^−/− mice were sacrificed at day 30 postinfection, and the average number of CFU in the lung was measured. There were no detectible bacteria in either the liver or the spleen of these mice, and there were no surviving Rip2/MyD88-deficient mice to analyze. These data are in contrast to immune-sufficient mice, which did not have detectible bacteria in the lung at day 30 postinfection. Each point represents data from a single MyD88^−/− mouse, and the line represents the average from all mice.

Host protection to L. pneumophila infection requires activation of innate signaling pathways in bone marrow-derived cells.To elucidate cell types in the lung that require MyD88 and Rip2 signaling to direct a protective host response during infection by L. pneumophila, chimeric mice were generated by irradiating wild-type C57BL/6 mice and repopulating these mice with bone marrow cells obtained from either wild-type mice, MyD88-deficient mice, or Rip2/MyD88-deficient mice. Data examining L. pneumophila numbers at day 5 postinfection revealed that the chimeric mice had similar phenotypes as their respective bone marrow donor mice (Fig. 5A). Bacterial loads in the lung and spleen were significantly higher in the chimeric mice receiving MyD88-deficient or Rip2/MyD88-deficient bone marrow cells than in control mice receiving wild-type bone marrow cells. Importantly, bacterial colonization in the chimeric mice was not significantly different from that of their respective bone marrow-donor mice. High IL-6 levels were present in BALF from chimeric mice receiving MyD88-deficient or Rip2/MyD88-deficient bone marrow cells but not in BALF from respective donor MyD88-deficient or Rip2/MyD88-deficient mice (Fig. 5B), which indicates that a significant amount of Rip2/MyD88-dependent production of IL-6 is mediated by a radiation-resistant population of cells in the chimeric mice. Thus, even though radiation-resistant cells are contributing to the immune response, innate immune signaling pathways in the bone marrow-derived cells are critical for host protection against L. pneumophila.

• Open in new tab
• Download powerpoint

FIG. 5.

Host protection from L. pneumophila infection requires activation of innate signaling pathways in bone marrow-derived cells. Bone marrow-chimeric mice were generated by irradiating wild-type (WT) mice and reconstituting them with donor bone marrow from either WT (WT>WT), MyD88-deficient (MyD88^−/−>WT), or Rip2/MyD88-deficient (Rip2^−/− MyD88^−/−>WT) mice. Mice were given a low intranasal dose (4 × 10⁴ CFU) of L. pneumophila ΔflaA. (A) Bacterial CFU in the lung and spleen was measured at 5 days postinfection for irradiated mice (open circles) receiving bone marrow from wild-type, MyD88^−/−, or Rip2^−/− MyD88^−/− donor mice and for mice with the same genetic background as the donors (filled circles). (B) BALF from infected mice was assayed for IL-6. Each point represents data from a single mouse. The lines indicate the means calculated from data for each group of mice. *, P < 0.05; **, P < 0.01.

A functional hierarchy of pattern recognition receptors generates a protective immune response against L. pneumophila.Rip2, MyD88, and Naip5/NLRC4 are activated during infection of macrophages by L. pneumophila, and each signaling pathway contributes to host protection. Although MyD88-dependent responses provide protection against flagellin-deficient L. pneumophila that fails to activate Naip5/NLRC4, flagellin-deficient L. pneumophila replicates to higher numbers and persists longer in the lungs of wild-type mice than isogenic L. pneumophila that produces flagellin (4, 5, 10, 34, 40), indicating that the contribution of Naip5/NLRC4 to bacterial clearance is not masked by a strong MyD88-dependent response. This is in contrast to results shown here for Rip2, where no significant difference was observed in L. pneumophila numbers when infected Rip2-deficient mice were compared to control wild-type mice. To determine whether the Naip5/NLRC4 pathway and the Rip2 pathway have functionally redundant or independent roles in providing host protection, immune-deficient mice were infected with a high dose of wild-type L. pneumophila or an isogenic ΔflaA strain, and bacterial colonization and host survival were compared.

Similar to what has been observed in wild-type mice, there were significantly more L. pneumophila ΔflaA bacteria in the lung of Rip2-deficient mice at day 5 postinfection than in wild-type L. pneumophila-infected mice (Fig. 6A). Rip2-deficient mice were able to control infection by either strain, as indicated by a decrease in the number of wild-type L. pneumophila and ΔflaA bacteria in the lung at day 5 compared to the count at 4 h postinfection (Fig. 6A). Additionally, all of the Rip2-deficient animals survived a high-dose infection by either wild-type L. pneumophila or the ΔflaA strain (Fig. 6B and C). In both MyD88-deficient and Rip2/MyD88-deficient mice, bacterial numbers were higher at day 5 than at 4 h postinfection, and the difference between the numbers of wild-type and ΔflaA bacteria was not as great as that which was observed in Rip2-deficient mice (Fig. 6A), indicating that MyD88-deficient mice are unable to suppress replication of these strains in vivo. Importantly, the majority of MyD88-deficient mice (9/10) survived a high-dose infection by wild-type L. pneumophila, whereas most Rip2/MyD88-deficient mice (8/10) died within the first 10 days of infection by this strain (Fig. 6B). Measurements of the number of wild-type L. pneumophila bacteria in the lungs of the surviving mice sacrificed 21 days after infection revealed that the Rip2-deficient mice were able to clear bacteria from the lungs, whereas persistent colonization of the lung by wild-type L. pneumophila was observed in the MyD88-deficient and Rip2/MyD88-deficient survivors (Fig. 6B). All of the MyD88-deficient and Rip2/MyD88-deficient mice receiving a high dose of the L. pneumophila ΔflaA strain died (Fig. 6C) although, consistent with what was observed for mice that succumbed following a low-dose infection, the Rip2/MyD88-deficient mice had a shorter mean time to death than the MyD88-deficient mice. Thus, Rip2 and Naip5/NLRC4 have additive and nonoverlapping roles in contributing to host protection following L. pneumophila infection.

• Open in new tab
• Download powerpoint

FIG. 6.

Cooperative signaling between multiple pattern recognition systems is important for a protective immune response against L. pneumophila. Rip2^−/−, MyD88^−/−, and Rip2^−/− MyD88^−/− mice in a C57BL/6 background were given a high intranasal dose (1 × 10⁶ CFU) of either wild-type L. pneumophila or L. pneumophila ΔflaA. (A) At day 5 postinfection bacterial CFU were measured in the lung. Data in the left panel (day 0) represent CFU measured in the lungs of control animals sacrificed 4 h after intranasal infection. Each point represents data from a single mouse. The lines indicate the means calculated from the data for the two groups of mice. The dashed line indicates the lower limit of detection. *, P < 0.05 (B) The graph on the left indicates the survival of Rip2^−/−, MyD88^−/−, and Rip2^−/− MyD88^−/− mice after a high intranasal dose (1 × 10⁶ CFU) of wild-type L. pneumophila (n = 9 mice for each group). The difference between the MyD88^−/− and Rip2^−/− MyD88^−/− mice was significant (P < 0.01). The graph on the right shows the number of L. pneumophila bacteria in the lungs of mice surviving until day 21. (C) Survival of Rip2^−/−, MyD88^−/−, and Rip2^−/− MyD88^−/− mice after a high intranasal dose (1 × 10⁶ CFU) of L. pneumophila ΔflaA (n = 3 mice for each group). The difference between the MyD88^−/− and Rip2^−/− MyD88^−/− mice was significant (P < 0.05).