ABSTRACT
Rhodococcus equi is an intracellular pathogen which causes pneumonia in young horses and in immunocompromised humans. R. equi arrests phagosome maturation in macrophages at a prephagolysosome stage and grows inside a privileged compartment. Here, we show that, in murine macrophages activated with gamma interferon and lipopolysaccharide, R. equi does not multiply but stays viable for at least 24 h. Whereas infection control of other intracellular pathogens by activated macrophages is executed by enhanced phagosome acidification or phagolysosome formation, by autophagy or by the interferon-inducible GTPase Irgm1, none of these mechanisms seems to control R. equi infection. Growth control by macrophage activation is fully mimicked by treatment of resting macrophages with nitric oxide donors, and inhibition of bacterial multiplication by either activation or nitric oxide donors is annihilated by cotreatment of infected macrophages with ferrous sulfate. Transcriptional analysis of the R. equi iron-regulated gene iupT demonstrates that intracellular R. equi encounters iron stress in activated, but not in resting, macrophages and that this stress is relieved by extracellular addition of ferrous sulfate. Our results suggest that nitric oxide is central to the restriction of bacterial access to iron in activated macrophages.
INTRODUCTION
Rhodococcus equi is a soil organism which belongs to the group of mycolic acid-producing actinomycetes (42) and is closely related to Mycobacterium tuberculosis. The facultative intracellular pathogen R. equi can cause pneumonia in young horses and immunocompromised humans when inhaled with contaminated dust (42) and primarily infects phagocytic monocytes and macrophages (15). Newly phagocytosed bacteria reside in a macrophage compartment which passes normally through the early phase of phagosome maturation and which is arrested in between an early and a late maturation stage. The resulting compartment, the R. equi-containing vacuole (RCV), is characterized by some markers of late phagosomes, such as Ras-like protein from rat brain (Rab) 7, bis(monoacylglycerol) phosphate (BMP), and lysosome-associated membrane proteins (LAMP) 1 and 2. However, the RCV acquires neither hydrolytic enzymes nor proton-pumping vacuolar ATPase (v-ATPase), both of which are characteristic for late phagosomes, nor does it acidify or fuse with lysosomes. The bacteria multiply in this unusual vacuole, and the host cell is eventually lysed by necrosis (10, 22). R. equi virulence largely depends on virulence-associated plasmids which may also determine host specificity (38). Virulence for foals requires a plasmid encoding the central virulence factor, virulence-associated protein A (VapA) (12, 17).
Whereas a resting macrophage is susceptible to infection with different intracellular pathogens, its antimicrobial activity is strongly enhanced by activation through proinflammatory host signals, e.g., interferons (IFN), or by various pathogen-associated molecular patterns, e.g., lipopolysaccharides (LPS) (35). Accordingly, R. equi cells multiply in resting macrophages, yet their multiplication is completely inhibited when macrophages are activated before infection (5). Activation induces a dramatic change in gene expression and increases microbicidal and antigen presentation capacities of the phagocyte (48).
Mechanisms which execute the killing functions in activated macrophages include enhanced phagolysosome formation, induction of autophagy, or production of toxic radicals. Vacuoles containing pathogenic mycobacteria (47, 50, 58), Legionella pneumophila (45), or Coxiella burnetii (11) are forced back into the normal degradation pathway when macrophages are activated before infection. Proteins that may mediate IFN-γ–LPS-induced enhanced phagosome maturation include the IFN-γ-induced immunity-related GTPases (IRG). One of them, Irgm1 (formerly known as LRG-47) (56), is a central player for controlling M. tuberculosis infections in mice. Irgm1−/− mice are unable to control M. tuberculosis replication. Bone marrow-derived macrophages (BMMs) from these animals are partly permissive for M. tuberculosis multiplication even when activated, and the bacteria are less frequently found in phagolysosomes (26). Irgm1 is also involved in infection control via autophagy (13, 51). Induction of autophagy in macrophages leads to increased formation of M. tuberculosis-containing phagolysosomes and to killing of the bacteria (13).
Production of nitric oxide by activated macrophages is crucial for infection control of Leishmania and Burkholderia mallei (18, 29), whereas killing of Listeria monocytogenes largely depends on superoxide (39). Both compounds seem to be important for elimination of Francisella tularensis (23). Likewise, control of R. equi infection by activated macrophages has been ascribed to killing of bacteria by peroxynitrite, a strongly bactericidal compound that results from reaction of superoxide with nitric oxide radicals (5).
Here, we analyzed the compartmentation and survival of R. equi in activated macrophages and the mechanisms behind intracellular growth restriction of R. equi by macrophage activation.
MATERIALS AND METHODS
Cultivation of macrophages and bacteria.J774E macrophage-like cells were cultivated in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum and 1% GlutaMax (all from Invitrogen, Karlsruhe, Germany), designated “DMEM complete” here. The knockout and transgenic mice used, including Irgm1−/−, GFP-LC3 (where GFP is green fluorescent protein), and gp91phox−/− mice, have been described (3, 34, 41). Bone marrow-derived macrophages (BMMs) were isolated and cultivated as described previously (10). All experiments involving mice were performed according to the German law for animal protection. The procedures were additionally approved by the local government according to the laws of the State of North Rhine-Westfalia, reference number 8.87-50.10.45.08.219.
If not stated otherwise, macrophages were activated by treatment with culture medium containing 500 U/ml recombinant IFN-γ (Tebu-bio, Offenbach, Germany) overnight and additionally with 250 ng/ml lipopolysaccharide (LPS; from Salmonella enterica serotype Typhimurium; Sigma-Aldrich, Taufkirchen, Germany) 2 h before infection. IFN-γ and LPS were left on activated macrophages the entire experiment. Rhodococcus equi 103+ (where “+” denotes the presence of a virulence-associated plasmid) has been isolated from a foal with R. equi pneumonia (6). The isogenic but avirulent strain R. equi 103− has been cured from the virulence-associated plasmid (10). Strains will be denoted as “103+” and “103−” here. Bacteria were grown on brain heart infusion (BHI; BD Diagnostic Systems, Sparks, MD) agar plates at 30°C for routine cultivation. Before infection experiments, bacteria were grown in BHI broth at 37°C overnight at 200 rpm on a rotary shaker to induce virulence gene expression.
Live-cell determination of intracellular R. equi.For determination of intracellular survival of R. equi, J774E cells were seeded into 24-well plates at 4 × 104 cells per well 2 days before infection with R. equi at a multiplicity of infection (MOI) of 0.25 for 30 min at 37°C. Samples were rinsed with phosphate-buffered saline (PBS) twice, and fresh DMEM complete containing 10 μg/ml gentamicin was added. In samples for 48 h of incubation, cells were supplied with fresh medium at 24 h. At the times indicated, medium was removed and macrophages were lysed in 1 ml PBS-0.1% (vol/vol) Triton X-100. Serial dilutions were plated onto 0.5× LB (Lennox) agar. Plates were incubated at 30°C for 30 h and CFU were counted. Numbers of less than 10 were neglected for analysis as long as they were not the only colonies grown in this dilution series. The mean number of CFU for each sample and time was determined and expressed as a percentage of the 0 h values.
Quantification of fusion of phagosomes with lysosomal or endocytic compartments.J774E macrophages were seeded onto coverslips in a 24-well plate at 2 × 105 cells per well.
