Mapping of the Denitrification Pathway in Burkholderia thailandensis by Genome-Wide Mutant Profiling

Burkholderia thailandensis is a soil-dwelling saprophyte that is often used as surrogate of the closely related pathogen Burkholderia pseudomallei, the causative agent of melioidosis and a classified biowarfare agent. Both organisms are adapted to grow under oxygen-limited conditions in rice fields by generating energy through denitrification. Microoxic growth of B. pseudomallei is also considered essential for human infections. Here, we have used a Tn-Seq approach to identify the genes encoding the enzymes and regulators required for growth under denitrifying conditions. We show that a mutant that is defective in the conversion of N₂O to N₂, the last step in the denitrification process, is unaffected in microoxic growth but is severely impaired in biofilm formation, suggesting that N₂O may play a role in biofilm dispersal. Our study identified novel targets for the development of therapeutic agents to treat meliodiosis.

ABSTRACT

Burkholderia thailandensis is a soil saprophyte that is closely related to the pathogen Burkholderia pseudomallei, the etiological agent of melioidosis in humans. The environmental niches and infection sites occupied by these bacteria are thought to contain only limited concentrations of oxygen, where they can generate energy via denitrification. However, knowledge of the underlying molecular basis of the denitrification pathway in these bacteria is scarce. In this study, we employed a transposon sequencing (Tn-Seq) approach to identify genes conferring a fitness benefit for anaerobic growth of B. thailandensis. Of the 180 determinants identified, several genes were shown to be required for growth under denitrifying conditions: the nitrate reductase operon narIJHGK2K1, the aniA gene encoding a previously unknown nitrite reductase, and the petABC genes encoding a cytochrome bc₁, as well as three novel regulators that control denitrification. Our Tn-Seq data allowed us to reconstruct the entire denitrification pathway of B. thailandensis and shed light on its regulation. Analyses of growth behaviors combined with measurements of denitrification metabolites of various mutants revealed that nitrate reduction provides sufficient energy for anaerobic growth, an important finding in light of the fact that some pathogenic Burkholderia species can use nitrate as a terminal electron acceptor but are unable to complete denitrification. Finally, we demonstrated that a nitrous oxide reductase mutant is not affected for anaerobic growth but is defective in biofilm formation and accumulates N₂O, which may play a role in the dispersal of B. thailandensis biofilms.

IMPORTANCE Burkholderia thailandensis is a soil-dwelling saprophyte that is often used as surrogate of the closely related pathogen Burkholderia pseudomallei, the causative agent of melioidosis and a classified biowarfare agent. Both organisms are adapted to grow under oxygen-limited conditions in rice fields by generating energy through denitrification. Microoxic growth of B. pseudomallei is also considered essential for human infections. Here, we have used a Tn-Seq approach to identify the genes encoding the enzymes and regulators required for growth under denitrifying conditions. We show that a mutant that is defective in the conversion of N₂O to N₂, the last step in the denitrification process, is unaffected in microoxic growth but is severely impaired in biofilm formation, suggesting that N₂O may play a role in biofilm dispersal. Our study identified novel targets for the development of therapeutic agents to treat meliodiosis.

INTRODUCTION

Bacteria inhabit various environmental niches with highly diverse physicochemical parameters, of which the availability of oxygen is a key factor. While oxygen respiration is the most efficient energy-generating reaction (1, 2), bacteria can often utilize a wide range of electron acceptors, including sulfate, ferric iron, and nitrogen oxides, or produce energy via fermentation (3–5) in the absence of oxygen to generate a proton motive force. This not only allows bacteria to colonize a wide range of environmental niches but is also important during infection (6–9). In fact, anaerobic respiration has been implicated in the pathogenesis of various bacterial infections and was shown to affect many functions, including biofilm formation, virulence, persistence, motility, and intracellular growth (10–13).

Under oxygen-limiting conditions, nitrate can be utilized to generate a proton motive force to power growth via denitrification. Denitrification involves the reduction of the N-oxides nitrate (NO₃⁻) and nitrite (NO₂⁻) to nitric oxide (NO), nitrous oxide (N₂O), and dinitrogen gases (N₂) (2, 14). A total of 9 mol of ATP per 2 mol of nitrate was estimated to be produced in the denitrification pathway through the proton activity of the nitrate reductase (15, 16). The energy yield for each step of the denitrification process remains elusive, although each step was predicted to be energetically favorable (17). In the commensal microbe Escherichia coli, energy is generated via formate oxidation and reduction of NO₃⁻ and NO₂⁻ into ammonium (NH₄⁺), a process known as dissimilatory NO₃⁻ reduction (18).

The genus Burkholderia sensu stricto comprises more than 100 species (19), which inhabit highly diverse ecological niches and have been isolated from soil, plants, insects, industrial settings, hospital environments, and infected humans (20, 21). While some Burkholderia strains have attracted considerable interest from the agricultural industry for bioremediation of recalcitrant xenobiotics, plant growth promotion, and biocontrol purposes, the genus also comprises two class 3 pathogens, Burkholderia mallei and Burkholderia pseudomallei, and a growing number of Burkholderia species have been reported as opportunistic pathogens in humans. B. pseudomallei is the causative agent of melioidosis, an infection of humans and animals that is often fatal (9, 22–24). The clinical manifestations of melioidosis are variable and may include pneumonia, septicemia with abscesses in multiple organs, and latent infections that can persist for many years. B. mallei, the etiological agent of glanders, is a highly evolved obligate pathogen of horses, mules, and donkeys with no other known natural reservoir (25, 26). The disease can also occur in humans with occupational exposure to glanderous animals. B. pseudomallei and B. mallei are potential biowarfare agents and have been included on the Centers for Disease Control and Prevention list of biothreat agents as category B agents (27, 28). Burkholderia thailandensis is a soil-dwelling saprophyte that is closely related to B. pseudomallei and B. mallei but is rarely associated with human infections (29–31). Given that B. thailandensis shows a high degree of genetic similarity, it is often used as surrogate for its pathogenic relatives (32, 33).

