A new assay for nitric oxide reductase reveals two conserved glutamate residues form the entrance to a proton-conducting channel in the bacterial enzyme

Abstract

A specific amperometric assay was developed for the membrane-bound NOR [NO (nitric oxide) reductase] from the model denitrifying bacterium Paracoccus denitrificans using its natural electron donor, pseudoazurin, as a co-substrate. The method allows the rapid and specific assay of NO reduction catalysed by recombinant NOR expressed in the cytoplasmic membranes of Escherichia coli. The effect on enzyme activity of substituting alanine, aspartate or glutamine for two highly conserved glutamate residues, which lie in a periplasmic facing loop between transmembrane helices III and IV in the catalytic subunit of NOR, was determined using this method. Three of the substitutions (E122A, E125A and E125D) lead to an almost complete loss of NOR activity. Some activity is retained when either Glu¹²² or Glu¹²⁵ is substituted with a glutamine residue, but only replacement of Glu¹²² with an aspartate residue retains a high level of activity. These results are interpreted in terms of these residues forming the mouth of a channel that conducts substrate protons to the active site of NOR during turnover. This channel is also likely to be that responsible in the coupling of proton movement to electron transfer during the oxidation of fully reduced NOR with oxygen [U. Flock, N. J. Watmough and P. Ädelroth (2005) Biochemistry 44, 10711–10719].

INTRODUCTION

N₂O (nitrous oxide) along with CO₂ (carbon dioxide) and methane is one of the three most important greenhouse gases. Although the volumes of N₂O discharged into the atmosphere are lower than those of the other two gases, the ability of N₂O to contribute to the global warming process is approx. 300 times greater than that of CO₂ because it is so difficult to break down. On a global scale the largest source of anthropogenic N₂O emissions is agriculture (>70%), because of the increased use in recent decades of artificial fertilizers that contain soluble nitrate (NO₃⁻) and nitrite (NO₂⁻). These are converted by some bacteria and fungi to atmospheric nitrogen in a process known as denitrification. Denitrification requires four specialist enzyme activities to catalyse the sequential conversion of nitrite ions to dinitrogen:

Fungi apparently lack an enzyme that can further reduce N₂O to dinitrogen and under aerobic conditions most denitrifying bacteria produce N₂O instead of dinitrogen partly as a consequence of the oxygen lability of nitrous oxide reductase [1]. Hence organisms that can reduce nitrate to N₂O exert some influence over the process of global warming, and the respiratory NOR (NO reductase) [2,3] which catalyses the final reaction in this pathway in bacteria is the particular focus of the present paper.

Three types of respiratory NOR have been reported in Eubacteria [3–6], but most detailed biochemical and spectroscopic information about NOR has been obtained using the cytochrome c-dependent NOR purified from Paracoccus denitrificans [7–9]. Complementary biochemical information has been reported for the closely related enzymes isolated from Pseudomonas stutzeri [10] and Paracoccus halodenitrificans [11]. The cytochrome c-dependent NORs are usually purified as a two-subunit complex, NorBC [7,12]. The NorC subunit is a mono-haem c-type cytochrome that possesses an N-terminal transmembrane helix that anchors the haem domain to the periplasmic face of the cytoplasmic membrane. In P. denitrificans the function of NorC is to serve as the immediate electron acceptor for two periplasmic electron donors: pseudoazurin and cytochrome c₅₅₀. The catalytic subunit, NorB, whilst functionally distinct, is structurally homologous with the catalytic subunit (subunit I) of the respiratory haem-copper oxidases (e.g. cytochrome aa₃ oxidase) [13,14]. Haemcopper oxidases catalyse the four-electron reduction of dioxygen to water and terminate the aerobic respiratory chains of both bacteria and mitochondria. Much of the energy liberated in this exogonic reaction is conserved by moving protons across the membrane to contribute to the electrochemical proton gradient which is subsequently used to drive ATP synthesis [15,16].

The bioenergetics of succinate-dependent NO respiration in membrane vesicles of Rhodobacter capsulatus was studied using a carotenoid bandshift assay to monitor formation of a transmembrane electrochemical potential [17]. Inhibiting the bc₁ complex of the main respiratory chain with myxothiazol abolished both succinate-dependent NO reduction and the generation of a transmembrane potential. However using ascorbate and the mediator TMPD (N,N,N′,N′-tetramethyl-p-phenylenediamine) to donate electrons directly to NOR in membranes inhibited with myxothiazol restored NOR activity, but a transmembrane potential was not generated. These observations led the authors to conclude that NOR is not electrogenic and that NO was reduced close to the periplasmic surface [17]. However the subsequent discovery of the relationship between NorB and subunit I of the haem-copper oxidases requires that this interpretation is re-evaluated.

