Identification, Functional Studies, and Genomic Comparisons of New Members of the NnrR Regulon in Rhodobacter sphaeroides

Abstract

Analysis of the Rhodobacter sphaeroides 2.4.3 genome revealed four previously unidentified sequences similar to the binding site of the transcriptional regulator NnrR. Expression studies demonstrated that three of these sequences are within the promoters of genes, designated paz, norEF, and cdgA, in the NnrR regulon, while the status of the fourth sequence, within the tat operon promoter, remains uncertain. nnrV, under control of a previously identified NnrR site, was also identified. paz encodes a pseudoazurin that is a donor of electrons to nitrite reductase. paz inactivation did not decrease nitrite reductase activity, but loss of pseudoazurin and cytochrome c₂ together reduced nitrite reduction. Inactivation of norEF reduced nitrite and nitric oxide reductase activity and increased the sensitivity to nitrite in a taxis assay. This suggests that loss of norEF increases NO production as a result of decreased nitric oxide reductase activity. 2.4.3 is the only strain of R. sphaeroides with norEF, even though all four of the strains whose genomes have been sequenced have the norCBQD operon and nnrR. norEF was shown to provide resistance to nitrite when it was mobilized into R. sphaeroides strain 2.4.1 containing nirK. Inactivation of the other identified genes did not reveal any detectable denitrification-related phenotype. The distribution of members of the NnrR regulon in R. sphaeroides revealed patterns of coselection of structural genes with the ancillary genes identified here. The strong coselection of these genes indicates their functional importance under real-world conditions, even though inactivation of the majority of them does not impact denitrification under laboratory conditions.

Nitric oxide (NO) is a diffusible signaling molecule in both eukaryotic and bacterial organisms (6, 43). Denitrification can be an important source of NO in the environment (7). During denitrification, NO is produced by nitrite reductase (Nir) and serves as a terminal oxidant during its reduction by nitric oxide reductase (Nor). Evidence indicates that NO is the effector of transcriptional regulators that regulate the genes required for its production and consumption during denitrification, including both Nir and Nor (24, 37). Prokaryotes contain a diverse array of transcriptional regulators whose function is modulated by NO. A number of these regulators are members of the FNR/CRP family (43). One member of this family, termed NnrR for nitrite and nitric oxide reductase regulator, is found most commonly in denitrifiers that utilize a copper-containing nitrite reductase instead of a heme-containing nitrite reductase (22). Members of the FNR/CRP family typically bind a small-molecule signal and are then activated for DNA binding (44). NnrR has not been shown to directly bind NO; however, available evidence strongly suggests that NO is the activating signal (12, 24). A related protein, DNR, has recently been shown to contain heme, which is the likely site of NO binding (14).

