Heme Oxygenase-1 and Carbon Monoxide Promote Burkholderia pseudomallei Infection

Melioidosis, caused by the environmental Gram-negative rod Burkholderia pseudomallei, is often associated with pneumonia and bacterial dissemination to distant sites. In Southeast Asia and northern Australia, where it is endemic and an important cause of sepsis, infection may be associated with high mortality despite appropriate antibiotic therapy (1–5). Infection with the facultative intracellular pathogen results from s.c. inoculation, inhalation, or ingestion. The bacterium expresses various pathogen-associated molecular patterns, including peptidoglycan, LPS, and flagella, which are recognized by TLRs (5). Components of the B. pseudomallei type 3 secretion system 3 can also activate the cytosolic inflammasome by NOD-like receptors, which leads to the activation of caspase-1 and promotes the maturation of IL-1β and IL-18 (6–8). IL-18 plays a protective role during melioidosis through induction of IFN-γ (8, 9). In contrast, IL-1β may play a deleterious role by causing excessive neutrophil recruitment and tissue damage, as well as by inhibiting the activation of IFN-γ. Although disproportionate neutrophil recruitment may be detrimental (8, 10), neutrophils play a critical role in early bacterial containment (11). IFN-γ is essential for resistance against B. pseudomallei infection, with additional crucial roles played by TNF-α, IL-12, and IL-18 (5, 12, 13). In this context, macrophages have been shown to be pivotal for early resistance against B. pseudomallei (14). NADPH oxidase–mediated production of reactive oxygen species (ROS) contributes to suppression of intracellular replication of B. pseudomallei in murine macrophages (14) and in human monocytes (15), but it is not sufficient to prevent intracellular multiplication. Although B. pseudomallei stimulates immune cells to release cytokine mediators and to produce ROS, triggering cellular defense mechanisms to clear the infection, an excessive inflammatory response and increase in oxidants is likely to contribute to pathogenesis. However, the role of antioxidant defense mechanisms has rarely been addressed in B. pseudomallei infection.

The activation of the transcription factor nuclear erythroid-related factor 2 (Nrf2) is an efficient antioxidant defense mechanism used by the host to counteract oxidative stress (16). Under normal physiological conditions, Nrf2 is sequestered in the cytoplasm by interaction with Kelch-like ECH-associated protein 1. When cells are exposed to oxidative stress, Nrf2 is released and translocates to and accumulates in the nucleus. There it binds specifically with transcription cofactors to the antioxidant response elements and thereby regulates the activation of most antioxidant and phase II enzyme genes (e.g., heme oxygenase-1 [HO-1], peroxiredoxins 1 and 6) to maintain cellular redox homeostasis. Disruption of Nrf2 increases mortality of mice in LPS- as well as cecal ligation and puncture (CLP)–induced sepsis (17–19). In Nrf2-deficient mice, LPS elicits stronger pulmonary inflammation associated with enhanced levels of inflammatory cytokines and ROS production (20).

HO-1 is a stress-responsive enzyme and is important for defense against oxidant-induced injury during inflammatory processes. HO-1 is highly inducible by a variety of stimuli, such as LPS, inflammatory cytokines, heat shock, heavy metals, oxidants, and its substrate heme. It catalyzes the first and rate-limiting step in the degradation of heme (iron protoporphyrin IX) into CO, biliverdin, and ferrous iron (21). Its metabolite CO has various biological functions, including cytoprotective and anti-inflammatory properties (22, 23). Biliverdin is converted by biliverdin reductase to bilirubin, an effective endogenous antioxidant with recently recognized anti-inflammatory effects (24). Excess free iron is toxic for host cells because it catalyzes the production of ROS, leading to oxidative damage. However, ferrous iron is normally sequestered by ferritin, a cytoprotective antioxidant compound.

The HO-1/CO system has been involved in many physiological and pathological processes by mediating cytoprotection (25) and regulating the host inflammatory response (22). The immunmodulatory effects can have both beneficial and detrimental consequences for the host immunity against infectious agents (21). HO-1 is able to promote Plasmodium liver infection (26), whereas it plays a favorable role in the host during cerebral malaria (27). HO-1 also enhances bacterial clearance during CLP-mediated polymicrobial sepsis (28) and Salmonella typhimurium–induced enterocolitis (29).

Despite the significance of Nrf2 and HO-1 as antioxidant or anti-inflammatory defense mechanisms, their specific role in the host–Burkholderia interaction has to date not been addressed. Thus, in this study we aimed to analyze the expression of HO-1 in response to B. pseudomallei and to identify upstream signaling pathways involved in its regulation. We further sought to characterize the role of the transcription factor Nrf2, HO-1, and its metabolite CO in microbial host defense during infection of primary macrophages with B. pseudomallei and during murine melioidosis.

All of the animal experiments described in the present study were conducted in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All studies were conducted under a protocol approved by the Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern.

C57BL/6J wild-type mice were purchased from The Jackson Laboratory (Bar Harbor, ME). C57B/L129 Nrf2 knockout (Nrf2^−/−) and C57B/L129 wild-type (Nrf2^+/+) mice (30) were provided by Andreas Teubner (Institute for Laboratory Animal Science, University Hospital/ RWTH, Aachen, Germany) with permission of Yuet W. Kan (Department of Medicine, University of California at San Francisco, San Francisco, CA). Mouse strains were bred by the Department of Laboratory Animal Science of the University Medicine of Greifswald and Micromun (Greifswald, Germany). Female mice (10–12 wk old) were housed in filter-top cages under standard conditions in physical containment level 3 laboratory facilities and provided with food and water ad libitum.

B. pseudomallei strain E8 is a soil isolate from the region of Ubon Ratchathani in northeastern Thailand (31) and was used throughout the study. Bacteria were grown on Columbia agar (BD Biosciences, Heidelberg, Germany) at 37°C for 24 h and adjusted to the desired concentration in Dulbecco’s PBS (D-PBS; Life Technologies, Darmstadt, Germany) or the respective cell culture medium. For experiments employing heat-inactivated B. pseudomallei, bacteria were suspended in D-PBS, incubated at 70°C for 20 min, and stored at −70°C until use.

