RegA, the Regulator of the Two-Component System RegB/RegA of Brucella suis, Is a Controller of Both Oxidative Respiration and Denitrification Required for Chronic Infection in Mice

Abstract

Adaptation to oxygen deficiency is essential for virulence and persistence of Brucella inside the host. The flexibility of this bacterium with respect to oxygen depletion is remarkable, since Brucella suis can use an oxygen-dependent transcriptional regulator of the FnrN family, two high-oxygen-affinity terminal oxidases, and a complete denitrification pathway to resist various conditions of oxygen deficiency. Moreover, our previous results suggested that oxidative respiration and denitrification can be simultaneously used by B. suis under microaerobiosis. The requirement of a functional cytochrome bd ubiquinol oxidase for nitrite reductase expression evidenced the linkage of these two pathways, and the central role of the two-component system RegB/RegA in the coordinated control of both respiratory systems was demonstrated. We propose a scheme for global regulation of B. suis respiratory pathways by the transcriptional regulator RegA, which postulates a role for the cytochrome bd ubiquinol oxidase in redox signal transmission to the histidine sensor kinase RegB. More importantly, RegA was found to be essential for B. suis persistence in vivo within oxygen-limited target organs. It is conceivable that RegA acts as a controller of numerous systems involved in the establishment of the persistent state, characteristic of chronic infections by Brucella.

INTRODUCTION

Brucellosis, a zoonosis encountered worldwide (1), is caused by the Gram-negative bacteria of the genus Brucella. Humans have been affected by this very ancient disease (2) most commonly consecutive to infection by the strains B. abortus, B. suis, and B. melitensis. Brucella is an obligate aerobe pathogen; nevertheless, at each time point of natural infection, this bacterium has to resist oxygen depletion. Most infections occur after ingestion of contaminated animal products, which implies a passage through the oxygen-deficient gastrointestinal tract, via Peyer's patches (3). Brucella, which behaves as a facultative intracellular parasite, uses macrophages to multiply and to spread throughout the organism. During intramacrophagic multiplication, bacteria reside inside a specific niche, the so-called Brucella-containing vacuoles (BCV) (4) derived from phagosomes, which are known to have oxygen concentrations lower than those found in the extracellular environment (5). Analysis of the intramacrophagic virulome of B. suis (6) showed that low levels of nutrients and oxygen are major features of its replicative niche. In the absence of treatment, brucellosis may tend to get chronic, resulting in localized infection of liver, spleen, or brain (7, 8, 9), where bacteria are located within granulomatous lesions. Granulomas are multitype immune cell structures in which microaerobic and anaerobic areas coexist and which possibly evolve to oxygen-deprived abscesses at the late stage of the disease in humans.

The pathogenicity of brucellae and chronicity of brucellosis are due to the ability of the pathogen to adapt to these harsh environmental conditions. In previous studies (10, 11, 12), we demonstrated that B. suis is able to use various systems to adapt to oxygen-limited conditions, in vitro as well as inside the host organism. The flexibility of this bacterium with respect to oxygen depletion is actually very high since it can grow under low oxygen concentrations (10) and survive under anaerobic denitrifying conditions (13) without repression of basic metabolic activities, as shown by our proteomic analysis (12). This was therefore in accordance with an earlier classification of several alphaproteobacteria, which distinguished B. melitensis (14) as a bacterium with an expected facultative/aerobic lifestyle, according to its rank 1 position with respect to the number of predicted highly expressed genes of different energy metabolism pathways, denitrification included. B. suis possesses an oxygen-dependent transcriptional regulator of the FnrN family that regulates two high-oxygen-affinity terminal oxidases, the cbb3-type cytochrome c oxidase and the cytochrome bd ubiquinol oxidase, the first one being strongly induced under microaerobiosis (11). Interestingly, contribution of the cbb3 oxidase to bacterial virulence in the mouse model of infection was specific to the sole stage of persistence and was correlated to the oxygen level of organs, as a mutant of B. suis lacking the cytochrome cbb3 oxidase was more attenuated in the liver than in the spleen (10). Moreover, B. suis possesses a complete denitrification pathway consisting of the four reductases Nar (nitrate reductase), Nir (nitrite reductase), Nor (nitric oxide [NO] reductase), and Nos (nitrous oxide reductase). This respiratory system allows Brucella to use nitrogen oxides as electron acceptors when oxygen becomes scarce. Denitrification can provide a double advantage to the bacterium: production of sufficient energy for bacterial persistence under low and very low oxygen conditions and elimination of toxic NO produced by the macrophage during the innate immune response (13).