Subunit I of the haem-copper oxidases comprises a minimum of 12 transmembrane helices that bind three metal centres: a magnetically isolated bis-histidine co-ordinated low-spin haem and an active site, comprised of a high-spin haem and a copper ion (Cu_B) site magnetically coupled to form a dinuclear centre [18,19]. NorB shares the same overall architecture and the seven conserved histidine residues responsible for ligating the metal centres in subunit I of the haem-copper oxidases are completely conserved in NorB [13,14]. The principle difference between NOR and the haem-copper oxidases is in the elemental composition of the dinuclear centre which in NOR contains an iron (Fe_B) rather than a copper (Cu_B) ion.

Movement of protons to the dinuclear centre of haem-copper oxidases, which lies buried in the hydrophobic interior of the membrane, needs to be assisted by pathways formed by ionizable amino acid side-chains. Structural studies of haem-copper oxidases identified two pathways, known as the D-channel and the K-channel, that could conduct protons from the negative face of the membrane towards the dinuclear centre [20–22]. The specific roles of these two channels have been elucidated by site-directed mutagenesis. The K-channel contains a conserved lysine residue (Lys³⁶² in the Rhodobacter sphaeroides cytochrome c oxidase) and appears to be used to transfer protons required for the (re)reduction of the dinuclear centre during turnover [23,24]. The D-channel is named after a conserved aspartate residue (Asp¹³²) at its entrance which when changed to either an alanine residue (D132A) or the corresponding amide (D132N) by site-directed mutagenesis results in a loss of activity [25].

The residues that contribute to the D- and K-channels in subunit I of the haem-copper oxidases are not conserved in NorB [3,26]. This is not surprising as NOR has no requirement to take protons from the cytoplasm for catalysis (see above). However, homology modelling of NorB using the amino acid sequence of NorB from Pseudomonas stutzeri and the co-ordinates of the X-ray structure of P. denitrificans cytochrome aa₃ [27] suggests that the haem::Fe_B dinuclear centre, at which NO is reduced to N₂O, is buried in the lipid bilayer. Consequently NorB is likely to contain structure(s) that allow protons consumed in catalysis to move from the periplasmic face of the enzyme to the buried active site. Consistent with this notion is a recent report that clearly demonstrates proton-coupled electron transfer to the active site of NOR in the reaction of the fully reduced enzyme with molecular oxygen [28].

Multiple sequence alignments of the derived amino acid sequences of NorB from both cytochrome c- and quinone-dependent NORs have shown there to be five glutamate residues that are highly conserved in the NorB subunit, but which are not found in the orthodox haem-copper oxidases [3,29]. We have proposed that some of these glutamate residues may form a proton channel leading from the periplasm to the active site. Secondary structure analysis places the glutamate residues at positions 122 and 125 in a periplasmic facing loop between transmembrane helices III and IV and so ideally placed to accept protons from the periplasm. The conserved glutamate residues at positions 198 (helix VI), 202 (helix VI) and 267 (helix VIII) are all candidates for terminating the channel close to the active site. Previous work in this laboratory has shown that substitution of two of these residues, Glu¹²⁵ and Glu¹⁹⁸, with alanine gives rise to catalytically inactive NOR [29].

One problem in assessing electron transfer to NOR is that although NOR activity can easily be measured using an amperometric steady-state assay, the assay conditions reported in the literature have used different mediators either singly or in combination with the chemical electron donor ascorbate. For example PMS (phenazine methosulfate) [30], TMPD [7] and a combination of horse heart cytochrome c plus TMPD [31], or DAD (2,3,5,6-tetramethylphenylenediamine) [32] have all been used to mediate electron transfer to NOR. Thus far no single definitive method for assaying NOR using a physiologically relevant electron donor has been reported.

The aim of the work described in the present paper was threefold. First, we wanted to devise a reliable steady-state assay of NOR activity using a physiologically relevant species as the immediate electron donor. The second goal was to optimize the assay to allow assay of recombinant NOR expressed in membranes of Escherichia coli, so that site-directed mutants could be rapidly and reliably screened for changes in catalytic activity. Finally, we wished to use this system to determine whether mutations in the two conserved glutamate residues, Glu¹²² and Glu¹²⁵, that lie in a periplasmic loop between helix III and helix IV, lead to changes in steady-state activity that would be commensurate with them forming the entrance to a channel that might couple the movement of protons from the periplasm to the active site with internal electron transfer during turnover.

EXPERIMENTAL PROCEDURES

Bacterial strains and growth

Bacterial strains and plasmids used in the present paper are summarized in Table 1. The method for the expression of recombinant NOR in E. coli has been reported previously along with the plasmids expressing the E125A, E198A and E202A variants of NorB [29]. Plasmids expressing new variants of NorB required by the present paper were constructed from the pNOREX vector by site-directed mutagenesis using the ‘Quickchange’ method (Stratagene). Silent restriction sites were incorporated into the mutagenic primers (Table 2) to allow rapid screening of the mutants. Restriction analysis followed by sequencing of the inserts confirmed the presence of correct mutations.

Table 1. Bacterial strains and plasmids used in this study.