Bacteria were grown on Columbia agar at 37°C for 24 h and adjusted to an OD at 650 nm of 0.01 in 15 ml Lennox Luria–Bertani (LB) broth and incubated with or without the HO-1 activator cobalt (III) protoporphyrin IX chloride (CoPPIX; 25 μM, Frontier Scientific, Livchem Logistics, Frankfurt am Main, Germany), the CO-releasing molecule (CORM) tricarbonyldichlororuthenium (II) dimer (CORM-2; 100 μM, Sigma-Aldrich Chemie, Taufkirchen, Germany) or corresponding vehicle at 37°C with shaking at 140 rpm. At 0, 3, 6, and 24 h the OD₆₅₀ of bacterial cultures was measured and 100 μl was collected, serially diluted 10-fold in D-PBS, and the number of viable cells (CFU/ml) was counted by plating on LB agar in duplicate. Plates were incubated at 37°C for 48 h.

Bone marrow–derived macrophages (BMM) were generated and cultivated in a serum-free cell culture system as recently described (32). Briefly, tibias and femurs were aseptically removed and bone marrow cells were flushed with sterile D-PBS and then centrifuged at 150 × g for 15 min. Cells were resuspended in RPMI 1640 medium (Life Technologies) containing 5% Panexin BMM (PAN Biotech, Aidenbach, Germany), 2 ng/ml recombinant murine GM-CSF (Cell Signaling Technology, New England Biolabs, Frankfurt am Main, Germany), and 50 μM 2-ME (Life Technologies) and cultivated at least 10 d at 37°C in a humidified atmosphere containing 95% air and 5% CO₂.

A nonsilencing small interfering RNA (siRNA) duplex was obtained from Qiagen (Hilden, Germany). siRNA oligonucleotides directed against mouse HO-1 were purchased from Dharmacon (GE Healthcare, Berlin, Germany). The duplexes were dissolved in RNase-free water and stored in aliquots at −20°C.

Delivery of siRNA into BMM was performed by electroporation using a modified protocol of Wiese et al. (33). siRNA duplexes (6 μg) and cell suspension containing 2 × 10⁶ BMM in Opti-MEM (Life Technologies) were transferred to a 0.4-cm Gene Pulser cuvette and pulsed in a Gene Pulser Xcell (Bio-Rad, München, Germany). After electroporation, cells were transferred to RPMI 1640 medium (Life Technologies) without supplements and seeded in 48-well plates (1.5 × 10⁵ cells/well) and 6-well plates (6.5 × 10⁵ cells/well). After 60 min an equal amount of RPMI 1640 medium containing 10% Panexin BMM (PAN Biotech) and 100 μM 2-ME (Life Technologies) was added, resulting in a final concentration of RPMI 1640 medium of 5% Panexin BMM and 50 μM 2-ME. Twenty-four hours after electroporation, culture medium was replaced and BMM were cultivated for a further 24 h followed by infection with B. pseudomallei as described below.

Eighteen to 20 h prior to infection, BMM were seeded in 48-well plates (1.5 × 10⁵ cells/well), and where applicable treated with Nrf2 activators sulforaphane (SFN; 20 μM, Sigma-Aldrich Chemie) and tert-butylhydroquinone (tBHQ; 50 μM, Enzo Life Sciences, Lörrach, Germany), CoPPIX (25 μM), or the corresponding vehicle. CORM-2 (50 μM) or vehicle was added 1 h prior to infection.

BMM were infected with B. pseudomallei at the indicated multiplicity of infection (MOI). Well plates were centrifuged for 4 min at 120 × g. After infection for 30 min, cells were washed twice with D-PBS and incubated in 100 μg/ml kanamycin-containing medium (to eliminate remaining extracellular bacteria) supplemented with SFN, tBHQ, CoPPIX, CORM-2, or vehicle. At indicated time points (time 0 was taken 30 min after incubation with antibiotic-containing medium), the number of intracellular CFU was determined. Consequently, cells were washed twice with D-PBS and subsequently lysed using 150 μl D-PBS–containing 0.5% Tergitol TMN (Fluka, Buchs, Switzerland) and 1% BSA (Carl Roth, Karlsruhe, Germany) per well. After 15 min of incubation, appropriate dilutions of lysates were plated on LB agar and incubated at 37°C for 48 h.

For pretreatment versus posttreatment studies, BMM were treated with CoPPIX (25 μM) at 18–20 h prior to infection and in kanamycin-containing medium (pretreatment) as described above. Posttreatment was performed by adding CoPPIX only in kanamycin-containing medium 0, 3, or 6 h postinfection.

To quantify the extent of membrane-damaged cells in response to bacterial infection, release of lactate dehydrogenase (LDH) in cell culture supernatants, bronchoalveolar lavage fluid (BALF), or serum was determined. BMM were seeded in 48-well plates (1.5 × 10⁵ cells/well) and infected at the indicated MOI with B. pseudomallei for 30 min. Cells were washed twice with D-PBS, and 400 μl medium containing 100 μg/ml kanamycin was added to each well to eliminate extracellular bacteria. At the indicated time points, cell culture supernatant was collected, and LDH activity was detected by using the CytoTox-ONE homogeneous membrane integrity assay (Promega, Mannheim, Germany) according to the manufacturer’s instructions. Briefly, 50 μl supernatant and BALF or 25 μl serum was added to the kit reagent and incubated for 10 min at room temperature. After addition of stopping solution, the fluorescence intensity was measured using the microplate reader Infinite M200 PRO (Tecan, Crailsheim, Germany) at an excitation wavelength of 560 nm and emission wavelength of 590 nm.