Under microaerobiosis, the presence of nitrate in the culture medium improves growth of the wild-type (WT) B. suis strain but not of a mutant devoid of cytochrome bd ubiquinol oxidase, which accumulates nitrites (10). This finding suggested that aerobic respiration and nitrite use could be coordinated and controlled by a common regulator. Interestingly, as many other alphaproteobacteria, Brucella possesses the two-component regulatory system RegB/RegA, extensively studied in the phylogenetically close species Rhodobacter capsulatus and Rhodobacter sphaeroides, whose homologue is named PrrB/PrrA (15). RegB is the membrane-spanning histidine sensor kinase, which autophosphorylates on a conserved histidine residue upon sensing a redox signal. It then transfers the phosphate to the cognate response regulator RegA, which binds DNA using the C-terminal helix-turn-helix (HTH) motif. Control of the RegB auto-kinase activity in response to redox changes could be ensured by three nonexclusive modes of regulation. Under oxidizing conditions, the fully conserved cysteine residue (Cys265) located in the cytosolic “redox-box” inhibits kinase activity by forming an intermolecular disulfide bond (16). The highly conserved ubiquinone binding site GGXXNPF, at the N-terminal membrane-spanning domain of RegB, has also been involved in redox sensing (17) by interacting with both oxidized and reduced forms of ubiquinone, only the former inhibiting kinase activity (18). Nevertheless, PrrB of R. sphaeroides, devoid of the proposed ubiquinone fixation site, was shown to still have its kinase activity inhibited by the oxidized ubiquinone (19). The authors defend a model that implies that the cbb3 oxidase acts as a redox sensor and generates a signal to PrrB (19, 20). All the functionally important motifs of RegB are highly conserved in B. suis (15). RegA (PrrA) regulates a considerable number of processes, all implied in energy-generating or energy-utilizing functions, including denitrification (21) and aerobic and anaerobic respiration. Transcriptomic studies have shown that approximately 25% of the R. sphaeroides genome is controlled either directly or indirectly by PrrA, which can regulate gene transcription both positively and negatively (22). Since amino acid sequence alignments have shown that the response regulators RegA of R. capsulatus, R. sphaeroides (PrrA), B. suis, and B. melitensis share a high degree of sequence identity (around 80%) and identical putative HTH motifs (15), we hypothesize that RegA in B. suis could play a role similar to that of R. capsulatus. The present study was undertaken to investigate the role of the transcriptional regulator regA in the coordinated regulation of both oxidative respiration and denitrification and in virulence in a murine in vivo model of infection.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The Brucella strains used in this study, B. suis 1330 (ATCC 23444) and derived mutants, were all grown in tryptic soy (TS) medium. Under standard aerobic conditions, cultures were grown to an optical density at 600 nm (OD₆₀₀) of 1 to 1.2 over a period of 20 h. To perform cultures under microaerobiosis or anaerobiosis, cultures of brucellae at an optical density of 0.2 or 0.5, respectively, were placed in a jar with GENbox generators of microaerobic (oxygen concentration ranging from 6.2 to 13.2% after 1 h) or anaerobic (oxygen concentration of <0.1% after 2.5 h) atmosphere (bioMérieux, Marcy l'Etoile, France). Slow agitation (100 rpm) was applied to the jar to ensure homogenization. Under microaerobiosis, bacteria were grown to an OD₆₀₀ of 0.8 to 1.2 over 30 h or 72 h of incubation. Anaerobic conditions were applied to cultures in TS medium supplemented with 10 mM NaNO₃ for 14 days, and the viability of bacteria was controlled by plating serial dilutions on TS agar. Production and utilization of nitrite by the bacteria were assessed by measuring the concentration of NO₂/NO (spontaneously transformed to NO₂) in the culture supernatant, using the Griess reagent (13).

Strains resistant to kanamycin or chloramphenicol were cultured in the presence of 50 μg ml⁻¹ or 25 μg ml⁻¹, respectively, of the antibiotic in the medium. Escherichia coli strain DH5α was used as the recipient strain for cloning and was routinely grown in Luria-Bertani (LB) medium. The appropriate antibiotics were added when needed: ampicillin, 50 μg ml⁻¹; kanamycin and chloramphenicol, as indicated above. The standard growth temperature for all bacterial strains was 37°C.

DNA manipulation.

Plasmid DNA was isolated from E. coli according to standard procedures. B. suis chromosomal DNA was prepared using the Wizard genomic DNA purification kit (Promega, Madison, WI, USA). DNA amplification products were obtained by using Goldstar DNA polymerase (Eurogentec, Belgium). DNA treatments with restriction and modification enzymes were performed according to the manufacturer's instructions (Invitrogen, United Kingdom).

For cloning, PCR products were purified with the QIAquick gel extraction kit (Qiagen, Germany).

Inactivation of B. suis genes by homologous recombination.