Despite the fact that many Burkholderia species live in oxygen-limited environments, only species of the B. pseudomallei lineage can reduce nitrate as a terminal electron acceptor to dinitrogen under anoxic conditions (11, 34, 35). In B. pseudomallei, the ability to form biofilm aggregates was found to be associated with the presence of a functional nitrous oxide reductase (36). More recently, it has been demonstrated that biofilm formation of B. pseudomallei is inhibited by exogenous addition of sodium nitrate and that the denitrification pathway affects biofilm growth dynamics through regulation of the intracellular levels of the secondary messenger c-di-GMP, which controls the transition between motile and sessile lifestyles (37). In B. thailandensis, disruption of the biosynthetic pathway for molybdopterin, a cofactor of the nitrate reductase, prevented growth under anoxic conditions with nitrate as the sole terminal electron acceptor and resulted in defects in motility and biofilm formation (11). In spite of the importance of denitrification both for persistence in the natural environment and during infection, knowledge of the underlying molecular mechanisms in members of the genus Burkholderia is scarce.

In this study, we employed transposon sequencing (Tn-Seq) (38) to identify genetic determinants that provide a fitness benefit for anaerobic growth of B. thailandensis with either NO₃⁻ or NO₂⁻ as an electron acceptor. This approach allowed us to identify not only the genes encoding the enzymes required for denitrification but also regulators of the anaerobic metabolism. The analysis of defined mutants revealed that (i) the conversion of NO₃⁻ to NO₂⁻ provides the cells with sufficient energy for fast growth, while the subsequent reduction of NO₂⁻ to molecular N₂ allowed only poor growth under denitrifying conditions, (ii) the denitrification process is regulated by a Fnr-type regulator (ANR) as well as the two-component system RoxRS, and (iii) the inactivation of nosZ does not affect growth but inhibits biofilm formation, possibly due to the accumulation of N₂O.

RESULTS

Bioinformatics analysis of the denitrification pathway in B. thailandensis E264.

A recent report revealed that the clinical isolate B. pseudomallei 1026b has the coding capacity for nitrate sensing, metabolism, and transport (37). In a first step, we determined whether orthologs of the genes predicted to be involved in denitrification in this pathogen are also present in B. thailandensis E264 by using a protein-protein Basic Local Alignment Search Tool (39). This strain contains two chromosomes with a total of 6.72 million base pairs and 5,712 predicted genes (40). In B. pseudomallei 1026b, two operons (Bp1026b_I1015–20 and Bp1026b_II1222–25) were found to encode nitrate reductases, which catalyze the first step of the denitrification pathway, namely, the reduction of NO₃⁻ to NO₂⁻ (41). Inspection of the B. thailandensis E264 genome revealed the presence of two orthologous gene clusters (BTH_I1851–56-narIJHGK2K1 [94.7 to 99.2% nucleotide identities] and BTH_II1249–52-narZYWV [90.8 to 96.4% nucleotide identities]). Furthermore, homologs of the two putative copper nitrite reductases of B. pseudomallei 1026b (Bp1026b_II1540 and Bp1026b_II1580), which convert NO₂⁻ to NO, are also present in B. thailandensis E264 (BTH_II0944 and BTH_II0881 [86 and 93.3% nucleotide identities]). The third step in the denitrification pathway, the reduction of NO to N₂O, is predicted to be performed by the nitric oxide reductase NorB (Bp1026b_I0974) in B. pseudomallei 1026b, and a homolog (BTH_I1813 [96.2% nucleotide identity]) is also found in the B. thailandensis E264 genome. A second gene, BTH_II0945 (Bp1026b_I0974 [32.3% nucleotide identity]), is annotated as a nitric oxide reductase gene in the B. thailandensis genome (41). Both nitric oxide reductases belong to the quinol-oxidizing single-subunit class. A homolog of the cytochrome c subunit NorC of the Pseudomonas aeruginosa PAO1 nitric oxide reductase (PA0523) is missing in the genomes of B. pseudomallei 1026b and B. thailandensis E264 (42). Finally, a gene coding for the putative nitrous oxide reductase NosZ (BTH_I2325, Bp1026b_I1546 [96.8% nucleotide identity]), which is required for the reduction of N₂O to N₂, could be identified in the B. thailandensis genome (41). Overall, the denitrification genes of B. pseudomallei 1026b share 86 to 99% nucleotide identities to B. thailandensis E264.

Tn-Seq experimental setup.

To identify genetic determinants required for anaerobic denitrification in B. thailandensis E264, a Tn-Seq approach was employed. We used plasmid pLG99, which includes a Tn5 derivative transposon (Tn23) carrying an outward-facing rhamnose promoter, to avoid off-target polar effects (38). A pool of B. thailandensis Tn23 mutants with approximately a half-million unique insertion sites in nonessential protein and RNA coding regions was generated (see Table S1 in the supplemental material). Our high-resolution Tn mutant library covers the full length of the B. thailandensis genome, with on average one insertion every 16 bp. The mutant library was grown under oxic and oxygen-depleted conditions with either NO₃⁻ or NO₂⁻ as a terminal electron acceptor for 5 to 7 generations (Fig. 1A).