(i) Phagosome-lysosome fusion.Macrophages were incubated with DMEM complete containing 30 μg/ml ovalbumin TexasRed (Invitrogen) overnight and rinsed with warm PBS, and the fluorescent conjugate was chased into lysosomes by incubation in fresh medium for 2 h. Cells were infected with ATTO-488 (Atto-Tec, Siegen, Germany)-labeled live or heat-killed (15 min, 85°C) R. equi 103+ or 103− (aliquots of 2 × 108 bacteria, incubated in 50 μg/ml of the fluor in 0.1 M NaHCO3 for 45 min on ice, followed by one wash in 20 mM Tris/HCl [pH 8] and two washes in PBS) at an MOI of 40 for 20 min, rinsed with PBS twice, and incubated in fresh medium without bacteria. After 2 h and 5 h, macrophages were fixed with PBS-3% formaldehyde.
(ii) Phagosome-endosome fusion.After overnight incubation, macrophages were infected with ATTO-488-labeled R. equi 103+ or 103− or heat-killed (10 min, 70°C) Escherichia coli DH5α labeled as described above at an MOI of 40 for 15 min, rinsed with PBS twice, and incubated in fresh medium for 2 h. Medium was replaced by DMEM complete containing 150 μg/ml dextran Texas Red conjugate (Invitrogen). After 2 h or 4 h, cells were rinsed with PBS and fixed in PBS-3% formaldehyde.
All samples were mounted onto glass slides in Mowiol, and colocalization of bacteria with lysosomal or endocytosed material was quantified using a confocal laser scanning microscope, model LSM510 (Zeiss, Oberkochen, Germany).
Microscopic and fluorimetric determination of phagosome pH.For determination of phagosome colocalization with LysoTracker, J774E cells were seeded at 1.5 × 105 cells/well onto glass coverslips and incubated overnight. They were infected with R. equi 103+ at an MOI of 1 for 30 min, rinsed with PBS twice, and incubated in fresh DMEM complete containing 10 μg/ml gentamicin. After 1.5 h or 23.5 h, medium was replaced by DMEM complete containing 100 nM LysoTracker (Invitrogen) and 3.3 μM of the DNA stain Syto13 (Invitrogen). Cells were incubated for 30 min, rinsed with PBS, and mounted onto glass slides with 5 mg/ml low-melting-point agarose (Sigma-Aldrich) in PBS. They were analyzed immediately using a confocal laser scanning microscope. Quantitative ratiometric determination of phagosome pH was as described in reference 54. Briefly, R. equi was labeled with a pH-insensitive fluor, allowing for the quantification of ingested bacteria, and with a pH-sensitive fluor, allowing for the quantification of pH. J774E cells in 24-well plates were infected with labeled bacteria for 15 min and chased in fresh medium. For each time, fluorescence in two wells of infected J774E cells was determined in assay buffer in an FLX800 microplate reader (Bio-Tek Instruments, Neufahrn, Germany). After quantification of fluorescence of the last sample, 10 μM of the K+/H+ antiporter nigericin was added to equilibrate phagosome pH with that of the assay buffer as the internal control. The system was calibrated by incubating infected macrophages with K+-containing buffers of defined pH containing 10 μM nigericin. For pH determinations of phagosomes containing killed R. equi cells, bacteria were treated before infections with heat (15 min, 85°C), UV irradiation (1 h on ice), formaldehyde (3% in PBS, 1 h at 4°C), or gramicidin (300 μg/ml in PBS, 2 h at 37°C). Bacteria were fluorescently labeled after killing, and bacteria with formaldehyde treatment were fluorescently labeled before killing. Efficient killing was tested by plating samples on LB agar plates.
Transmission electron microscopy.To analyze the ultrastructural appearance of the RCV in resting and activated macrophages, 5 × 106 J774E cells were seeded into 53-mm cell culture dishes and incubated overnight. They were infected with R. equi 103+ at an MOI of 20 for 30 min, rinsed with PBS twice, and incubated in fresh medium containing 150 μg/ml gentamicin. After 1 h, medium was replaced by DMEM complete containing 10 μg/ml gentamicin.
After different times of incubation, cells were fixed in PBS containing 2% formaldehyde and 0.5% glutaraldehyde for 1 h and prepared for electron microscopy as described previously (10).
Microscopic analysis of intracellular multiplication.J774E (1.5 × 105 cells per well) or bone marrow-derived (1 × 105 cells per well) macrophages were seeded in a 24-well plate containing coverslips and incubated overnight. They were infected with R. equi 103+ at an MOI of 1 for 30 min, rinsed with PBS twice, and incubated in fresh medium containing 10 μg/ml gentamicin. Macrophages were treated with different buffers or chemicals before and/or after infection. Hanks' balanced salt solution (HBSS; 0.137 M NaCl, 5.4 mM KCl, 0.25 mM Na2HPO4, 0.44 mM KH2PO4, 1 mM MgSO4, 1.3 mM CaCl2, 4.2 mM NaHCO3, 5.6 mM glucose)-treated cells were rinsed with PBS twice and incubated with HBSS 2 h before and always after infection (then with 10 μg/ml gentamicin). 3-Methyladenine (3MA; Sigma-Aldrich; dissolved in medium with heating) was added at 10 mM for 2 h before and at 1 mM after infection; for analysis of intracellular multiplication of 103−, 3MA was added only after infection at 10 mM. Earle's balanced salt solution (EBSS; 117.2 mM NaCl, 5.3 mM KCl, 1.0 mM NaH2PO4, 1.0 mM MgSO4, 1.8 mM CaCl2, 5.6 mM glucose, 26.2 mM NaHCO3)-treated cells were rinsed with PBS twice and incubated with EBSS 1 h before and 7 h after infection (then with 10 μg/ml gentamicin), before the buffer was replaced by DMEM complete containing 10 μg/ml gentamicin. Rapamycin (200 nM; Merck, Darmstadt, Germany; dissolved in dimethyl sulfoxide [DMSO]) was added 1 h before and always after infection, and DMSO was added as the corresponding control. Macrophages pretreated with 100 μM S- nitroso-N-acetyl-d,l-penicillamine (SNAP; Sigma-Aldrich; dissolved in PBS-30 mM NaOH) were incubated with the compound in DMEM complete 24 h before infection, rinsed with PBS twice, and incubated in medium containing 10 μg/ml gentamicin without SNAP after infection. Alternatively, SNAP was added only after infection at 100 μM. The reactive oxygen species inhibitor N-acetyl-l-cysteine (Enzo Life Sciences, Lörrach, Germany; dissolved in sterile water) was added at 5 mM 2 h before and always after infection. Compounds that were added only after infection were 50 nM wortmannin (Merck; solved in DMSO), 500 μM NG-methyl-l-arginine (NMLA; Sigma-Aldrich; order number M7033; dissolved in PBS), 50 μM diethylenetriamine-nitric oxide adduct (DETA-NO; Sigma-Aldrich; order number D185; dissolved in PBS-10 mM NaOH), and different concentrations of FeSO4 (Roth, Karlsruhe, Germany; order number P015-1; dissolved in medium). At 2 h and 24 h of infection, samples were fixed with PBS-3% formaldehyde and stained with the DNA stain Syto13 in PBS at 1.67 μM for 30 min. Cells were rinsed with PBS four times and mounted onto slides in Mowiol for microscopic analysis. The numbers of intracellular bacteria were counted for at least 50 infected macrophages per sample. Since numbers of bacteria in large clusters cannot be properly quantified, samples were analyzed for robust bacterial multiplication by determining the percentage of macrophages containing more than 10 bacteria (5).