Mutants of B. suis harboring a deleted cydB or fnrN gene were previously obtained by allelic exchange between chromosomal genes and deleted PCR products of these genes cloned into a suicide plasmid (11). Chloramphenicol and kanamycin resistance genes were inserted in deletions of cydB and fnrN, respectively. To obtain the ΔregΑ mutant, B. suis DNA fragments comprising upstream and complete coding sequences of regA were produced by PCR using total genomic DNA as the template. The sequences of the primers (Sigma-Genosys, United Kingdom) and their targets are as follows: regA5, CGCTATCGTCGTATTGCGCTG; regAXba, TGCTCTAGAGCAATCGGAAAACGCACTGC. The PCR product was then cloned into pUC18 using the SphI and XbaI restriction sites, the first one located 67 bp downstream of the regA5 primer and the second in the designed regAXba primer (underlined). The 1.2-kb blunted kanamycin resistance gene (kan) from plasmid pUC4K (Pharmacia Biotech, France) was inserted into the cloned sequence after introduction of the following deletion: 464-bp NsiI + XcmI fragment of regA. B. suis was transformed with this suicide vector by electroporation. Clones of B. suis that had integrated the inactivated regA gene into the chromosome by allelic exchange were selected by their resistance to kanamycin. Isolated clones were screened for the loss of the plasmid-encoded ampicillin resistance marker. Gene amplifications from chromosomal DNAs of the wild-type and of the mutant were performed to verify that adequate allelic exchange took place. A single PCR product was obtained, whose size was increased by the Kan^r cassette in the mutant.

The ΔregA strain of B. suis was complemented in trans with the native regA gene produced from the initial pUC18-clone (see above), subcloned as a 709-bp SspI + ClaI fragment into the replicative plasmid pBBR1MCS, under the control of the lacZ promoter.

Isolation of total RNA from B. suis strains.

For each B. suis 1330 wild-type or mutant strain, RNA samples from three independent cultures were prepared after addition of 1.1 ml ethanol-phenol solution (9:1) to 10 ml of the cultures, to stop de novo RNA synthesis. RNA extractions were performed with the RNeasy minikit from Qiagen according to the manufacturer's instructions (Qiagen, Germany) with a modified lysis step, consisting of an additional incubation with 10% SDS and proteinase K (20 mg/ml), for 10 min at 25°C. A first DNase I treatment (Qiagen) was performed on RNA bound to the column and a second on purified RNA samples with RNase-free Turbo DNase I (Ambion), according to the manufacturer's instructions (Life Technologies SAS, United Kingdom). RNA concentration was determined using the NanoDrop ND-1000 (Labtech), and controlled qualitatively using the Agilent 2100 bioanalyzer. To check the absence of contaminating DNA, RNA samples (0.5 μg) were submitted to direct amplification with fnrN primers (11). After 40 cycles, a fnrN PCR product was not detected in any of the RNA samples used in this study.

RT-qPCR analysis.

Expression levels of nirK, cydD, ccoN, and fnrN genes were measured by quantitative analysis of RNA using a two-step reverse transcription quantitative PCR (RT-qPCR) technique. Denatured RNA samples (0.5 μg; 5 min at 65°C) were reverse transcribed using a 6-mer random primers mix (150 ng), 0.5 mM deoxynucleoside triphosphate (dNTP), and 200 units of SuperScript II reverse transcriptase (RT) (Invitrogen, United Kingdom), at 42°C for 50 min, according to the manufacturer's instructions.

Quantitative PCR experiments were performed using the Light Cycler 480 with SYBR green chemistry to monitor and quantify the amplification rate (Roche). Primers (Sigma Genosys) were designed using Primer 3 Software (Table 1), defining an identical annealing temperature (60°C) for all the primers. Aliquots of cDNA samples were diluted 20-fold and 2,000-fold to quantify expression of the targeted mRNAs cited above and of the constitutively expressed 16S rRNA, respectively, the latter being used to normalize expression values. The expression level of the genes under study was given relative to that of the 16S rRNA, as the ratio between the concentrations (ng/μl) of the specific cDNA and of the 16S cDNA (multiplied by 100 since 16S cDNA is 100-fold more diluted). Concentrations were determined using standard qPCR curves established with a calibrating genomic DNA of B. suis wild type.

Table 1.

Primers used for real-time PCR

Primer	Sequence	PCR product size (bp)
fnrN left	CATGGCTGTGTGTTCCGCACT	139
fnrN right	TGCCGGAGGTCAGGCTATAGA
cydD left	GTGCCTTATCTGTCGCGCTTC	224
cydD right	CGGTCGAGCAGAAAGCCATTC
ccoN left	CTCAACCTCCAGCCCTACTTC	168
ccoN right	CCCAGAATACGAACCAGGCGAGA
nirK left	CCCAGAATACGAACCAGGCGAGA	132
nirK right	ATCGTCCTGATGCACGACCACGA
16S left	GGGGAGCAAACAGGATTAGAT	224
16S right	ATGTCAAGGGCTGGTAAGGTT

Infection and growth of B. suis strains in the BALB/c murine model.

Survival of wild-type, ΔfnrN, and ΔregA strains was measured in 7-week-old female BALB/c mice inoculated intraperitoneally with 10⁵ CFU of either wild-type B. suis or the fnrN or regA mutants. At different time points, on days 7, 28, 56, and 84, spleens and livers of 6 mice per B. suis strain were aseptically harvested. After homogenization of the whole organ in phosphate-buffered saline, serial dilutions were plated on TS agar to determine bacterial counts.

Statistical methods.

Student's t test was applied to the two sets of three independent experiments to be compared. Differential expression was considered to be statistically significant at P values of ≤0.05. For mouse survival, analysis of variance using the Fisher protected least-significant-difference test and the Tukey-Kramer test was performed to determine the level of significance (P) of differences in CFU between the wild-type strain and the mutant of interest.