FIG 1.

Identification of fitness determinants affecting anaerobic growth under denitrifying conditions by Tn-Seq. (A) Illustration of the Tn-Seq experimental procedure. The mutant pool was grown aerobically (control sample) or in the absence of oxygen with either NO₃⁻ or NO₂⁻ as an electron acceptor. (B) Numbers of genetic determinants identified that provide a fitness benefit under anoxic condition with NO₃⁻ and/or NO₂⁻ as a terminal electron acceptor. (C) Plot that shows the fitness values of the cells grown with NO₃⁻ (x axis) or NO₂⁻ (y axis). The 85 NO₃⁻ and 120 NO₂⁻ fitness determinants are shown in blue and green, respectively. Genes of interest are depicted.

After sequencing of the Tn junctions (38), we compared the relative abundances of insertions in the mutant pools obtained under aerobic and anaerobic growth conditions. Genes with a log₂ fold change of at least −0.5 between the aerobic and anaerobic samples were considered to provide a fitness benefit for growth under anoxic conditions (see Materials and Methods for details). Using these settings, about 3% of the predicted B. thailandensis genes were found to play a role in anaerobic growth with NO₃⁻ or NO₂⁻ as a terminal electron acceptor.

Fitness determinants required for anaerobic growth in the presence of NO₃⁻ or NO₂⁻.

With the threshold applied, our genome-wide fitness profiling identified 85 and 120 genetic determinants that are required for anaerobic growth with NO₃⁻ and NO₂⁻ as electron acceptors, respectively. Of these determinants, a set of 25 core genes was found to be important for growth under both anoxic growth conditions (Fig. 1B and C; see Tables S2 to S4 in the supplemental material). To gain insight into the functions required for anaerobic growth with NO₃⁻ and NO₂⁻, the identified genes were assigned to four main functional categories of Clusters of Orthologous Groups (COGs: “Cellular processes and signaling,” “Information storage and processing,” “Metabolism,” and “Poorly characterized”) (43; http://clovr.org/docs/clusters-of-orthologous-groups-cogs) (see Fig. S1 in the supplemental material). This analysis revealed that approximately one-fourth of the fitness genes were not categorized or were coding for hypothetical proteins. Several COGs in the main category “Metabolism” were overrepresented, namely, those for energy production (C) and coenzyme metabolism (H). For example, Tn mutants in the nitrate reductase-encoding genes narIJHG (BTH_I1851 to BTH_I1854 [COG C]) and the associated molybdenum synthesis genes (e.g., moeA; BTH_I1704 [COG H]), were depleted under both anoxic conditions (in the presence of NO₃⁻ or NO₂⁻) (Table 1 and Tables S2 to S4). The importance of these genes with nitrite as a terminal electron acceptor is possibly caused by traces of nitrate present in the rich medium used for mutant selection (15). norB, which codes for a putative nitric oxide reductase (BTH_I1813), was also within the core genes. This result suggests that NO, a toxic intermediate of the denitrification process, needs to be converted into the less toxic N₂O, similar to what has been reported for P. aeruginosa (44). Interestingly, the gene coding for the nitrous oxide reductase NosZ (BTH_I2325) was not among the genetic determinants that provide a fitness benefit for anaerobic growth. In line with this finding, it was shown that a P. aeruginosa PAO1 nosZ mutant is unaffected in growth under the anoxic condition (15). To validate our Tn-Seq results, we constructed B. thailandensis E264 mutants in which narG (BTH_I1854), norB (BTH_I1813), or nosZ (BTH_I2325) has been inactivated. While in the presence of NO₃⁻, both the narG and the norB mutants exhibited reduced anaerobic growth compared to the wild type after 24 h (Fig. 2A); the narG mutant was able to grow with NO₂⁻ as the electron acceptor (Fig. 2B). However, both mutants were able to reduce NO₃⁻ to N₂ (Fig. 3), suggesting that the narIJHG paralogs BTH_II1249–52 (narZYWV) and the norB ortholog BTH_II0945 are functionally redundant. The nosZ mutant was not affected in anaerobic growth (Fig. 2A), despite its inability to reduce N₂O (Fig. 3C). These results are in good agreement with the literature, highlighting the validity of our Tn-Seq approach.

TABLE 1.

List of fitness determinants predicted to be involved in the denitrification metabolism of B. thailandensis E264^a

