Involvement of the PrrB/PrrA Two-Component System in Nitrite Respiration in Rhodobacter sphaeroides 2.4.3: Evidence for Transcriptional Regulation

Abstract

Rhodobacter sphaeroides strain 2.4.3 is capable of diverse metabolic lifestyles, including denitrification. The regulation of many Rhodobacter genes involved in redox processes is controlled, in part, by the PrrBA two-component sensor-regulator system, where PrrB serves as the sensor kinase and PrrA is the response regulator. Four strains of 2.4.3 carrying mutations within the prrB gene were isolated in a screen for mutants unable to grow anaerobically on medium containing nitrite. Studies revealed that the expression of nirK, the structural gene encoding nitrite reductase, in these strains was significantly decreased compared to its expression in 2.4.3. Disruption of prrA also eliminated the ability to grow both photosynthetically and anaerobically in the dark on nitrite-amended medium. Complementation with prrA restored the wild-type phenotype. The PrrA strain exhibited a severe decrease in both nitrite reductase activity and expression of a nirK-lacZ fusion. Nitrite reductase activity in the PrrA strain could be restored to wild-type levels by using nirK expressed from a heterologous promoter, suggesting that the loss of nitrite reductase activity in the PrrA and PrrB mutants was not due to problems with enzyme assembly or the supply of reductant. Inactivation of prrA had no effect on the expression of the gene encoding NnrR, a transcriptional activator required for the expression of nirK. Inactivation of ccoN, part of the cbb₃-type cytochrome oxidase shown to regulate the kinase activity of PrrB, also caused a significant decrease in both nirK expression and Nir activity. This was unexpected, since PrrA-P accumulates in the ccoN strain. Together, these results demonstrate that PrrBA plays an essential role in the regulation of nirK.

Denitrifiers can use nitrogen oxides as alternative electron acceptors under oxygen-limited conditions (55). The expression of genes required for denitrification is consistent with denitrification being an alternative form of respiration. Maximum expression of these genes requires that oxygen concentrations be limiting and that a nitrogen oxide be present (4). Nitrogen oxides are not used as terminal oxidants when oxygen levels are sufficient for aerobic respiration.

During denitrification, nitrate is reduced to nitrite, nitric oxide (NO), nitrous oxide, and then nitrogen gas (N₂). Each intermediate is obligatory and freely diffusible. As expected, denitrifiers contain several nitrogen oxide stimulons, since multiple intermediates are produced during denitrification. Nitrate, nitric oxide, and nitrous oxide each activate the expression of a distinct cluster of genes (4, 20). The mechanisms used by denitrifiers to detect the presence of each nitrogen oxide are beginning to be understood. According to recent findings, Fnr-like regulators and two-component systems have been implicated in nitrate sensing (14, 51), while gene activation in the presence of nitric oxide requires either an activator belonging to the FNR family (21, 47, 50) or one belonging to the recently described RpoN-dependent family (35). Less is known about the mechanism for the sensing of nitrous oxide, although genes required for this process have been identified (6).

In addition to detecting nitrogen oxides, denitrifying bacteria also require a means of regulating genes that are preferentially expressed under anaerobic conditions. The transcriptional regulator Fnr is probably the best-studied regulator of this type. It functions as an oxygen-dependent regulator (36). When oxygen concentrations become limiting, an iron-sulfur cluster in Fnr undergoes a structural change causing Fnr to bind to its target DNA (36). Fnr-like proteins are required for anaerobic growth by denitrifiers, but the role of these proteins seems to differ among the various denitrifiers (1, 48, 49).

Rhodobacter sphaeroides strain 2.4.3 is a photosynthetic bacterium capable of dissimilatory reduction of nitrogen oxides under anaerobic conditions (27). Two essential enzymes in this process are nitrite reductase (Nir) and nitric oxide reductase (Nor), encoded by the nirK gene and norCBQD operon, respectively. It has been shown that NnrR, a member of the FNR/Crp family of transcriptional regulators, is necessary for the expression of nirK and nor during denitrification (44, 45). Studies of regulation by NnrR demonstrated that the presence of NO is required for transcriptional activation during denitrification, yet the mechanism whereby NnrR senses NO is not known (21). To date, no other regulatory factors involved in the regulation of nirK have been described.

Bacterial photosynthesis is another anaerobic process in which gene expression is regulated in response to low oxygen tensions. In addition to Fnr, genes for anaerobic growth in many photosynthetic bacteria are regulated by a two-component sensor kinase system (12, 25, 29). In R. sphaeroides, this system is composed of the sensor kinase PrrB and the response regulator PrrA. In Rhodobacter capsulatus, these proteins have been designated RegB and RegA (29). Inactivation of prrB in R. sphaeroides 2.4.1 causes impaired photosynthetic growth, while inactivation of prrA eliminates photosynthetic growth (10, 12). Genes regulated by PrrBA include puc, puf, and puh, all of which are required for photosynthesis, along with genes required for heme biosynthesis, CO₂ fixation, and N₂ fixation (13, 16, 31). Recently, a model for regulation by PrrBA was presented that suggests that PrrB receives a signal from the cbb₃-type cytochrome c oxidase with the assistance of the protein PrrC (11, 32). When oxygen concentrations are high, the reductant flow through the cbb₃-type cytochrome complex results in PrrB, which is believed to have both kinase and phosphatase activities (10), receiving an inhibitory signal transmitted through PrrC (11). This inhibitory signal causes the equilibrium of PrrB, whose default state is in the kinase mode, to shift to the phosphatase mode, resulting in dephosphorylation of PrrA (31, 33, 34). When the inhibitory signal is removed by decreased activity of the oxidase, the kinase activity of PrrB predominates, causing PrrA-P levels to increase.