Control for effects of treatments interfering with autophagic pathways.GFP-LC3 bone marrow-derived macrophages were seeded at 1 × 105 cells per well and activated with IFN-γ–LPS where indicated. After overnight incubation, one sample of each resting or activated macrophage was left untreated. To investigate the effects of these treatments on LC3 localization, resting macrophages were rinsed twice with PBS and were incubated in the starvation buffers HBSS or EBSS or with 200 nM rapamycin in DMEM complete. Activated macrophages were incubated in 50 nM wortmannin or in 10 mM 3MA in DMEM complete. After 2 h of incubation, all samples were fixed with 3% formaldehyde in PBS, embedded in Mowiol, and investigated with a laser scanning microscope using the same microscope settings for all samples.
Quantification of nitric oxide.For analysis of nitric oxide production, J774E macrophages were seeded into 24-well plates at 8 × 104 cells per well 2 days before infection. Resting or activated macrophages were infected at an MOI of 1 for 1 h and chased in fresh medium containing 10 μg/ml gentamicin. At the times indicated, medium was transferred into tubes and centrifuged at 140 × g for 5 min. The supernatants were stored at −20°C until analysis for nitrite with the Griess reagent system (Promega, Madison, WI) by following the manufacturer's instructions. For determination of nitric oxide release from nitric oxide donors, these were dissolved as described above. SNAP was diluted in DMEM complete with or without 100 μM ferrous sulfate at 100 μM and incubated in 15-ml polypropylene tubes at 37°C. At different times, samples were taken and stored at −20°C until analysis with Griess reagent. DETA-NO was diluted in DMEM complete at 50, 100, or 200 μM and incubated and analyzed as described for SNAP. A concentration of 50 μM DETA-NO was chosen for infection experiments to achieve a comparable final concentration of nitrite as an indication for similar amounts of released nitric oxide.
Viability and VapA expression of SNAP-treated R. equi.R. equi was grown at 30°C overnight, pelleted, and resuspended in DMEM complete with 100 μM SNAP or solvent alone at an optical density at 600 nm (OD600) of 0.1. Bacteria were incubated in a rotary shaker at 200 rpm and 37°C, and OD600 was recorded over time. After 24 h of incubation, expression of VapA in samples was analyzed by immunoblotting (see Western blot analysis) after heating equal numbers of bacteria to 95°C for 10 min in 2× Laemmli buffer (5× stock contains 60 mM Tris-HCl [pH 6.8], 25% glycerol [vol/vol], 2% sodium dodecyl sulfate [wt/vol], 5% β-mercaptoethanol [vol/vol], 0.1% bromphenol blue [wt/vol]), followed by removal of debris by centrifugation. A 0.3 OD600 culture equivalent was applied to a 15% SDS-polyacrylamide gel for each sample. To determine the effect of nitric oxide on R. equi growth under conditions of limited iron concentrations, R. equi 103+ was grown in minimal medium [30 mM K2HPO4, 16.5 mM KH2PO4, 78 mM (NH4)2SO4, 0.85 mM sodium citrate, 1 mM MgSO4, 0.02% glycerol, 0.1 mM thiamine, 20 mM acetate, supplemented with 1× trace metal solution (60) adjusted to different concentrations of ferrous sulfate] containing 100 μM ferrous sulfate at 30°C overnight. Bacteria were pelleted, washed with PBS once, and used for inoculation of minimal media with different iron concentrations at an OD600 of 0.1. Suspensions were mixed with no or 100 μM SNAP. Optical density was determined before and after 24 h of incubation at 37°C in a rotary shaker.
Western blot analysis.Western blot analysis was as described in reference 10. Antibodies used were mouse anti-VapA (Mab10G5; 1:5.000; kindly supplied by Shinji Takai, Kitasato University, Aomori, Japan [55]) and Odyssey goat anti-mouse IRDye 800CW (1:10.000; Li-Cor, Bad Homburg, Germany). Western blots were analyzed with the Odyssey infrared imaging system (Li-Cor).
Determination of infection cytotoxicity.J774E macrophages were seeded into 24-well plates at 8 × 104 cells per well 2 days before infection. They were infected with R. equi 103+ at an MOI of 30 for 1 h. To remove extracellular bacteria, macrophages were rinsed with PBS twice and incubated in DMEM complete containing 150 μg/ml gentamicin for 1 h before replacing the medium with DMEM complete containing 10 μg/ml gentamicin. Where indicated, cells were additionally treated with a mix of antibiotics (100 μM each hygromycin B, penicillin G, and apramycin, all from Sigma-Aldrich) at different times of chase to kill intracellular bacteria. After 24 h of chase, an uninfected sample was lysed with 1% Triton X-100 (vol/vol). Release of cytosolic lactate dehydrogenase (LDH) activity from this sample was set as 100%, and activities in all samples were expressed relative to this. Sample supernatants were collected, centrifuged at 6,150 × g for 5 min, and analyzed for LDH release with the cytotoxicity detection kit (Roche Diagnostics, Mannheim, Germany) by following the instructions of the manufacturer.
Immunofluorescence microscopy.Fixed and quenched (PBS-50 mM NH4Cl for 30 min) macrophages on coverslips were incubated in blocking buffer (PBS containing 5% bovine serum albumin and 0.1% saponin, both from Sigma-Aldrich). Samples were sequentially stained with first and secondary antibodies in blocking buffer, embedded in Mowiol, and analyzed using the confocal laser scanning microscope LSM510 (Zeiss) and the same microscope settings for all samples in any one experiment. Antibodies used were goat anti-Irgm1 (1:50; Santa Cruz Biotechnology), mouse anti-R. equi 103− serum (1:60; produced against complete fixed R. equi 103−), donkey anti-mouse-Cy3 (1:50; Dianova, Hamburg, Germany), and donkey anti-goat-Cy3 (1:50; Dianova).
RNA isolation from intramacrophage bacteria.R. equi was grown in 10-ml BHI broth portions at 37°C for 16 h at 200 rpm. Cultivation at iron-limited conditions was in minimal medium with acetate containing 2,2-dipyridyl (Sigma-Aldrich, Munich, Germany). Concentrations for activation with IFN-γ and LPS overnight were as above, and IFN-γ and LPS were left on J774E macrophages during the entire experiment. Infection was performed in 6-cm dishes at a concentration of 2 × 106 macrophages per dish at an MOI of 10 for 30 min at 37°C, followed by two rinses in PBS and the addition of fresh DMEM containing 10 μg/ml gentamicin. For iron-rich conditions, medium was additionally supplemented with 50 μM FeSO4. Cells were kept at 37°C and 10% CO2. After the indicated time periods of infection, macrophages were rinsed with warm PBS and were lysed in 1 ml 0.5% Triton X-100 in water. The obtained solution was shaken on ice for 10 min and spun at 4,000 rpm for 5 min to harvest intracellular bacteria. RNA from these bacteria or bacteria grown in minimal medium was isolated with the ZR fungal/bacterial RNA miniprep kit (Hiss Diagnostics GmbH, Freiburg, Germany) according to the manufacturer's instructions.