Locus tag	Orthologb	Gene product	Description	COGa	Log2 FCc
					NO3−	NO2−
BTH_I1851	Bp1026b_I1015	NarI	Respiratory nitrate reductase subunit gamma	C	-2.63	-1.18
BTH_I1852	Bp1026b_I1016	NarJ	Nitrate reductase subunit delta	C	-1.51	-0.86
BTH_I1853	Bp1026b_I1017	NarH	Nitrate reductase subunit beta	C	-2.08	-1.19
BTH_I1854	Bp1026b_I1018	NarG	Nitrate reductase subunit alpha	C	-1.89	-0.83
BTH_I2975	Bp1026b_I3350	PetC	Ubiquinol-cytochrome c reductase, cytochrome c1	C	N/A	N/A
BTH_I2976	Bp1026b_I3351	PetB	Ubiquinol-cytochrome c reductase, cytochrome b	C	-0.53	-1.53
BTH_I2977	Bp1026b_I3352	PetA	Ubiquinol-cytochrome c reductase, iron-sulfur subunit	C	-0.046	-1.45
BTH_II0881	Bp1026b_II1580	AniA	Multicopper oxidase domain-containing protein	C	0.10	-1.88
BTH_I1813	Bp1026b_I0974	NorB	Nitric oxide reductase	P	-0.98	-2.15
BTH_I1850	Bp1026b_I1014	NarX	Nitrate/nitrite sensory protein	T	-1.35	-0.32
BTH_I1849	Bp1026b_I1013	NarL	DNA-binding response regulator	T	-2.38	-0.95
BTH_I0162	Bp1026b_I3313	RoxS	Sensor histidine kinase	T	-2.81	-2.82
BTH_I0163	Bp1026b_I3312	RoxR	DNA-binding response regulator	T	-1.88	-2.63
BTH_II0035	Bp1026b_II0032	ANR	Crp/FNR family transcriptional regulator	K	0.003	-0.61
BTH_I1856	Bp1026b_I1020	NasA	Nitrate/nitrite transporter	G	-0.89	-0.31

^aCOG, Cluster of Orthologous Groups; C, energy production and conversion; P, inorganic ion transport and metabolism; T, signal transduction mechanisms; K, transcription; G, carbohydrate transport and metabolism.

^bB. pseudomallei 1026b.

^cThe library of Tn mutants was grown with either NO₃⁻ or NO₂⁻ as the terminal electron acceptor. The log₂ fold change (FC) of the nUID between the aerobic and the anaerobic (NO₃⁻ or NO₂⁻) samples is shown. N/A, not available (i.e., the FC could not be calculated because of zero values).

FIG 2.

Growth of the wild type and various mutant strains under denitrifying conditions. The arrows indicate the time points when the library of B. thailandensis mutants was sampled for Tn-Seq analysis. The wild-type strain and defined mutants, in which narG, petB, aniA, norB, nosZ, roxS, and anr had been inactivated, were grown in the absence of oxygen with either NO₃⁻ (A and C) or NO₂⁻ (B and D) as the terminal electron acceptor. OD₆₀₀, optical density at 600 nm. For each experiment, three independent replicates (n = 3) were performed and the standard error of the mean (SEM) is shown. An asterisk represents a significant difference between the wild type and the respective mutant (P < 0.05 according to two-way analysis of variance [ANOVA], Dunnett’s posttest).

FIG 3.

Concentrations of denitrification products in the supernatants of the B. thailandensis wild type, the narG, petB, aniA, norB, and nosZ denitrification mutants, and the roxS and anr regulatory mutants. NO₃⁻ and NO₂⁻ concentrations were measured with the Brucine method as described by Nicholas and Nason (71) (A and B). Gas chromatography was employed to measure the quantity of the gaseous compounds N₂O and N₂ (C and D). Three independent experiments (n = 3) were performed, and the SEM of each independent experiment is shown. An asterisk represents a significant difference between the wild type and the respective mutant (P < 0.05 according to two-way ANOVA, Dunnett’s posttest). All mutants produced significantly less N₂ compared to the wild type, as shown in panel D.

BTH_II0881 (AniA) and BTH_I2976 (PetB) are involved in the conversion of nitrite to nitric oxide.

While the nitrate reductase catalyzing the first step of denitrification in B. thailandensis was identified previously (11), the enzyme required for the reduction of NO₂⁻ to NO is not known. Two classes of structurally different respiratory nitrite reductases have been described in the literature: the multiple copper cluster (CuNiR) and the c-type cytochromes/d₁ heme cofactor (cd₁NiR) (45). Our Tn-Seq analysis identified several genes that are potentially involved in the reduction of NO₂⁻. BTH_II0881 codes for a periplasmic multicopper oxidase domain-containing protein, which is located in an operon with genes that encode an SCO1/SenC family copper chaperone (BTH_II0879) and a cytoplasmic membrane protein of undetermined function (BTH_II0880, PF03781). BTH_II0881 does not show nucleotide sequence similarity to any protein in P. aeruginosa PAO1 (37) but shares more than 61% sequence identity (60% coverage) with the copper-containing nitrite reductase AniA of Neisseria gonorrhoeae NCCP11945 (46). Furthermore, two genes (BTH_I2976–77) that are organized in an operon appear to be important fitness determinants for anoxic growth with NO₂⁻ as an electron acceptor. BTH_I2977 is annotated as petA, a potential ubiquinol-cytochrome c reductase gene, and BTH_I2976 is an ortholog of BPSL3122 (petB) of B. pseudomallei K96243 (99% sequence identity, 100% coverage) encoding a cytochrome b (41). To corroborate our Tn-Seq data, we constructed a B. thailandensis E264 petB (BTH_I2976) mutant and an aniA (BTH_II0881) mutant. Both knockout strains showed reduced growth rates compared to the wild-type strain under anoxic conditions with NO₃⁻ and NO₂⁻ as terminal electron acceptors. The most severe growth defect was observed for the petB and aniA mutants in the presence of NO₂⁻ (Fig. 2A and B). To gain further insights into the functions of aniA and petB, the mutants were grown under the anoxic condition and the concentrations of NO₃⁻ and NO₂⁻ were determined. Both mutants transiently accumulated NO₂⁻ and showed reduced levels of N₂ production after 96 h of incubation (Fig. 3), suggesting that they are impaired in the reduction of NO₂⁻ to N₂. This may explain the growth defect of the mutants with NO₂⁻ as an electron acceptor.