Rhodobacter sphaeroides 2.4.3, sometimes referred to as strain 17025, is a purple nonsulfur bacterium that is capable of denitrification and also has a host of metabolic attributes that include aerobic respiration and anoxygenic photosynthesis. Denitrification is utilized only when oxygen is limiting and respirable nitrogen oxides, like nitrate or nitrite, are readily available (50). Oxygen controls expression of nitrate reductase in 2.4.3 (A. Hartsock and J. P. Shapleigh, unpublished data) and has been shown to control expression of nirK, the structural gene for Nir, via the PrrBA global two-component response regulatory system (27). NnrR mediates the nitric oxide-dependent regulation of nirK and norCBQD, the operon encoding Nor (45, 46). The synthesis of nitrite reductase results in production of NO, which acts as an activating signal for NnrR. This creates a feedback loop in which enhanced NO production leads to increased expression of nirK and norCBQD. Presumably, coordinated expression of these key denitrification enzymes is required to prevent accumulation of toxic levels of NO. NnrR also activates expression of nnrS, which encodes a heme-copper membrane protein with an unknown function, and also negatively autoregulates its own expression (4, 45). The promoters of genes under NnrR control in 2.4.3 share a sequence (5′-TTG[C/T]GNNNNC[G/A]CAA-3′) that is required for expression and is predicted to be the NnrR binding site (46). This site has a high degree of similarity to the known FNR binding site (5′-TTGATNNNNATCAA-3′), supporting assignment of the consensus sequence to NnrR (46). Mutagenesis of the predicted NnrR binding site of nnrS into the FNR site resulted in nnrS becoming a member of the FNR regulon, demonstrating that the identified consensus sequence is a binding site for a member of the FNR family (4). This paper describes experiments that identified and characterized the function of additional genes in the NnrR regulon of 2.4.3. Five new genes were identified. Three of these genes were shown to contribute significantly to the denitrifying physiology of strain 2.4.3.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Denitrifying strain 2.4.3 (= ATCC 17025) of R. sphaeroides was used in this study along with the partial denitrifier strain 2.4.1 (= ATCC 17023). Mutants of these strains are listed in Table 1. Escherichia coli strain DH5αF′ was used for transformation, and S17-1 was used for biparental conjugation.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or description	Source or reference
Strains
DH5αF′	E. coli cloning host; F′ φ80lacZM15 redA endA1 gyrA96 thi-1 hsdR17(rk mv) supE44 relA1 deoR (lacZYA-argF)U169
S17-1	For conjugal transfer of plasmids; recA thi pro hasdRM+ RP4:2-Tc:Mu:Km:TnZ
C58	Wild-type strain of Agrobacterium tumefaciens	ATCC 33970
2.4.3	Wild-type strain of Rhodobacter sphaeroides	ATCC 17025
2.4.1	Wild-type strain of Rhodobacter sphaeroides	Type strain
nnrR	2.4.3 derivative with ΩSm/Sp in nnrR; NnrR mutant	45
prrA	prrA::aph, 2.4.3 derivative, Kmr	27
fnrL	FnrL-deficient strain of 2.4.3	18
nirK	nirK::Tn5, nitrite reductase mutant, Tpr	46
cycA	cycA::aph, 2.4.3 derivative, Kmr	26
ΔnorEF	2.4.3 derivative with ΩSm/Sp in norEF, partial deletion	This study
Δpaz	2.4.3 derivative with ΩSm/Sp in paz, partial deletion	This study
Δpaz ΔcycA	Double mutant, combined Δpaz and ΔcycA	This study
ΔnnrV	2.4.3 derivative with ΩSm/Sp in nnrV, partial deletion	This study
Δcdg1	2.4.3 derivative with ΩSm/Sp in cdgA, partial deletion	This study
Plasmids
pUC19	Used for cloning in E. coli DH5-α (Apr)	48
pRK415	Broad-host-range plasmid (Tcr)	20
pBBR1MCS-5	Broad-host-range plasmid (Gmr)	23
pSUP202-1	Mobilizable suicide vector (Tcr Apr Cmr)	42
pKOK-6	Source of lacZ-Km cassette (Tcr Kmr Apr)	21
pHP45Ω	Source of aadA+-Smr cassette (Apr Smr)	36
pAK1	pRK415 with nirK promoter and open reading frame (Tcr)	24
pNIR298	pRK415 with 298 bases of the nirK promoter fused to lacZ, transcriptional reporter (Tcr Kmr)	27
pIT9	pRK415 with the nor operon promoter fused to lacZ, transcriptional reporter (Tcr Kmr)	45
pYSW35	pRK415 carrying the prrnB ribosomal promoter (Tcr)	26
pWLNIR	pRK415 carrying the prrnB ribosomal promoter fused to nirK (Tcr)	27
pnirkOXP	pBBR1-MCS5 with prrnB-nirK fusion (Gmr)	This study
pnorEFZ	pRK415 with norEF-lacZ transcriptional fusion (Tcr Kmr)	This study
pnorEFKO	pSUP202-1 with norEF::ΩaadA+ (Tcr Smr)	This study
pnorEF	pBBR1-MCS5 with norEF operon (Gmr)	This study
pnorE	pBBR1-MCS5 with norE gene (Gmr)	This study
ppazZ	pRK415 with paz-lacZ transcriptional fusion (Tcr Kmr)	This study
ppazKO	pSUP202-1 with paz::ΩaadA+ (Tcr Smr)	This study
ppaz	pRK415 with paz gene (Tcr)	This study
pnnrVZ	pRK415 with nnrV-lacZ transcriptional fusion (Tcr Kmr)	This study
pnnrVKO	pSUP202-1 with nnrV::ΩaadA+ (Tcr Smr)	This study
pnnrVOXP	YSW35 with prrnB′-nnrV fusion (Tcr)	This study
pnnrV	pRK415 with nnrV gene (Tcr)	This study
pcdg1Z	pRK415 with cdgA-lacZ transcriptional fusion (Tcr Kmr)	This study
pcdg1KO	pSUP202-1 with cdg1::ΩaadA+ (Tcr Smr)	This study
pcdg1OXP	YSW35 with prrnB′-cdgA fusion (Tcr)	This study
ptatAZ	pRK415 with tatA-lacZ transcriptional fusion (Tcr Kmr)	This study

Rhodobacter strains were grown in Sistrom's medium (31) at 32°C, and antibiotics were added when necessary at the following concentrations: tetracycline, 1.0 μg/ml; kanamycin, 25 μg/ml; streptomycin, 50 μg/ml; and gentamicin 20 μg/ml. For denitrifying growth, nitrate (KNO₃) was added as a substrate to a final concentration of 12 mM. Growth conditions have been described previously (46). All E. coli strains were grown in Luria-Bertani medium (32) at 32°C with aeration, and, when necessary, the medium was supplemented with antibiotics at the following concentrations: ampicillin, 100 μg/ml; tetracycline, 10 μg/ml; streptomycin, 25 μg/ml; and kanamycin, 25 μg/ml.

Construction of plasmids and strains.

Chromosomal DNA was isolated from strain 2.4.3 using a Puregene DNA isolation kit and was subsequently used for PCRs. Oligonucleotide primers were purchased from Integrated DNA Technologies, and, if the primers were used to produce products that were cloned, restriction sites were added to facilitate cloning. Plasmid isolation was done by the alkaline lysis method (32). Plasmids are listed in Table 1. Standard methods were used for restriction digestion, ligation, and biparental conjugation. Transformation into E. coli strains was done by the TSS method (9).

