Innate Immune Responses to Rhodococcus equi⁴

Rhodococcus equi is a Gram-positive coccobacillus that resides primarily if not exclusively within host tissue macrophages (1). This organism is becoming more frequently diagnosed as an opportunistic pathogen of immunocompromised patients, especially those infected with HIV (2). In these patients, the most often observed presentation of rhodococcal disease is severe pyogranulomatous pneumonia. Treatment of R. equi pneumonia is often prolonged, and relapses at distant sites are not uncommon (2). R. equi is a common soil organism, which frequently infects young horses (3), where similar pulmonary involvement, including abscess formation and cavitary pneumonia are most frequently observed (4). Virtually all isolates obtained from diseased animals, and a subset of R. equi isolated from infected people express virulence-associated protein A (VapA),³ a surface-localized lipoprotein, the presence of which appears to be essential for intramacrophage replication (5). Infections in either humans or young horses can result in tissue damage and occasionally reduced lung function (6).

The TLRs are critical components of the innate immune response (7). Signaling through the TLRs gives rise to NF-KB activation and a myriad of other cellular signaling events (8) that result in the up-regulation of costimulatory molecules on APCs and the production of cytokines that are instrumental in initiating and shaping adaptive immune responses (9 , 10). Many of the cytokines that are produced subsequent to TLR engagement are proinflammatory and carry the unwanted side effect of causing tissue damage. For these reasons, the production of these cytokines can be a two-edged sword. On one hand they are a necessary component of host defense, but on the other the price paid for their production is the destruction of host tissue. Macrophages are a rich source of inflammatory mediators, including cytokines, lipid derivatives, and toxic metabolites of oxygen and nitrogen. The engagement of TLRs on macrophages has been shown to be a particularly efficient way to induce the production of these mediators. A variety of Gram-positive and -negative bacteria have been shown to induce cytokine production from macrophages as a result of TLR activation (11, 12). TNF is the prototypical proinflammatory cytokine (13). The overproduction of TNF has been associated with a number of autoimmune pathologies (14), and interfering with TNF production often ameliorates autoimmune manifestations (15). IL-12 production is associated with Th1 responses, and its role in host defense to intracellular pathogens has been well established (16). IL-12 induces the production of IFN-γ from T and NK cells. IFN-γ in combination with TNF induces macrophage activation and the killing of intracellular pathogens, including R. equi (17). In murine systems, these cytokines stimulate the production of NO from macrophages, a molecule that has been implicated in macrophage-mediated killing of R. equi (17).

In the present work, we show that R. equi is an efficient inducer of TNF, IL-12, and NO from macrophages. This induction correlates with NF-κB translocation as a result of TLR activation. We show that the R. equi surface lipoprotein, VapA, can activate TLR2 and that the interaction of R. equi with TLR2 on macrophages is one of the major mechanisms by which these cells respond to R. equi. Finally, we demonstrate the importance of TLR2 signaling during in vivo infection.

R. equi strains 238 and 103+ are virulent clinical isolates obtained from pneumonic foals. They were provided by the Veterinary Microbiology Laboratory (New Bolton Center, University of Pennsylvania, Kennett Square, PA) and by Dr. J. F. Prescott (University of Guelph, Guelph, Canada), respectively. Bacteria were streaked on chocolate agar plates and grown at 37°C for 36 h. For all assays, one isolated colony of bacteria was inoculated into 10 ml of Mueller-Hinton broth (Difco Laboratories, Detroit, MI) and cultured overnight at 37°C with constant shaking (150 rpm) to a final density of ∼1 × 10⁸/ml. Before use, bacteria were harvested by centrifugation (1800 × g, 15 min, 4°C), washed twice in PBS (Mediatech, Herndon, VA), and incubated with 10% normal mouse serum as a source of opsonizing complement for 10 min at 37°C.

For some experiments (Figs. 1 and 2), LPS, derived from Escherichia coli 0127:B8 (Sigma-Aldrich, St. Louis, MO), was reconstituted in PBS, stored at 4°C, and used to stimulate macrophages at a final concentration of 100 ng/ml. For the rest of the experiments, twice-extracted LPS (18) was generously provided by Dr. S. Vogel (University of Maryland Medical School, Baltimore, MD). R. equi acetone-precipitated, Triton X-114-extracted protein (APTX) was used as a source of R. equi-derived surface proteins and included the virulence-associated protein called VapA. APTX was obtained from R. equi strain 103+ by Triton X-114 phase partitioning as described (19). The precipitate was resuspended in Tris-buffered saline (TBS) and stored frozen at −70°C. Protein content was measured using the BCA protein assay (Pierce, Rockford, IL).