In a screen for mutants of R. sphaeroides 2.4.3 that are unable to grow by using nitrite as their sole terminal electron acceptor, four strains were isolated that contained insertions within the gene encoding PrrB. Study of these mutants led us to investigate the role of the PrrBA two-component system in denitrification. Here we provide evidence that the PrrBA two-component system exerts transcriptional control over nirK. The requirement for PrrBA suggests that nirK expression is under the control of both NnrR, which regulates only those genes specific for nitric oxide metabolism, and the global regulator PrrA, which adjusts cellular physiology in response to redox status.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The strains and plasmids used in this work are listed in Table 1. Escherichia coli strain DH5α was used as the maintenance strain for plasmids. E. coli S-17-1 was used as the donor for matings (41). R. sphaeroides 2.4.3 (ATCC 17025) was the denitrifying wild-type strain, while R. sphaeroides 2.4.1 was the nondenitrifying type strain of R. sphaeroides. The broad-host-range plasmid pRK415 was used for transferring genes from E. coli to R. sphaeroides (18). Plasmids containing either prrB or prrA that were used in complementation assays were derived from R. sphaeroides 2.4.1. Plasmid pUI1641 (12) contained all of prrB and most, but not all, of prrC; pUI1621 (13) contained prrA; and pUI1645 (12) carried an XhoI-XhoI fragment that contained only prrC. The plasmids pJS84, carrying a nirK-lacZ fusion (44), and pIT53, carrying an nnrR-lacZ fusion (45), have been described elsewhere. pAK1 was used to express the nirK gene in R. sphaeroides 2.4.1 derivatives (22).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or descriptiona	Reference or source
Strains
DH5αF′	Host for E. coli cloning; F′ φ80 lacZΔM15 recA1 endA1 gyrA96 thi-1 hsdR17(rK− mK+) supE44 relA1 deoR (ΔlacZYA-argF)U169	5
S17-1	For conjugal transfer of plasmids from E. coli into R. sphaeroides; recA thi pro hsdRM+ RP4:2-Tc::Mu::Km::Tn7
2.4.3	Wild-type denitrifying strain of R. sphaeroides	ATCC 17025
2.4.1	Wild-type nondenitrifying strain of R. sphaeroides	W. Sistrom
R125	2.4.3 derivative with ΩSm/Sp in upstream nnrR gene; Nnr mutant	45
5.11	prrB358::Tn5; 2.4.3 Tn5 derivative unable to grow on nitrite; Tpr	This study
14.3	prrB155::Tn5; 2.4.3 Tn5 derivative unable to grow on nitrite; Tpr	This study
133.2	prrB81::Tn5; 2.4.3 Tn5 derivative unable to grow on nitrite; Tpr	This study
37	prrB851::Tn5; 2.4.3 Tn5 derivative unable to grow on nitrite; Tpr	This study
PRRA	prrA::aph; 2.4.3 derivative; Kmr	This study
CCON	ccoN; 2.4.3 derivative with pRK415 Campbell-type insertion into ccoN; Tcr	This study
CCONΔ	ΔccoN; 2.4.1 derivative	32
CCOQΔ	ΔccoQ; 2.4.1 derivative	32
Plasmids
pT7/T3-19U	Used for cloning in E. coli	BRL
pUC19	Used for cloning in E. coli	52
pUC4K	Carries aph gene; used as restriction site mobilizing element	Pharmacia
pRK415	Broad-host-range plasmid (Tcr)	18
pSUP202	Mobilizable suicide vector	41
pKOK6	Source of lacZ-Km cassette	19
pUI1124	Plasmid containing P rrnB promoter	53
pJS84	pRK415 with nirK-lacZ transcriptional fusion (Tcr Kmr)	44
pIT53	pRK415 with nnrR-lacZ translational fusion (Tcr)	45
pUI1621	pRK415 with prrA from strain 2.4.1 (Tcr)	13
pUI1641	pRK415 with prrB from strain 2.4.1 (Tcr)	12
pUI1645	pRK415 with prrC from strain 2.4.1 (Tcr)	12
pWL-PRRA1	Derivative of pUC19 with 860-base prrA-containing fragment from 2.4.3	This study
pWL-PRRA2	Derivative of pSUP202 with 860-base prrA-containing fragment from 2.4.3	This study
pWL-PRRA3	Derivative of pWL-PRRA2 with 1.2-kb aph cassette from pUC4K disrupting prrA	This study
pYSW35	pRK415 with P rrnB cloned into HindIII/BamHI	Wang and Shapleigh
pWLNIR	Derivative of pYSW35 with P rrnB′-nirK fusion	This study
pWL-CYCA	Derivative of pYSW35 with P rrnB′-cycA fusion	This study
pNIR298	pRK415 nirK′-lacZ containing 298 bp upstream of the transcriptional start (Tcr Kmr)	This study
pNIR84	pRK415 nirK′-lacZ containing 84 bp upstream of the transcriptional start (Tcr Kmr)	This study

^aAbbreviations: Tp, trimethoprim; Km, kanamycin; Tc, tetracycline.

E. coli strains were grown in Luria-Bertani medium (24) supplemented with antibiotics at the following concentrations: ampicillin, 100 μg ml⁻¹; tetracycline, 10 μg ml⁻¹; streptomycin-spectinomycin, 25 μg (each) ml⁻¹; and kanamycin, 25 μg ml⁻¹. Rhodobacter strains were grown in Sistrom's medium at 30°C (23) and, when necessary, antibiotics were added at the following concentrations: tetracycline, 1 μg ml⁻¹; kanamycin, 25 μg ml⁻¹; and streptomycin-spectinomycin, 50 μg (each) ml⁻¹. For photosynthetic growth, cells were cultured over incandescent light in a jar that was made anaerobic by using a BBL GasPak (Becton Dickinson). For denitrifying growth, cultures were grown microaerobically in the presence of 12 mM nitrate, while other culture conditions were described previously (44, 45).