Mutant strains.

norEF (Rsph17025_3679 and Rsph17025_3678), paz (Rsph17025_2686), nnrV (Rsph17025_0969), and cdgA (Rsph17025_3576) were inactivated by deletion of a central portion of each gene via double homologous recombination events that replaced the portion with a streptomycin resistance cassette. In each case, the suicide vector was constructed and transformed into E. coli S17-1. This strain was then used for conjugal transfer of the vector into R. sphaeroides 2.4.3 by biparental mating. For the Δpaz ΔcycA strain, the paz suicide vector was conjugated into the existing cycA strain. Mutant strains were verified by PCR.

The norEF mutant was generated by amplifying the norEF genomic region using a single primer pair, and the amplicon was digested with EcoRI and PstI and cloned into pUC19. The resulting clone was digested with BamHI, and the streptomycin resistance cassette, aadA, from pHP45Ω was ligated into the vector. There were two BamHI sites in the norEF amplicon, resulting in deletion of 522 bp from the center of the operon. The resulting norEF::aadA fragment was cloned into the suicide vector pSUP202-1 via EcoRI and PstI. The paz mutant was generated by amplifying the ends of the paz gene using two separate primer pairs. The resulting amplicons were cloned into pUC19 at EcoRI/BamHI and BamHI/PstI sites. The two fragments were then joined at the BamHI site in a new pUC19 construct, and aadA was cloned into the BamHI site. The paz::aadA fragment was then cloned into pSUP202-1 via EcoRI and PstI. The nnrV mutant was generated by amplifying the genomic region surrounding the gene using two primer pairs. The fragments were cloned into pUC19 by using HindIII/EcoRI and EcoRI/PstI. The aadA cassette was cloned into the EcoRI site. The resulting nnrV::aadA construct was cloned into pSUP202-1 using HindIII and PstI sites. The cdgA mutant was generated by amplifying the genomic region surrounding the gene using two primer pairs. The resulting amplicons were cloned into pUC19 using PstI/BamHI and BamHI/EcoRI sites. The aadA cassette was then cloned into the BamHI site. The cdgA::aadA fragment was cloned into pSUP202-1 using PstI and EcoRI.

Overexpression constructs.

Overexpression constructs were made using YSW35, which carries the Rhodobacter rRNA promoter prrnB′. Genes of interest, each containing its native ribosome binding site and open reading frame, were cloned downstream of the prrnB′ promoter. In each case a single primer pair was used to produce the fragment used in construction of the overexpression construct.

lacZ expression fusions.

Expression fusions were made by fusing the predicted promoter of the gene of interest to the lacZ gene. Fusions were cloned into the broad-host-range vector pRK415 and conjugated into relevant strains. Initially, promoters were amplified from 2.4.3 genomic DNA and cloned into pUC19. The promoter fragment was then transferred into pRK415, and the lacZ cassette was added using the PstI fragment of pKOK-6, which contained the lacZ gene in addition to a kanamycin resistance marker. The norEF promoter fragment contained 263 bp upstream of the predicted translation start site for norE. The paz promoter contained 245 bp upstream of the predicted translation start site. The nnrV promoter contained 528 bp upstream of the predicted translation start site. The cdgA promoter contained 476 bp upstream of the predicted translation start site. The tatA promoter contained 598 bp upstream of the predicted translation start site.

Complementation constructs.

norEF, norE, and paz complementation constructs were made to confirm mutant phenotypes. In each case, the open reading frame along with a significant upstream sequence was PCR amplified and cloned into the broad-host-range vector pRK415. The resulting construct was then conjugated into the relevant mutant background.

Enzymatic assays.

β-Galactosidase assays were used to determine relative gene expression levels and were performed as previously described (32). Activity was determined for three independently grown cultures. Samples were taken at various time points throughout growth, and the values for each relevant growth phase were averaged. Standard deviations were also determined.

Whole-cell Nir activity was determined using a previously described colorimetric assay (27). Activity was determined for three independently grown cultures. Samples were taken throughout growth, and the highest values were recorded. Values were averaged, and standard deviations were also determined.

Denitrification enzyme activity was measured by gas chromatography using a GOW-MAC series 550 thermal conductivity chromatograph with a 1-m by 2-mm column packed with 120/140 Carbosieve S (Supelco Inc, Bellafonte, PA). The carrier gas was helium at a flow rate of 50 ml min⁻¹. For these assays, cells were grown microoxically with nitrate to induce denitrification. Cells were harvested at an optical density at 600 nm of 0.8 to 1.0, concentrated, and resuspended in 5 ml of Sistrom's medium containing 1 mM KNO₂. The cells were sealed in 15-ml vials that were flushed with N₂ gas for 30 s and then incubated with shaking at 32°C. The headspace was sampled periodically to detect production of N₂O. The data below indicate the concentration of N₂O in the headspace.

Taxis assay.

Taxis assays were performed as previously described (28) using nitrite plugs which contained 360 mM nitrite. Experiments were done in a sealed jar under an N₂ atmosphere.

RESULTS

Identification of potential NnrR-dependent promoters.