RESULTS

The high-oxygen-affinity cytochrome bd ubiquinol oxidase is required for nitrite reductase expression.

Our earlier results (10) strongly suggested that oxidative respiration and denitrification can be used simultaneously by B. suis under low oxygen tension, as indicated by the increased growth of wild-type B. suis when nitrate was added as an alternative electron acceptor to the culture medium. Unexpectedly, the growth of the ΔcydB mutant lacking functional bd oxidase was not improved under these conditions. Moreover, during cultures under microaerobiosis in TS broth supplemented with 10 mM (10) or 20 mM (Fig. 1A) sodium nitrate, nitrites produced by the nitrate reductase (13) were utilized by the wild-type strain and totally disappeared from the culture supernatant at 72 h (Fig. 1A). In contrast, the ΔcydB strain was unable to use nitrites, which could reflect a tight link between the bd oxidase and nitrite reductase activities. To test the possibility that nitrite reductase expression needs native bd oxidase, we compared the expression levels of the nirK gene, producing the nitrite reductase enzyme, in the wild-type strain and in the cydB mutant of B. suis, using nirK-specific RT-qPCR. The transcript level of nirK (Fig. 1B) was strongly reduced in the ΔcydB strain, exhibiting an mRNA concentration approximately 4.5-fold lower than that of the wild-type strain at 48 h (P = 0.002), the point at which about one-half of the nitrites were consumed by the latter (Fig. 1A). In contrast, the regA transcription level was found unchanged with respect to that observed in the wild type (not shown). Expression of the nitrite reductase therefore depends on a functional bd oxidase. This result suggested that aerobic respiration and denitrification pathways are tightly coordinated and could be under the control of a common regulator. Interestingly, Elsen et al. (15) identified in B. suis the homologues of the two-component system RegB/RegA (BR0133/BR0137), also existing in B. melitensis and B. abortus (23). This led us to consider RegA to be the candidate transcriptional factor that might control various respiratory systems in B. suis.

Fig 1.

Nitrite utilization and expression of nitrite reductase under microaerobic conditions. Wild-type (●) and cydB mutant (▲) bacteria were cultivated in TS broth with or without (corresponding open symbols) 20 mM NaNO₃ in a jar maintaining a microaerobic atmosphere. At various time points, the nitrite concentration was measured in an aliquot of the culture supernatant (A). At 48 h of incubation, total RNA from both strains was reverse transcribed to quantify nirK mRNA by use of nirK-specific qPCR (B). The means ± standard deviations (error bars) of three independent experiments are shown.

RegA induces nitrite reductase expression.

We investigated the potential role of RegA in the control of nitrite utilization by B. suis under conditions of low oxygen tension, since this bacterium was found to be unable to use nitrites under aerobiosis (not shown). To this aim, the behaviors of wild-type and ΔregA mutant strains were compared under microaerobiosis and anaerobiosis in the presence or absence of additional nitrate (Fig. 2). Production and utilization of nitrite by the bacteria were assessed by measuring the concentration of NO₂ in the culture supernatant. Moreover, the transcription of the nirK gene was analyzed by performing RT-qPCR in harvested wild-type and ΔregA bacteria to correlate nitrite consumption patterns with expression levels of nirK in both strains and therefore to estimate the role of RegA in regulation of Nir activity. Experiments were performed also using the ΔregA strain complemented with the native regA gene to check the specificity of the phenotypic features of the regA mutant.

Fig 2.

Growth or survival and nitrite reductase activity of wild-type and ΔregA mutant strains of B. suis under microaerobiosis (A, B, C) and under anaerobic denitrifying conditions (D, E, F). Wild-type (●), ΔregA (▲), and complemented ΔregA (★) strains were cultivated in TS broth with (A, D) or without (dotted lines, empty symbols) (A, B) 10 mM NaNO₃. At various time points, the nitrite concentration was measured in the supernatant (B, E). nirK transcription in wild-type (white bars), ΔregA (gray bars), and complemented ΔregA (C-ΔregA) (black bars) strains was assessed by RT-qPCR using cDNAs obtained from total RNAs prepared from bacteria harvested after 6 h of microaerobic culture (μO₂) (C), with (+NO₃⁻) or without nitrates, and after 3 days under anaerobic denitrifying conditions (F). Means and standard deviations of three independent experiments are reported.