The denitrification pathway is controlled by a complex regulatory network.

Two-component signal transduction systems (TCSs) are used by bacteria to sense and respond to various environmental stimuli. TCSs are in most cases composed of a membrane-anchored sensor histidine kinase and a cytosolic response regulator (47). Our Tn-Seq analysis identified two TCSs that provide a fitness benefit under denitrifying conditions: narXL (BTH_I1849–50), the P. aeruginosa homologs of which have been demonstrated to regulate denitrification (48), and BTH_I0162–63, which are similar to the RoxSR TCS of P. aeruginosa MH38 (64% sequence identity, 87% coverage), which is involved in redox signaling and can activate expression of several oxidases, such as the terminal cbb_3-1, terminal cbb_3-2, bo₃, and the cyanide-insensitive oxidases (7). To confirm our Tn-Seq results, we constructed a B. thailandensis E264 roxS (BTH_I0162) mutant and tested the strain for growth under denitrifying conditions. The mutant exhibited a prolonged lag phase with NO₃⁻ and was unable to grow with NO₂⁻ as a terminal electron acceptor (Fig. 2). In line with these results, we observed that the roxS mutant showed a delayed reduction of NO₃⁻, an accumulation of N₂O, and a reduced production of N₂ compared to the wild-type strain (Fig. 3). These results suggest that the TCS RoxSR of B. thailandensis E264 may be involved in the regulation of NO₂⁻ reduction. Furthermore, our Tn-Seq analysis identified an additional regulatory gene, BTH_II0035, which encodes a Crp (cAMP receptor protein)-Fnr (fumarate and nitrate reductase) family-type transcriptional regulator, which was found to be important for anoxic growth in the presence of NO₂⁻ (Fig. 1C and 2D). BTH_II0035 is homologous to the transcriptional regulator ANR (anaerobic regulation of arginine deiminase and nitrate reduction) of P. aeruginosa PAO1 (41% sequence identity, 86% coverage). In P. aeruginosa, Crp-Fnr family transcriptional regulators control expression of the denitrification enzymes via a complex regulatory cascade (7). Consistent with these findings, we found that a B. thailandensis E264 anr (BTH_II0035) mutant showed a growth defect under anoxic conditions (Fig. 2C and D), accumulated NO₂⁻, and produced significantly less N₂ relative to the wild type (Fig. 3B and D). To investigate whether this regulator directly controls genes involved in denitrification, we screened for the presence of consensus FNR-binding motifs (49) in the promoter regions of candidate genes identified in our study. Sequences with high similarity to the consensus sequence were found in the promoter regions of aniA (BTH_II0881), petB (BTH_I2976), norB (BTH_I1813), and nosZ (BTH_I2325) (see Fig. S2 in the supplemental material), suggesting that the transcriptional regulator ANR is involved in the regulation of these denitrification genes in B. thailandensis E264.

The B. thailandensis nosZ gene is not important for denitrification but is important for biofilm formation.

Intriguingly, our list of genes required to sustain growth of B. thailandensis E264 under denitrifying conditions did not contain a nitrous oxide reductase, which converts N₂O to N₂. A putative nitrous oxide reductase gene (BPSL1607) was previously shown to be upregulated in a biofilm overproducer strain of B. pseudomallei (36), suggesting a role of this enzyme in cell aggregation. To shed light on the function of the homologous gene in B. thailandensis E264, BTH_I2325 (nosZ), we investigated its expression by quantitative reverse transcription-PCR (qRT-PCR). The nosZ gene was found to be 10-fold upregulated when the strain was grown anaerobically in the presence of NO₃⁻, similar to the denitrification genes narG, petB, aniA, and norB, as well as the anaerobic regulator genes roxS and anr (see Fig. S3 in the supplemental material). This suggests that despite its coexpression under denitrifying conditions, nosZ is not required for anaerobic growth. To investigate the role of nosZ in biofilm formation and swimming motility, we constructed a respective B. thailandensis E264 mutant and tested it along with all other denitrification mutants constructed in this study. With the exception of the narG and aniA mutants, all strains formed significantly less biofilm than the wild-type strain, with the strongest defect observed with the nosZ mutant (Fig. 4A). Swimming motility was significantly reduced in the narG, petB, roxS, and anr mutants relative to the wild-type strain, whereas the nosZ mutant was unaffected (Fig. 4B). Collectively, these findings suggest that (i) the denitrification pathway has a major influence on whether B. thailandensis cells are in the sessile or motile mode of life and (ii) N₂O appears to play a role in biofilm dispersal.

FIG 4.

A The B. thailandensis nosZ mutant is defective in biofilm formation but not in swimming motility. The wild-type strain (WT), the narG, petB, aniA, norB and nosZ denitrification mutants, and the roxS and anr regulatory mutants were tested for biofilm formation (A) and swimming motility (B). Three independent experiments (n = 3) were performed, and the SEM of each independent replicate is shown. Mutants with asterisks (P < 0.05) are significantly affected in biofilm formation or swimming motility relative to the wild-type strain, using an unpaired t test.