18-20 h prior to infection, BMM (6.5 × 10⁵ cells per well) were seeded in 6-well plates, and infected for 30 min with B. pseudomallei at the indicated MOI followed by washing with D-PBS and incubation in 100 μg/ml of kanamycin-containing medium. When indicated, BMM were pretreated for one hour with the appropriate inhibitors LY294002 (PI3K, 10-40 μM), SB202190 (p38 MAPK, 1-4 μM), Ro318220 (PKC, 0.25-1 μM) (Enzo Life Sciences) or the corresponding vehicle, followed by infection with B. pseudomallei. BMM were treated with SFN (5-20 μM), tBHQ, (12.5-50 μM), CoPPIX (0.2 M NaOH pH 7.2 in D-PBS, 25 μM), the HO-1 inhibitor zinc (II) protoporphyrin IX (ZnPPIX; 50 mM Na₂CO₃ pH 7.2 in D-PBS, 25 μM, Frontier Scientific) or the corresponding vehicle for 18–20 h and in kanamycin-containing medium.

Total RNA was isolated by TRIzol reagent following the manufacturer’s instructions. Reverse transcription of 1 μg total RNA was performed using the Moloney murine leukemia virus reverse transcriptase (Promega) and 0.5 μg oligodeoxythymine [oligo(dT)] primer (Microsynth, Balgach, Switzerland). Quantitative real-time PCR (qRT-PCR) was performed using the LightCycler 480, and amplification products were detected by the Maxima Probe qPCR master mix (Thermo Scientific Fermentas, Schwerte, Germany). TaqMan PCR probes and gene-specific primer pairs were generated by Microsynth. Data were analyzed with LightCycler software version 1.5. The reference gene RPLP0 served for the standardization of the individual PCR reactions. All assays were performed in duplicate and repeated as indicated in the figure legends.

Eighteen to 20 h prior to infection, BMM were seeded in six-well plates (6.5 × 10⁵ cells/well) and infected for 30 min with B. pseudomallei at the indicated MOI followed by washing with D-PBS and incubation in 100 μg/ml kanamycin-containing medium. When indicated, BMM were treated with SFN (5–20 μM), tBHQ (12.5–50 μM), CoPPIX (25 μM), ZnPPIX (25 μM), or the corresponding vehicle. For analysis of phosphorylated kinases, BMM (6.5 × 10⁵ cells per well) were seeded in six-well plates and infected for 5–60 min with B. pseudomallei at an MOI of 50 without subsequent incubation in kanamycin-containing medium. Proteins of BMM were prepared by TRIzol reagent according to manufacturer’s instructions. Protein content was determined using the Bradford method. Equal amounts of protein were separated by SDS-PAGE and transferred onto nitrocellulose membranes by electroblotting. Membranes were blocked with 1× Roti-Block (Carl Roth) for 1 h at room temperature and subsequently incubated overnight at 4°C with a rabbit anti-GAPDH (AbFrontier/Acris Antibodies, Herford, Germany), rabbit anti-Nrf2 (BioWorld Medical, Meggen, Switzerland), rabbit anti–HO-1 (Enzo Life Sciences), rabbit anti–phospho-PI3K p85 (Tyr⁴⁵⁸)/p55 (Tyr¹⁹⁹), rabbit anti–phospho-Akt (Thr³⁰⁸), rabbit anti–phospho-p38 MAPK (Thr¹⁸⁰/Tyr¹⁸²), or rabbit anti–phospho-PKC (pan) (βII Ser⁶⁶⁰) (Cell Signaling Technology/New England Biolabs) Ab (in 20 mM Tris, 138 mM NaCl [pH 7.6], 5% [w/v] BSA, 0.1% [v/v] Tween 20). HRP-conjugated anti-rabbit IgG (in 1× Roti-Block) was used as a secondary Ab for 1 h at room temperature. The LumiGLO system (Cell Signaling Technology/New England Biolabs) was used for detection.

BMM (4 × 10⁶) were seeded in 10-cm dishes and infected for 30 min with B. pseudomallei at the indicated MOI. Twenty-four hours after infection BMM were harvested in ice-cold D-PBS. After being washed, BMM were incubated in hypomolar HEPES buffer (10 mM HEPES [pH 7.6], 15 mM KCl, 2 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT, cOmplete mini EDTA-free protease inhibitor mixture [Roche]) for 10 min on ice. BMM were lysed by the addition of Nonidet P-40. After the nuclei were collected by centrifugation, the supernatant, which contained the cytosolic proteins, was recovered. The pelleted nuclei were rinsed with ice-cold D-PBS and nuclear proteins were extracted subsequently in hypermolar HEPES buffer (20 mM HEPES [pH 7.9], 0.42 M NaCl, 25% glycerol [v/v], 0.2 mM EDTA, 0.5 mM DTT, cOmplete mini EDTA-free protease inhibitor mixture) and agitated for 15 min at 4°C. Protein content was determined using the Bradford method.

BMM were seeded on coverslips in 24-well plates (2 × 10⁵ cells/well) and infected for 30 min with B. pseudomallei at an MOI of 25. Twenty-four hours after infection BMM were washed with ice-cold D-PBS, incubated for 10 min in ice-cold methanol, and washed three times with immunofluorescence (IF) buffer (0.2% [w/v] BSA, 0.05% [w/v] saponin, 0.1% [w/v] sodium azide in D-PBS [pH 7.4]). To block nonspecific Ab binding, BMM were incubated for up to 1 h in IF buffer followed by an overnight incubation at 4°C in a humidity chamber with polyclonal rabbit anti-Nrf2 Ab (in IF buffer, Santa Cruz Biotechnology, Heidelberg, Germany) and a wash in IF buffer. The immunoreacted primary Ab was visualized with green fluorescent Cy2-conjugated goat anti-rabbit IgG (in IF buffer) by incubating for 1 h at room temperature in the dark followed by another wash in IF buffer. For staining of both B. pseudomallei and Nrf2, monoclonal mouse anti–B. pseudomallei 3015γ2b (34) and green fluorescent Alexa Fluor 488 anti-mouse IgG2b as well as a polyclonal rabbit anti-Nrf2 Ab and red fluorescent Cy3-conjugated goat anti-rabbit IgG (in IF buffer) were used as primary and secondary Abs, respectively. Slices were covered with Fluoprep (bioMérieux, Nürtingen, Germany) and observed by fluorescence microscopy with a BZ-9000 microscope (Keyence, Neu-Isenburg, Germany).