Mutant isolation and characterization.

The procedures used to isolate strains unable to grow with nitrite as their terminal electron acceptor have been described previously (44). The site of transposon insertion for three strains was determined by the amplification of genomic ends (28). Chromosomal DNA from the mutant was isolated and digested with either PstI or BamHI, which cut within the Tn5 and within the adjacent chromosomal DNA. The restriction fragments were ligated into pUC19 (54), and a product was amplified by using an oligomer that binds to pUC19 and an oligomer that binds to the Tn5. The amplified fragment was restricted and cloned into pT7-19U and then sequenced from both ends. Sequences from these three mutants confirmed the presence of the Tn5 and its site of insertion. The insertion site within the fourth strain was determined by digesting genomic DNA from strain 37 with BamHI, ligating the digested DNA into pUC19, and transforming into DH5α. Transformants were selected on Luria-Bertani agar supplemented with ampicillin and trimethoprim, and plasmid DNA was isolated. Sequencing of the plasmid was carried out at the Cornell University BioResource Center with a primer that annealed to the Tn5, and it revealed that the Tn5 was located within the prrB gene.

Construction of vectors and strains.

The vector used to inactivate prrA in 2.4.3 was constructed by cloning a DNA fragment containing a portion of prrC and most of prrA into pSUP202 (41). This region was amplified from strain 2.4.3, and the 860-bp amplification fragment, which contained restriction sites for EcoRI and KpnI, was digested and cloned into pUC19 (52) to form pWL-PRRA1. This fragment was moved into the mobilizable suicide vector pSUP202 (43) by digesting pWL-PRRA1 with EcoRI and PstI to create pWL-PRRA2. pWL-PRRA3 was created by digesting pWL-PRRA2 with XhoI, which cuts within the prrA reading frame, and ligating into it a 1.2-kb aph containing a SalI fragment from pUC4K (Pharmacia).

A PrrnB′-cycA fusion was constructed by first isolating a HindIII/BamHI fragment from pUI1124 (53) containing the promoter region of the R. sphaeroides rrnB gene and then cloning it into pRK415 (18) to create pYSW35 (Y. Wang and J. P. Shapleigh, unpublished data). The cycA gene was amplified from pC2P404.1 (5) to generate a 553-bp fragment with BglII and EcoRI restriction sites at the ends. This was digested with BglII and EcoRI and cloned into BamHI/EcoRI-digested pYSW35 to create pWL-CYCA.

A similar strategy was used to construct a PrrnB′-nirK fusion. In this case, the nirK gene was amplified from pNIRB (44) to generate a 1,381-bp fragment with BamHI and KpnI restriction sites. This fragment was digested and cloned into BamHI/KpnI-digested pYSW35 to create pWLNIR.

Truncated nirK-lacZ fusions were created such that the promoter extended 84 or 298 bases upstream of the transcriptional start site of nirK. For pNIR84, the nirK promoter was amplified with an upstream primer that annealed 84 nucleotides upstream and a downstream primer that annealed 199 nucleotides downstream of the transcriptional start site. The amplification product was digested with KpnI/PstI and cloned into pRK415. A 4.7-kb lacZ cassette from pKOK6 (19) was moved into the PstI site of the resulting plasmid, and the proper orientation of the cassette was confirmed by using an EcoRI digest. pNIR298 was constructed by using the same strategy, except in this case a primer was used that annealed 298 nucleotides upstream of the transcriptional start site.

Absorbance spectrum of crude extracts.

Cells were grown microaerobically, harvested, resuspended in an equal volume of phosphate buffer (pH 7.4), and disrupted by passage through a French pressure cell. Cell extracts were obtained by centrifugation for 10 min at 16,000 × g. To confirm the presence of cytochrome c₂, the dithionite-reduced minus air-oxidized spectrum was obtained by monitoring the absorbance of the sample between 400 and 700 nm with a Beckman DU 640 spectrophotometer.

Assays for enzymatic activities.

β-Galactosidase activities were determined in duplicate on at least two independently grown cultures as previously described (45). Samples were taken at successive times during growth, and the highest values obtained before the cells stopped growing were used to determine the reported values. To measure Nir activity, 500 μl was removed from cultures grown under denitrifying conditions, added to microcentrifuge tubes, washed two times in an equal volume of phosphate buffer (pH 7.4), and resuspended in 500 μl of phosphate buffer (pH 7.4). Then, 36 nmol of sodium nitrite was added to each tube, and the cells were incubated at 30°C. In a modification of a previously described protocol for quantifying nitrite accumulation (42), we performed a colorimetric assay on the reactions to measure a decrease in nitrite concentration over time, which correlates to Nir activity. Nitrite concentration in this assay is proportional to the A₅₄₀ of a sample. Thus, by subtracting the A₅₄₀ obtained during a test reaction from the A₅₄₀ of a standard containing 36 nmol of sodium nitrite in the absence of cells, we were able to quantify the amount of nitrite reduced. Activity was calculated by using the formula units = 100 · ΔA₅₄₀/t · A₆₀₀, where ΔA₅₄₀ represents nitrite consumed during the course of incubation (A_540standard − A_540test), t is the time of incubation in minutes, and A₆₀₀ is the density of cells in a culture.