	Characteristics	Reference or source
Strain
E. coli GM119	rm−,dam−,dcm−	Lab stock
E. coli JM109	RecA endA gyrA thi hsdR supE relA Δ(lac-proAB) {F’traD proAB+lacIqZΔM15}	Lab stock
E. coli BL21(DE3)	BF−dcm ompT hsdS (rB−mB−) galλ (DE3)	Novagen
P. denitrificans Pd1222	Spr Rifr, restriction deficient	[36]
P. denitrificans Pd21.31	Pd1222 cycA	[36]
P. denitrificans IP1013	Pd1222 pazS::Kmr	[36]
P. denitrificans IP1121	Pd21.31 pazS::Kmr	[36]
P. denitrificans GB1	Pd1222 norB::Ω-Km	[29]
Plasmid
pEC86	Plasmid containing ccm gene cluster	[40]
pJR2	Plasmid containing paz gene	[35]
pET24d	Cloning vector	Novagen
pET-psaz	Pseudoazurin expression vector	The present study
pNOREX	6.1 kb NOR-cycA fusion cloned into pUC18	[29]
pNORxx16	1.6 kb Xba1-Xho1 nor fragment cloned in pBluescript KS+	[29]
pNOR122A	E122A NorB mutation in pNORXX16	The present study
pNOR122D	E122D NorB mutation in pNORXX16	The present study
pNOR122Q	E122Q NorB mutation in pNORXX16	The present study
pNOR125A	E125A NorB mutation in pNORXX16	[29]
pNOR125D	E125D NorB mutation in pNORXX16	The present study
pNOR125Q	E125Q NorB mutation in pNORXX16	The present study
pNOR198A	E198A NorB mutation in pNORXX16	[29]
pNOR202A	E202A NorB mutation in pNORXX16	[29]
pNOR267A	E267A NorB mutation in pNORXX16	[29]
pE122Aex	E122A NorB mutation in pNOREX	The present study
pE122Dex	E122D NorB mutation in pNOREX	The present study
pE122Qex	E122Q NorB mutation in pNOREX	The present study
pE125Aex	(formally p125EX) E125A NorB mutation in pNOREX	[29]
pE125Dex	E125D NorB mutation in pNOREX	The present study
pE125Qex	E125Q NorB mutation in pNOREX	The present study
pE198Aex	(formally p198EX) E198A NorB mutation in pNOREX	[29]
pE202Aex	(formally p202EX) E202A NorB mutation in pNOREX	[29]
pE267Aex	(formally p267EX) E267A NorB mutation in pNOREX	[29]

Table 2. Primer design for site-directed mutagenesis of NorB.

	Primers	Restriction enzyme incorporated
E125D	5′-CGCGAGTTCCTTGATCAGCCGAAATGG-3′
	3′-GCGCTCAAGGAACTAGTCGGCTTTACC-5′	SpeI
E125Q	5′-CGCGAGTTCCTGCAGCAGCCGAAATGG-3′
	3′-GCGCTCAAGGACGTCGTCGGCTTTACC-5′	PstI
E122A	5′-AAGGAGGGGCGCGCGTTCCTGGAACAG-3′
	3′-TTCCTCCCCGCGCGCAATGACCTTGTC-5′	BssHI
E122D	5′-AAGGAGGGTCGCGACTTCCTGGAACAG-3′
	3′-TTCCTCCCAGCGCTGAAGGACCTTGTC-5′	NruI
E122Q	5′-AAGGAGGGGCGCCAATTCCTGGAACAG-3′
	3′-TTCCTCCCCGCGGTTAAGGACCTTGTC-5′	NarI

Small-scale batch cultures (25 ml) used for growth curves and whole cell assays of P. denitrificans 1222 and related strains were routinely grown at 37 °C without aeration on succinate medium [25 mM KNO₃, 30 mM succinate, 55 mM Na₂HPO₄, 11 mM KH₂PO₄, 6 mM NH₄Cl, 0.4 mM MgSO₄ and 2 ml/l Vishniac's trace elements solution (pH 7.6)] [33]. The medium was supplemented with antibiotics as required at the following concentrations: 25 μg/μl kanamycin and 25 μg/μl spectinomycin. Native NOR was purified from P. denitrificans 93.11 (ΔctaDI, ΔctaCII qoxB::kan^R) grown on minimal medium under anaerobic denitrifying conditions in a 90 litre Bioflow-5000 fermentor (New Brunswick Scientific) as previously described [8].

Strains of E. coli expressing recombinant P. denitrificans NOR were grown in 15×1 litre batch cultures of E. coli JM109 transformed with pNOREX and pEC86 as previously described [29]. The structural gene for pseudoazurin from Paracoccus pantotrophus (pazS) was cloned into a pET-24d expression vector exactly as described by Pauleta and colleagues [34]. The resulting vector (pET-psaz) was used to transform E. coli BL21(DE3) and grown aerobically at 37 °C in Luria–Bertani medium supplemented with 30 μg/μl kanamycin and 0.5 mM CuSO₄ for 24 h without induction. Cells were harvested by sedimentation using a Beckman Avanti J-20 centrifuge (JLA 8.1000 rotor) at 5500 g for 20 min.