Experimental melioidosis was induced by intranasal inoculation of C57BL/6 mice with B. pseudomallei. For intranasal infection, mice were anesthetized i.p. by ketamine/xylazine and inoculated with 30 μl bacterial suspension (500 CFU in D-PBS). Mice were monitored every 24 h for survival postinfection. To evaluate the bacterial burden of internal organs, mice were sacrificed 24 or 48 h postinfection, and their lungs, livers, and spleens were homogenized in D-PBS containing 0.5% Tergitol and 1% BSA and plated onto Ashdown agar in appropriate dilutions. Data are presented as the total bacterial count (CFU) per organ.

SFN (5 mg/kg) and tBHQ (50 mg/kg) or vehicle (DMSO) were administered i.p. once daily starting 2 d before B. pseudomallei inoculation and continued until sacrifice. CoPPIX (10 mg/kg) or vehicle (0.2 M NaOH [pH 7.2] in D-PBS) was given i.p. once daily starting 2 d before inoculation and continued until sacrifice or every 2 d (survival studies). CORM-2 (12.8 mg/kg) or vehicle (DMSO) was administered i.p. 30 min before and 24 h after infection.

Blood was collected from the orbital sinus in ketamine/xylazine–anesthetized C57BL/6 mice 24 or 48 h postinfection. Animals were sacrificed and the tracheas were exposed through midline incision and cannulated with a sterile 20-gauge Introcan Safety i.v. catheter (Braun, Melsungen, Germany). The lungs were instilled and aspirated serially with 1-ml aliquots of sterile D-PBS for a total of 2 ml. BAL cells were pelleted by centrifugation and resuspended in autoMACS buffer (Miltenyi Biotec, Bergisch Gladbach, Germany). The resulting supernatant (BALF) was plated onto Ashdown agar in appropriate dilutions. The BALF was stored at −70°C until use.

Lymphocytes from spleen were prepared using modified protocols as previously described (35). Spleens were removed and splenocyte suspensions were produced by passing them through sterile 70-μm cell strainers. Cells were washed and resuspended in autoMACS buffer.

Nonspecific Ab binding was blocked with FcR blocking reagent (Miltenyi Biotec). Abs used for cell-surface staining were CD45-allophycocyanin-Cy7, F4/80–Pacific Blue (BioLegend, Fell, Germany), Gr-1–allophycocyanin, CD11b–PerCP–Vio 700, Siglec-F–PE, MHC class II–FITC, CD11c-FITC, and CD49b-PE (Miltenyi Biotec). Cells were blocked, stained with Abs according to the manufacturer’s protocol, and fixed for 20 min in 2.25% formaldehyde (in D-PBS). Cells were washed with autoMACS buffer and analyzed using the MACSQuant analyzer (Miltenyi Biotec). Gating strategy for identification of different subsets was based on Hackstein et al. (36). Briefly, cells were gated based on light-scattering (forward scatter, side scatter) characteristics, and viable singlet leukocytes were identified by CD45 expression. Of the CD45⁺ cells, neutrophils were identified by Gr-1^brightCD11b^bright expression. Of the neutrophil-negative fraction, eosinophils were identified by Siglec-F⁺MHC class II⁻ expression. Of the neutrophil and eosinophil negative fraction, macrophages were identified as Siglec-F⁺⁺F4/80⁺ double-positive cells. Total dendritic cells were identified as CD11c⁺Siglec-F⁻CD49b⁻ leukocytes. Cell numbers of subsets were calculated per organ.

Protein levels of cytokines (IL-6, MCP-1, IFN-γ, and TNF-α) in serum and BALF were quantitatively measured using the cytometric bead array mouse inflammation kit (BD Biosciences) and flow cytometry with a MACSQuant analyzer (Miltenyi Biotec) according to the manufacturers’ protocols.

Figures were constructed and statistical analyses performed using GraphPad Prism 5.0. Comparisons between groups were conducted using a Student t test or one-way ANOVA parametric test followed by the Dunnett or Bonferroni post hoc test for multiple comparisons as specified in the figure legends. The Dunnett test was used to compare infected groups with a single noninfected control group. Comparisons between infected groups were performed with the Bonferroni test. Survival curves were compared using the log-rank Kaplan–Meier test. A p value <0.05 was considered to be statistically significant.

We first evaluated whether HO-1 mRNA and protein expression changes in response to infection of primary macrophages with live or heat-inactivated B. pseudomallei E8. Following 6 and 24 h of infection, HO-1 transcripts were significantly induced 2- or 6-fold over control levels, a regulation that was dependent on the viability of bacteria at 24 h (Fig. 1A). Correspondingly, HO-1 protein levels were found to be higher in infected than in both noninfected macrophages and macrophages infected with heat-inactivated B. pseudomallei.

We next investigated the role of several kinases in the regulation of HO-1 expression in response to B. pseudomallei. To this end, macrophages were infected with B. pseudomallei for various periods of time, and cell lysates were analyzed for phosphorylated kinases by Western blot as shown in Fig. 1E. Phosphorylation of PI3K became apparent at 5–15 min postinfection and declined thereafter, whereas phosphorylation of downstream kinase Akt (protein kinase B [PKB]) could be detected after 30 min. Infection caused an activation of p38 MAPK within 10 min and protein kinase C (PKC) within 30 min. GAPDH did not show any change in expression at the various time points (Fig. 1E). In subsequent experiments, BMM were treated with increasing concentrations of respective kinase inhibitors for 1 h followed by infection with B. pseudomallei and incubation for 24 h. As shown by qRT-PCR, LY294002, which specifically inhibits PI3K (Fig. 1B), and SB202190, an inhibitor of p38 MAPK (Fig. 1C), led to a dose-dependent reduction of HO-1 mRNA induction, whereas JNK inhibitor SP600125 and MEK1-p44/42 MAPK inhibitor PD98059 had no effect, although both kinases were phosphorylated following infection (data not shown). Inhibition of PKC by Ro318220 resulted in a dose-dependent decrease of HO-1 mRNA expression as well (Fig. 1D). Thus, our findings suggest that HO-1 expression is induced in response to infection with live B. pseudomallei, and PI3K, p38 MAPK, and PKC are upstream signaling kinases required for HO-1 induction.