As previously described (10), addition of 10 mM NaNO₃ to the TS medium had a beneficial effect on the microaerobic growth of the wild-type strain (Fig. 2A), the culture of which reached an optical density of approximately 1.2 at 30 h, instead of 0.8 when nitrate was absent. This effect disappeared in the cultures of the ΔregA mutant strain (Fig. 2A), which showed a rather identical growth rate with or without supplemented nitrate. After a 17-h incubation, the nitrite concentration in the culture supernatant of the ΔregA mutant remained at the maximal level (more than 9 mM) observed at 6 h for both strains and decreased only slightly at 30 h (Fig. 2B). In contrast, a concentration of only 3.7 mM nitrites was measured at 17 h in the culture medium of the wild-type strain. Nitrites were undetectable at 24 h, most likely because nitrogen oxide reductases were fully active. As expected under anaerobic denitrifying conditions, the wild-type B. suis strain is able to survive over a period of 14 days in TS medium containing 10 mM (Fig. 2D) or 20 mM (13) NaNO₃. The ΔregA strain, in contrast, was characterized by a severe loss of viability starting on day 3, as shown by the approximately 4-log-fold reduction in viable bacteria at day 14 (Fig. 2D). In a manner similar to cultures under microaerobiosis, measurement of the nitrite concentration in the medium demonstrated that the wild-type strain very rapidly produced nitrites (day 1) and utilized them until almost total consumption at day 3 (Fig. 2E). In contrast, the ΔregA mutant accumulated nitrites, which reached the maximal concentration of 10 mM at day 1 and remained unchanged until day14 (Fig. 2E). The complementation of the regA mutant led to the recovery of nitrate-dependent improved growth and survival under microaerobiosis and anaerobiosis, respectively, identical to that of the wild-type strain (Fig. 2A and D). Utilization of nitrite was also restored in the mutant strain complemented in trans (Fig. 2B and E).

These results showed that RegA was necessary for the optimal growth of B. suis under microaerobiosis with nitrate and for its survival under anaerobic denitrifying conditions. Under both conditions of low oxygen tension, absence of RegA provoked nitrite accumulation, indicating that this potential transcriptional regulator was essential for nitrite utilization by the denitrification pathway. This was further analyzed by nirK-specific qPCR performed with total RNA prepared from wild-type and ΔregA bacteria harvested during the above-described experiments. In 6-h microaerobic cultures, the expression level of nirK was highly induced in the WT strain consecutively to the 10 mM NaNO₃ supplement in the medium (approximately 3-fold [P = 0.01]), but this transcription activity was extremely decreased in the ΔregA mutant (approximately 30-fold [P < 0.01]) (Fig. 2C). The basal expression rate measured in the absence of nitrate was also affected by the inactivation of regA (P < 0.05). The complementation of the ΔregA mutant restored the nitrate-dependent induction of nirK transcription observed in the WT strain (Fig. 2C). At day 3 of anaerobic denitrifying conditions, the ΔregA strain expressed nirK at an extremely low rate, showing a 10⁴-fold decrease in mRNA levels compared to the WT strain (P < 0.001), whereas the complemented mutant strain recovered nirK expression (Fig. 2F). The specific effect of RegA on the optimal expression of nirK in the parental B. suis strain, as demonstrated by the complementation of the expression defects in the regA mutant, showed its probable role in nitrite reductase upregulation under oxygen deficiency.

RegA is involved in induction of oxidative respiration.

Involvement of RegA in the control of oxidative respiration using as high-oxygen-affinity terminal oxidases the bd or the cbb3-type oxidase was also investigated under microaerobiosis, in TS broth with or without 10 mM NaNO₃. Expression levels of both the bd and cbb3-type terminal oxidases were evaluated by RT-qPCR using as targets the first gene of the corresponding operons cydDCAB and ccoNOQP, respectively. The expression level of cydD in the wild-type B. suis strain was extremely low under aerobiosis (Fig. 3A), and microaerobiosis resulted in a very slight induction, which is in accordance with the low-level expression previously observed in vitro with a transcriptional gfp fusion to the cyd promoter (11).When nitrates were present in the medium, expression of cydD in the wild type remained very low under aerobiosis; in contrast, following 6 h of microaerobiosis this expression became actually significant, showing an approximately 5.5-fold-higher level (P < 0.001) than that obtained without nitrates (Fig. 3A). Under those conditions, the ΔregA strain displayed a strong decrease in microaerobic expression level of the cydDCAB operon, compared to that observed in the wild type (10-fold; P < 0.001). When microaerobic cultures were performed in the absence of nitrates, expression of the cyd operon was also affected in the ΔregA mutant but to a lesser extent (approximately 2-fold, P < 0.001). Complementation of the regA mutant strain also restored microaerobic cydD expression levels observed in the parental strain, in the presence or in the absence of nitrates (not shown). These results showed that RegA is necessary under microaerobiosis for the nitrate-dependent activation of the operon encoding the bd ubiquinol oxidase.

Fig 3.

Expression of the two high-oxygen-affinity terminal oxidases and of the FnrN oxygen-dependent regulator in wild-type and ΔregA strains of B. suis. Expression levels of genes encoding the bd ubiquinol oxidase (A), the cbb3-type cytochrome c oxidase (B), and FnrN (C) were determined by RT-qPCR targeting cydD, ccoN, and fnrN, respectively. Wild-type (empty bars) and ΔregA (filled bars) strains were cultivated in TS broth under aerobic (O₂) (A) or microaerobic (μO₂) (A, B, and C, left ordinate) conditions, in the presence (+NO₃⁻) or absence of 10 mM NaNO₃ or under anaerobic denitrifying conditions (−O₂) (C, right ordinate). Standard deviations are reported for the means of three independent experiments.