RESULTS

Isolation of mutants.

A pool of approximately 20 R. sphaeroides mutants , unable to grow anaerobically in the dark on nitrite-amended medium, was isolated by using a modified Tn5 as previously described (40, 44). Characterization of several photosynthetically competent mutants revealed insertions within either nirK or the nor operon (2, 44). The mutant pool also contained several strains that were unable to grow photosynthetically. These strains were studied in more detail due to the likelihood that they carried the Tn5 insertion within genes whose effect on nitrite respiration has yet to be characterized.

Tn5 insertion sites within three independent strains, 14.3, 5.11, and 133.12, were determined to lie near the 5′ end of the prrB gene (Table 1). In a fourth strain, designated strain 37, the insertion site of the transposon was found to lie midway between the 5′ and 3′ ends of the open reading frame (Table 1). This places the insertion downstream of the histidine proposed to be the site of autophosphorylation, whereas in the other three mutants, insertion occurs upstream of this residue (12). Analysis of the DNA sequence obtained from the amplified products showed that sequence from strain 2.4.3 had a >90% identity with the same region of the 2.4.1 strain (data not shown). The gene order of the prrABC region is conserved among a wide range of photosynthetic bacteria, with prrC and prrA being adjacent and transcribed in the same direction while prrB lies upstream and is divergently transcribed from prrC (Fig. 1A) (25). Sequence analysis showed that this gene order is conserved in R. sphaeroides 2.4.3 (Fig. 1A).

FIG. 1.

Genetic organization of the prrB/prrA and nirK regions. (A) Map of the prrB, prrC, and prrA genes. Triangles represent the approximate locations of Tn5 insertion within the prrB gene of strains 37, 5.11, 14.3, and 133.12 and the approximate location of aph cassette insertion within the prrA gene of strain PRRA. (B) Genetic map of the nirK region, including the various nirK-lacZ fusions used in this study. pJS84 includes a nirK-lacZ fusion with approximately 2 kb of DNA upstream of the transcriptional start. pNIR298 and pNIR84 include nirK-lacZ fusions truncated 298 and 84 bases upstream of the transcriptional start site (t) of nirK, respectively.

Complementation was used to confirm that the inhibition of anaerobic growth on nitrite-amended medium was due solely to disruption of prrB rather than to extragenic mutations or polar effects. Plasmid pUI1641 (12), carrying an intact copy of prrB, was moved into strains 14.3 and 37, and its ability to complement the prrB mutation was assessed by growing the strains both anaerobically in the presence of nitrite and photosynthetically. Growth of both strains was similar to that of 2.4.3 when pUI1641 was present, but in the absence of this plasmid, both strains exhibited much slower growth (Table 2).

TABLE 2.

Complementation of prrB and prrA mutant strains^a

Strain	Genotype	Plasmid	Growth on nitriteb	Photosynthetic growthc
2.4.3	Wild type	None	+++	+++
14.3	prrB	None	−/+	−/+
37	prrB	None	−	−/+
PRRA	prrA	None	−	−
14.3	prrB+	pUI1641	+++	+++
37	prrB+	pUI1641	+++	+++
PRRA	prrA+	pUI1621	+++	+++

^aGrowth of mutant strains was compared with that of strain 2.4.3. pUI1641 contains a wild-type copy of prrB, while pUI1621 contains a wild-type copy of prrA. Strong growth, +++; very weak growth, −/+; undetectable growth, −.

^bCells were struck onto Sistrom's agar plates supplemented with 0.5 g of NaNO₂ per liter. Plates were placed in an anaerobic bell jar, which was flushed with argon three times and incubated in the dark at 30°C.

^cCells were struck onto Sistrom's agar plates and placed in a bell jar made anaerobic by using a BBL GasPak (Becton Dickinson). This was incubated over incandescent light until robust growth was observed for strain 2.4.3.

Nitrite reductase activity of the PrrB mutants.

Assays of nitrite reductase activity from whole cells revealed that Nir activity was over fivefold lower in strain 14.3 than in strain 2.4.3 when grown under conditions optimal for Nir expression (Table 3). Nir activity in strain 37 was even lower than that of 14.3, having activity that was only one-twentieth that of strain 2.4.3 (Table 3). The lower activity in strain 37 was consistent with the observation that this strain accumulated more nitrite during microaerobic growth in medium amended with nitrate (Table 3).

TABLE 3.

Nitrite accumulation and whole-cell Nir activity of PrrB/PrrA strains

Strain	Genotype	Plasmid	Nitrite accumulationa	Nir activityb
2.4.3	Wild type	None	−	10.7 ± 0.7
14.3	prrB	None	+	1.7 ± 0.3
37	prrB	None	+++	0.4 ± 0.2
PRRA	prrA	None	+++	0.6 ± 0.2
2.4.3	Wild type	pWLNIR	−	9.5 ± 0.5
37	prrB	pWLNIR	−	9.6 ± 0.1
PRRA	prrA	pWLNIR	−	9.5 ± 0.4
14.3	prrB	pWL-CYCA	+++	1.3 ± 0.2
CCON	ccoN	None	++	0.8 ± 0.2

^aNitrite accumulation after overnight microaerobic growth at 30°C in the presence of nitrate. Nitrite accumulation levels: very high, +++; moderate, ++; none, −.

^bIn vivo activity of nitrite reductase from cultures grown as described for nitrite accumulation. Data represent activity from at least two independently grown cultures. Units are defined in Materials and Methods, “Assays for enzymatic activities.”

Expression of nirK in the prrB background.