We previously demonstrated that Nrf2 is activated in primary macrophages in response to LPS (37). Using IF staining and subcellular fractionation, we now investigated the nuclear accumulation of Nrf2 in macrophages during infection with B. pseudomallei at the indicated MOI as shown in Fig. 2. Immunocytochemical staining indicated that in uninfected macrophages, Nrf2 is localized in both the cytoplasm and the nucleus. However, after bacterial infection (Fig. 2A) or stimulation with the Nrf2 activators SFN and tBHQ as positive controls (Supplemental Fig. 1A), Nrf2 accumulates within the nucleus. Western blot analyses of corresponding nuclear extracts confirmed nuclear accumulation of Nrf2 in response to B. pseudomallei in an MOI-dependent manner (Fig. 2B).

To assess whether infection with B. pseudomallei also leads to enhanced Nrf2 gene expression and whether Nrf2 activation is essential for the B. pseudomallei–induced increase in HO-1 expression, macrophages from Nrf2^−/− and Nrf2^+/+ mice were infected with B. pseudomallei and analyzed after 24 h. Fig. 2C shows that transcription of Nrf2 was doubled in response to infection of wild-type macrophages, whereas both basal and B. pseudomallei–induced gene expression was strongly reduced in Nrf2-deficient macrophages, as expected. Moreover, induction of HO-1 mRNA and protein expression was significantly diminished in Nrf2^−/− compared with Nrf2^+/+ cells. The basal expression levels of HO-1 mRNA did not significantly differ among the macrophages of the respective mouse strains (Fig. 2D). To verify the obtained results, macrophages were treated with increasing concentrations of SFN (Fig. 2E) and tBHQ (Fig. 2F) for 24 h. As expected, the mRNA and protein expression of HO-1 was increased in a dose-dependent fashion by both Nrf2 activators. Treatment of Nrf2^−/− macrophages with SFN or tBHQ did not significantly change HO-1 mRNA and protein levels (Supplemental Fig. 1B). These data strongly suggest that the induction of HO-1 expression by infection with B. pseudomallei is controlled by a mechanism involving Nrf2 activation.

To further characterize the role of Nrf2 during bacterial infection, we investigated the invasion and the course of intracellular bacterial burden in macrophages from Nrf2^−/− and Nrf2^+/+ mice as well as from SFN- and tBHQ-treated C57BL/6 mice at different time points after B. pseudomallei infection. As shown in Fig. 2, there was no difference in both bacterial invasion and bacterial load at 6 h postinfection between Nrf2-deficient and wild-type macrophages (Fig. 2G) and in cells treated with Nrf2 activators or vehicle (Fig. 2H, 2I). However, 24 h postinfection, the deficiency of Nrf2 significantly enhanced bacterial clearance of B. pseudomallei (Fig. 2G), whereas activation of Nrf2 by SFN (Fig. 2H) or tBHQ (Fig. 2I) was associated with an increase in intracellular bacterial numbers in response to infection. Using the LDH activity assay, we determined whether Nrf2 affects the viability of uninfected or infected macrophages and found that LDH release was significantly reduced in Nrf2-deficient compared with wild-type macrophages at 6 and 24 h postinfection (Fig. 2J). In contrast, LDH release from SFN-treated (Fig. 2K) and tBHQ-treated (Fig. 2L) macrophages was only slightly reduced 6 h postinfection, but it was not affected 24 h postinfection compared with control macrophages. Thus, our results indicate that B. pseudomallei–induced Nrf2 activation may facilitate bacterial survival in primary macrophages.

To analyze a potential in vivo role of the transcription factor, Nrf2^−/− and Nrf2^+/+ mice as well as C57BL/6 mice treated with either SFN (5 mg/kg), tBHQ (50 mg/kg), or corresponding vehicle were intranasally infected with B. pseudomallei. Supplemental Fig. 2 demonstrates that neither deficiency of Nrf2 (Supplemental Fig. 2A, 2D) nor administration of SFN (Supplemental Fig. 2B, 2E) or tBHQ (Supplemental Fig. 2C, 2F) led to significantly changed bacterial numbers in lung, liver, and spleen 48 h postinfection. Additionally, concentrations of serum IL-6, MCP-1, TNF-α, and IFN-γ were not significantly altered. These data suggest that the Nrf2-mediated effects, observed in macrophages in vitro (Fig. 2), might be neutralized by other Nrf2-mediated mechanisms during murine melioidosis.

Synthetic metalloporphyrins such as CoPPIX are widely used as inducers of HO-1 activity. We first examined the expression of HO-1 after exposure of macrophages to CoPPIX (25 μM). As shown in Fig. 3A, treatment with CoPPIX significantly increased HO-1 mRNA and protein expression in a time-dependent manner. Infection of CoPPIX-pretreated cells with B. pseudomallei did not further enhance HO-1 expression after 6 and 24 h compared with vehicle-pretreated cells (Fig. 3B). To address whether HO-1 possesses host defense functions during B. pseudomallei infection, we evaluated the effect of induction of HO-1 on bacterial growth by pretreating macrophages with CoPPIX. Fig. 3C illustrates that invasion of B. pseudomallei was not affected by treatment. However, 6 and 24 h postinfection, the number of intracellular bacteria in CoPPIX-treated macrophages was significantly higher than in controls. Furthermore, we evaluated the effect of CoPPIX on cytotoxicity by measuring LDH release of B. pseudomallei–infected macrophages. Treatment with CoPPIX did not significantly change LDH levels in infected or noninfected macrophages compared with vehicle-treated cells (Fig. 3D).