DISCUSSION

Members of the B. pseudomallei complex (Bpc) (50), which includes B. pseudomallei, B. thailandensis, and B. mallei, among a few other species, are the only Burkholderia species for which denitrification has been demonstrated (11, 51). While disruption of the molybdenum cofactor biosynthesis pathway was shown to affect anaerobic viability of B. thailandensis (11), the components and regulation of the denitrification pathway in members of the Bpc remained unknown. In this study, we employed a Tn-Seq approach to comprehensively identify genetic determinants required for growth under denitrifying conditions using NO₃⁻ or NO₂⁻ as the terminal electron acceptor. Among the 180 fitness determinants, we identified genes encoding so far unknown components of the denitrification pathway, including enzymes for the conversion of NO₂⁻ to NO, the nitrite reductase AniA (BTH_II0881), and the PetCBA cytochrome bc₁ (BTH_I2975–77), as well as novel regulators of denitrification in the Bpc, including the TCSs RoxSR (BTH_I0162–63) and NarXL (BTH_I1849–50) and the Crp-Fnr family-type transcriptional regulator ANR (BTH_II0035) (Fig. 5).

FIG 5.

Schematic model of the denitrification pathway and its regulation in B. thailandensis E264. The identified regulators NarXL (BTH_I1850-49), RoxSR (BTH_I0162-63), and ANR (BTH_II0035) are thought to respond to different stimuli (NO₃⁻, redox status, and low O_2, respectively) and to induce the biosynthesis of the denitrification enzymes: the nitrate (NarIJHG; BTH_I1851–54, NarZYWV; BTH_II1249–52), nitrite (AniA; BTH_II0881, BTH_II0944), nitric oxide (NorB; BTH_I1813, BTH_II0945), and nitrous oxide (NosZ; BTH_I2325) reductases. The cytochrome bc₁ PetABC (BTH_I2975–77) potentially transfers electrons to AniA for the reduction of NO₂⁻ to NO. Genes that have been mutated in this study are underlined.

The first step of the denitrification pathway is the reduction of NO₃⁻ into NO₂⁻ (15). Expectedly, our Tn-Seq fitness screen revealed that the NarIJHG nitrate reductase and the molybdopterin biosynthesis enzyme MoeA (BTH_I1704), which is required for NarIJHG functioning, are important for anaerobic growth in the presence of NO₃⁻. An narG mutant had a reduced growth rate with NO₃⁻ as the terminal electron acceptor but reached a similar culture density to the wild type after 48 h (Fig. 2A), suggesting that the orthologous cytoplasmic membrane complex nitrate reductase NarZYWV can functionally substitute for NarIJHG. This is different from the situation in B. pseudomallei K96243, where an narG mutant (BPSL2309) was shown to be unable to grow anaerobically (52), although the orthologous narZ (BPSS1159) gene was found to be upregulated in hypoxia (53). Likewise, in P. aeruginosa PAO1, the absence of the NarIJHG complex was shown to be compensated for by a second nitrate reductase enzyme, the periplasmic NapABC (2).

Two classes of bacterial dissimilatory nitrite reductase enzymes, which catalyze the reduction of NO₂⁻ to NO, were previously defined on the basis of whether they contain copper (CuNiR) or heme c and heme d₁ (cd₁NiR) cofactors (45). The accumulation of NO₂⁻ and the growth defect observed with a defined aniA (BTH_II0881) knockout strain (Fig. 2 and 3B) suggested that this gene encodes the nitrite reductase of B. thailandensis E264. Moreover, a petB (BTH_I2976) mutant, which is defective in the synthesis of a cytochrome bc₁, was unable to metabolize NO₂⁻ under anoxic conditions, while aerobic growth was found to be unaffected (see Fig. S4 in the supplemental material). Expression of the orthologous petB gene (BPSL3121) in B. pseudomallei K96243 was previously shown to be induced under a hypoxic condition (53), but a function in denitrification has not been reported before. Cytochromes c or copper-containing azurins are required for electron transfer to CuNiR (54), suggesting that the cytochrome bc₁ synthesized by PetCBA (BTH_I2975-77) might be involved in the transfer of electrons to the CuNiR AniA in B. thailandensis.

The growth behavior of the aniA and the petB mutants under denitrifying conditions indicated that nitrate reduction generates sufficient energy for growth (Fig. 2A), similar to what has been reported for a P. aeruginosa nir mutant (15, 55). This result is of particular interest, given that diverse Burkholderia strains, including Burkholderia ambifaria AMMD, Burkholderia ubonensis Bu, Burkholderia multivorans D2214, and Burkholderia vietnamiensis G4, Burkholderia gladioli BSR3, and Burkholderia glumae PG1 and BGR1, possess only the nitrate reductase genes narIJHG or narZYWV but not the additional genes required for complete denitrification (41). The presence of the nitrate reductase in these bacteria suggests that these strains could survive in an anoxic environment by using nitrate reduction, similarly to the facultative anaerobe E. coli, which converts NO₃⁻ into NH₄⁺ via dissimilatory nitrate reduction (56). This is an important aspect since pathogenicity has been previously associated with denitrification in clinical isolates from cystic fibrosis patients, among which B. multivorans was able to grow anoxically by NO₃⁻ respiration (57). Accordingly, nitrate-reducing species may have an increased pathogenicity resulting from their ability to grow and proliferate in hypoxic environments.