The motif discovery program MEME was used to identify shared motifs in the nirK, norCBQD, and nnrS promoters (3). As expected, one of these motifs was identical to the previously predicted NnrR binding site (46). The genome of 2.4.3 was searched for occurrences of this motif using the program FIMO (3). Motifs that did not contain the 5′-TTGN₈CAA-3′ motif conserved in members of the FNR/CRP family were eliminated. MEME analysis indicated that the G and C at positions 5 and 10 of the 14-base motif were conserved, so motifs without these bases were eliminated. Using these criteria, four additional occurrences of this sequence were discovered in putative promoter regions. In addition, the previously identified site upstream of norCBQD was suggested to be involved in expression of a divergently transcribed gene that has been designated nnrV. No occurrences of the consensus sequence were found in predicted protein-encoding open reading frames. The predicted motifs and adjacent genes are listed in Table 2. Explanations for gene names are provided below. The proximity of the binding site to the predicted start of translation for norEF, paz, nnrV, and cdgA suggests that there is class II-type regulation by NnrR at these promoters (5). For tatA, the proximity of the binding site to the start of translation does not discriminate the type of NnrR regulation.

TABLE 2.

Predicted members of the NnrR regulon

Locus tag(s)	Gene(s)	NnrR binding site	Positiona
Rsph17025_3679, Rsph17025_3678	norEF	TTGCGN4CGCAA	−72
Rsph17025_2686	paz	TTGCGN4CACAA	−55
Rsph17025_0969	nnrV	TTGCGN4CACAA	−73
Rsph17025_3576	cdgA	TTGCGN4CGCAA	−58
Rsph17025_1983	tatA	TTGTGN4CGCAA	−115

^aThe position is the distance from the predicted start of translation to the 3′ end of the NnrR binding site.

A similar search for this motif was done with the other three strains of R. sphaeroides whose genomes are available, 2.4.1, 2.4.9 (= 17029), and KD131 (8, 30). Each of these strains is a partial denitrifier (40). In every instance where a FIMO-predicted member of the NnrR regulon of 2.4.3 is found in one of these other strains the ortholog also contains a consensus NnrR binding site upstream of the predicted start of translation (Table 3). In addition, in all three of these strains an NnrR consensus binding site is located upstream of a gene designated bolA (gene RSP_2952 in 2.4.1) (47). While this gene is conserved in strain 2.4.3, the NnrR binding site is not present in its promoter. Experiments using a transcriptional fusion between the promoter of bolA from 2.4.1 and lacZ showed that bolA expression was not dependent on NnrR or nitrogen oxides (data not shown).

TABLE 3.

Comparison of the predicted NnrR regulons in four strains of R. sphaeroides

Strain	Genes in NnrR regulona
	nnrR	norCBQD	nnrS	nnrV	nirKV	paz	norEF	cdgA
2.4.3	+	+	+	+	+	+	+	+
KD131	+	+	+	+	+	+	−	−
2.4.1	+	+	+	+	−	−	−	−
2.4.9	+	+	+	+	−	−	−	−

^a+, present; −, absent.

Expression of the identified genes in 2.4.3.

To determine if the motifs identified genes in the NnrR regulon of 2.4.3, a number of lacZ expression fusions were generated. Expression was monitored under oxic, microoxic, and denitrifying (microoxic plus 12 mM nitrate) conditions. The expression pattern for genes under NnrR control is denitrifying ≫ microoxic > oxic (46). The norEF, paz, and nnrV expression fusions had expression patterns consistent with inclusion of these genes in the NnrR regulon (Fig. 1). To confirm this, expression was assessed in an nnrR mutant. As expected, expression of these genes did not increase under denitrifying conditions in the absence of NnrR.

FIG. 1.

Expression of lacZ fusions expressed as a percentage of wild-type denitrifying expression. For each expression fusion data for both the wild type (wt) and the nnrR mutant are shown. Expression was measured by determining β-galactosidase activity. For each condition, activity was determined in triplicate, and the results were averaged; the error bars indicate one standard deviation. Absolute values were converted to percentages of wild-type denitrifying expression. Open bars indicate oxic conditions, gray bars indicate microoxic conditions, and black bars indicate denitrifying conditions. The absolute values for fusions under oxic, microoxic, and denitrifying conditions were as follows: for norEF::lacZ wild type, 13, 40, and 309, respectively; for norEF::lacZ nnrR, 7, 14, and 17, respectively; for paz::lacZ wild type, 16, 386, and 2,375, respectively; for paz::lacZ nnrR, 13, 28, and 59, respectively; for nnrV::lacZ wild type, 4, 61, and 232, respectively; for nnrV::lacZ nnrR, 2, 3, and 6, respectively; for cdgA::lacZ wild type, 134, 238, and 280, respectively; for cdgA::lacZ nnrR, 113, 194, and 204, respectively; for tatA::lacZ wild type, 1,574, 1,802, and 1,775, respectively; and for tatA::lacZ nnrR, 1,520, 1,627, and 1,580, respectively.

The cdgA fusion showed a slightly different expression pattern. As found with the other NnrR-regulated genes, maximal expression was observed under denitrifying conditions. However, expression increased only modestly (∼20%) when nitrate was added to microoxic cultures. In the nnrR mutant, expression of cdgA increased under microoxic conditions, but there was no nitrate-dependent increase. The expression results suggest that a decrease in the oxygen level may be the predominant signal for increased cdgA expression. Two global oxygen responsive regulators in R. sphaeroides are PrrA and FnrL (13, 49). To test whether these regulators directly or indirectly contribute to cdgA expression, the fusion was mobilized into prrA and fnrL mutant backgrounds. The cdgA expression in both mutant backgrounds was similar to the wild-type expression (data not shown). This suggests that NnrR enhances expression of cdgA in conjunction with some unknown oxygen-dependent regulator, but NnrR is not an absolute requirement for expression, as seen for norEF, paz, and nnrV.