Preparation of whole cells, membrane vesicles and proteins

Whole cells for assay of NOR activity were harvested from 25 ml of culture by sedimentation (6000 g for 20 min). In order to remove excess nitrate, the cells were then washed three times by re-suspension in 1.5 ml nitrate-free medium and subsequent collection by sedimentation (10600 g for 5 min) in a benchtop micro-centrifuge. After the final wash the cells were resuspended in 1.5 ml nitrate-free medium prior to use. Membrane vesicles were prepared by passing cells twice through a French Pressure Cell (Aminco), at 75.9 MPa, followed by ultracentrifugation on a Beckman Optima XL-100K at 30000 rev./min (45Ti rotor) for 2 h at 4 °C. Membranes were then resuspended and washed in 50 mM Tris/HCl (pH 7.6), 0.5 M NaCl and sodium-deoxycholate (0.05% w/v) and sedimented by centrifugation at 180000 g for 2 h at 4 °C. The washed membranes were resuspended in 50 mM Tris/HCl (pH 7.6), pelleted and resuspended in the same buffer to approx. 20 mg/ml determined by the bicinchoninic acid method with BSA as a standard.

NOR from P. denitrificans was purified as previously described by Hendricks et al. [7]. Pseudoazurin was purified as described by Leung et al. [35]. The purity of NOR and pseudoazurin was assessed by SDS/PAGE and UV-visible spectra which were recorded using a Hitachi U3100 spectrophotometer. The concentrations of oxidized NOR and pseudoazurin were calculated on the basis of their molar absorption coefficients: NOR, ϵ₄₁₁=3.11×10⁵ M⁻¹·cm⁻¹; pseudoazuruin, ϵ₅₉₀=1360 M⁻¹·cm⁻¹. Whole cells, membrane vesicles and purified protein were stored at −80 °C until needed.

Determination of NOR concentrations in whole cells and membranes

Proteins were separated by SDS/PAGE, blotted onto nitrocellulose and blocked overnight at 4 °C in PAWS [pre-antibody wash solution; 10 mM Tris/HCl, 150 mM NaCl (pH 7.5) supplemented with 5% (w/v) non-fat dried milk powder]. The nitrocellulose was probed for 3 h at 4 °C with an antibody (1:10000 dilution in PAWS) raised in rabbit and directed against a synthetic peptide corresponding to the NorB sequence. Excess primary antibody was removed by washing the membrane three times with PAWS buffer. The nitrocellulose was subsequently probed with a goat-anti-rabbit IgG conjugated to alkaline phosphatase (1:7500 in PAWS) for 3 h at 4 °C with agitation. The membrane was then washed extensively with PAWS buffer followed by distilled water. The blot was developed using the Sigma fast BCIP (5-bromo-4-chloroindol-3-yl phosphate)/NBT (Nitro Blue Tetrazolium) system. Known amounts of purified NOR from P. denitrificans were run as standards alongside samples of membranes and cells. The intensity of the standards on the developed blots was analysed using Kodak ID 3.5 software and used to construct a standard curve from which the concentration of NOR in the membrane could be calculated.

Amperometric assays of NO consumption

The consumption of NO was measured anaerobically using a modified oxygen electrode (Oxytherm, Hansatech Instruments). The platinum cathode was polarized at −0.8 V with respect to the Ag:AgCl reference electrode in order to increase sensitivity towards NO. The electrolyte was 1 M KCl and a PTFE (polytetra-fluoroethylene) membrane (12.5 μM thickness) was used as a gas-permeable barrier to protect the electrodes from the reactants in the chamber.

NO stock solutions were prepared by addition of 8 ml NO gas to 3 ml of nitrogen-sparged water (pH 3.0). The concentration of NO was confirmed to be 1.8 mM by titration against reduced myoglobin: ϵ_421nm=114 mM⁻¹·cm⁻¹ at 20 °C.

Assay of NOR activity in whole cells

The reaction chamber was filled with 2 ml of nitrate-free succinate medium [30 mM succinate, 55 mM Na₂HPO₄, 11 mM KH₂PO₄, 6 mM NH₄Cl, 0.4 mM MgSO₄ and 2 ml/l of Vishniac's trace element solution (pH 7.6)] [33] and warmed to 30 °C with stirring. The lid of the chamber was added and approx. 0.1 mg of pre-prepared whole cells were added using a Hamilton syringe. The cells were allowed to respire and consume the oxygen in the media. Once the conditions in the chamber were anaerobic, NO was added to a final concentration of 100 μM and an immediate increase in the signal was observed with the subsequent decay being due to the succinate-dependent NOR activity of the cells. The initial rate of NO consumption was calculated from the time taken for 20% of the NO to be consumed. Note that the response of the electrode in the presence of whole cells to the addition of 100 μM NO is less than in the presence of purified enzyme which we attribute to non-enzymic consumption of NO.