We also compared expression levels of ccoN in the ΔregA and wild-type strains of B. suis in bacteria grown under microaerobiosis, a condition inducing maximal activation of the cbb3-type oxidase (11). The use of ccoN-specific RT-qPCR showed that in the regA mutant, expression levels of the cco operon were reduced by a factor of 2 (P ≤ 0.001), in a nitrate-independent manner, after overnight incubation under low oxygen tension (Fig. 3B). Therefore, the RegA transcriptional regulator is an activator of the cbb3-type terminal oxidase under these conditions. Addition of nitrates in the culture medium resulted in a 1.8-fold and a 1.6-fold decrease of ccoNOQP transcription levels, respectively, in both the wild-type and ΔregA strains (P < 0.005). This indicates that, in contrast to the cytochrome bd oxidase, the cytochrome cbb3 oxidase was repressed by nitrates.

We have previously shown that the promoter of the fnrN gene, encoding the transcriptional regulator FnrN, was itself submitted to regulation when oxygen concentration became low, as depicted by its increased activity in bacteria exposed to microaerobiosis (11). We therefore examined if fnrN transcription could be controlled by RegA under low oxygen tension. fnrN mRNA levels were compared in the wild-type and ΔregΑ strains following 6 h of microaerobiosis or 3 days of anaerobiosis under denitrifying conditions. Results showed in the mutant an approximately 4-fold and a 22-fold (P < 0.001) decrease of fnrN transcript level under microaerobiosis (Fig. 3C, left ordinate) and anaerobiosis (Fig. 3C, right ordinate), respectively. Although fnrN was considerably less expressed in the absence of oxygen, its expression was still subjected to positive control by RegA. Therefore, the transcriptional regulator fnrN was activated by RegA under both conditions of oxygen deficiency.

Inactivation of fnrN affected virulence of B. suis in the mouse model of infection.

Our earlier results have revealed that survival of the fnrN-deleted B. suis strain was not altered inside human THP-1 macrophage-like cells (11). In another study, however, we obtained evidence that this model of in vitro cellular infection does not reflect in vivo oxygenation conditions (10). We therefore decided to investigate the possible impact of FnrN on B. suis pathogenicity in a mouse model of infection. The behavior of the wild-type B. suis strain was compared to that of the ΔfnrΝ mutant during the course of infection by determining bacterial replication and survival within target organs harvested at different time points. In the spleen (Table 2), the difference in the bacterial load was less than 1 log between the two strains at 1 week postinfection. This indicated that FnrN had little effect during the initial phase of multiplication. Moreover, the ΔfnrΝ strain exhibited persistence similar to that of the parental strain at 12 weeks postinfection. In the liver (Table 2), the two strains exhibited a similar level of multiplication at 1 week postinfection (0.7-log difference). At week 12, the ΔfnrN mutant displayed an attenuated phenotype with bacterial counts reduced by more than 2 log (P < 0.05) compared to the wild type. At this time point, ΔfnrN was eliminated from 50% of the infected mice, whereas the wild-type strain was eliminated from only 16% of the mice. This result indicated that fnrN was involved in B. suis persistence in the liver only, probably because the FnrN transcriptional regulator modulated expression of persistence genes under very low oxygen concentration, which is the case in the liver but not in the spleen (24).

Table 2.

Participation of FnrN in B. suis virulence in BALB/c mice

Organ	No. of wks postinfection	Mean log CFU ±SD
		B. suis 1330	B. suis ΔfnrN
Spleen	1	7.34 ± 0.14	6.49 ± 0.11
	12	4.03 ± 0.59	3.61 ± 0.42
Liver	1	6.00 ± 0.31	5.29 ± 0.13
	12a	2.38 ± 0.62	0.18 ± 1.017

^aSignificant difference (P < 0.01).

RegA is required for chronic infection in BALB/c mice.

In order to evaluate a possible role of the transcriptional regulator RegA in virulence of B. suis, we examined whether deletion of regA had an impact on bacterial survival in mice. In fact, inactivation of regA was without any effect on B. suis multiplication inside human THP-1 and murine J774A.1 macrophage-like cells (not shown). BALB/c mice were infected with wild-type B. suis 1330 and the ΔregΑ mutant, and the numbers of viable bacteria were determined inside spleens and livers at different time points. In the spleen, the two strains behaved similarly, showing curves consisting of three phases: multiplication during the first week, reduction in CFU numbers between 1 and 4 weeks, and persistence until 12 weeks postinoculation (Fig. 4A). In the liver, however, the two strains displayed different behaviors after a similar phase of initial multiplication (0.6-log difference) (Fig. 4B). Absence of functional regA severely impaired persistence of B. suis, since the ΔregA strain showed a continuous decline in CFU numbers from week 4 until total elimination at week 12 (Fig. 4B). In contrast, the parental strain persisted until week 8, when bacterial counts were 2.5 log higher than for the ΔregA strain. At the end of the experiment, the wild type survived 30-fold better than the mutant. Under our conditions, i.e., at late time points of infection, immune activation could elicit formation of granulomas. Within these organized structures of immune cells, oxygen tension can become extremely low inside the liver, a poorly oxygenated organ (24), making regA essential for bacterial persistence.