In subsequent experiments we tested the consequences of CoPPIX pretreatment versus posttreatment on intracellular bacterial growth. Thus, macrophages were exposed to CoPPIX (25 μM) 18–20 h prior to infection and in kanamycin-containing medium (pretreatment) as described above. Posttreatment was performed by adding CoPPIX only in kanamycin-containing medium 0, 3, or 6 h postinfection. Fig. 3E indicates that the number of intracellular bacteria in all CoPPIX-treated macrophages was significantly higher compared with the respective vehicle-treated cells at 24 h postinfection, although CoPPIX posttreatment (0, 3, or 6 h) tended to a lower increase of CFU in comparison with CoPPIX pretreatment.

HO-1 and its metabolites have been shown to exhibit beneficial anti-inflammatory properties in pure inflammatory models of disease such as endotoxemia (38). However, inhibition of the inflammatory response to bacterial infection by the HO-1 system might disrupt the ability of the immune system to eliminate invading bacteria. To test this hypothesis, we examined the gene expression of several proinflammatory mediators in CoPPIX-treated (25 μM) and vehicle-treated macrophages postinfection with B. pseudomallei. We observed that induction of HO-1 by CoPPIX decreased mRNA levels of cytokines TNF-α, IL-1β, IL-6, and IL-12 both at 6 and 24 h postinfection (Fig. 4), whereas gene expression of the neutrophil chemoattractants keratinocyte-derived chemokine (KC; CXCL1) and MIP-2 (CXCL2) as well as MCP-1 (CCL2) were not influenced by CoPPIX (data not shown). Thus, pharmacological induction of HO-1 can increase susceptibility to B. pseudomallei infection by manipulating immune effector functions of macrophages.

Additionally, we examined the impact of a HO-1 gene knockdown on intracellular bacterial counts and expression of cytokines as well. Macrophages were treated with either a nonsilencing siRNA (small interfering nontemplate control) or an siRNA directed against HO-1 (siHO-1) and infected with B. pseudomallei. As shown in Fig. 5A, delivery of siHO-1 resulted in a significant decrease of HO-1 mRNA and protein expression compared with the nonsilencing siRNA both in noninfected and B. pseudomallei–infected macrophages. The rate of bacterial uptake and intracellular bacterial counts at 6 h postinfection were not affected by siHO-1, whereas B. pseudomallei counts recovered from macrophages after 24 h of infection were significantly reduced in HO-1 knockdown versus control macrophages (Fig. 5B). However, silencing of HO-1 did not significantly alter gene expression of TNF-α, IL-1β, IL-6, and IL-12 at 24 h postinfection (Fig. 5C). Our data suggest that HO-1 gene knockdown results in improved control of B. pseudomallei infection in macrophages.

We further analyzed the expression of HO-1 in response to intranasal infection of C57BL/6 mice with B. pseudomallei. Twenty-four and 48 h postinfection, HO-1 mRNA levels were increased in lung and liver in response to B. pseudomallei. Moreover, we detected a higher mRNA expression of HO-1 in the spleen of infected mice after 24 h, whereas no differences could be observed after 48 h (Fig. 6A). To determine the role of HO-1 in host defense during in vivo infection, we treated mice with CoPPIX (10 mg/kg) by i.p. injection followed by pulmonary infection with B. pseudomallei. As shown in Fig. 6B, repeated administration of CoPPIX resulted in a strong induction of HO-1 mRNA expression in lung, liver, and spleen of mice 48 h postinfection.

In accordance with the observed effects of CoPPIX treatment on bacterial growth in macrophages, administration of CoPPIX to mice resulted in significantly enhanced bacterial numbers in lung, liver, and spleen, as well as BALF compared with vehicle-treated mice after 24 and 48 h (Fig. 7B). This was associated with a lower survival rate (Fig. 7A) and strongly increased IL-6, TNF-α, and MCP-1 levels in serum (Fig. 7D) and BALF (Fig. 7F), as well as reduced secretion of IFN-γ 48 h after B. pseudomallei challenge. To verify the results, we determined gene expression of several inflammatory mediators in lung, liver, and spleen of CoPPIX- and vehicle-treated mice after bacterial infection and found that cytokines TNF-α, IL-1β, IL-6, and IL-12 as well as chemokines KC and MIP-2 were significantly increased in liver and/or spleen. Furthermore, IL-1β, IL-12, and MIP-2, but not TNF-α, IL-6, or KC mRNA levels were significantly higher in the lung of CoPPIX-treated mice (Table I). Measurement of LDH resulted in slightly higher relative fluorescence units in serum and BALF (Fig. 7C). Recruitment of neutrophils into spleen was slightly decreased in CoPPIX-treated mice, whereas dendritic cells were significantly increased (Fig. 7E). Thus, our results provide evidence that the CoPPIX-mediated induction of HO-1 is detrimental during murine melioidosis.

CORMs serve as carriers for the delivery of controlled amounts of CO in biological systems (39). To evaluate whether the effect of HO-1 on B. pseudomallei infection could be mediated by its metabolite CO, we used the lipid-soluble tricarbonyldichlororuthenium (II) dimer (CORM-2), which releases CO at a rapid rate. For in vitro studies, primary macrophages were exposed to 50 μM CORM-2 or vehicle 1 h prior to infection with B. pseudomallei. As shown in Fig. 8, bacterial invasion and bacterial load at 6 h postinfection were not markedly influenced by treatment with CORM-2 (Fig. 8A). However, 24 h postinfection CORM-2–treated macrophages showed significantly higher intracellular numbers than did controls. CORM-2 tended to result in slightly more LDH production at 6 h, whereas treatment did not alter LDH release by macrophages 0 and 24 h postinfection (Fig. 8B).

In line with our in vitro studies, the in vivo administration of CORM-2 enhanced the bacterial load in lung (p < 0.001), liver, and spleen (Fig. 8C) and tended to result in higher serum levels of TNF-α, IL-6, MCP-1, and IFN-γ (Fig. 8D). These data suggest that the effect of HO-1 in B. pseudomallei infection is the result of the action of at least one of the HO-1 enzymatic end products.