Assay of NOR activity in P. denitrificans membranes

Resuspended membrane vesicles (100 μl) containing approx. 0.2 mg total protein were added to 2 ml of the same reaction medium in a stirred reaction chamber thermostatically maintained at 30 °C. The lid was secured and oxygen removed from the chamber by the glucose (16 mM)/glucose oxidase (4 units/ml)/catalase (20 units/ml) system. Once the system was anaerobic, 5 mM nitrogen-sparged ascorbate and 60 μM pseudoazurin were added. When a steady baseline was observed, 100 μM NO was added. The initial rate of reaction was calculated by the time taken for 20% of the NO to be consumed.

Assay of NOR activity in E. coli membranes

The reaction chamber was filled with 2 ml of 20 mM phosphate buffer [50 mM KCl (pH 7.6)] at 30 °C and the lid attached. Subsequent additions were made using Hamilton syringes. Resuspended membrane vesicles (100 μl) containing approx. 400 ng of NOR were added to the stirred reaction chamber. Anaerobic conditions were achieved in the chamber by a glucose (16 mM)/glucose oxidase (4 units/ml)/catalase (20 units/ml) system. Once the system was anaerobic, 5 mM nitrogen-sparged ascorbate and 1.85–200 μM pseudoazurin were added. When a steady baseline was observed, 100 μM of NO was added. Again the initial rate of consumption of NO was measured by the time taken for 20% of the NO to be consumed.

Assay of purified NOR

NOR (25 nM), in 2 ml of 20 mM phosphate buffer [50 mM KCl (pH 7.6)] supplemented with 0.02% (w/v) dodecyl-β-D-maltoside, was added to the 30 °C stirred reaction chamber. The lid was fastened and oxygen removed from the chamber by a glucose (16 mM)/glucose oxidase (4 units/ml)/catalase (20 units/ml) system. Once the system was anaerobic, 5 mM ascorbate and 5–200 μM pseudoazurin were added. Once a steady baseline was observed, 100 μM of saturated NO was added to initiate the reaction. The rate of NO consumption was measured over the initial part of the reaction.

RESULTS

NOR activity in strains in which the gene encoding either pseudoazurin or cytochrome c₅₅₀ has been deleted

At the outset we wished to compare two natural electron donors with NOR, pseudoazurin and cytochrome c₅₅₀, in order to determine which might be the most appropriate mediator to use with ascorbate for in vitro assays of NOR activity. Strains of P. denitrificans in which the genes that encode cytochrome c₅₅₀ and pseudoazurin had been disrupted, either individually or both together, have been described previously [36]. Comparison of these strains with the parent P. denitrificans strain and the ΔNorB strain [37] showed that disruption of either the gene encoding cytochrome c₅₅₀ (cycA⁻) or the gene encoding pseudoazurin (pazS⁻) had little effect on anaerobic growth (Figure 1A), and these mutations only slightly attenuated succinate-dependent NOR activity (Figure 1B) when compared with the parent strain (Pd1222). Interestingly, using the same criteria, a strain (cycA⁻ pazS⁻) in which the genes encoding both cytochrome c₅₅₀ and pseudoazurin had been disrupted gave a phenotype that was indistinguishable from that of ΔNorB (Figures 1A and 1B).

Figure 1. Succinate-dependent NOR activity in whole cells.

(A) Growth curves for batch cultures of P. denitrificans; Pd1222 (–––) and mutant strains cycA⁻ (– – –), pazS⁻ (-------), cycA⁻ pazS⁻ (– -- –) and ΔNorB (– - –). Each point is the mean of three replicate observations. (B) Succinate-dependent NOR activity measured using whole cells. NO consumption was measured anaerobically using a Clark-type platinum electrode polarized at −0.8 V with respect to a silver reference electrode. In a 2 ml reaction volume, approx. 120 mg of whole cells was added to 2.0 ml of a 20 mM Hepes buffer containing 20 mM succinate (pH 7.2 at 30 °C). Once cells had consumed all of the oxygen in the reaction, NO was added to a final concentration of 100 μM. Pd1222 (–––) and mutant strains cycA⁻ (– – –), pazS⁻ (——-), cycA⁻ pazS⁻ (– – –) and ΔNorB (– - –). (C) Western blot of prepared membranes probed with an anti-NorB antibody. Lane 1, loading control (1.4 μg of purified NOR); lane 2, molecular mass markers; lane 3, Pd1222; lane 4, cycA⁻; lane 5, pazS⁻; lane 6, cycA⁻PazS⁻; lane 7, ΔNorB.

Analysis of protein expression by SDS/PAGE followed by immunostaining using a primary antibody directed against NorB confirmed the presence of NorB in the parent strain as well as in the strains lacking a single electron donor (Figure 1C), but not in the double mutant or ΔNorB strains. These data are consistent with the proposal that pseudoazurin and cytochrome c₅₅₀ act as alternative electron donors to NOR. However, the absence of antigen in the Western blot of the double mutant may arise because electron donation to cytochrome cd₁ (nitrite reductase) is not possible, preventing NO production. Transcription of the nor genes is activated by NNR (nitrite reductase and nitric oxide reductase regulator) under anaerobic growth conditions in response to NO or a related species. If NO is not produced by cytochrome cd₁ the NNR genes will not be activated and NOR will, in turn, not be expressed [38,39].