Female C57BL/6 and C3H/HeJ mice (a LPS-hyporesponsive strain), 4–6 wk old were purchased from The Jackson Laboratory (Bar Harbor, ME). The MyD88 knockout mice (20) and the TLR2 knockout (21) were kindly provided by Dr. S. Akira (Osaka University, Osaka, Japan). TLR4-deficient mice were generated by Dr. Akira (22) and generously provided by Dr. R. Medzhitov (Yale University School of Medicine, New haven, CT). For infections, mice were infected i.v. with 1–3 × 10⁵ R. equi. At 2 or 3 days and at 8 days postinfection, mice were euthanized, and the number of bacterial colonies in the liver, lung, and spleens were determined as previously described (17).

Murine bone marrow-derived macrophages (BMMφ) were established as previously described (23). Briefly, bone marrow was flushed from the femurs of mice with cold Dulbecco’s PBS (PD) using a 23-gauge needle and collected by centrifugation. Bone marrow cells were resuspended in DMEM (Mediatech) supplemented with 10% heat-inactivated (HI)-FCS, 2 mM glutamine, 100 μg/ml streptomycin, and 100 U/ml penicillin G, and 20% L-929-conditioned medium as a source of M-CSF. Cells were incubated for 5–7 days at 37°C in 5% CO₂ in plastic petri dishes. BMMφ were removed from plates by incubation in 5 mM EDTA at 37°C for 10 min. The murine macrophage-like cell line RAW264.7 was obtained from the American Type Culture Collection (ATCC) (Manassas, VA) and maintained in RPMI 1640 (Mediatech) containing 10% HI-FCS, 2 mM glutamine, 100 μg/ml streptomycin, and 100 U/ml penicillin G (R-10). RAW267.4 cells were seeded onto tissue culture plates at a density of 2–5 × 10⁵/ml and split biweekly.

Bone marrow-derived dendritic cells were prepared as described by Lutz et al. (24), with minor modifications. Briefly, bone marrow was flushed from the femurs and tibias from mice and plated in 100-mm petri dishes at a concentration of 1 × 10⁶ cells/ml in RPMI supplemented with 10% FCS, penicillin-streptomycin, HEPES, glutamine, 50 μM 2-ME, and 35 μg/ml recombinant murine GM-CSF (R&D Systems, Minneapolis, MN). On days 3 and 6, one-half of the medium was removed and replaced with fresh medium. On day 9, nonadherent cells were removed and replated on 100-mm tissue culture-treated dishes overnight to further remove adherent cells. Nonadherent cells were removed from dishes and used for experiments. By flow cytometry, cells were generally 80% CD11c⁺ pure and had a CD11c⁺CD11b⁺CD8α⁻ phenotype, with moderate expression of MHC class II.

The CHO-K1 fibroblast-derived reporter cell lines, CHO/CD14.ELAM.Tac (clone 3E10), and CHO/CD14mut/TLR2.ELAM.tac (clone 7.19/TLR2), were engineered as described (25, 26). The LPS-responsive 3E10 reporter cell line expresses an inducible CD25 (the α-chain of the IL-2 receptor) under control of a portion of the human endothelial leukocyte adhesion molecule (ELAM) promoter containing NF-κB-binding sites. The LPS-nonresponsive mutant termed 7.19 was derived from the 3E10. This cell line contains a mutation in MD-2 (27), a molecule that associates with TLR4 and enables LPS responsiveness. Consequently, these cells neither translocate NF-κB nor express CD25 in response to endotoxin; however, they do express CD25 in response to other stimuli such as TNF or IL-1. Both the 3E10 and the 7.19 reporter cell lines used in this study stably express the human TLR2 receptor (26). All Chinese hamster ovary (CHO) reporter cell lines were grown in Ham’s F-12 medium (Invitrogen Life Technologies, Carlsbad, CA) supplemented with 10% HI-FCS, 2 mM glutamine, 100 μg/ml streptomycin, 100 U/ml penicillin G, and 400 U/ml hygromycin B (Invitrogen Life Technologies). TLR2-expressing cell lines were grown in the presence of 0.5 mg/ml geneticin (Invitrogen Life Technologies). Cell lines were maintained at 37°C with 5% CO₂ and were removed from plates by adding a trypsin (0.05%)-EDTA (0.2 mM) solution (Invitrogen Life Technologies) for 10 min at 37°C.