Expression of a nirK-lacZ fusion was used to determine whether the decrease in Nir activity in the prrB mutants reflected a change in the transcription of the gene encoding Nir. When cultures were grown microaerobically in medium amended with nitrate, expression of the fusion was 3.3-fold lower in strain 14.3 and over 10-fold lower in strain 37 than in strain 2.4.3 (Fig. 2). When grown microaerobically in unamended medium, both of the prrB mutants had approximately half the expression of 2.4.3. For comparison, the nirK-lacZ activity of strain R125, in which nnrR had been interrupted, was 18-fold lower than that of 2.4.3 in the nitrate-amended media and 3-fold lower in unamended media (Fig. 2).

FIG. 2.

Expression of a nirK-lacZ fusion under microaerobic conditions in various backgrounds. Results for cells grown in Sistrom's medium are indicated with cross-hatched bars, and those for cells grown in Sistrom's medium amended with 12 mM nitrate are indicated with solid bars. The strains are as follows: 2.4.3, wild-type denitrifier; R125, nnrR mutant; 14.3, prrB Tn5 mutant; 37, prrB Tn5 mutant; CCON, ccoN insertional mutant (expression in CCON was monitored in nitrate-amended medium only). For further description of strains, see Table 1. Activity represents data from duplicate assays on at least two independently grown cultures.

Disruption of PrrA.

Previously, Eraso and Kaplan noted that the disruption of prrA resulted in photosynthetic incompetence, while prrB mutants could grow photosynthetically but required high light intensity (10). Further studies monitoring the transcription of PrrA-regulated promoters revealed that the expression of puc and puf was substantially lower in a PrrA mutant than in a PrrB mutant. The difference in expression was attributed to cross talk between PrrA and other kinases in the absence of PrrB (10). Since PrrB seemed to be involved in the expression of nirK, we predicted that inactivation of prrA would more severely affect nirK expression. The prrA gene in 2.4.3 was amplified with primers designed from the 2.4.1 sequence. Cloning and sequencing of this gene revealed that it had 90% DNA sequence identity to the prrA gene from 2.4.1 and encoded a product with 99% identity (data not shown). Using this DNA fragment, the prrA gene was disrupted with a kanamycin cassette. The prrA mutant strain, designated PRRA, had a pale color characteristic of previously described prrA mutants (12) and was impaired in photosynthetic growth and anaerobic growth on nitrite (Table 2). Complementation of PRRA with pUI1621, which contains the prrA gene from 2.4.1, restored the photosynthetic growth of the strain and its growth on nitrite (Table 2), confirming that the observed phenotype of these mutants was due solely to an interruption of prrA.

Expression of the nirK-lacZ fusion was monitored in strains 2.4.3 and PRRA under various conditions to evaluate the impact of prrA inactivation on nirK expression. When grown aerobically in unamended Sistrom's medium, both strains had roughly equal levels of nirK-lacZ expression (Fig. 3). The expression of the nirK-lacZ fusion in strain 2.4.3 was threefold higher than that in PRRA during microaerobic growth in Sistrom's medium (Fig. 3). As shown in Fig. 3, when microaerobic cultures contained nitrate, nirK-lacZ expression was 10-fold higher in strain 2.4.3 than in PRRA. Surprisingly, culturing either strain under strictly anaerobic conditions with nitrate led to poor expression of the fusion (Fig. 3). This result was unexpected, because denitrification is typically considered an anaerobic process, and the expression of the nirK-lacZ fusion in the 2.4.3 background was expected to lead to high activity.

FIG. 3.

Expression of a nirK-lacZ fusion under various conditions in strains 2.4.3 and PRRA. The strains carried the nirK-lacZ fusion on plasmid pJS84 (Fig. 1). Results for 2.4.3 are indicated with cross-hatched bars, and results for PRRA, the prrA mutant strain, are indicated with solid bars. The nirK-lacZ activity of both strains was measured under aerobic, microaerobic, microaerobic plus nitrate, and anaerobic plus nitrate conditions (see Materials and Methods for growth conditions).

Expression of nnrR.

One mechanism whereby PrrBA could regulate nirK expression is by acting on the expression of the nnrR gene, which encodes the transcriptional activator NnrR. Decreased expression of nnrR in a prrA or prrB strain could explain both the accumulation of nitrite in nitrate-grown cultures and the reduced expression from the nirK-lacZ fusion. In order to assess the effect of the PrrBA system on nnrR expression, plasmid pIT53 (45), which contains an nnrR-lacZ fusion cloned into the broad-host-range plasmid pRK415 (20), was moved into strains 2.4.3, PRRA, and 14.3. The β-galactosidase activity of each strain was monitored by using cultures grown under microaerobic conditions in nitrate-amended medium. Under these conditions, the expression of the nnrR-lacZ fusion in strains 14.3 and PRRA resulted in 185 ± 7 and 188 ± 3 Miller units of activity, respectively, while the expression in 2.4.3 was 115 ± 6 Miller units. The lower expression exhibited by strain 2.4.3 is consistent with the previous observation that NnrR negatively autoregulates its own expression in the presence of nitric oxide (45). The lack of negative autoregulation in strains 14.3 and PRRA likely results from decreased production of nitric oxide, which is a consequence of their impaired Nir activity. The fact that nnrR-lacZ expression was not impaired in either mutant strain argues against the PrrBA system indirectly affecting nirK expression via nnrR expression.

Effect of nirK expression from a heterologous promoter.