An appropriate balance of the inflammatory and redox states is essential to resolve most infections and finally the infectious process (21). However, the components of the inflammatory and oxidative stress response during melioidosis are unknown. In the present study, we demonstrated that infection with B. pseudomallei induces mRNA and protein expression of HO-1 in primary murine macrophages (Fig. 1A). Additionally, we observed a significant increase in HO-1 expression in both murine macrophage RAW264.7 and hepatoma Hepa1-6 cells (unpublished data). This is consistent with previously published studies demonstrating an upregulation of HO-1 mRNA and/or protein expression in response to bacterial (29, 40–45) or parasitic (26, 27, 46) infections and might be a general response of infected host cells.

Expression of HO-1 is regulated essentially at the level of gene transcription. A number of intracellular signaling molecules, including PI3K/PKB, MAPKs, and PKC have been shown to play major roles in controlling LPS-induced HO-1 gene expression (47–49). However, the mechanisms that regulate HO-1 induction in response to bacterial infections are almost unknown. Although multiple protein kinases are phosphorylated following infection, we found that PI3K, p38 MAPK, and PKC are upstream kinases required for B. pseudomallei–induced HO-1 expression (Fig. 1B–D). This is partially consistent with a recently published study showing that PI3K/phosphatidylinositol 3,4,5-trisphosphate–dependent PKB activity is involved in HO-1 induction in response to S. aureus infection in a murine peritonitis model (42). In the present study, we found an increased transcription (Fig. 2C) and nuclear accumulation (Fig. 2A, 2B) of Nrf2 in primary macrophages following B. pseudomallei infection, which contributes to HO-1 induction (Fig. 2D). MacGarvey et al. (42) observed increased hepatic Nrf2 mRNA expression and nuclear Nrf2 protein expression in S. aureus septic mice as well as nuclear translocation of Nrf2 after S. aureus peptidoglycan administration in human HepG2 cells. However, because HO-1 expression is slightly upregulated in Nrf2 KO macrophages after infection, HO-1 seems to be induced by Nrf2-independent pathways as well (Fig. 2D). As infection with B. pseudomallei induces degradation of the NF-κB inhibitor IκBα in macrophages within 30 min (data not shown), the transcription factor NF-κB might also play a role in HO-1 induction. Both NF-κB and AP-1 have been shown to contribute to regulation of HO-1 expression (50).

In several experimental models, including LPS- and CLP-induced septic shock (17–19) or hyperoxia-induced acute lung injury (51), Nrf2 deficiency was associated with increased inflammation and tissue injury. Absence of Nrf2 was shown to suppress the alveolar transcriptional network for mitochondrial biogenesis and anti-inflammation, which worsens acute lung injury in an S. aureus pneumonia model (40). Moreover, Nrf2 deficiency enhanced inflammation in the lung as well as oxidative stress and inflammatory cytokine expression in alveolar macrophages, promoting susceptibility to Pseudomonas aeruginosa infection following hyperoxic lung injury (51). Targeting Nrf2 signaling seems to improve bacterial clearance by alveolar macrophages in patients with chronic obstructive pulmonary disease as well as in a mouse model of P. aeruginosa or nontypeable Haemophilus influenzae (52).

For several viruses, the Nrf2-mediated induction of cytoprotective genes has been considered to protect infected cells from oxidative damage (53–56). A recent study described the direct interaction of a Marburg virus protein with the Nrf2 inhibitor Kelch-like ECH-associated protein 1 to activate the cytoprotective antioxidant response pathway as part of their replication strategy (57). Comparably, our data show that activation of Nrf2 by SFN and tBHQ increases growth of B. pseudomallei in macrophages (Fig. 2H, 2I), together with slightly reduced or unchanged LDH release (Fig. 2K, 2L), whereas absence of Nrf2 leads to reduced intracellular bacterial counts (Fig. 2G). Thus, B. pseudomallei–induced Nrf2 activation may promote bacterial survival in primary macrophages most likely by induction of cytoprotective antioxidant proteins, such as HO-1 (Fig. 1A) or Prx1 and Prx6 (unpublished data). Although gene expression of HO-1 was upregulated in lung, liver, and spleen of B. pseudomallei–infected mice as well (Fig. 5A), mRNA expression of Nrf2 as well as Prx1 and Prx6 was differentially regulated in vivo with an increased expression in spleen, but markedly decreased hepatic levels (unpublished data). This might also contribute to the different effects of Nrf2 observed in macrophages and mice (Supplemental Fig. 2).

In the present study, we demonstrate that HO-1 gene knockdown results in improved control of infection (Fig. 5B), whereas CoPPIX-treated macrophages display higher bacterial loads (Fig. 3C) without significant changes in LDH release (Fig. 3D). Therefore, B. pseudomallei–mediated HO-1 induction might support bacterial survival and/or replication within host macrophages (Fig. 3). This is consistent with studies showing that HO-1 can induce dormancy-associated genes in Mycobacterium tuberculosis, leading to latency and survival of the pathogen inside macrophages (41, 45). Additionally, adenovirus-mediated overexpression or CoPPIX-mediated induction of HO-1 promoted persistence of Leishmania chagasi in macrophages (46) and in murine Plasmodium liver infection (26), and it was required to protect infected host cells by controlling the inflammatory response. Recent work provided evidence that induction of HO-1 by S. typhimurium exerts detrimental effects in the control of infection, as HO-1 knockdown induced antibacterial immune effector pathways of macrophages and promoted bacterial elimination (58). In our study, the significantly better control of intracellular B. pseudomallei in HO-1 knockdown macrophages was not associated with changes in gene expression of proinflammatory cytokines. However, we found that the higher B. pseudomallei load in macrophages after induction of HO-1 by CoPPIX was accompanied by reduced levels of cytokines (Fig. 4), but the regulatory mechanism is still unclear. We propose that reduced immune effector mechanisms of macrophages during HO-1 induction contribute to the detrimental role of HO-1 during Burkholderia infection. Wang et al. (59) observed that the HO-1/CO pathway is able to suppress LPS-induced TLR4 signaling, including the production of TNF-α and IL-6 in murine macrophages, by regulating the interaction of TLR4 with caveolin-1. Whereas both TLR4 and TLR2 have been found to contribute to cellular responsiveness to B. pseudomallei in vitro, only TLR2 might have an impact on the immune response of the intact host in vivo (60).