Approximately 2.5–5 × 10⁵ BMMφ were added per well in 24-well plates and primed with 100 U/ml IFN-γ for 15 h. Macrophages were stimulated with R. equi APTX (1 μg/ml) or E. coli LPS (100 ng/ml) or infected with serum-opsonized R. equi (multiplicity of infection (MOI), 5–40). Unbound bacteria were washed from wells 1 h later, and cell culture supernatants were harvested at 24 h postinfection. Levels of secreted IL-12 p40 and TNF protein concentrations in appropriately diluted cell supernatants were determined by ELISA according to protocols provided by BD PharMingen (San Diego, CA). Capture and detection (biotinylated) Ab pairs (IL-12 p40, C15.6 and C17.8; TNF-α, G281-2626 and MP6-XT3) and recombinant cytokine protein standards were purchased from BD PharMingen. Streptavidin-alkaline phosphatase and p-nitrophenyl phosphate substrate were purchased from Southern Biotechnology Associates (Birmingham, AL). To determine NO production by macrophages, Griess reagent was added 1:1 (v/v) to culture supernatants, and nitrite content was calculated by comparison with a sodium nitrite standard curve by measuring the OD₅₅₀ as previously described (28).

To determine NF-κB-dependent luciferase activity, RAW264.7 cells were transiently transfected with a plasmid encoding the luciferase gene downstream of five tandem repeats of the NF-κB binding site derived from the IL-6 promoter (Stratagene, La Jolla, CA). Constructs encoding dominant negative forms of TLR2 and TLR4 were previously described (29). Transient transfections were conducted by electroporation as previously described (30). Briefly, RAW264.7 cells were resuspended in serum-free RPMI 1640 at 2.5 × 10⁷/ml in 400 μl of cell suspension and added to 0.45-cm Gene Pulser II electroporation cuvettes (Bio-Rad, Hercules, CA) containing 50 μg of total DNA, (2 μg of the NF-κB-luciferase plasmid or 5 μg of the TNF-luciferase plasmid, with 3 μg of CMV-β-galactosidase, and pBluescript as nonspecific, carrier DNA). Cells were electroporated in a Gene Pulser II electroporator (Bio-Rad) at 320 V and 975 μF. Cells transfected with identical plasmids were pooled to prevent experimental differences caused by transfection efficiency variation in parallel cuvettes. Pooled cells (2 ml) were added to wells of a 24-well plate and allowed to recover overnight at 37°C in 5% CO₂.

For luciferase assays, transfected cells were pretreated with 1.2% DMSO for 24 h and IFN-γ for 15 h before the addition of stimuli. Cells were harvested 5–7 h after stimulation and resuspended in 100 μl of lysis buffer (125 mM Tris (pH 8.0), 10 mM DTT, 10 mM 1,2-cyclohexylenedinitrilo)tetraacetic acid, 50% glycerol, and 5% Triton X-100). Cellular debris was removed by centrifugation, and 20 μl of cell lysate was transferred to a new tube containing 100 μl of luciferase buffer (20 mM tricine, 1.07 mM MgCO₃, 2.67 mM MgSO₄, 0.1 mM EDTA. 33.3 mM DTT, 270 μM coenzyme A, 530 μM ATP, 470 μM d-luciferine potassium salt) immediately before measuring relative luciferase units in a FB12 tube luminometer (Zylux, Maryville, TN).

Intracellular staining for the p65 subunit of NF-κB was conducted 45 min after unprimed BMMφ adhered to glass coverslips were stimulated with LPS (100 ng/ml) or R. equi APTX (1 μg/ml) or infected with 5-chloromethylfluorescein diacetate-labeled (Molecular Probes, Eugene, OR), serum-opsonized R. equi (MOI 40) bacteria. Bacteria were labeled with 25 μM CellTracker Green (Molecular Probes) for 45 min at 37°C followed by washing and an additional 30 min of incubation. Macrophage monolayers were washed three times with warm PBS and fixed in cold methanol for 20 min at 4°C. Monolayers were incubated in blocking buffer containing 1% normal goat serum in PBS for 15 min, before being stained with a mouse monoclonal anti-NF-κB p65 (F-6) Ab (Santa Cruz Biotechnology, Santa Cruz, CA). F-6 was used at a final concentration of 1 μg/ml in blocking buffer. A 1/100 dilution of FITC-goat anti-mouse IgG Ab (Jackson ImmunoResearch Laboratories, West Grove, PA), was added for 1 h at room temperature with gentle agitation. Macrophages were counterstained for 15 min with 0.008% Evans blue (Sigma-Aldrich). Coverslips were mounted on slides using SlowFade antifade reagent (Molecular Probes) and visualized at either ×400 or ×630 (oil immersion) magnification using a Zeiss Axioplan 2 fluorescent-imaging research microscope and Zeiss KS300 imaging software.