Transcription of nirK is part of a positive feedback loop. The enzyme encoded by this gene, nitrite reductase, catalyzes the reduction of nitrite to nitric oxide. This latter molecule is the signal for the transcriptional activator of nirK, NnrR (21). Because of this positive feedback loop, assessing the role of PrrBA in modulating nirK expression is not straightforward. For instance, it is known that PrrBA (or RegBA) is involved in the transcriptional activation of a number of genes encoding components of the electron transport chain. In particular, the cycA and cycY genes, encoding cytochromes c₂ and c_Y, respectively, are partially under the regulation of this system (17, 43). Urata and Satoh have shown that cytochrome c₂ donates electrons to nitrite reductase in some Rhodobacter species (46), while preliminary evidence suggests that cytochrome c_Y may also donate electrons to nitrite reductase (W. P. Laratta and J. P. Shapleigh, unpublished data). Thus, impaired electron flow in strain PRRA could result in the decreased Nir activity observed, which would lead to a decrease in nirK-lacZ expression.

To determine whether decreased nirK expression was due to impairment of nitrite reductase activity, the transcription of this gene was uncoupled from Nir activity. Plasmid pWLNIR was constructed so that the nirK gene would be expressed from the PrrnB ribosomal promoter of R. sphaeroides 2.4.1 (53). The plasmid was moved into strain 11.10, a nirK-deficient strain, to verify that nitrite reductase was produced. After overnight growth in nitrate-amended medium, 11.10 pWLNIR was tested for nitrite accumulation and for in vivo nitrite reductase activity. Both tests confirmed that the PrrnB′-nirK fusion could complement the nitrite reductase deficiency (data not shown).

Given that the PrrnB′-nirK fusion of pWLNIR could complement the nirK deficiency in 11.10, the plasmid was moved into strains 2.4.3 and PRRA. These strains were grown microaerobically in nitrate-amended medium and assayed for nitrite accumulation and nitrite reductase activity. Table 3 shows nitrite accumulation and nitrite reductase activity for strains 2.4.3, PRRA, 2.4.3 pWLNIR, and PRRA pWLNIR. Only strain PRRA accumulated nitrite, while the strains complemented with pWLNIR had nitrite reductase activities similar to that of strain 2.4.3.

As an additional test to determine whether changes in cytochrome content might be influencing nitrite reductase activity, a plasmid carrying a PrrnB′-cycA fusion was moved into strain 14.3. Since cytochrome c₂ has been shown to donate electrons to Nir (48), its overexpression should relieve problems associated with the limitation of reductant. Spectrophotometry confirmed that the presence of the PrrnB′-cycA fusion in strain 14.3 resulted in the production of a cytochrome that was spectroscopically indistinguishable from native cytochrome c₂ (data not shown). However, even with excess cytochrome, the in vivo nitrite reductase activity of strain 14.3 remained impaired (Table 3).

Truncation of the nirK promoter.

The experiments described above are not consistent with PrrBA indirectly affecting nirK expression by controlling the expression of regulatory factors, Nir assembly, or the supply of reductant. This suggests that PrrBA may directly regulate nirK expression. Previous studies of PrrA and its homologue RegA suggest that these proteins often act in concert with other regulators to activate gene transcription (8, 31, 54). If this occurs with nirK, then PrrA might be predicted to have a binding site upstream of the transcription start site. Previous work has identified the transcription start site of nirK and identified a putative NnrR binding site centered 43.5 bases upstream (44). Similar binding sites have been identified upstream of two other sets of genes, the nor operon and the nnrS gene, which lie within the NnrR regulon. Studies on the promoter region of the nor operon demonstrate that its promoter retains full expression when truncated to within 41 bases of the center of the putative NnrR binding site (2). Similarly, significant expression of the nnrS promoter is retained when it is truncated to within 21 bases of the center of its putative NnrR binding site (3). Comparable truncation experiments were carried out with nirK to determine whether there might be additional sequence elements required for its expression. For the truncation experiments, plasmid pJS84, which contains a nirK-lacZ fusion with approximately 2 kb of DNA upstream of the transcriptional start, was used as a control. The expression from pJS84 was compared to those from plasmids pNIR298 and pNIR84, which include 298 and 84 bases upstream of the nirK transcription start, respectively (Table 1 and Fig. 1B). Cultures of strain 2.4.3 and PRRA containing these plasmids were grown overnight under microaerobic conditions in media amended with nitrate. As shown in Table 4, strain 2.4.3 carrying the pNIR298 nirK-lacZ fusion had 85% of the activity of the full-length fusion. Under the same conditions, strain 2.4.3 carrying the pNIR84 nirK-lacZ fusion had only 3% of the activity of the pJS84 nirK-lacZ fusion (Table 4). As expected, only limited expression of the fusions was observed in strain PRRA (Table 4). These data suggest that there are additional regulatory sites within the nirK promoter that are not present in other genes in the NnrR regulon.

TABLE 4.

Expression of truncated nirK-lacZ fusions

Straina	Fusion truncationb	β-Galactosidase activity (Miller units)c	% of full length
2.4.3	−2,000	1,884 ± 107	100.0
2.4.3	−298	1,603 ± 53	85.0
2.4.3	−84	64 ± 1	3.0
PRRA	−2,000	167 ± 3	9.0
PRRA	−298	41 ± 3	2.0
PRRA	−84	28 ± 1	1.0

^a2.4.3 is the wild-type denitrifying strain of R. sphaeroides, while PRRA contains an ΩSm/Sp cassette in the open reading frame of prrA. Strains were microaerobically grown in medium amended with 12 mM nitrate.

^bFusion truncation represents the distance upstream of the transcription start for nirK. The putative NnrR binding site spans the region from −51 to −37 relative to the nirK transcriptional start.

^cActivities ± standard deviations were determined in duplicate assays of at least two independently grown cultures.

Role of the cbb₃-type cytochrome oxidase in nirK expression.