The tatA expression fusion showed no NnrR-dependent expression. Expression was high under all conditions, and there was no significant change in the nnrR mutant. The tatA gene is part of a predicted operon that encodes the proteins required for the Tat transport system along with a protein with an unknown function. This operon is conserved in all strains of R. sphaeroides; however, the putative NnrR binding site is present only in strain 2.4.3. Genes upstream of the tat operon are transcribed in the same orientation as tat, indicating that the putative NnrR binding site could only be involved in tat regulation. The results of the expression studies support inclusion of norEF, paz, nnrV, and cdgA in the NnrR regulon of strain 2.4.3. These genes and their products were characterized further.

norEF.

Rsph17025_3679 was designated norE due to significant similarity to other known norE genes in alphaproteobacteria (11). Typically, norE is found in association with a gene designated norF (11). The norF product is not highly conserved, even among closely related bacteria; however, most putative NorF proteins have two predicted transmembrane regions. Since Rsph17025_3678 encodes a small hydrophobic protein predicted to have two transmembrane regions, it was designated norF. Only 3 bp separate norE and norF, suggesting that they form an operon.

NorE proteins have homology to subunit III of the cytochrome oxidase (51). This suggests that NorE may associate with the Nor enzyme complex since Nor is in the cytochrome oxidase superfamily. This conclusion is supported by the observation that the norEF genes are commonly close to the norCBQD operon (11). This is not the case in 2.4.3 since norEF is on chromosome II, while the norCBQD cluster is on chromosome I. None of the other R. sphaeroides strains have the norEF genes despite the fact that they all have the nor operon (Table 3). A norEF mutant was constructed by insertional inactivation, which resulted in deletion of the last 512 nucleotides of norE and the first 269 nucleotides of norF. The Nir activity of the ΔnorEF mutant was about one-half that of the wild type under microoxic conditions with and without nitrate amendment (Fig. 2). Reduced Nir activity was not due to a decrease in nirK expression since a nirK-lacZ fusion was expressed at wild-type levels during denitrifying growth (data not shown). nor was also expressed at wild-type levels in the ΔnorEF mutant. Wild-type Nir activity was restored in the ΔnorEF mutant by addition of plasmid-borne norEF but not by addition of norE alone (Fig. 2), suggesting that both gene products are required for optimal denitrification in strain 2.4.3. The norEF genes were also mobilized into strain 2.4.1 along with nirK. When grown under denitrifying conditions, the 2.4.1 strain with nirK plus norEF had 2- to 3-fold higher Nir activity than the 2.4.1 strain with nirK alone (data not shown).

FIG. 2.

Nir activity of 2.4.3, the norEF mutant, the norEF mutant complemented with norE and norF (norEF+EF), and the norEF mutant complemented with only norE (norEF+E). All strains were grown under microoxic conditions with 10 mM nitrate. The values are the highest activity observed for each strain.

To gain additional insight into how inactivation of norEF impacts denitrification, the cellular response to nitrite was assessed using a taxis assay. Wild-type strain 2.4.3 accumulates in a well-defined ring around a concentrated nitrite source, and this behavior requires Nir activity (28). The region where cells accumulate likely corresponds to a concentration of nitrite at which NO production and consumption are optimal. Like the wild type, the ΔnorEF mutant was found to produce a well-defined ring around a nitrite source (Fig. 3). However, the mutant consistently accumulated further from the source than the wild type. The ring formed by the wild type had a radius of 1.16 ± 0.15 cm, while the ring formed by the ΔnorEF mutant had a radius of 2.03 ± 0.06 cm. The complementation strain containing norEF in trans showed a wild-type response to nitrite and formed a ring with a radius of 1.43 ± 0.11 cm, while cells containing norE alone formed a ring with a radius of 2.26 ± 0.06 cm, indicating that both genes are needed to restore the taxis response.

FIG. 3.

Taxis responses of the 2.4.3 strain, the norEF mutant (norEF), the norEF mutant complemented with norEF (norEF+EF), and the norEF mutant complemented with only norE (norEF+E). The nitrite source was an agar plug located in the middle of the plate. The ring in each plate is where cells have accumulated.

The taxis and Nir assays indicated that inactivation of norEF impacts the cell's ability to respire nitrite. This was investigated further using gas chromatography to assess the ability of the mutant to reduce nitrite to nitrous oxide (N₂O). N₂O is the major end product of denitrification in strain 2.4.3 likely due to a frameshift in gene Rsph17025_3318, which is a putative nosR ortholog (7). The wild-type strain generated N₂O rapidly over the first 10 to 15 min of incubation, after which the levels did not change significantly (Fig. 4). In contrast, the ΔnorEF mutant significantly lagged in N₂O production, taking approximately 10 times longer to accumulate N₂O levels comparable to those accumulated by the wild type. A wild-type rate of N₂O production was restored by complementation with norEF.

FIG. 4.