RT-PCR and real-time PCR.Concentrations of extracted RNAs were normalized, and RNA was used for reverse transcription (RT) with the SuperScript VILO cDNA synthesis kit (Invitrogen, Darmstadt, Germany) by following the manufacturer's recommendations. Resulting cDNA was directly used as the template for amplification using the Maxima SYBR green/ROX quantitative PCR (qPCR) Master Mix (Fermentas, St. Leon-Rot, Germany) with the following primers: 16srRNA200F and 16srRNA200R (for 16SrRNA) (31) and SID4-193F and SID4-193R (for iupT) (32). Amplification and analysis were performed in the LightCycler 480 SW 1.5 (Roche Applied Science, Mannheim, Germany) instrument.
Statistics.Data are expressed as means ± standard deviations. Significance of differences was analyzed by a two-tailed unpaired Student's t test. Differences were termed as significant when P values were ≤0.05.
RESULTS
R. equi survives for at least 24 h in activated macrophages.Activated macrophages have an increased microbicidal capacity (35). We investigated the viability of virulent R. equi 103+ in resting and IFN-γ–LPS-activated J774E macrophages by CFU quantification. Bacteria multiplied in resting macrophages over the course of 48 h (Fig. 1A). In contrast, in activated J774E macrophages, R. equi did not multiply but survived for at least 24 h (recovery of 106% of bacterial load at 0 h) and was reduced to less than 20% of the initial load by 48 h (Fig. 1A).
Virulent R. equi cells survive in activated J774E macrophages for at least 24 h. (A) Resting and activated J774E macrophages were infected with R. equi 103+ at an MOI of 0.25 for 30 min and incubated in fresh medium containing gentamicin to kill extracellular bacteria. Infected macrophages were lysed at different times of infection, and serial dilutions were plated. CFU were counted and calculated as percentage of samples plated immediately after infection. Data represent means and standard deviations of results from four independent experiments. (B) Resting and activated J774E cells were infected with R. equi 103+ at an MOI of 30 and incubated in fresh medium containing a low concentration (10 μg/ml) of gentamicin to kill extracellular bacteria. At different times of infection, a mix of antibiotics (AB), including hygromycin B, penicillin G, and apramycin (each at 100 μM), was added to kill intracellular bacteria. After 24 h, the supernatants of all samples were collected and release of cytosolic lactate dehydrogenase was quantified. Cytotoxicity is expressed relative to an uninfected sample lysed with Triton X-100. Data are means and standard deviations of results from four independent experiments. (C) J774E macrophages were infected with R. equi 103+ at an MOI of 0.25 for 30 min and chased in fresh medium containing 10 μg/ml gentamicin to kill extracellular bacteria (Control) or gentamicin and the mix of high-dose antibiotics (AB) described in the legend to panel B. Cells were lysed at different times of infection, and serial dilutions were plated. CFU were counted and calculated as percentage of samples plated at 0 h postinfection. Data represent means and standard deviations of results from three or four independent experiments. *, P ≤ 0.05.
Infection with viable but not with dead R. equi at high MOI is cytotoxic for macrophages (22). Therefore, levels of R. equi cytotoxic effects on macrophages can be used as a measure of bacterial intracellular viability. Cytotoxicity of infection with 103+ was little but significantly decreased in IFN-γ–LPS-activated J774E cells compared with that of resting cells at 24 h after infection (39% versus 47%; P = 0.02) (Fig. 1B). To analyze whether necrosis of macrophages was due to the intracellular survival of 103+ or due to activation-mediated cytotoxic host cell responses to infection, bacteria were killed intracellularly with a mix of antibiotics at different times after infection. The combination and concentrations of antibiotics used here killed 94% of all intracellular R. equi within 8 h, as determined by CFU counting (Fig. 1C). At 24 h postinfection, necrotic death of infected resting and activated J774E cells treated with antibiotics was reduced compared to that of untreated cells depending on treatment time (Fig. 1B), strengthening the interpretation that R. equi retains metabolic activity in activated macrophages for at least several hours.
R. equi compartmentation does not change upon activation.To test whether control of R. equi infection by activated J774E cells was correlated with altered intracellular vesicle trafficking, we compared key features of maturation of R. equi-containing phagosomes in resting and activated macrophages. Frequencies of fusion of RCVs with lysosomes was unaltered after macrophage activation, regardless of whether virulent, avirulent, or heat-killed R. equi were contained (Fig. 2A). Moreover, RCVs in activated macrophages were still as accessible to newly endocytosed molecules as in resting cells (Fig. 2B).
Communication of the R. equi-containing vacuole with endosomes or lysosomes in resting or activated macrophages. (A) Lysosomes of resting or IFN-γ–LPS-activated J774E macrophages were preloaded with the fluid phase marker ovalbumin-TexasRed (OvTR) and infected with ATTO-488-labeled R. equi 103+ or 103− or heat-killed 103+ (103+ΔT). Samples were fixed at 2 h or 5 h postinfection. (B) Resting or IFN-γ–LPS-activated J774E cells were infected with ATTO-488-labeled R. equi 103+ or 103− or heat-killed E. coli (E.c.ΔT) and incubated for 2 h before medium was exchanged for medium containing dextran TexasRed (DexTR). Samples were incubated at 37°C for another 2 h or 4 h before fixation. Samples were microscopically analyzed for colocalization of phagosomes with the fluorescent tracers. Data represent means and standard deviations of results from three or four experiments.
Phagosome acidification by v-ATPase precedes phagosome-lysosome fusion and is part of the macrophage killing arsenal. To determine the effect of macrophage activation on pH of the RCV, we used a calibrated microplate-based assay early after infection and a microscopic assay based on accumulation of LysoTracker fluor in acidic compartments for early and late times. The pH of phagosomes containing virulent R. equi dropped initially to pH 6.1 before it slowly rose to neutral pH. At 3 h postinfection, RCV pH had stabilized at 7.4 (Fig. 3A). Phagosome pH neutralization strictly depended on the virulence plasmid and on bacterial viability, as phagosomes containing the plasmid-cured strain or heat-killed R. equi acidified to pH 5.0 and pH 5.2 at 3 h postinfection, respectively (Fig. 3A). Regardless of whether R. equi had been killed by UV irradiation or formaldehyde or gramicidin addition, their phagosomes acidified (Fig. 3B). Macrophage activation with IFN-γ–LPS before and during infection with viable, virulent R. equi had no significant effect on RCV pH compared to that in resting J774E cells (Fig. 3C). Even at 24 h after infection, only very few RCVs were positive for LysoTracker in resting and activated macrophages (Fig. 3D and E).