Macrophage nuclear extraction was conducted as described previously (28). Approximately 5 × 10⁶ BMMφ in 60-mm petri dishes were infected with complement-opsonized R. equi (MOI 25). At 0, 15, 30, 45, 60, or 75 min postinfection, macrophages were washed with warm TBS, harvested by scraping, and resuspended in 400 μl of ice-cold buffer containing 10 mM HEPES (pH 7.9), 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, and 0.5 mM PMSF. Cells were lysed by the addition of 0.6% Nonidet P-40 followed by vigorous vortexing. Intact nuclei were pelleted and the supernatants (cytoplasmic extract) were transferred to new tubes and frozen at −80°C. Nuclei were washed before being lysed by vigorous shaking for 15 min at 4°C in 50 μl of ice-cold buffer containing 20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and 1 mM PMSF. Nuclear debris was pelleted by cold centrifugation, and the supernatant (nuclear extract) was used immediately or frozen at −80°C.

Cytoplasmic and nuclear extracted proteins were separated using the Laemmli system of SDS-PAGE (31) as described previously (32). Proteins were electroblotted to polyvinylidine difluoride membranes (Millipore, Bedford, MA) which were blocked with 5% milk in TBS containing 0.05% Tween 20 (TBS-T; Sigma-Aldrich). The following primary Abs were purchased from Santa Cruz Biotechnology and used at a concentration of 1 μg/ml diluted in TBS-T, 1% milk: mouse monoclonal anti-NF-κB p65 (F-6); mouse monoclonal anti-c-Rel (B-6); and rabbit polyclonal anti-IκB-α (C-21). After extensive washing, membranes were incubated with HRP-conjugated secondary Abs diluted 1/4000 in TBS-T, 5% milk. Membranes were then washed thoroughly with TBS-T followed by TBS before the addition of a chemiluminescent substrate (Amersham Pharmacia Biotech, Piscataway, NJ) and exposure to film.

Flow cytometric analysis of CHO reporter cell lines was conducted as previously described (25, 26). CHO reporter cell lines were plated at a density of 2.5–5 × 10⁵ cells/well in six-well plates. The following day, cells were washed and stimulated with R. equi at an MOI of 40:1 or LPS at a concentration of 200 ng/ml, or IL-1 at a concentration of 5 ng/ml, or R. equi APTX at a concentration of 1 μg/ml. After 20 h of stimulation, monolayers were washed three times with warm PD, harvested with trypsin-EDTA, and labeled with FITC anti-CD25 in PD containing 1% HI-FCS for 45 min on ice. After labeling, cells were washed twice with PD, 1% HI-FCS and fixed with 1% paraformaldehyde for 30 min on ice.

Surface expression of costimulatory molecules on DC was also measured by flow cytometry. Bone marrow derived DC were stimulated with 1 μg/ml LPS plus 5 μg/ml CpG oligonucleotides or with 3 μg/ml rVapA. The CpG oligonucleotides were generously provided by Dr. R. Seder (National Institutes of Health, Bethesda, MD) (33). Parallel populations of resting or stimulated DC were stained in separate tubes with an FITC-conjugated rat Ab to murine CD86 (GL1), a PE-conjugated rat Ab to CD40 (3/23), a PE-conjugated hamster Ab to CD11c (HL3), and an FITC-conjugated mouse Ab to IA^D (Ams-32.1). All primary Ab staining was performed in the presence of a rat Ab to CD16/CD32 (F_cBlock). All Abs were purchased from BD Biosciences PharMingen. After fixation, cells were washed and analyzed by flow cytometry using a FACSCalibur benchtop flow cytometer and CellQuest software (BD Biosciences, San Jose, CA).