Production of nitrous oxide over time in 2.4.3 (•), the ΔnorEF mutant (⧫), and the ΔnorEF mutant complemented with norEF (▪). Cells were incubated under an N₂ atmosphere in Sistrom's medium containing 1 mM nitrite. The headspace was sampled over time, and nitrous oxide was detected by gas chromatography.

paz.

The paz gene codes for the blue copper protein pseudoazurin, a known donor of electrons to nitrite reductase (34). Commonly, Nir has been found to have more than one electron donor, usually a cytochrome and a pseudoazurin (34, 50). In 2.4.3 it has been shown that cytochrome c₂ donates electrons to Nir in vitro (26). However, inactivation of the gene encoding cytochrome c₂ did not decrease Nir activity, suggesting that 2.4.3 also has multiple donors of electrons to Nir. To determine if paz is involved in electron transfer to Nir, a strain lacking paz and a strain lacking both paz and cycA, which encodes cytochrome c₂, were constructed. The Δpaz mutant had levels of Nir activity similar to the levels of the wild type and the cycA mutant (Fig. 5). However, the Nir activity in the strain lacking both cycA and paz was more than 10-fold lower than that in the wild type. Nir activity was restored to nearly wild-type levels in the double mutant if the paz gene was present in trans (data not shown). These results were supported by the cell taxis response to nitrite. The Δpaz ΔcycA double mutant showed no clearing or ring formation, which is consistent with a lack of Nir activity (not shown) (28).

FIG. 5.

Nir activity of 2.4.3, the cycA mutant (cycA), the Δpaz mutant (paz), the Δpaz ΔcycA double mutant (Δpaz/cycA), and the double mutant expressing nirK constitutively (Δpaz/cycA+cnirK). Activity was determined for cells taken from cultures grown microoxically with 10 mM nitrate.

The large decrease in Nir activity of the Δpaz ΔcycA double mutant does not conclusively demonstrate that c₂ and pseudoazurin are the donors of electrons to Nir. Inactivation of cycA and cycY, which encodes cytochrome c_Y, also resulted in a >10-fold decrease in Nir activity (26). Expression of nirK from a heterologous promoter restored Nir activity in this double mutant, indicating that the decrease in Nir activity was a result of a decrease in nirK expression, not a result of a decrease in electron flow to Nir. To control for decreased nirK expression in the Δpaz ΔcycA strain, a construct that constitutively expresses nirK was mobilized into the double-mutant background. Changing the regulation of nirK did not restore Nir activity in the Δpaz ΔcycA mutant, demonstrating that the lack of Nir activity was due to a lack of suitable donors of electrons to Nir (Fig. 5).

Recent work has shown that paz in Agrobacterium tumefaciens is regulated by FnrN and the ActRS two-component system (2). In 2.4.3, the ortholog of the ActRS system is PrrAB, and nirK expression has been shown to require this system (27). Expression of paz in 2.4.3 decreased significantly in a prrA mutant background to 30% of the wild-type level of expression. While this suggests that paz regulation is dependent on PrrA, interpretation of this result is complicated by the fact that the function of NnrR, which regulates paz, is dependent on Nir activity. Since nirK expression relies on PrrA, the decrease in expression may be due to a decrease in Nir activity, limiting the function of NnrR. To reduce the interdependent nature of the regulation, expression of paz was measured in a prrA mutant strain that contained constitutively expressed nirK. The expression of paz in this strain was 70% of the wild-type level of expression. These trends in expression are similar to those observed with norCBQD, which is not under PrrA control in 2.4.3 (27). Therefore, it seems likely that paz is not under direct control of PrrA.

nnrV.

The locus designated Rsph17025_0969 in the genome sequence does not have a known function and is not a member of a known gene family; therefore, it was given the generic designation nnr since it is a member of the NnrR regulon. nnrV is predicted to encode a 68-amino-acid protein with an unknown function and is present in the four available R. sphaeroides genomes (Table 3). NnrV orthologs are not widespread. Most of them have been found in alphaproteobacterial denitrifiers, and there are a few putative orthologs in nondenitrifying clostridia. Examination of the nnrV locus in other alphaproteobacteria revealed that in almost every case nnrV is within one or two genes of an FNR-type regulator and either nirK or norCBQD genes, suggesting that there is a link with denitrification.

In order to assess the functional activity of NnrV, nnrV was insertionally inactivated. Under denitrifying conditions with nitrate as a substrate, the ΔnnrV mutant had wild-type nitrite reductase activity. Similarly, under denitrifying conditions with nitrite as a substrate, the ΔnnrV mutant had wild-type rates of N₂O production, indicating that it had wild-type Nir and Nor activities. Consistent with the wild-type denitrifying ability, the expression of nirK and nor in the nnrV mutant was identical to the wild-type expression. The ΔnnrV mutant also had a wild-type taxis response to nitrite. Since no phenotype was associated with deletion, nnrV was placed under control of the strong, constitutive prrnB′ promoter to determine if overexpression impacted denitrification. However, there was no change in any of the phenotypes assessed as a consequence of the change in regulation of nnrV.

cdgA.