In the present paper we elected to use pseudoazurin as the electron donor to NOR for two reasons. First, unlike cytochrome c₅₅₀, pseudoazurin can be expressed at high levels in E. coli [35] without the need for a helper plasmid containing genes expressing proteins required for cofactor maturation [40]. The subsequent purification of recombinant pseudoazurin is both well established and straightforward [35]. Secondly, the UV-visible spectrum of pseudoazurin does not overlay that of NOR, which will greatly simplify interpretation of any future experiments in which the progress of the reaction is monitored optically.

A pseudoazurin-dependent amperometric assay of purified NOR

We decided to develop an amperometric assay of NOR using ascorbate as the bulk electron donor and pseudoazurin as the mediator. Clark-type oxygen electrodes have been used to detect NO consumption in a number of studies of NOR. Typically the Pt cathode is polarized at −800 mV with respect to the Ag:AgCl reference electrode rather than −700 mV used for measurements of dioxygen consumption to increase the sensitivity of the system to NO. Moreover anaerobic conditions are required so that the electrode only responds to NO and not O₂. These were achieved by the use of degassed buffers from which the final traces of oxygen were removed using glucose/glucose oxidase/catalase scrubbing system. The reduction of NO by ascorbate was absolutely dependent upon the presence of both NOR and pseudoazurin in the assay medium (Figure 2A).

Figure 2. Ascorbate/pseudoazurin-dependent amperometric assay of purified NOR.

(A) NO consumption by purified NOR was measured anaerobically using a Clark-type platinum electrode polarized at −0.8 V with respect to a silver reference electrode. The reaction mixture (2.0 ml) comprised of 50mM Hepes and 0.02% (w/v) dodecyl-β-D-maltoside (pH 7.6). The system was made anaerobic by the addition of 16 mM glucose, 4 units/ml of glucose oxidase and 20 units/ml of catalase. Ascorbate (2 mM), NO (100 μM), NOR (25 nM) and pseudoazurin (5–200 μM) were added as indicated. (B) The effect of varying the concentration of pseudoazurin on the initial rate of NO consumption. Measurements were made in triplicate and the data points represent the means±S.E.M. A line of best fit to the Michalis–Menten equation yielded an apparent K_m of 27 μM and a V_max of 904 min⁻¹.

Assays were done at a range of different concentrations of pseudoazurin and an excess of ascorbate using purified NOR from P. denitrificans. The apparent K_m for pseudoazurin is estimated to be 27 μM (Figure 2B) and the maximum turnover of the enzyme 904 min⁻¹ (15 s⁻¹). The data presented in Table 3 show that the maximum turnover observed with pseudoazurin is comparable with rates observed with the chemical mediators used in previous studies.

Table 3. A comparison of the NOR activity of NorBC purified from P. denitrificans measured using different combinations of mediators in the presence of 5 mM ascorbate.

Mediator	Maximum turnover number (min−1)
Pseudoazurin (200 μM)	801±20 (n=3)
TMPD (2.3 mM)	190±21 (n=3)
Horse heart cytochrome c (50 μM)	106±5 (n=5)
TMPD (2 mM)+horse heart cytochrome c (50 μM)	170±15 (n=3)
Hexa-amineruthenium-(III)-chloride (4 mM)	316±31 (n=6)
Phenazine ethylsulfate (20 μM)	812±138 (n=3)
Phenazine methosulfate (10 μM)	640±24 (n=3)

NOR activity in P. denitrificans and E. coli membrane vesicles

Purifying NOR from membrane vesicles is an expensive and labour-intensive process, not least for the recombinant enzyme expressed in E. coli. Moreover, yields of purified enzyme may fall if a particular mutation destabilizes the enzyme outside its membrane environment. A convenient alternative to purifying the enzyme is specifically to measure NOR activity in membrane vesicles using the ascorbate/pseudoazurin amperometric assay. Such an assay requires any endogenous NOR activity in the membranes to be inhibited, and pseudoazurin to donate electrons to NOR as if it were an isolated system. In membrane vesicles isolated from P. denitrificans, succinate-dependent NOR activity is halted in the presence of myxathiazol, a specific inhibitor of the bc₁ complex (Figure 3A). Subsequent addition of ascorbate/pseudoazurin to the system restores NOR activity (Figure 3A).

Figure 3. NOR activity of cytoplasmic membranes of P. denitrificans and E. coli expressing NOR.