The vapA gene, without its first 93 bases (encoding the signal peptide) from R. equi 103+ was amplified by PCR using primers VapA3 (5′-GTCGCTAATGGATCCGTTCTTGATT) and VapA3c (5′-GTGCAGCAAGCTTGGCGTTGTGCC) to obtain a product of 516 bp. The amplicon was digested with BamHI and HindIII and cloned into similarly digested pQE30 vector (Qiagen, Chatsworth, CA) to obtain pSJ11. E. coli M15 transformed with pSJ11 were grown in Luria Bertani medium containing carbenicillin (100 μg/ml) to an OD₆₀₀ of 0.7–0.8 at 37°C. Expression of vapA was induced by addition of IPTG to a final concentration of 1 mM, and the incubation was continued for an additional 2.5 h. One gram of cell pellet was suspended in 2 ml of lysis buffer, consisting of 50 mM phosphate buffer (pH 7), 300 mM NaCl, 10 mM imidazole, followed by addition of lysozyme to 1 mg/ml. The cell suspension was incubated on ice for 1 h followed by sonication (3 pulses of 30 s each). After clarification, the cell-free extract was loaded on a 0.5-ml Ni-NTA column (Qiagen) that had been pre-equilibrated with 10 column volumes (CV) of the lysis buffer. The column was washed with 5 CV each of three wash buffers containing 20, 40, and 60 mM concentrations of imidazole. Bound protein was eluted from the column using 10 CV of elution buffer (50 mM phosphate buffer (pH 7), 300 mM NaCl, 500 mM imidazole). The various fractions were analyzed by 15% SDS-PAGE, and those containing the purified VapA were pooled and concentrated on Vivaspin 20 (Vivasciences, Edgewood, NY) columns with a molecular mass cutoff of 10 kDa. The protein was stored in 50% glycerol at −80°C. The amount of LPS in the rVapA preparation was determined to be 100 pg/ml, as determined by the Limulus amebocyte assay. We have previously determined (34) that this amount of LPS was insufficient to induce cytokines from macrophages.

The NF-κB family of transcription factors is involved in the regulation of numerous components of the innate immune response. In the resting state, NF-κB is sequestered in the cytoplasm bound to its inhibitor, IκB-α. IκB-α is degraded on stimulation, allowing NF-κB to translocate into the nucleus to activate gene transcription (35). The translocation of NF-κB from the cytoplasm into the nucleus of macrophages after R. equi infection was determined by immunofluorescence (Fig. 1) or Western blotting (Fig. 2). For immunofluorescence, the p65 component of NF-κB was stained with an FITC-conjugated (green) mAb to p65, and macrophages were counterstained in red. In untreated macrophages, NF-κB remained sequestered in the cytoplasm where it appears greenish yellow, whereas the nuclei of these cells remain dark red (Fig. 1,A). At 45 min postinfection with R. equi (Fig. 1,B), virtually all of the p65 had translocated to the nuclei as indicated by the bright green nuclear staining within a dark red cytoplasm. Stimulation of cells with LPS, as a positive control (Fig. 1 C) resulted in a similar degree of NF-κB translocation. Identical results were observed with several other strains of R. equi, including strains that contained or lacked the virulence-associated plasmid (data not shown). All R. equi strains examined induced a rapid and substantial translocation of NF-κB into the nuclei of macrophages during the process of infection.

Nuclear translocation of NF-κB was confirmed by Western blotting. Nuclear and cytoplasmic fractions were extracted from R. equi-infected macrophages and immunoblotted for NF-κB and IκB-α, respectively. The NF-κB family members p65 and c-Rel were readily detectable in the nuclear extracts of macrophage-infected R. equi or stimulated with LPS (Fig. 2,A). Infection of macrophages with R. equi caused a time-dependent decrease in the amount of IκB-α detectable in the cytoplasmic extract and a reciprocal increase in the amount of NF-κB p65 detectable in the nuclear extract (Fig. 2,B). To show that the translocated NF-κB was transcriptionally active, RAW264.7 cells were transiently transfected with a luciferase reporter gene driven by a promoter containing tandem repeats of an NF-κB-binding consensus sequence. Transfected cells were infected with R. equi and luciferase activity was measured (Fig. 2 C). R. equi infection stimulated levels of NF-κB-dependent luciferase activity comparable with that seen with LPS stimulation.