Our findings in a murine model of melioidosis show that HO-1 induction by CoPPIX increases bacterial load in organs and BALF (Fig. 6B), and it is associated with reduced survival (Fig. 7A). We also provide evidence that mRNA induction of HO-1 in organs upon B. pseudomallei infection (Fig. 6) is accompanied by increased transcription of diverse proinflammatory cytokines in organs (Table I), lower IFN-γ levels, but higher concentrations of TNF-α, IL-6, and MCP-1 in serum and BALF (Fig. 7D, 7F) of CoPPIX-treated mice. Several reports indicated that IFN-γ is essential for the early control of B. pseudomallei infection in mice (14, 61), whereas high amounts of proinflammatory cytokines are detrimental and contribute to the immunopathogenesis of infection (62). Elevated serum levels of TNF-α, IL-6, and IL-10 have been associated with disease severity in patients with melioidosis (63). Mortality in human melioidosis was also predicted by the presence of serum bilirubin among other factors (64), which may serve as a marker for increased HO-1 activity and inducibility in tissues (65). Consistent with our results, increased concentrations of HO-1 have also been found during Leishmania infection of mice, and elevated serum levels of TNF-α have been associated with human visceral leishmaniasis (46). The high amounts of cytokines might be a secondary effect of elevated bacterial counts in vivo and can induce higher levels of HO-1 as a counterregulatory response.

In the present study, we also observed an increased recruitment of dendritic cells into spleen of CoPPIX-treated mice, whereas neutrophils were decreased (Fig. 7E). Recent research demonstrated that B. pseudomallei–infected dendritic cells can facilitate the systemic spread of the pathogen, although dendritic cells have been previously shown to kill intracellular B. pseudomallei in vitro (66). Additionally, although disproportionate neutrophil recruitment can be detrimental in melioidosis (6, 8), activated neutrophils play a critical role in early bacterial control (11). In agreement with our results, HO-1 has been previously shown to suppress the infiltration of neutrophils in rat liver during CLP sepsis by inactivation of p38 MAPK (67). HO-1 inhibition by zinc deuteroporphyrin 2,4-bis-glycol enhanced neutrophil migration into the bronchoalveolar spaces and improved the outcome of murine K. pneumoniae–induced sepsis (68). Conversely, administration of the HO-1 inducer CoPPIX caused a reduction in neutrophils in BALF after LPS challenge, whereas no obvious difference could be observed with the HO-1 inhibitor ZnPPIX (69). Similarly, we observed a significant decrease in bacterial burden only in spleen of ZnPPIX-treated mice, whereas there were no significant changes in lung and liver. Proinflammatory cytokines tended to demonstrate higher serum concentrations as well. Gene expression analysis of respective organs revealed that HO-1 is upregulated in lung and liver of ZnPPIX-treated mice, but it was not affected in spleen (unpublished data). Furthermore, HO-1 mRNA and protein expression were elevated after treatment of macrophages with ZnPPIX, leading to strongly reduced cytokine expression as well (Supplemental Fig. 4). Thus, it is likely that ZnPPIX is not an appropriate inhibitor of HO-1.

It is well recognized that the properties of HO-1 are closely associated with the release of CO, which exerts important biological activities, including vasorelaxation, anti-inflammatory as well as antiapoptotic and cytoprotective actions in various models of disease (22, 23). CO may effectively interact with several intracellular heme-containing targets (e.g., soluble guanylate cyclase, cytochrome c oxidase, NADPH oxidase, potassium channels) to transduce important signals within cells. Several studies demonstrated that CO derived from CORMs can protect mice from lethal endotoxemia and polymicrobial sepsis induced by CLP (28, 70–73). Recently, CORMs have been described to exert bactericidal (CORM-2, -3) or bacteriostatic (CORM-A1) activities against P. aeruginosa growth in vitro. However, only ruthenium-containing CORM-2 and -3 have been shown to possess potential therapeutic properties by increasing the survival of mice infected with P. aeruginosa (74, 75). CORM-2 significantly reduced bacterial numbers in Salmonella-infected macrophages, as well (58). In contrast, our data indicate that the rapidly CO-releasing molecule CORM-2 as well as CoPPIX have no effect on B. pseudomallei growth in vitro (Supplemental Fig. 3), and they promote survival of the pathogen in macrophages and in mice (Figs. 3, 7, 8). In accordance with the observed effects of CoPPIX-mediated HO-1 induction on bacterial growth, CO might contribute to the establishment of Burkholderia infection in macrophages by decreasing the host inflammatory response. Furthermore, CO and iron have been shown to enhance coagulation and to attenuate fibrinolysis (76). Studies by Shiloh et al. (45) and Kumar et al. (41) revealed that in addition to HO-1, CO can induce the M. tuberculosis dormancy regulon as well. Another study observed that treatment of mice with exogenous CO increases Plasmodium liver load (26), but fully prevents cerebral malaria and malaria-associated acute lung injury incidence in mice (27, 77).

In summary, our study shows that upregulation of HO-1 expression occurs in B. pseudomallei infection, and both HO-1 and HO-1–derived CO favor the establishment of Burkholderia infection, an effect that seems to be mediated by the control of immune effector functions, including the inflammatory response of macrophages. Further studies are necessary to investigate the role of the HO-1 metabolites iron and biliverdin/bilirubin during infection with B. pseudomallei. Our results also indicate that pharmacological targeting of HO-1 activity in melioidosis might have the potential to modulate disease severity.

We are grateful to Eylin Topfstedt and Claudia Weber for excellent technical assistance.

The authors have no financial conflicts of interest.