Oh et al. have presented a model for PrrA regulation whereby the sensor kinase, PrrB, receives a signal regarding cellular redox status from the cbb₃-type cytochrome oxidase (30, 32). In this model, the flow of reductant through the cbb₃-type cytochrome oxidase results in an inhibitory signal being sent to PrrB via PrrC. As the electron flow decreases, this signal is relieved, resulting in an increase in PrrB kinase activity with a concomitant increase in the level of PrrA-P (11, 31, 33). This explains why inactivation of the structural genes for the oxidase causes aerobic expression of some of the genes involved in photosynthesis. Given these results, it might be possible that inactivation of the cbb₃-type cytochrome oxidase would have an observable effect on nirK expression.

To test whether inactivation of the cbb₃-type cytochrome oxidase affected nirK expression, two 2.4.1 cbb₃-type cytochrome oxidase mutants, CCONΔ and CCOQΔ, which contain in-frame deletions of the ccoN and ccoQ genes, respectively, were utilized (32). Plasmid pAK1 (22), which carries a wild-type copy of nirK, was moved into each of these strains. Kwiatkowski et al. demonstrated that the addition of pAK1 to strain 2.4.1, a strain with all the machinery necessary for denitrification except for nirK, would allow it to denitrify (22). Unexpectedly, when CCONΔ pAK1 and CCOQΔ pAK1 were grown microaerobically in nitrate-amended medium, both strains accumulated nitrite, suggesting that Nir was not active (data not shown).

To further study why nitrite accumulated in the 2.4.1 derivatives, we disrupted the ccoN gene of the denitrifying strain 2.4.3. As with 2.4.1, the 2.4.3 ccoN derivative, CCON, also accumulated nitrite after microaerobic growth in nitrate-amended medium. Measurement of its in vivo nitrite reductase activity revealed that it was similar to that found in the PrrA mutant (Table 3). Next, pJS84, which contains the nirK-lacZ fusion, was moved into CCON, and β-galactosidase activity was monitored. Consistent with the decrease in Nir activity, there was >10-fold less expression of the fusion in CCON than in strain 2.4.3 (Fig. 2). These results demonstrate that the ccoN background has a negative effect on nirK expression.

DISCUSSION

In this work, we have shown that the PrrBA system is involved in the transcriptional regulation of nirK. Previously, Tosques et al. demonstrated that nirK expression required NnrR, a transcriptional activator of a small set of genes whose products appear to be involved in nitric oxide metabolism (44). While the data presented here demonstrate that PrrBA must be added to the nirK regulatory model, they argue against the notion that this regulation occurs by PrrBA affecting transcription of nnrR. Importantly, this work suggests that PrrA-P acts to activate nirK in concert with NnrR. Such cooperation by multiple factors appears to be a common theme for genes controlled by PrrA or its orthologues. In R. sphaeroides 2.4.1, numerous genes, including hemA, puc, hemN, hemZ, and bchE, require both FnrL and PrrA-P for expression (31, 54). In R. capsulatus, a number of genes in the RegBA regulon require the presence of additional activators or repressors. For instance, nifA2 requires RegA-P and NtrC for expression, while hupSLC, which is activated by HupR, is repressed by RegA-P (8).

Phenotypic characterization of the four randomly generated prrB mutants demonstrated there were two distinct phenotypes. The first phenotype evidenced only a moderate decrease in nitrite reductase expression, as demonstrated by the accumulation of low levels of nitrite when the strain was cultured in nitrate-amended medium (Table 3) and only a modest decrease in nirK-lacZ expression (Fig. 3). The second phenotype, seen in strain 37, was more severe. It resulted in the accumulation of high levels of nitrite (Table 3) and significantly lower nirK-lacZ activity (Fig. 3). The N-terminal region of PrrB (residues 1 to 182) is a six-helix, membrane-spanning domain, while the C-terminal region (residues 183 to 462) forms a cytoplasmic domain that has kinase and phosphatase activity (34). Mutants with the less severe phenotype contain disruptions in the region predicted to encode the membrane-spanning domain of PrrB. Strain 37, the sole isolate displaying the severe phenotype, has an insertion at the codon for residue 283 of PrrB, which is predicted to lie within the cytoplasmic domain. While it is not clear why there are two distinct phenotypes for the PrrB mutants, it is interesting that the Tn5 was located in the 5′ end of prrB for the mutants with a less severe phenotype whereas in strain 37 it was located in the center of the gene.

The absence of PrrA also resulted in a severe effect on nitrite reductase expression. In the prrA background, nirK-lacZ expression was lower than in mutants containing insertions in the 5′ region of the prrB open reading frame. This result is consistent with previous work studying the expression of the reaction center genes in strain 2.4.1. In that work, Eraso and Kaplan observed that a PrrA strain experienced a more drastic decrease in puc and puf operon expression than did a PrrB strain (10). The expression patterns for the various 2.4.3 PrrBA strains suggest that in those strains carrying insertions near the 5′ end of prrB, sufficient PrrA-P is present to allow a modest level of nirK expression. In the absence of PrrA, there is a severe decrease in nirK expression, similar to that observed in a strain lacking NnrR (45). The requirement for PrrA indicates that nirK expression will occur only under conditions where PrrA-P is present, irrespective of the activity of NnrR.