Fig 4.

Course of infection of B. suis wild-type strain (●) and the regA mutant (△) in spleens (A) and livers (B) of BALB/c mice. Bacteria were recovered from the organs at 7, 28, 56, and 84 days postinfection. The arrows indicate the dose (10⁵ CFU) inoculated intraperitoneally. Error bars represent the standard deviations of the means. Results of statistical analysis are indicated by asterisks (**, P < 0.01; ***, P < 0.001).

DISCUSSION

Inside the host, brucellae have to adapt to low and variable oxygen concentrations, particularly when pathogens enter a phase of persistence within various low-oxygenated organs. Both microaerobic respiration (10) and denitrification pathways (13) were found to be active during infection of macrophage-like cells and essential for B. suis virulence in a murine model of in vivo infection. In B. neotomae, denitrification was also shown to favor long-term persistence in mice (25).

Brucella is therefore able to use different respiratory systems implying a tightly coordinated control of regulatory systems that ensure fast adaptive response to the evolving encountered conditions. Importantly, under in vitro microaerobic conditions, nitrate conferred an advantage to the growth of wild-type B. suis but not of the ΔcydΒ mutant (10). Both strains transformed nitrate into nitrite, but only the wild-type strain consumed the latter. Here, we showed the requirement of a functional cytochrome bd oxidase for high levels of nirK expression. These results indicated the simultaneous use of oxidative respiration and denitrification, with a probable linkage of these two pathways. In the current study, we further investigated the possible role of the two-component system RegB/RegA in the coordinated control of the two respiratory systems. In Rhodobacter species, RegB/RegA is a redox-responding system involved in the regulation of numerous systems, all implied in energy-generating and energy-utilizing processes (15).

We gained evidence that RegA is necessary for the use of nitrite by the denitrification pathway, as demonstrated by its requirement for the nitrite reductase expression under microaerobic as well as anaerobic conditions. This explains why a B. suis strain devoid of RegA lost growth advantage under microaerobiosis in the presence of nitrates or its capacity to survive under anaerobic denitrifying conditions. Its inability to perform complete denitrification prevents the production of energy necessary for the bacteria under these conditions. Another explanation could be the toxicity of various nitrogen oxide intermediates accumulated by the ΔregΑ strain. However, the mutant showed a slower but continuous growth under microaerobiosis even when nitrite concentration became high, which was evidenced by determining viable bacteria counts (not shown). Therefore, neither nitrites nor possible low levels of toxic NO produced by the residual nitrite reductase activity in ΔregA displayed bactericidal effect. In fact, the regA deletion strongly affected expression of the nor and nos operons (not shown), although this could be also the consequence of substrate limitation due to lack of nitrite reductase activity. At least a low NO reductase activity persisted in the ΔregΑ strain, consuming 0.8 mM NO₂/NO during the course of microaerobic incubation. The positive control exerted by RegA on nirK expression could be either direct or indirect via the R. sphaeroides Nnr homologue NnrA, which has been shown in B. melitensis to activate expression of the last three reductases of the denitrification pathway and to participate in virulence (26). Direct control may be possible, as potential recognition sites of RegA (27) were found upstream of the start codon of nirK, as well as of that of the nor and nos operons (not shown). Moreover, nnrA was found to be poorly expressed at day 3 of incubation under anaerobic denitrifying conditions and to be not significantly controlled by RegA at both 6 h and overnight time points of microaerobiosis in the presence of nitrates (not shown). Both NnrA and RegA regulators might be independently involved in nitrite reductase expression of B. suis, as is the case in R. sphaeroides (21). Nevertheless, since putative fixation sites for RegA were also found upstream of nnrA (not shown), a possible control of nnrA expression by RegA under certain conditions cannot be excluded at present.

We also demonstrated that RegA induced expression of both the bd and the cbb3-type terminal oxidases of high affinity for oxygen, as shown for R. capsulatus (28), but in the latter species both oxidases are highly expressed under microaerobic conditions. Interestingly, in wild-type B. suis, expression of cydD, the first gene of the operon encoding the bd oxidase, was more responsive to nitrates than to oxygen deficiency, since no marked variation of its expression level was observed following exposure to microaerobiosis. Activation of the bd oxidase expression by nitrates also strengthened the RegA-mediated link between the oxidative respiration and denitrification pathways, as suggested by the fact that this activation required regA. The oxygen-dependent transcriptional regulator FnrN was unlikely to be involved in this nitrate-dependent induction, since it was previously shown to repress in vitro expression of the bd oxidase (11), similarly to the repression exerted by FnrL in R. capsulatus under microaerobic conditions (28).

The cbb3-type terminal oxidase was also found to be positively controlled by RegA, acting either directly or indirectly, or both under microaerobiosis. In a previous work, we showed that microaerobic induction of the cco operon is strictly dependent on fnrN, although the expression levels were measured indirectly by using a transcriptional gfp fusion to the cco promoter (11). The identification of two potential RegA binding sites in the upstream sequences of ccoN (not shown) may suggest a direct regulation by RegA. In contrast to the bd oxidase, the ccb3-type cytochrome c oxidase was repressed by nitrates. Since nitrates still repressed ccoN expression in the absence of RegA, this transcriptional regulator was probably not involved in this repression.