(A) Myxathiazol inhibits succinate-dependent NO reduction in P. denitrificans membranes and the use of ascorbate/pseudoazurin restores activity by donating electrons directly to NOR. (B) NOR activity can be measured directly in E. coli membranes and depends on the expression of NOR from pNOREX. Dotted line, JM109(pEC86)pUC18; solid line, JM109(pEC86)pNOREX. The inset of (B) shows the effect of varying the concentration of pseudoazurin on the initial rate of NO consumption. Each data point represents the means±S.E.M. of at least three independent observations. A line of best fit to the Michalis–Menten equation yielded an apparent K_m value of 35 μM and a V_max of 4877 min⁻¹.

The next step was to apply this assay to measure NOR activity in membrane vesicles isolated from strains of E. coli expressing recombinant NOR. In principle this is straightforward because E. coli does not possess either a bc₁ complex or a membrane-bound respiratory NOR. The electrons for its terminal respiratory complexes are derived directly from the quinone pool. Figure 3(B) shows that NOR activity in E. coli membranes is absolutely dependent upon heterologous expression of NOR by the pNOREX plasmid. To ascertain that pseudoazurin was interacting with recombinant NOR in the membrane vesicles in a similar fashion to the purified enzyme, we investigated the relative NOR activity as a function of pseudoazurin concentration. The apparent K_m of the membrane-associated recombinant NOR for pseudoazurin was estimated to be 35 μM (Figure 3B inset) suggesting a similar mode of interaction to that we established between the purified native NOR and pseudoazurin (Figure 2B). Interestingly the estimated V_max of 4877 min⁻¹ is approx. 5-fold higher than the value obtained with purified enzyme, suggesting some loss of activity associated with removal of the enzyme from its membrane environment.

Alanine scanning mutagenesis of five conserved glutamate residues in NorB

Each of the five conserved glutamate residues in NorB was systematically replaced by alanine residues and the variant NORs expressed in E. coli. The level of expression of NorC was assessed by SDS/PAGE followed by staining for covalently bound haem [41] (Figure 4A) and the level of expression of NorB evaluated by SDS/PAGE followed by immunostaining (Figure 4B). In each case, levels of expression in the membranes of both subunits of the variant NOR was comparable with that of the wild-type enzyme.

Figure 4. SDS/PAGE analysis of E. coli cytoplasmic membranes expressing recombinant wild-type and mutant NOR.

(A) Haem staining confirms the presence of normal amounts of NorC. (B) Western blot of prepared membranes probed with an anti-NorB antibody confirms the presence of NorB in the membranes. For each sample 10 μl of resuspended membranes were added to 90 μl of sample buffer [6 M urea, 5% (w/v) SDS, 5% (w/v) glycerol and 0.05% Bromophenol Blue]. These were incubated at room temperature (22 °C) for 30 min and aliquots analysed by SDS/PAGE. Each lane contained approximately 100 μg of total protein for haem staining (A) or 20 μg of total protein for immunostaining (B). Lane 1, purified P.denitrificans NOR; lane 2, wild-type recombinant NOR; lane 3 E122A; lane 4 E122D; lane 5 E122Q; lane 6 E125A; lane 7 E125D; lane 8 E125Q; lane 9 E198A; lane 10 E202A; lane 11 E267A.

The mutation essentially abolished activity in each of the variants except E202A which retained approx. 50% of the activity of the wild-type enzyme. Since the reduction of NO to N₂O is a proton-consuming process it was proposed that lowering the pH of the reaction would increase the rate of reaction of NOR. Decreasing the pH of the medium to see whether activity could be recovered had little or no effect in all of the mutants except for the E125A variant in which the specific activity at pH 8.5 was 0.5 nmol NO per mg of protein (±0.07, n=3) compared with 0.9 nmol NO per mg of protein (±0.09, n=3) at pH 6.0. This contrasts strongly with the enhanced rate determined for recombinant wild-type NOR expressed in E. coli membranes [19.42 nmol NO per mg of protein (±2.22, n=3) at pH 8.5, compared with 31.92 nmol NO per mg of protein (±1.06, n=3) at pH 6.0].

Further analysis of the surface glutamate residues Glu¹²² and Glu¹²⁵

The periplasmic facing loop between transmembrane helices III and IV is far removed from the active site, and so these glutamates are unlikely to serve as ligands to the cofactors and the loss of activity associated with alanine residue substitution is likely to be related to their proposed function in proton uptake. Consequently it might be expected that replacement of these residues with an amino acid with a functionally related side-chain might lead to some recovery of activity. Again, analysis by SDS/PAGE followed by staining for covalently bound haems showed that NorC was present in the membranes of E. coli strains expressing the E125D/Q and E122D/Q variants (Figure 4A). Parallel experiments in which NorB was visualized by immunostaining showed that NorB was present in each of the membrane vesicle samples (Figure 4B). Assaying NOR activity at pH 7.2 with pseudoazurin (200 μM) and ascorbate (5 mM) showed that substitution of both Glu¹²² and Glu¹²⁵ with glutamine yielded a form of the enzyme that had approx. 10% of the specific activity of the wild-type enzyme. Substitution of Glu¹²² with an aspartate residue gave an enzyme that had approx. 80% of the specific activity of the wild-type enzyme, but rather surprisingly the E125D variant showed virtually no residual activity (Table 4).