Diversion of phagosome pH by R. equi depends on possession of a virulence-associated plasmid and bacterial viability but not on host activation status. (A) J774E cells were infected with fluorescently labeled virulent (103+), avirulent (103−), or heat-killed R. equi (103+ΔT) for 15 min, and fresh medium was added. At different times after infection, fluorescence was quantified and used to calculate phagosome pH (pH at 0 min corresponds to that of the external buffer). At 180 min (A and B) or 300 min (C), nigericin was added (black arrows) to equilibrate phagosome pH with that of the surrounding buffer to control for correct calibration. (B) As described for panel A, but infection was with 103+ that had been killed by gramicidin, formaldehyde (FA), or UV irradiation before infection. (C) As described for panel A, but infection was with 103+ only of resting versus IFN-γ–LPS-activated macrophages. (D and E) Resting or activated J774E cells were infected with 103+ at an MOI of 1 for 30 min, and fresh medium was added. At 1.5 h and 23.5 h, medium was exchanged for DMEM complete containing LysoTracker (red) and Syto13 (green), which was left on the cells for 30 min to visualize acidic compartments and bacteria. Samples were mounted onto slides, and colocalization of bacteria with LysoTracker was quantified by laser scanning microscopy. Panel E shows representative micrographs of results shown in panel D (black bars, 10 μm). The filled arrowhead points to bacteria colocalizing with LysoTracker, and open arrowheads point to bacteria that do not. Data in panels A to D represent means and standard deviations of results from three independent experiments.
To identify possible differences between intracellular life of R. equi in resting and activated macrophages, the ultrastructure of infected cells was analyzed. Bacteria were internalized by resting and activated macrophages via distinct and slender pseudopods (Fig. 4A and B). At 2 h postinfection, phagosome membranes were tightly attached to R. equi (Fig. 4C and D). As infection progressed, phagosomes became more spacious and began to fill with small vesicles and membranes in both resting and activated macrophages (Fig. 4E and F). At 24 h postinfection, R. equi in resting and activated macrophages resided in large vacuoles that were filled with vesicular electron-dense and electron-lucent structures (Fig. 4G and H) (10). The high numbers of microorganisms observed in activated macrophages were not the result of bacterial multiplication but were due to the high MOI of 20, which was necessary to visualize phagosomes by transmission electron microscopy. The cytoplasm of many macrophages in activated samples was electron dense, yet this was independent of infection (Fig. 4H).
Ultrastructural morphology of R. equi-containing phagosomes in resting and activated J774E macrophages. Resting or IFN-γ–LPS-activated macrophages were infected with 103+ for 1 h, and medium was replaced by fresh medium without bacteria. Samples were fixed at 0 h (A, B), 2 h (C, D), 8 h (E, F), or 24 h (G, H) postinfection and prepared for transmission electron microscopy. (I, J) At 24 h after infection, many macrophages were necrotic yet contained R. equi within phagosomes with intact membranes. Phagosome membranes are indicated by arrowheads. Bars, 1 μm.
In all samples investigated, R. equi cells were enclosed by vacuoles with apparently intact single membranes. Even at 24 h postinfection, when a considerable number of resting and activated macrophages were evidently necrotic and had lost most of their cytosolic contents, many phagosomes still had an intact membrane (Fig. 4I and J). In summary, no obvious differences were observed between the ultrastructures of RCVs in activated versus resting macrophages.
Growth restriction by activated macrophages is not due to immunity-related GTPase Irgm1.The interferon-induced GTPase Irgm1 is involved in protection against infection with multiple intracellular bacteria, including M. tuberculosis (25). However, activated primary bone marrow-derived macrophages (BMMs) of Irgm1−/− mice were as effective in controlling infection with R. equi 103+ as were activated wild-type BMMs (Fig. 5A). The knockout phenotype of Irgm1−/− macrophages and increased Irgm1 expression in wild-type BMMs upon activation were confirmed by immunofluorescence microscopy (Fig. 5B).
Irgm1 is not involved in control of R. equi infection in activated macrophages. (A) Resting or IFN-γ- or IFN-γ–LPS-activated BMMs derived of B6 wild-type (B6-wt) or Irgm1−/− mice were infected with 103+ at an MOI of 1 for 30 min and incubated for 2 h or 24 h before fixation. DNA was stained with Syto13, and intracellular bacteria were counted to determine the percentage of macrophages containing more than 10 bacteria. For each sample and experiment, bacteria in at least 50 infected macrophages were analyzed. Data represent means and standard deviations of results from three or four experiments. *, P ≤ 0.05. (B) Immunofluorescence of uninfected resting or IFN-γ–LPS-activated B6-wt and Irgm1−/− BMMs treated with anti-Irgm1 and using the same microscope settings for all samples (bars, 20 μm).
Growth inhibition by activated macrophages is not due to autophagy.Autophagy is a cellular mechanism which is induced upon macrophage activation and helps to control macrophage infection with mycobacteria (7, 13). Autophagy can be inhibited by phosphoinositol 3-kinase inhibitors such as 3-methyladenine (3MA) or wortmannin (WM). Indeed, 3MA largely restored multiplication of R. equi in activated J774E cells (Fig. 6A), but WM did not (Fig. 6A). If inhibition of autophagy promoted infection, activation of autophagy could possibly suppress it. However, multiplication of R. equi in resting J774E cells was normal after induction of autophagy with rapamycin or by starvation in HBSS or EBSS (Fig. 6B). Conversely, treatment of resting macrophages with 3MA did not promote multiplication of avirulent R. equi 103− (Fig. 6C). The effects of the different treatments on macrophage autophagy were confirmed by analysis of the distribution of a marker for autophagic vacuoles, microtubule-associated protein light chain 3 (LC3) in resting and activated BMMs of GFP-LC3 transgenic mice (not shown).
Autophagic pathways and R. equi infection control. (A and B) IFN-γ–LPS-activated or resting J774E cells were infected with 103+ at an MOI of 1 for 30 min and incubated in fresh medium for 2 h or 24 h. Cells were left untreated (control) or incubated with different compounds which manipulate autophagic pathways: the autophagy inhibitors 3-methyl adenine (3MA; 10 mM for 2 h before/1 mM after infection) or wortmannin (WM; 50 nM, with DMSO control) or with the autophagy inducers EBSS, HBSS, or rapamycin (RM; 200 nM, with DMSO control). Samples were fixed at 2 h and 24 h of infection and stained with Syto13, and the percentages of macrophages containing more than 10 bacteria were determined. (C) Experimental setup as described for panel A, but resting J774E cells were infected with 103+ or 103− and either not treated or treated with 10 mM 3MA. (D) Resting or IFN-γ–LPS-activated bone marrow-derived macrophages of GFP-LC3 mice were infected with R. equi 103+ as described for panel A and fixed at different times after infection. Samples were stained with antibodies against R. equi to visualize bacteria, and colocalization of R. equi with LC3 was determined in at least 50 infected macrophages per sample. (E) Representative fluorescence micrograph of experiments shown in panel D from a 10-min, activated sample (black bar, 10 μM). Red, R. equi 103+; green, GFP-LC3. The inset shows a closeup of the LC3-positive phagosome. All data in panels A to D represent means and standard deviations of results from three to nine independent experiments. *, P ≤ 0.05.
To directly investigate RCV intersection with autophagic compartments, RCV colocalization with GFP-LC3 was determined in infected resting and activated GFP-LC3 BMMs. There was infrequent but significant association of the RCV with LC3 early after infection, which was more pronounced in activated than in resting macrophages (19.2 versus 4.9% colocalization at 10 min; P = 0.049) (Fig. 6D and E). GFP-LC3 rarely colocalized with RCVs at later times after infection (Fig. 6D).