Although our bioinformatics analysis identified two potential nitric oxide reductases in the genome of B. thailandensis E264 (BTH_I1813 and BTH_II0945), our Tn-Seq data provided evidence that only the enzyme encoded by BTH_I1813 (norB) is required for denitrification (Table 1). This enzyme is predicted to be a single-subunit enzyme since no norC homolog could be found in the genome (42). Interestingly, a norB mutant was unable to grow anaerobically with NO₂⁻ as the terminal electron acceptor (Fig. 2B), suggesting that in the absence of a functional nitric oxide reductase, toxic NO accumulates. For P. aeruginosa, it has been reported that the regulatory protein NirQ controls nir and nor transcription to avoid overproduction and toxicity of NO (58). B. thailandensis does not possess a NirQ homolog but has the Crp-Fnr family transcription regulator BTH_II0035 (ANR). In contrast to P. aeruginosa PAO1, in which ANR is essential for denitrification (59), we observed that the B. thailandensis E264 anr mutant transiently accumulated NO₂⁻ and was delayed in the production of N₂, possibly by regulating expression of aniA, petB, norB, and nosZ.

We identified the TCS RoxSR (BTH_I0162–63) as a key regulator for anaerobic metabolism in B. thailandensis E264. In fact, an roxS mutant was unable to grow with NO₂⁻ as a terminal electron acceptor (Fig. 2D) and showed a reduced rate of NO₃⁻ reduction (Fig. 3A), indicating a role in the denitrification of NO₂⁻. The RoxSR system is homologous to the PrrBA and RegBA systems in photosynthetic bacteria, which activate transcription of photosynthesis genes and control expression of nitrite reductase genes under oxygen-limited conditions (7, 60). In P. aeruginosa, RoxSR regulates the terminal cbb_3-1, terminal cbb_3-2, bo₃, and cyanide-insensitive oxidases by sensing the redox state of the respiratory chain (7). Further studies are needed to investigate whether the cytochrome bc₁ PetCBA and the nitrite reductase AniA are similarly regulated by RoxSR in B. thailandensis.

Our Tn-Seq analysis provided evidence that, in analogy to P. aeruginosa (15), the last denitrification step (N₂O to N₂) is not required for anaerobic growth of B. thailandensis. However, the analysis of a defined nosZ mutant revealed that the strain was severely affected in biofilm formation (Fig. 4A) and accumulated N₂O (Fig. 3C), suggesting that this molecule may play a role in cell dispersal. In support of this hypothesis, it has been demonstrated that in a B. pseudomallei biofilm overproducer strain, nosZ (BPSL1607) is upregulated relative to a non-producer strain (36). It is also interesting to note that various studies have demonstrated that low, nontoxic levels of NO, rather than N₂O, induce biofilm dispersal in many bacterial species (61). NO has been reported to trigger the transition from the sessile to the planktonic mode of growth via ligation to one of the two types of characterized NO sensors, NosP or H-NOX (62). Binding of NO to these sensors either reduces the cellular levels of the secondary messenger cyclic di-GMP concentrations or modulates quorum sensing, both eventually leading to dispersal. However, we were unable to identify homologs of NosP or H-NOX in the B. thailandensis genome. Future work will therefore aim at identifying the N₂O sensor and the downstream regulatory cascade leading to the dispersal of B. thailandensis biofilms. Moreover, it will be interesting to see whether N₂O, an easily accessible medical gas that is commonly used for sedation and pain relief, also could serve as a dispersal signal in bacteria outside the Bpc lineage, which may open novel avenues for the treatment of biofilm infections.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

The bacterial strains, plasmids, and primers used in this study are listed in Table S5 in the supplemental material. Bacterial strains were routinely grown in lysogeny broth (LB) (63) under oxic (21% O₂) or anoxic (<1.0% O₂; GasPak EZ Anaerobe Container System [BD]) conditions. In the absence of oxygen, media were supplemented with 10 mM NaNO₃ or 5 mM NaNO₂ (Sigma). Suitable antibiotics were added when required at the following concentrations: for Escherichia coli, 30 μg/ml chloramphenicol (Cm), 25 μg/ml kanamycin (Kan), and 50 μg/ml trimethoprim (Tp); for B. thailandensis, 80 μg/ml Cm and 100 μg/ml Tp. All experiments were performed at 30°C, and the aerobic cultures were shaken at 200 rpm.

Tn-Seq methodology in B. thailandensis E264.

To construct a Tn library of mutants, 40 matings were performed by mixing the B. thailandensis E264 wild-type strain (64) with an equivalent volume of E. coli containing the plasmid pLG99, which carries the Tn5-based transposon Tn23 (38). The mixed suspension was plated on LB medium and incubated for 24 h at 30°C. Next, the mixture was resuspended in 1.5 ml 0.9% NaCl and 10 μl (13 per mating) was plated on Pseudomonas isolation agar (PIA) (Sigma) supplemented with 100 μg/ml Tp (total of 493 plates). After 48 h of incubation, the colonies were collected using 250 μl LB supplemented with 50 μg/ml Tp, mixed with 50% glycerol, and stored at −80°C. The Tn mutants were grown in LB medium under either oxic or anoxic (10 mM NaNO₃ or 5 mM NaNO₂) conditions for 5 to 7 generations in the presence of 0.2% rhamnose (Sigma) before being pelleted and stored at −80°C.