We evaluated innate immune responses to R. equi infection in wild-type macrophages and in macrophages from mice lacking MyD88^−/− (Fig. 3). MyD88 is a critical adaptor protein that couples TLR ligation to the activation of NF-κB (36). Recent studies have show TLR2 signaling to be dependent on MyD88, whereas some components of TLR4 signaling can be compensated for by other non-MyD88 adaptor molecules (8 , 37). In wild-type macrophages (Fig. 3, ▪), R. equi infection or stimulation with an acetone-precipitated extract from the surface of R. equi, designated APTX, which contains the surface-expressed VapA molecule, resulted in the rapid and efficient production of TNF (Fig. 3,A), IL-12 p40 (Fig. 3,B), and NO (Fig. 3,C). Similar cytokine responses also occurred in murine alveolar macrophages infected with R. equi (data not shown). In MyD88^−/− cells, however, R. equi infection failed to induce the production of any of these mediators (Fig. 3, □). Similarly, R. equi infection of macrophages from MyD88^−/− mice failed to induce the nuclear translocation of NF-κB (Fig. 3,D). MyD88^−/− macrophages retained some responsiveness to LPS. They translocated NF-κB in response to LPS (Fig. 3,D), as previously described (20). They also unexpectedly produced IL-12 (p40) (Fig. 3 B). These cells were primed overnight with IFN-γ and then stimulated with 100 ng/ml LPS for 24 h before cytokine measurement.

We next evaluated the innate immune response to R. equi infection by C3H/HeJ macrophages (Fig. 4). C3H/HeJ mice carry a mutation in TLR4 making them hyporesponsive to LPS (38). As expected, LPS stimulation, as a control, failed to elicit an inflammatory response from C3H/HeJ macrophages (Fig. 4, right). In contrast, R. equi invoked robust dose-dependent production of TNF (Fig. 4,A), IL-12 (Fig. 4,B), and NO (Fig. 4,C) from these macrophages. Additionally, although C3H/HeJ macrophages were refractory to NF-κB activation upon LPS stimulation (Fig. 4,D, right) they were fully capable of translocating NF-κB p65 in response to R. equi infection (Fig. 4 D, left). Thus, the response of C3H/HeJ macrophages to R. equi infection is similar to wild-type macrophages, indicating that the loss of TLR4 function does not abolish the ability of the macrophage to respond to R. equi.

We next examined TLR2 signaling in response to R. equi. Bacteria or bacterial extracts were added to CHO reporter cell lines that express CD25 under control of the NF-κB-dependent E-selectin promoter (25). The expression of surface CD25 was measured by flow cytometry as described (26). The parental cell line, 3E10, which lacks TLR2, failed to respond to R. equi infection by up-regulating CD25 (Fig. 5,A). LPS stimulation of these cells led to an increase in CD25 expression (Fig. 5,A) as a result of the endogenous hamster TLR4 (26). NF-κB-dependent CD25 expression in TLR2-transfected reporter cells (3E10/TLR2) was up-regulated 8- to 10-fold over untreated cells after infection with R. equi or stimulation with APTX (Fig. 5,B). Similar increases in the levels of CD25 expression were observed in 7.19/TLR2 cells (Fig. 5 C). The 7.19 cell line carries a mutation in MD-2 (27) rendering them hyporesponsive to LPS while remaining fully responsive to IL-1. Thus, the expression of CD25 on 7.19/TLR2 cells in response to R. equi is not due to contamination of LPS in our preparations.

To substantiate a role for TLR2 in response to R. equi, RAW-264 cells were transfected with constructs encoding TLRs with alterations in their C termini preventing them from transducing signals and causing them to act as dominant negatives (29). These cells were infected with R. equi or stimulated with APTX. Seven hours later, luciferase driven by the ELAM promoter (ELAM-Luc) was analyzed. RAW-264 cells transfected with the ELAM-Luc plasmid alone were used as the positive controls. These control cells responded well to all of the agonists used (Fig. 6, □). For these studies, LPS and AraLAM were used as known agonists for TLR4 and 2, respectively. Cells transfected with the TLR2-DN (Fig. 6, ▪) exhibited a reduced response to arabinofuranosyl-terminated lipoarabinomannan (AraLAM), and also to two clinical strains of R. equi, designated 238 and 103. Similarly, stimulation with APTX was significantly reduced by the TLR2 dominant negative form (TLR2-DN). Cells transfected with the TLR4-DN alone (Fig. 6, ▦) responded normally to R. equi. Furthermore, luciferase production by cells expressing both TLR2- and TLR4-DN (Fig. 6, ▨) was comparable with that by cells transfected with TLR2-DN alone. Taken together, these results indicate a major role for TLR2, but not TLR4, in cellular responses to R. equi.