Nitric oxide restricts R. equi multiplication in activated macrophages.Inducible nitric oxide synthase (NOS2; formerly i-NOS) is central for control of R. equi infection by activated macrophages (5). However, macrophage activation boosts not only production of nitric oxide after phagocytosis but also that of superoxide (35). To distinguish effects that might be caused by nitric oxide or superoxide alone, we treated infected macrophages with different compounds that affect production of either radical in vitro. The addition of the nitric oxide donor S-nitroso-N-acetyl-d,l-penicillamine (SNAP) to the medium of infected resting J774E cells strongly inhibited multiplication of R. equi (Fig. 7A). However, when J774E cells were treated with SNAP for 24 h before but not during and after infection, R. equi multiplied approximately as often as in untreated resting cells (Fig. 7B).
Multiplication of R. equi in resting or activated macrophages treated with compounds acting on release of nitric oxide or superoxide. (A) Resting J774E cells were infected with 103+ at an MOI of 1 for 30 min in the absence of relevant reagents and incubated in fresh medium containing the nitric oxide donor SNAP or solvent alone (Control) for 2 h or 24 h. Samples were fixed and stained with Syto13 to quantify the percentage of macrophages with more than 10 bacteria. (B) Infection and analysis of resting J774E cells was the same as described for panel A, but samples were treated with 100 μM SNAP or solvent alone (Control) for 24 h before infection and in medium without SNAP or solvent after infection (SNAP b.i., 24 h). (C) Infection and analysis of IFN-γ–LPS-activated J774E were the same as described for panel A, but samples were incubated in the absence or presence of 500 μM the NOS2 inhibitor NG-methyl-l-arginine (NMLA) after infection or treated with 5 mM the superoxide scavenger N-acetyl-l-cysteine (NAC) 2 h before and always after infection. (D) Experimental setup was as described for panel A but using resting or IFN-γ–LPS-activated BMMs from B6 wild-type (B6-wt) or B6-derived NADPH oxidase knockout (gp91phox−/−) mice. Data shown in panels A to D represent means and standard deviations of results from three independent experiments. *, P ≤ 0.05.
Inhibition of nitric oxide synthesis in activated macrophages with NG-methyl-l-arginine (NMLA) allowed R. equi multiplication, as had been described previously (5) (Fig. 7C). In contrast, treatment with N-acetyl-l-cysteine, a membrane-permeable compound which quenches reactive oxygen species stoichiometrically (59), did not restore R. equi intracellular multiplication in activated macrophages (Fig. 7C), making a killing role of these compounds unlikely. This is in line with the results from bone marrow-derived macrophages of wild-type and transgenic mice (gp91phox−/− mice) which are deficient in the major superoxide generator in immune cells, the NADPH oxidase complex. Activation with IFN-γ and LPS reduced multiplication of R. equi in resting macrophages of both genotypes as effectively as treatment with the nitric oxide donor SNAP (Fig. 7D). Thus, nitric oxide is essential for control of R. equi infection, while NADPH oxidase activity is dispensable.
It has been reported that R. equi is resistant to high concentrations of nitric oxide (5). However, since the kinetics of nitric oxide release, the redox state of nitric oxide, and the production of specific by-products can vary considerably between different nitric oxide donors (9, 21), we examined whether there was a direct bactericidal effect of SNAP or its products on R. equi. Bacteria were cultivated in DMEM complete in the presence of SNAP for 24 h without any loss in viability (Fig. 8A). Expression of the essential virulence factor VapA by SNAP-treated R. equi was the same as that by untreated bacteria (Fig. 8B), underscoring bacterial fitness.
Effects of nitric oxide treatment on R. equi viability and VapA expression. (A) Sensitivity of R. equi 103+ to treatment with nitric oxide was assessed by cultivation in DMEM complete containing 100 μM SNAP or solvent only (Control) at 37°C. Culture optical density at 600 nm was determined at various times. Data represent means and standard deviations of results from three independent experiments. (B) Samples from control and SNAP-treated 103+ grown for 24 h and taken from the three experiments (1, 2, and 3) shown in panel A were applied to SDS-PAGE, Western blotted, and decorated with anti-VapA. Numbers indicate molecular weight in kDa, and “VapA” indicates the migration position of VapA.
SNAP and IFN-γ–LPS-activated macrophages have similar quantities and kinetics of nitric oxide release.To compare the amounts of nitric oxide to which R. equi was exposed under the different experimental conditions, we quantified nitric oxide release from (i) SNAP, (ii) resting J774E macrophages, (iii) J774E cells activated with IFN-γ alone, or (iv) J774E cells activated with IFN-γ–LPS. All conditions were tested with uninfected and infected macrophages. No nitric oxide production was detected in resting macrophages, regardless of whether these were infected or not (Fig. 9A). Activation with IFN-γ alone led to production of small amounts of nitric oxide, which were massively increased in response to infection with either virulent or avirulent R. equi (Fig. 9B). J774E cells activated with IFN-γ–LPS produced nitric oxide faster than cells activated with IFN-γ alone, yet final nitrite concentrations were lower than with IFN-γ-activated cells (Fig. 9C). Even uninfected macrophages produced large amounts of nitric oxide when they had been activated with IFN-γ and LPS (Fig. 9C). Infection with plasmid-cured R. equi reproducibly resulted in a slightly higher release of nitric oxide by macrophages activated with IFN-γ–LPS than infection with virulent bacteria (Fig. 9C; P = 0.038 and 0.008 at 24 h and 48 h after infection, respectively).
Quantification of nitric oxide release by resting or IFN-γ- or IFN-γ–LPS-activated J774E macrophages or by SNAP. Resting (A) or IFN-γ (B)- or IFN-γ–LPS (C)-activated J774E cells were left uninfected or infected with 103+ or 103− at an MOI of 1 for 30 min and incubated in fresh medium. At the times indicated, supernatants were collected, and nitrite as a stable end product of nitric oxide was quantified using Griess reagent. (D and E) SNAP or DETA-NO were dissolved in DMEM at 100 μM (SNAP; Control; solvent only) or at 50, 100, or 200 μM (DETA-NO) and incubated at 37°C. Samples were taken at different times and analyzed for nitric oxide as described for panels A to C. All data are means and standard deviations of results from three to five experiments. (F) Resting J774E cells were infected with 103+ at an MOI of 1 for 30 min in the absence of relevant reagents and then cultivated in fresh medium without (Control) or with the nitric oxide donors SNAP (100 μM) or DETA-NO (50 μM) for 2 h or 24 h. Samples were fixed and stained with Syto13 to quantify the percentage of macrophages with more than 10 bacteria. Data shown represent means and standard deviations of results from four independent experiments. A P value of ≤0.05 is indicated by the following symbols: #, uninfected versus 103+; °, uninfected versus 103−; *, 103− versus 103+ in panels A to C or nitric oxide donor(s) versus the control in panels D and F.
In contrast to nitric oxide production by macrophages, nitric oxide release from SNAP was detectable already at 4 h (Fig. 9D). However, from 8 h onward, both the kinetics and the concentrations of nitric oxide were very similar to those produced by IFN-γ–LPS-treated macrophages (Fig. 9C and D).