In conclusion, RegA is necessary for the induction of the two high-oxygen-affinity terminal oxidases. However, they are differentially controlled with respect to their maximal induction, the cbb3 oxidase exhibiting the highest level of expression under microaerobiosis alone, whereas the cytochrome bd oxidase required the addition of nitrates. The opposite effects of nitrate on expression levels of the two oxidases also reinforced the idea that they could be differentially used by B. suis in vitro (20). Our previous studies have clearly demonstrated their divergent roles in virulence in vivo (10, 11). More strikingly, only the cbb3 oxidase is essential for persistence in mouse organs. The correlation between the contribution of the cbb3 oxidase to bacterial persistence and the oxygen levels in the organs and its oxygen-dependent activation via FnrN are consistent with a role of this high-oxygen-affinity terminal oxidase in optimum use of oxygen for successful adaptation to oxygen-limited conditions inside the host. Taking into account the fact that activation of nirK, encoding nitrite reductase, needs a functional bd oxidase, and by analogy to the model proposed for R. sphaeroides, we hypothesize that this terminal oxidase might play a role in redox signal transmission to RegB. In contrast, the cbb3 oxidase, but not the bd oxidase, is implied in both redox sensing and modulating the PrrB activity in R. sphaeroides (19) and, moreover, is required for nirK expression (21). In Fig. 5, we propose a tentative model for coordinated regulation of both oxidative respiration and denitrification by RegA in B. suis. If nitrate is present in the low-oxygenated environment of the bacteria, the bd oxidase will be active, following induction by RegA. As oxygen tension decreases, the electron flow through the bd oxidase decreases, and this slowing down might be sensed by RegB, which would then further activate RegA. RegA then induces expression of the target genes nirK and fnrN, encoding a transcriptional regulator itself, as shown in the present results under various conditions of oxygen deficiency, and it also induces directly or indirectly the cco operon. As a potential oxygen sensor, the transcriptional regulator FnrN may also be directly activated by the drop of the oxygen pressure and acts on the genes encoding the two terminal oxidases, activating and repressing the cco and cyd operons, respectively. The bd oxidase of B. suis might be induced to function as a redox sensor and a signal transmitter to RegB rather than as a “true” high-oxygen-affinity terminal oxidase. The fact that the noticeable induction of the bd oxidase needed nitrates could be a means to activate nirK via RegA only when complete denitrification is possible. When oxygen tension is low and consequently nitrite reductase activity is efficient, the bd oxidase would become dispensable and may be repressed by FnrN. The model described above obviously necessitates further detailed studies to determine the mechanisms of fine-tuning involved in the coordination of respiratory systems by RegA. Nevertheless, it sheds light on some of the questions pointed out (29) regarding recent data on the function of the NtrY/NtrX two-component system as a redox sensor involved in microaerobic induction of the denitrification genes (30). Not only is the nitrite reductase under the positive control of both the NtrYX and the RegBA/PrrBA redox signal transduction systems, but the two last enzymes of the denitrification pathway also depend on these regulatory systems. Searching for a link between these two systems (29), it is important to point out that we recently found that RegA induces expression of ntrY, as well as of the nifR3 and ntrB genes of the NtrBC two-component system (our unpublished results).

Fig 5.

Global regulation by RegB/RegA in B. suis under oxygen deficiency. On the basis of high homology and functional similarity of RegB/RegA in Brucella and Rhodobacter species, the existence of a signal (dashed arrow) generated by the bd oxidase and detected by the redox sensor RegB can be postulated (see the text for further explanations). Positive or negative regulation is indicated.

In the context of chronic infection by Brucella, our main result concerns the impact of RegA on B. suis persistence in vivo, which is correlated to the oxygenation level of the target organ. Therefore, RegA is indeed a global regulator of adaptation to oxygen deficiency, essential for persistence of B. suis within low-oxygenated niches of the host organism. Moreover, analysis of the intramacrophagic B. abortus proteome has revealed the highest increase of PrrA/RegA concentration at a late state of infection (44 h), after full adaptation to low oxygen tension (20 h), when bacteria have reached a phase of extensive multiplication (31). The involvement of the potential direct oxygen sensor FnrN in B. suis persistence within murine livers also underlined the importance of a low-oxygen environment in the establishment of a persistent state. Oxygen depletion has been identified as one of the main signals inducing nonreplicating persistence (NRP), which allows immune containment at the chronic stage of infection by Mycobacterium tuberculosis (32), another obligate aerobic bacterium. PrrA was described as a master controller modulating expression of 25% of the genome in R. sphaeroides, not only of genes directly implicated in energy-related processes interacting to preserve the redox balance, but also of genes belonging to all functional categories (22). According to these data, the identification of numerous RegA-dependent systems in B. suis involved in the establishment of the persistent state, characteristic of chronic infections, is expected and is presently under investigation.