The gene located on chromosome II at locus Rsph17025_3576 was designated cdg since its product is predicted to either produce or degrade cyclic diguanosine monophosphate (c-di-GMP) and gene A since there are multiple genes encoding proteins belonging to this family in 2.4.3 that have yet to be characterized. cdgA is predicted to encode a modular protein containing four distinct domains. There are two N-terminal PAS domains, followed by a GGDEF domain and an EAL domain. The more N-terminal PAS domain is predicted to bind heme and may confer a sensory function (15). The GGDEF domain is predicted to be functional for diguanylate cyclase activity, generating c-di-GMP (38). The EAL domain is predicted to be functional for phosphodiesterase activity degrading c-di-GMP (39). No obvious orthologs of cdgA are found in the other R. sphaeroides strains (Table 3).

No growth phenotype was observed for a cdgA mutant strain under aerobic, denitrifying, or photosynthetic conditions (data not shown). There was also no impact on nirK expression or Nir activity, and the taxis response to nitrite was identical to that of the wild type. In addition, there was no detectable change in cell morphology or motility as assessed by phase-contrast microscopy. It is not uncommon for inactivation of genes encoding diguanylate cyclases and phosphodiesterases to result in an undetectable phenotype. In some cases, a phenotype was observed if the gene was overexpressed (41). However, expression of cdgA from the prrnB′ promoter did not result in a detectable phenotype, and the strain exhibited wild-type growth, morphology, and denitrification enzyme activity and gene expression. To determine if the protein encoded by cdgA has diguanylate cyclase or phosphodiesterase activity, an indirect means of assessing activity was employed. Previously, it has been shown that high levels of c-di-GMP promote cellulose biosynthesis in exponentially growing A. tumefaciens (1). To determine if cdgA could impact cellulose biosynthesis, the cdgA overexpression construct was conjugated into wild-type A. tumefaciens, where the expression of genes from the Rhodobacter prrnB′ promoter is constitutive (not shown). Expression of cdgA resulted in visible cell clumping during exponential growth under aerobic conditions. This clumping indicated that there was increased cellulose biosynthesis, which was likely a result of the diguanylate cyclase activity of CdgA (data not shown).

DISCUSSION

By searching for a conserved sequence motif, five genes in strain 2.4.3 were identified and then confirmed to be in the NnrR regulon. While the function of every member of the NnrR regulon has not been determined, the functions of members that have been determined are involved in denitrification and more specifically NO metabolism. Given these results, it is likely that the NnrR regulon is small. In Neisseria meningitidis, the NO-responsive regulator NsrR also regulates a small set of genes involved in denitrification and NO detoxification, suggesting that NO metabolism during denitrification requires only a modest complement of proteins (17). Members of the NnrR regulon in 2.4.3 are found on both chromosome I and chromosome II, and most of the members are in single-gene units or small operons. This is in contrast to the situation in many denitrifiers, in which denitrification genes are clustered in one area of the genome (35).

Using the MEME program, the promoter regions of all of the genes identified as members of the NnrR regulon were aligned to further refine the NnrR binding motif and to determine if there were other common motifs (3). The only sequence identified that was an inverted repeat and present in all of the promoter regions was the NnrR binding site. The conserved motif predicted using all the sequences has been extended to 20 bases by inclusion of an A and a T at positions 1 and 20, respectively (Fig. 6). Since the G+C content of the genome of 2.4.3 is ∼69%, the conservation of A or T at these positions is likely to be structurally significant.

FIG. 6.

NnrR binding motif predicted by aligning MEME-identified sequences from the promoters of known NnrR-dependent genes from 2.4.3. The motif was generated using the MEME program and the nirKV, norCBQD, nnrS, norEF, cdgA, and paz promoters (3).

The sequence upstream of the tat operon matched this extended motif. The Tat transport system is important under all growth conditions, including denitrifying conditions, since the molybdopterin binding subunit of the respiratory nitrate reductase, Nir, and NirV all have the twin-arginine motif in their signal sequences. While expression of tat is essential under all conditions, it is possible that the pathways regulating this expression change with changing environmental cues. This could mean that NnrR-dependent expression occurs only under unique conditions not encountered in laboratory cultures.

A computational analysis of the putative NnrR regulon in 2.4.1 identified NnrR binding sites upstream of norCBQD, nnrS, and nnrR (37). The only other gene predicted to be part of the regulon was hemN. The motif in the hemN promoter does not match the motif predicted from this analysis. Instead, it is made up of an upstream half-site that is a perfect FNR site, TTGAT, while the downstream half-site matches the NnrR consensus. hemN, which is essential for photosynthesis, has been shown to be regulated in response to oxygen by FnrL (33). However, it is not known if this gene is in the NnrR regulon in 2.4.3 or 2.4.1.

While the size of the regulon was extended, only the function of pseudoazurin was definitively assigned. Pseudoazurin and cytochrome c₂ are the donors of electrons to Nir. This results in one donor of electrons to Nir, cytochrome c₂, which donates electrons to many proteins under a wide range of growth conditions, whereas the other donor, pseudoazurin, is required only during denitrification (10, 16, 26). This arrangement suggests that under certain growth conditions, such as during photosynthesis in the presence of nitrate, cytochrome c₂ is used to support cyclic electron transfer, while psuedoazurin is used for electron transfer to Nir. However, the Nir activity in the Δpaz mutant was similar to that in the wild type under photosynthetic conditions (data not shown). The physiological conditions under which the cell benefits from having redundant donors of electrons to Nir remain elusive.