Different nitric oxide donors do not only differ in kinetics of release but also in the redox state of nitric oxide produced (9). We therefore investigated the effect of a nitric oxide donor from a compound class different from that of SNAP, diethylenetriamine-nitric oxide adduct (DETA-NO), on R. equi intracellular multiplication. Within 1 h, a concentration of 50 μM DETA-NO in DMEM complete already yielded nitric oxide amounts similar to those after 24 h of incubation of 100 μM SNAP (Fig. 9E). However, the restriction of R. equi growth in resting macrophages by DETA-NO was as pronounced as that by treatment with SNAP (Fig. 9F).
Iron supplementation abrogates activation- or nitric oxide-mediated growth restriction.Iron availability is a limiting factor during infection of activated macrophages (4). To test the possible role(s) of iron limitation during the infection with R. equi, we took advantage of the R. equi iupT gene which is involved in catecholate siderophore synthesis and which is highly upregulated in iron-depleted medium (32). Its transcription was clearly induced in R. equi inside activated macrophages at 2 h of infection, whereas by 8 h of infection, the upregulation had ceased (Fig. 10). Remarkably, the strong upregulation at 2 h could be completely annihilated by the addition of ferrous sulfate to the macrophage medium. In resident macrophages, R. equi iupT expression was not upregulated at either time of infection. 2,2-Dipyridyl (2,2-bipyridin) is a membrane-permeable iron chelator which inhibits R. equi growth in liquid medium at high concentrations (200 μM) but not at lower concentrations (80 μΜ) (31). We observed that R. equi strain 103 did not multiply in minimal medium at 150 μM 2,2-dipyridyl and, hence, used supplementation with 0, 50, or 80 μM this compound as a control for iupT induction. Whereas 50 μM 2,2-dipyridyl did not very much increase iupT transcription, a 14-fold relative induction was observed using 80 μM (Fig. 10). In summary, these data suggested a strong iron removal response of R. equi early in infection of activated but not resident macrophages.
Real-time PRC analysis of iupT expression inside resident or activated macrophages. R. equi was grown for 16 h in minimal medium (Min. medium) with acetate and 0 μM (DIP 0), 50 μM (DIP 50), or 80 μM (DIP 80) 2,2-dipyridyl or isolated from resting or activated macrophages that had been infected for 2 h or 8 h in the absence or presence (+FeSO4) of 50 μM ferrous sulfate. Expression of iupT is shown relative to expression of 16S-rRNA (fold change expression). The mean values from two independent determinations are shown.
To further analyze whether this iron removal stress could contribute to killing of R. equi by activated macrophages, we added 100 μM ferrous sulfate extracellularly to infected IFN-γ–LPS-treated J774E cells, which completely restored multiplication of R. equi (Fig. 11A and B). The effect was concentration dependent and was not observed after the addition of equimolar amounts of magnesium sulfate (Fig. 11A and B). Importantly, the same phenomena were observed with SNAP-treated macrophages (Fig. 11A).
Inhibition of multiplication of R. equi in activated or in SNAP-treated resting J774E cells can be abrogated by the addition of iron sulfate. (A) Resting or IFN-γ–LPS-activated J774E cells were infected with 103+ at an MOI of 1 for 30 min and incubated in fresh medium containing no (control) or different additives for 2 h or 24 h. These were 100 μM the nitric oxide donor SNAP, 100 μM ferrous sulfate (FeSO4), or 100 μM magnesium sulfate (MgSO4). Samples were fixed and stained with Syto13 to quantify the percentage of macrophages with more than 10 bacteria. *, P ≤ 0.05. (B) Experimental setup as described for panel A, but activated macrophages were treated with no (Control) or different concentrations of ferrous sulfate after infection. *, P ≤ 0.05. (C and D) Effects of ferrous sulfate treatment on viability (C) and nitric oxide production (D) of J774E cells: IFN-γ–LPS-activated macrophages were treated with different concentrations of iron sulfate. At 24 h of incubation, sample supernatants were collected and the release of lactate dehydrogenase (C) and nitric oxide (D) was quantified. (E) 103+ was grown in minimal medium with 100 μM ferrous sulfate overnight and used for inoculation of minimal medium with different iron concentrations at an OD600 of 0.1. Suspensions were mixed with no (Control; solvent only) or 100 μM SNAP. Optical density was determined before and after 24 h of incubation at 37°C. A P value of ≤0.05 is indicated by the following symbols: *, versus control-250 μM ferrous sulfate; #, versus SNAP-250 μM ferrous sulfate. Data of all graphs represent means and standard deviations of results from three or four experiments.
Restoration of R. equi intracellular multiplication by ferrous sulfate could have been caused indirectly, e.g., by a decrease in macrophage viability by ferrous sulfate or by a decreased nitric oxide production in the presence of iron ions, but neither was the case (Fig. 11C and D). Furthermore, ferrous ions can enhance the decomposition of SNAP (9), but here release of nitric oxide from SNAP did not change in the presence of 100 μM ferrous sulfate (not shown), excluding this as a cause of the observed effects.
Another possible explanation for the observed effects was a direct action of nitric oxide on bacterial iron. To test the sensitivity of R. equi to nitric oxide under iron-limiting conditions, we grew the bacteria in minimal (Fig. 11E) rather than rich (Fig. 8A) medium to be able to control for iron availability. Bacteria were grown in minimal medium containing 100 μM ferrous sulfate overnight, pelleted, and cultivated in minimal media with different iron concentrations in the presence or absence of 100 μM SNAP for 24 h (Fig. 11E). At a concentration of 1 μM added ferrous sulfate, bacterial growth was significantly decreased by approximately 20% compared to growth in medium with 250 μM iron (Fig. 11E), indicating that iron was at a growth-limiting level. However, even at this low level of free iron, growth of SNAP-treated R. equi was the same as that in the untreated control samples (Fig. 11E).
DISCUSSION
This study analyzed the mechanisms by which activated macrophages inhibit growth of the intracellular pathogen R. equi. It had been proposed that IFN-γ–LPS activation of macrophages before infection leads to a fast killing of R. equi by peroxynitrite (5). However, in our experimental system, R. equi survived in activated macrophages for at least 24 h, followed by elimination within the following 24 h. At this time, IFN-γ–LPS activation had already resulted in significant macrophage cytotoxicity. Therefore, the decrease in numbers of recovered bacteria may either reflect their clearance or macrophage death, leading to exposure of formerly intracellular R. equi to gentamicin in the medium.
There are precedents for pathogens which do not multiply in activated macrophages yet survive for considerable periods of time, such as M. avium or L. pneumophila (45, 47). It has been suggested that growth restriction of these pathogens is mediated by enhanced formation of phagolysosomes. Increased phagosome-lysosome fusion and/or phagosome acidification upon macrophage activation has been reported for several other intracellular pathogens that do not grow in or are killed by activated host cells (19, 46, 50, 58), underscoring a close connection between the loss of their privileged intracellular compartment and infection control.
The effects of macrophage activation on RCV properties were investigated by comparison of hallmark features of phagosome maturation in resting and activated J774E cells. The kinetics of pH in vacuoles containing virulent R. equi matched the RCV maturation profile (10): the initial acidification to the pH of early endosomes observed in this study is paralleled by the published normal acquisition and loss of early maturation stage markers (10). The subsequent failure of the RCV to acquire v-ATPase and to fuse with lysosomes (10) coincided with a rise in phagosome pH. Similar interference with phagosome acidification has been observed with other members of the mycolata, suggesting related mechanisms, i.e., virulent but not avirulent strains of Nocardia asteroides reside in neutral-pH vacuoles (1), and phagosomes containing pathogenic mycobacteria have a pH of 6.2 to 6.5 (44, 52, 63).