Previous work has shown that PrrBA controls the expression of other terminal electron acceptors besides Nir. These include the cbb₃-type cytochrome c oxidase in 2.4.1 and R. capsulatus (31, 33, 43) and a ubiquinol oxidase in R. capsulatus (43). Many processes regulated by PrrA share components of the electron transport chain. For instance, cytochrome c₂ shuttles electrons between the cytochrome bc₁ complex and the reaction center during photosynthesis, and it serves as an electron carrier for various respiratory processes (7, 17). Placing the regulation of terminal electron acceptors under the control of a single global regulator creates a regulatory hierarchy, ensuring the presence of specific electron transport components and enzymes to maintain optimal metabolic activity (16, 33, 37). Bioenergetically, respiration of oxygen is the preferred growth mode (56), but when oxygen tensions become lower, PrrA phosphorylation allows any number of genes involved in anaerobic growth to become poised for expression. Then, based on the availability of various factors (e.g., electron acceptors, light, ammonia levels, nitrogen oxides, etc.), additional regulatory elements become involved, ensuring the expression of the enzymes and electron carriers necessary for optimal growth.

Recently, Emmerich et al. defined the imperfect repeat 5′-GCGGCNNNNNGTCGC-3′ as a binding site for RegR, which is a PrrA homologue in Bradyrhizobium japonicum (9). A similar site, 5′-G(C/T)G(G/C)G(G/C)ANN(T/A)(T/A)NNC(G/A)C-3′, was defined by aligning the RegA-protected regions of R. capsulatus (43). The nirK promoter region contains four sites that are similar to those listed above. Each site lies within a region between 50 and 160 bp upstream of the nirK transcriptional start (data not shown). Truncation of the nirK promoter region indicates that DNA upstream of the predicted NnrR binding site is required for expression. This distinguishes nirK from the two other sets of genes in the NnrR regulon, the nor operon and nnrS, whose promoter regions can be truncated near the NnrR binding site with only limited loss of expression (2, 3). The requirement for additional DNA upstream of the NnrR box is consistent with the regulation of nirK being more complex than that of other genes in the NnrR regulon.

The observation that the nor promoter region can be truncated closer to the putative NnrR binding site than can the nirK promoter suggests that nor may not directly require PrrBA for expression. However, directly assessing nor expression in a PrrA mutant is difficult, because activation by NnrR requires the presence of NO (21), a molecule that is lacking in the Nir-deficient prrA background. Initial studies using an exogenous nitric oxide donor, sodium nitroprusside, suggest that nor expression in a prrB mutant background is significantly more responsive to the presence of exogenous NO than is nirK expression (Laratta and Shapleigh, unpublished). Moreover, unlike the nirK promoter region, the nor promoter has no sites with similarity to the imperfect repeats proposed as binding sites for PrrA homologues. Together, these observations suggest that the PrrA may not directly regulate norB expression.

While it is obvious that PrrA-P is required for nirK expression, it is surprising that the inactivation of ccoN, which has been shown to result in expression of PrrA-P-regulated genes (31), prevents transcription of nirK. This suggests that under certain conditions, the PrrBA system might also repress nirK transcription, though it is not clear whether this is a direct or indirect effect. Interestingly, Fig. 3 shows that there is little expression of nirK under fully anaerobic dark conditions, even in the presence of N-oxides. The lack of oxygen in fully anaerobic cultures should physiologically mimic the deletion of the cbb₃-type cytochrome oxidase, thus providing an explanation of why there is limited expression of nirK under anaerobic, dark conditions. However, R. sphaeroides 2.4.3 has been reported to reduce nitrite when cultured under anaerobic light conditions (27). This raises the question of why nirK is expressed anaerobically in the presence of light, but not in its absence. Previous observations suggest that during photosynthetic growth, reductant flows through the cbb₃-type cytochrome oxidase, although the substrate for the enzyme in the absence of oxygen has not been identified (30, 32). Presumably, as long as there is some electron flow through the cbb₃-type cytochrome oxidase, enough of an inhibitory signal on PrrB exists to allow nirK expression. When the cbb₃-type cytochrome oxidase is inactive, either due to genetic disruption or under anaerobic dark conditions, the inhibitory signal is fully relaxed and nirK expression is impaired.

With the data presented here, a tentative model integrating the PrrBA system in the expression of nirK can be postulated. Under aerobic conditions, when PrrA is in its unphosphorylated form, there is no expression from the nirK promoter. As oxygen tensions decrease, the kinase activity of PrrB increases relative to its phosphatase activity, resulting in an increase in the concentration of PrrA-P. Under these conditions, if nitric oxide is present, NnrR, together with PrrA-P, activates transcription of nirK. As conditions become more anaerobic or when the cbb₃-type cytochrome oxidase is deleted, nirK-lacZ expression is repressed, though at this point it is not clear whether this is a direct or indirect result of PrrA-P accumulation. An important prediction for this model is that all other genes in the NnrR regulon are indirectly under the control of PrrBA, since their expression depends on nitrite reductase activity, which is directly regulated by PrrBA.

Finally, the question of why nirK expression, and therefore denitrification, is regulated by PrrBA must be addressed. Elsen et al. suggest that during photosynthesis, RegBA, and by analogy PrrBA, control processes involved in the production and consumption of reducing equivalents, ensuring that the cell does not become overreduced (8). During photosynthesis, the redox state of the ubiquinone pool must be optimally maintained to support cyclic electron transport, and anaerobic respiratory pathways provide a useful sink for excess reducing equivalents (15, 26, 38, 39). Thus, the regulation of nirK by PrrBA suggests that the primary function of nitrogen oxide respiration is redox balancing rather than providing for robust growth as seen in many denitrifiers. This unique regulation distinguishes R. sphaeroides from other “true” denitrifiers whose regulation of nitrogen oxide reductases has been studied in detail. These bacteria use denitrification primarily to support anaerobic growth, and the regulation of their denitrification genes is consistent with this function.