To determine whether TLR2 was the primary receptor involved in R. equi responsiveness, cytokine production by IFN-γ-primed macrophages from TLR2^−/− mice was measured and compared with parallel monolayers of primed macrophages from wild-type mice. Macrophages from TLR2^−/− mice failed to make significant amounts TNF, even in response to high MOIs of R. equi (Fig. 7,A, □). The control for these studies was AraLAM, a known TLR2 agonist, which also failed to induce TNF production in the TLR2^−/− macrophages. The TLR2-deficient macrophages also made substantially reduced amounts of IL-12 and NO in response to R. equi infection following IFN-γ priming (Fig. 7, B and C). Priming with IFN-γ induced a small amount of IL-12 and NO production in response to AraLAM, which was comparable with that produced in response to low and intermediate MOIs of R. equi. At high MOIs, however, the levels of IL-12 and NO made in response to R. equi infection were slightly higher than the levels induced by AraLAM. These levels of cytokine production were substantially less than that produced by wild-type macrophages, suggesting that TLR2 is the major, although not exclusive TLR mediating macrophage inflammatory responses to R. equi.

The lipoprotein VapA is a major virulence factor located on the surface of R. equi (5, 39). Deletion of vapA from bacteria renders them avirulent, and complementation of these organisms with vapA restores virulence (5). To determine whether VapA was a TLR agonist, purified rVapA was added to macrophages and dendritic cells. Dendritic cells responded to rVapA by up-regulating the expression of surface costimulatory molecules, CD86 and CD40 (Fig. 8). The extent of up-regulation was comparable with that observed on DC stimulated with LPS-CpG (Fig. 8). Macrophages also responded to purified VapA by the production of inflammatory cytokines and NO (Fig. 9,A). Finally, rVapA induced the translocation of NF-κB in macrophages (Fig. 9 B). Thus, VapA, one of the molecules required for R. equi pathogenesis, can directly activate mammalian TLRs.

The rVapA that was used for these studies contains low levels (100 pg/ml) of LPS. To show that the cellular responses to rVapA were not due to LPS contamination, rVapA was used to stimulate DC from the LPS-hyporesponsive C3H/HeJ mice (Fig. 10,A). These cells responded to rVapA by up-regulating CD40 (Fig. 10,A) and CD86 (data not shown). Furthermore, the LPS nonresponder, TLR2 reporter cell line (7.19/TLR2) up-regulated CD25 in response to rVapA (Fig. 10,B), and the degree of CD25 up-regulation by rVapA was comparable with the positive control, IL-1 (Fig. 10,B). These cells failed to substantially induce CD25 up-regulation in response to LPS, confirming their LPS hyporesponsiveness (Fig. 10,B). Finally, we treated rVapA with polymyxin B before its addition to a TLR4/TLR2 reporter cell line. Polymyxin B only minimally decreased cellular responsiveness to rVapA (Fig. 10,C); however, it was quite effective at eliminating the TLR-4-dependent response of these cells to relatively high concentrations of LPS (Fig. 10 C). Thus, VapA is sufficient to induce TLR2 activation, and it does so in a TLR4/LPS-independent manner.

Our in vitro data predicted a role for TLR2 in initiation of innate immune responses to R. equi. To address the requirement of an intact Toll signaling pathway in R. equi infection, we used a murine challenge model and i.v. infected wild-type C57BL/6, TLR4^−/−, and TLR2^−/− mice with virulent R. equi and followed bacterial clearance over time (Fig. 11). This murine model of infection has previously been used as an accurate measure of R. equi virulence (5), although immunocompetent mice infected with sublethal doses of R. equi typically do not develop overt rhodococcal pneumonia. The bacteria will replicate in the short term, increasing by ∼10-fold during the first 2 days of infection, after which time the bacterial burdens plateau (40). In normal mice, bacterial clearance is initiated at approximately day 5 postinfection, and then bacterial numbers quickly decline (41, 42). In our experiments, mice were infected with ∼2 × 10⁵ CFU of R. equi, seeding the spleen with ∼10⁴ CFU (data not shown). At 2 days postinfection, bacterial numbers had increased equally (∼10-fold) in the spleens of both C57BL/6 and TLR4-deficient mice. On day 8, clearance was evident as the bacterial burdens had comparably decreased by 2 logs in both C57BL/6 and the TLR4^−/− mice (Fig. 11,A). In contrast, clearance was not observed in mice deficient in TLR2 (Fig. 11,B). At 8 days postchallenge, the number of bacteria in the spleen of the TLT2^−/− mice was equivalent to that observed at 3 days postchallenge and was 2 logs higher than that of the similarly infected C57BL/6 control mice (Fig. 11 B). Thus, in the absence of TLR2, but not TLR4, in vivo clearance of R. equi is compromised.