Surprisingly, the unusual progression of R. equi phagosome pH was also observed in activated macrophages, and there was no enhanced RCV-lysosome fusion. Unlike M. avium-containing phagosomes (40), RCVs in activated macrophages remained accessible to endocytosed material. Hence, growth restriction in activated macrophages was likely not caused by alterations in RCV compartmentation characteristics.
Activation of macrophages can induce a change in not only intracellular trafficking but also autophagy, a cellular starvation and stress response which restricts multiplication of intracellular mycobacteria (13). The autophagy inhibitor 3MA significantly increased R. equi multiplication in activated macrophages, yet no other established inhibitor (wortmannin) or inducer (rapamycin or starvation) of autophagy altered bacterial growth. Furthermore, no bacteria were observed in autophagic structures by transmission electron microscopy. Colocalization of the RCV with the autophagy marker LC3 was infrequent and occurred only early in infection, as has also been observed with latex bead phagosomes in resting, nonstarved cells (49), essentially excluding a specific activation-related role of LC3 in R. equi control. The results obtained with 3MA treatment therefore likely reflect an autophagy-independent activity of this compound similar to those observed by others (33).
An important defense mechanism of activated macrophages depends on immunity-related GTPases (IRG), including Irgm1 (LRG-47) (57). In contrast to the situation with Salmonella (14) or Mycobacterium (26), activated Irgm1−/− primary macrophages inhibited R. equi multiplication as much as wild-type phagocytes. As Irgm1 has been correlated to autophagic mechanisms (13) and acts independently of nitric oxide production (26), this result supports the concept of an autophagy-independent, nitric oxide-dependent (as discussed below) mode of R. equi growth restriction. It also suggests that IRG proteins other than Irgm1 are not involved either, since Irgm1 is crucial for correct localization of at least two other immunity-related GTPase family members (16).
Nitric oxide production by NOS2 is critically required for control of R. equi infection by activated macrophages (5). SNAP (100 μM) produced nitric oxide in similar quantities and with similar kinetics as IFN-γ–LPS-activated macrophages infected with 103+, suggesting that R. equi was exposed to similar concentrations of nitric oxide in SNAP-treated and in activated macrophages. SNAP addition fully mimicked the growth-inhibiting effect of J774E activation, and so did addition of the nitric oxide donor DETA-NO. DETA-NO differs from SNAP in that nitric oxide is released very fast and in different redox form(s) (9), suggesting that exact kinetics of nitric oxide release and/or the amounts of particular nitric oxide redox variants are not relevant for inhibition of R. equi intracellular multiplication.
Growth restriction of R. equi by SNAP or by macrophage activation was independent of NADPH oxidase activity, in agreement with earlier studies (5). However, NADPH oxidase knockout mice succumb to R. equi infection (5). This may result from NADPH oxidase knockout effects on other cells, e.g., neutrophils which respond to infection with a strong, gp91phox-containing NADPH oxidase (20)-depending oxidative burst and which are indispensable for clearance of R. equi infection (27). Also, treatment of activated macrophages with the reactive oxygen species quencher N-acetyl-l-cysteine (NAC) did not restore R. equi intracellular growth. We conclude that growth inhibition by SNAP was not mediated by peroxynitrite, whose production requires both nitric oxide and superoxide, but was rather caused by nitric oxide alone or its superoxide-independent reaction products.
The precise identity of the nitric oxide target remains elusive. Our and others' (5) data show that, unlike M. tuberculosis (24) or L. pneumophila (53), R. equi is not sensitive to direct treatment with nitric oxide even at low iron supply. Further, treatment of macrophages with SNAP before infection did not restrict R. equi intracellular growth, indicating that nitric oxide action is required only after infection. R. equi iupT is a gene whose expression is strongly upregulated in iron-limiting conditions, and it is also upregulated in activated but not in resident macrophages. Hence, nitric oxide production is likely part of an activation-mediated host iron restriction system. Supplementation with abundant iron from the extracellular space annihilates the macrophages' effort to restrict iron, and hence intracellular R. equi cells do not upregulate the iron acquisition machinery. A multitude of data from other groups suggest a general regulatory effect of macrophage activation and particularly nitric oxide on intracellular iron levels. IFN-γ activation reduces free iron in macrophages and their phagosomes (36, 53, 61), and nitric oxide affects cellular iron uptake and availability (30, 43, 62). Growth inhibition by macrophage activation of several intracellular pathogens, such as L. pneumophila or S. Typhimurium, has been correlated with their limited access to iron (2, 37).
A particularly surprising finding of this study was that R. equi apparently did not suffer from iron removal stress in resident macrophages, as reflected in low iupT expression. It seems as if intracellular R. equi cells manage to readily acquire iron in sufficient quantities, possibly through a constitutively expressed iron acquisition system (28) or through the baseline expression of the iup genes or another inducible iron acquisition system.
In summary, this study shows that R. equi cells survive for at least 24 h in activated macrophages, where they are able to establish their unusual compartment. Unlike in the case with several other intracellular pathogens, such as M. tuberculosis, growth restriction by immune activation is not dependent on autophagy, Irgm1, or phagolysosome formation. Inhibition of growth was completely mimicked by treatment of resting macrophages with nitric oxide donors, demonstrating the central role of nitric oxide and the dispensability of induction of hundreds of genes by IFN-γ–LPS (8) in macrophage defense against R. equi infection. Mechanisms restricting multiplication of R. equi could be annihilated by the addition of ferrous sulfate. Macrophage activation and nitric oxide synthesis likely act in a regulatory way on the producing J774E cell itself by (i) making the macrophages more capable of storing iron in a way that it cannot readily be accessed by R. equi, (ii) making the macrophage able to prevent iron from being scavenged by bacterial siderophores, or (iii) inactivating a host-driven pathway to directly supply the intraphagosome space with iron.
ACKNOWLEDGMENTS
We thank Sabine Spürck and Ulrike Karow for expert technical assistance, Shinji Takai for the anti-VapA antibody, Noburo Mizushima (Tokyo Metropolitan Institute of Medical Sciences) for the GFP-LC3 mice, Ralf Brandes (University of Frankfurt Faculty of Medicine) for the gp91phox−/− mice, and Stefanie Riesenberg and Joachim Schultze (LIMES, Bonn) for access to and help with the Roche LightCycler. We acknowledge valuable discussions with Jonathan Howard and Simon Newman.
This study was supported by a fellowship of the German National Academic Foundation to K.V.B., by an NIH grant (AI57831) and a VA Merit Review grant to G.T., and by a collaborative research center (SFB 670) grant from the Deutsche Forschungsgemeinschaft (German Research Foundation) to O.U. and A.H.
FOOTNOTES
- Received 9 September 2010.
- Returned for modification 2 October 2010.
- Accepted 16 February 2011.
- Accepted manuscript posted online 7 March 2011.
- Copyright © 2011, American Society for Microbiology