The only other locus whose function could be linked to denitrification is norEF. A role for the products of these genes in nitrogen oxide metabolism was suggested by the results of the taxis assay, which showed that strains lacking NorEF preferred areas of the plate with lower nitrite concentrations than the wild type. As shown by the absence of any taxis response in the Δpaz ΔcycA double mutant, nitrite in the absence of an active nitrite reductase does not result in a taxis response (28). This indicates that the response requires production of NO, a well-known inhibitor of respiration. In the presence of an active nitrite reductase, the region where cells accumulate is likely a region where the conditions allow a high rate of electron flow to the terminal oxidant without accumulation of inhibitory compounds, like NO. The preference of the ΔnorEF mutant for regions with lower nitrite concentrations indicates that higher nitrite levels lead to suboptimal respiration, possibly due to NO accumulating to inhibitory levels. This may indicate that NorEF is required for optimal Nor activity. In the absence of NorEF Nor would be less active, leading to accumulation of NO and reducing the overall respiration rate. Attempts to demonstrate high levels of NO in cells using the NO sensor Cu-fluorescein were unsuccessful even though the dye was able to detect the high levels of NO produced by a Nor mutant (data not shown) (29). The inability to detect NO in the ΔnorEF strain does not eliminate the possibility that NO accumulation leads to the taxis phenotype. NO is preferentially soluble in membranes where it can impact components of the respiratory chain (19), and the increase in the NO level above normal levels required to observe the phenotype in the ΔnorEF strain might be small.

Another phenotypic change associated with inactivation of norEF was a decrease in Nir and Nor activity (Fig. 2 and 4). Previous studies of norEF from Paracoccus denitrificans also noted a decrease in denitrifying enzyme activity, specifically Nir and Nor activity, in the absence of NorEF (11). A coordinate decrease in Nir and Nor activity alone seems unlikely to lead to the altered taxis response. In the ΔnorEF mutant the rate of disappearance of nitrite, measured by the Nir assay, and the rate of appearance of the product (N₂O), measured by gas chromatography, suggest that there was an imbalance in the activities compared to the rates seen in the wild type. In the Nir assay nitrite is reduced at a rate that is one-half the wild-type rate. In contrast, N₂O is produced almost 10-fold more slowly from nitrite in the absence of norEF. The relative differences in the changes in Nir and Nor activity after loss of norEF are consistent with conditions under which NO would accumulate, accounting for the observed taxis response.

The coselection of certain sets of genes in the NnrR regulon is clear (Table 3). paz and nirKV are coselected, but the conditions resulting in this strong coselection remain elusive. Nevertheless, this set of genes appears to be sufficient to produce a functional Nir. norCBQD, nnrS, and nnrV also appear to be coselected since they are retained in strains lacking Nir and since they are tightly linked genetically (Table 3). This suggests that this set of genes is required for NO reduction. However, nnrV may actually be coselected with nnrR, since it is frequently closely associated with genes encoding members of the FNR/CRP family (data not shown). These groupings indicate that nirK nirV, paz, nnrS, nnrV, and norCBQD, along with nnrR, are the genes required for R. sphaeroides to be physiologically competent for nitrite reduction to NO and NO reduction to N₂O. Within this set, four genes, paz, nirV, nnrS, and nnrV, can be inactivated with no loss of denitrification capacity under laboratory conditions (4, 18). In the case of paz, the presence of a functionally redundant protein accounts for its lack of phenotype. In the case of the other three genes, it is unclear if they are functionally redundant or if laboratory conditions mask any potential phenotypic changes. Nevertheless, the strong coselection of these genes with the structural genes for the nitrogen oxide reductases indicates that they have functional importance under real-world conditions.

The one notable exception to this strong coselection is norEF; 2.4.3 is the only strain with this operon (Table 3). In nirK-containing alphaproteobacteria, genomic analysis has shown that norEF has strong genetic linkage to the norCBQD operon; the former is either immediately upstream or immediately downstream of the latter. The one exception to this linkage is the family Rhodobacteraceae; some members of this family contain norCBQD, but either norEF is not present or norEF is located some distance from the norCBQD operon. Absence of norEF in a common ancestor of the four R. sphaeroides strains included in Table 3 followed by acquisition of norEF via lateral gene transfer is the most parsimonious explanation for why 2.4.3 is the only member of the Rhodobacteraceae family with both nirK and norEF. Recent work suggests that chromosome II, which is where norEF is located, is evolving more rapidly than chromosome I and that lateral gene transfer from organisms with similar G+C contents into this chromosome is relatively frequent (8). Since norEF has a significant impact on the denitrification ability of 2.4.3, it provides a selectable benefit and likely is retained. In 2.4.1 and 2.4.9, which lack nirK and in which Nor is likely used to mitigate the toxic effects of NO, it is possible that the lack of norEF is also advantageous. The lack of norEF may enhance the NO stress response, ensuring that cells avoid environments with toxic levels of NO. The same would be true of KD131 except that it would be responsive to both nitrite and NO, since it retains nirK. Further research should help to determine how the NnrR regulon is adapted to best fit the physiological roles of nitrogen oxide reduction in each strain of R. sphaeroides.