Innate recognition of pathogens by phagocytic cells is critical to the initiation of adaptive immune responses to infectious microorganisms (10). Phagocytes recognize a large number of diverse pathogens without the luxury of an enormous repertoire of immune receptors. The discovery of TLRs has elucidated the mechanisms by which microbes or microbial products can induce innate immunity leading to the activation of cytokine production and costimulatory molecule expression. In the present work, we demonstrated that infection of macrophages with R. equi induces the rapid translocation of NF-κB, resulting in the production of high levels of inflammatory cytokines. Cytokine production was comparable to or in excess of that induced by bacterial LPS. The production of cytokines by macrophages after infection by R. equi confirms and extends previous studies by Giguere and Prescott (43), who demonstrated that R. equi infection of macrophages resulted in cytokine synthesis and secretion.

The present studies sought to identify the mechanism whereby R. equi induced macrophage cytokine production. The lack of cytokine production by infected macrophages from MyD88^−/− mice established a role for the TLR pathway in this event. The use of C3H/HeJ mice demonstrated that mice lacking TLR4 were fully capable of responding the R. equi. Furthermore, cells transfected with a dominant negative form of TLR4 induced cytokines as well as nontransfected cells. Thus, TLR4 was not a primary receptor involved in macrophage responses to R. equi. Several experimental approaches were taken to implicate TLR2 in R. equi responses. First, CHO cells expressing human TLR2 became activated in response to R. equi or an acetone precipitate containing surface proteins of R. equi (APTX). Secondly, cytokine responses to R. equi were strongly diminished in macrophages from mice lacking TLR2. Finally, the expression of dominant negative forms of TLR2 inhibited the response of RAW-264 macrophage-like cells to R. equi or APTX. Taken together, these observations indicate that TLR2 is the main receptor mediating the innate response of mammalian cells to R. equi. Some modest cytokine responses to R. equi were retained in cells lacking TLR2. Whereas TNF production was virtually ablated in macrophages from TLR2^−/− mice, other cytokines, such as IL-12 and NO, were produced in reduced amounts after infection with high MOI of R. equi, suggesting that other TLRs may play minor roles in triggering macrophage responses to R. equi. Consistent with the marked reductions in in vitro macrophage cytokine production, TLR2-deficient mice failed to clear an in vivo challenge of R. equi as efficiently as wild-type C57BL/6 mice.

R. equi disease frequently occurs in neonatal hosts (foals) that may have immature immune function. It is known that human newborns frequently do not mount efficient Th1 responses (44, 45), and it was recently shown that TLR-induced IL-12 responses of cord blood cells were diminished as compared with that of adult blood cells (46). Perhaps inefficient TLR2 signaling contributes to the enhanced susceptibility of neonatal horses to R. equi.

To begin to address the bacterial molecules responsible for TLR2 activation, VapA was examined. This molecule was selected because it is expressed on the surface of bacteria, it is a bacterial lipoprotein, and to date it is the only virulence factor to be definitively identified in this organism (5). Bacteria lacking VapA fail to grow in macrophages and the complementation of deficient bacteria with the vapA gene restores virulence (5). rVapA induced a rapid translocation of NF-κB in macrophages and induced inflammatory mediator production. This molecule also activated a TLR2 reporter cell line that had a mutation in MD-2 causing it to be nonresponsive to LPS, and it also activated dendritic cells from C3H/HeJ mice. These data indicate that VapA is sufficient and capable of activating macrophage cytokine responses via TLR2. Previous studies (47, 48) have indicated that lipidation of bacterial proteins was required for TLR2 activation. Studies are under way to determine the nature and extend of lipidation of the rVapA used in these studies. To determine whether VapA was required for cytokine induction, bacteria lacking vapA (5) or the entire virulence plasmid upon which the vapA gene resides were compared with wild-type bacteria. All strains activated macrophage cytokine production to equivalent degrees (data not shown). Thus, although VapA is capable of inducing cytokine production from macrophages, the presence of VapA on R. equi is not necessary for TLR-mediated activation of innate immunity by R. equi.

The observation that inflammatory cytokine production follows TLR2 stimulation by R. equi implies that a way to limit the pathology associated with infection by this organism may be to specifically block the TLR2 pathway. Future studies will examine disease progression, adaptive immunity, and bacterial replication in mice lacking key components of the TLR pathway. The goal of these future studies is to determine the extent to which innate immune responses to this organism contribute to adaptive immunity, and whether they are linked to bacterial-induced tissue pathology.