The sequencing of the DNA libraries was performed by using the circle method (38, 65). The Tn-Seq fitness analysis was performed as previously described (66). Briefly, the Tn-Seq Explorer (67) open source software was used to map trimmed forward reads to the B. thailandensis E264 genome (41) and calculate the unique insertion density (UID: number of unique insertion counts per gene divided by length in bases). The UID was normalized by the total number of unique insertions, a value termed “normalized unique insertion density” (nUID). As a result, fitness determinants required for anaerobic competence were selected with a log₂ fold change (FC) of ≤−0.5 and a difference of at least 0.0045 in the nUID relative to the control aerobic sample as a refined stringency to restrict the number of potential hits and maximize the likelihood to identify genes that govern a strong fitness phenotype. To estimate a log₂ FC threshold that separates the genes conferring a fitness benefit from those that do not, we plotted the log₂ fold change of the nUID values of anaerobic versus aerobic growth against the frequency at which they appear (see Fig. S5 in the supplemental material). In this type of plot, a bimodal distribution is observed, whereby genes conferring a fitness benefit are found in the small left peak and genes that can tolerate transposon insertions are in the main peak. It is at the crossover point (−0.5) that the predicted threshold for essential genes is taken. The “ggplot2” and “Venn” packages from R were used to draw the plot and the Venn diagram (68, 69). The EggNOG 4.5.1 mapper was employed to assign the genes to COGs (43), and the assessment of the possible overrepresentation of the categories was performed online (Fisher tests; https://www.graphpad.com/quickcalcs/contingency1.cfm). GraphPad Prism 5 software (version 5.01) was used to analyze the data (growth, biofilm, swimming, and qRT-PCR). An unpaired t test or a two-way analysis of variance (ANOVA) was performed to compare the wild type against each mutant. A P value of <0.05 indicates a significant difference.

Construction of mutants.

To validate the Tn-Seq fitness data, insertion mutants were constructed by single crossover with the broad-host-range suicide plasmid pSHAFT2 (70). This vector carries an outward-facing promoter to allow for transcription of downstream genes when inserted into the chromosome. Briefly, an internal fragment of each gene was amplified by PCR using HF Phusion polymerase (Thermo Scientific). The plasmid DNA was extracted with the QIAprep Spin Miniprep kit (Qiagen, catalog no. 27106). The resulting plasmids with the ligated gene fragment were transformed by heat shock into E. coli cc118 λpir. After triparental mating with the helper E. coli pRK2013 strain, the donor E. coli cc118 λpir strain, and the recipient B. thailandensis wild-type strain, transconjugants were selected by plating on PIA medium supplemented with 80 μg/ml Cm. After two purification steps, the correct insertion in B. thailandensis was checked by PCR. The wild-type and mutant strains were grown as described above.

Quantification of denitrification products.

Bacterial strains were grown anoxically in LB medium with 10 mM KNO₃. Butyl rubber stoppers were used to seal the Hungate tubes, which were filled with argon gas. After 24, 72, and 96 h, the remaining NO₃⁻ and NO₂⁻, N₂O, and N₂ were measured. The Brucine method was used to analyze the NO₃⁻ concentration, and NO₂⁻ was measured as described by Nicholas and Nason (71). Gas chromatography (GC-8AIT; Shimadzu) was used to measure N₂O and N₂ gases, with helium as the carrier gas. The detection of N₂ and N₂O was done by using a Molecular Sieve 5A column and a Shincarbon ST column, respectively (72).

Biofilm formation and swimming assays.

Biofilm formation was assessed with crystal violet in a microtiter plate as described by Huber et al. (73) after 48 h with the following modifications: bacteria were grown in ABC medium (74) amended with 10 mM NaNO₃, and 120 μl of dimethyl sulfoxide (DMSO) was added to solubilize the attached bacteria after crystal violet staining. The biofilm was quantified by using the Biofilm Index (BI) (75). Swimming motility was tested by inoculating the bacterial suspensions onto a 0.2% LB agar plate (76).

qRT-PCR.

Total RNA was extracted from the B. thailandensis wild type grown under oxic and anoxic conditions to early stationary phase (77). An RNeasy kit (Qiagen, catalog no. 74104) was used to purify the samples. After DNase treatment, a control PCR (40 cycles) was performed to check the complete removal of the undesired DNA. The first strand of cDNA was synthesized as previously described (35). For the purification of the cDNA, the Qiagen MinElute PCR purification kit (catalog no. 28004) was used, and the cDNA product was quantified by determination of the absorbance at 260 nm (Nanodrop 2000; Thermo Scientific). qRT-PCR was performed by using the 2× SYBR green III master mix (Agilent), 5 μM primers, and 3.75, 7.5, or 15 ng/μl of cDNA for each reaction (triplicates) in a total volume of 30 μl. An AMx3000P instrument (Agilent, Switzerland) was used to quantify expression of those genes. The threshold cycle (ΔΔC_T) method was used to calculate the fold changes in expression normalized to the expression values obtained for the sigma factor-encoding gene rpoD (BTH_I0621) (78).

Data availability.

The raw FASTQ files created by the MiSeq Illumina platform are publicly accessible on the NCBI Short Reads Archive (SRA) platform and are found in the Bioproject “Burkholderia thailandensis E264 Tn-Seq (Aerobic versus Anaerobic),” under accession no. PRJNA604112 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA604112). Each sample can be found under the following accession numbers: aerobic data set 1, SRX7654887; aerobic data set 2, SRX7654888; anaerobic NO₃⁻ data set, SRX7654889; and anaerobic NO₂⁻ data set, SRX7654890.