Table 4. The effects of amino acid substitutions in NorB on the ascorbate/pseudoazurin-dependent NOR activity of P. denitrificans NOR expressed in E. coli.

	Turnover number (min−1; means±S.E., n=3)	Activity relative to wild-type NOR (%)
Wild-type	4223±120	100
E122A	612±68	14.5
E122D	3511±363	83.2
E122Q	53±3	1.3
E125A	311±29	7.4
E125D	76±17	1.8
E125Q	566±71	13.4
E198A	61±39	1.5
E202A	1660±134	39.3
E267A	155±6	3.6

DISCUSSION

Initial experiments in which succinate-dependent NOR activity was measured in whole cells showed that the ability of NOR to reduce NO was only slightly attenuated in strains lacking either pseudoazurin of cytochrome c₅₅₀. This observation is consistent with an earlier proposal that in P. denitrificans both of these periplasmic proteins can transfer electrons between the bc₁ complex and three respiratory enzymes involved in denitrification: cytochrome cd₁ nitrite reductase, NOR and nitrous oxide reductase [36].

Further evidence for the participation of pseudoazurin in the transfer of electrons between the bc₁ complex and NOR comes from experiments with membrane vesicles prepared from P. denitrificans grown under anaerobic denitrifying conditions. In these vesicles, which contain an intact electron-transfer chain, it is possible to measure succinate-dependent NOR activity (Figure 3A). Electrons are transferred from succinate dehydrogenase to NOR via the ubiquinone pool, the bc₁ complex and a periplasmic electron-transfer protein (cytochrome c₅₅₀ or pseudoazurin). The preparation of membrane vesicles inevitably leads to the loss of periplasmic proteins. Consequently it was not surprising that supplementing the assay medium with 60 μM pseudoazurin further increased the rate of NO reduction (results not shown).

The current view of intra-molecular electron transfer in NorBC based on both measurement of the reduction potentials of the metal centres [8] and kinetic measurements of electron backflow [42] is that electrons move from the c-type haem in NorC to the dinuclear centre via the low-spin b haem in NorB. The presence of the c-type haem in a periplasmic domain of NorC suggests that this centre serves to accept electrons from pseudoazurin/cytochrome c₅₅₀, and that interaction of pseudoazurin is primarily with NorC. The interaction of pseudoazurin with cytochrome c peroxidase from P. denitrificans is believed to occur through ionic interactions between a patch of acidic residues on the surface of the peroxidase and a group of positively charged lysine residues on pseudoazurin [34]. Experiments using purified NOR in the pseudoazurin-dependent steady-state assay showed that NOR activity is also dependent on the ionic strength of the medium with the optimal buffer conditions containing 50 mM KCl (results not shown). Under these conditions we established an apparent K_m for pseudoazurin of 27 μM. This is similar to the apparent K_m for pseudoazurin reported for two other Paracoccus respiratory enzymes, the di-haem cytochrome c peroxidase (13 μM [34]) and cytochrome cd₁ nitrite reductase (9.7 μM [43]).

The participation of cupredoxins as electron carriers is well documented in both photosynthetic and bacterial respiratory electron-transfer chains [44]. The question as to why P. denitrificans appears to require two periplasmic electron carriers remains unresolved, but the work in the present paper provides a sound basis for our use in vitro of pseudoazurin as a physiologically relevant and specific electron donor to NOR expressed in E. coli membranes.

Previously reported amperometric methods to measure the NOR activity of NorBC have relied upon the use of small molecules to mediate electron transfer between a chemical reductant, typically ascorbate, and the enzyme. Different workers have used PMS [30], TMPD [7] or a combination of horse heart cytochrome c plus DAD [32], TMPD [31] or TMPD and PMS [9] to mediate electron transfer from ascorbate to NOR. At the outset of this study we had some concerns about the use of PMS, because ascorbate/PMS represents an efficient system for donating electrons to the qNOR from Ralstonia eutropha [45]. This enzyme lacks a c-type haem and derives electrons for NO reduction directly from the ubiquinone pool. Consequently it is conceivable that in NorBC, PMS can donate electrons directly to NorB, bypassing the haem c of NorC. Thus although 10 μM PMS as a mediator gave rates of NO reduction comparable to 200 μM pseudoazurin (Table 3), it may not be an appropriate choice of mediator.

Another consideration in the choice of electron donors arises from a study of the qCu(A)NOR from Bacillus azotoformans, a Gram-positive denitrifying soil bacterium. This NOR is apparently able to receive electrons directly from the quinone pool via menaquinol and also from a membrane-bound cytochrome c₅₅₁. It is proposed the menaquinol pathway, which has a 4-fold greater maximal activity than the pathway via cytochrome c₅₅₁, is used for NO detoxification, whereas electron donation via the endogenous cytochrome c involves the cytochrome bf₆ complex to serve the bioenergetic needs of the organism [46]. These observations raise the question as to whether the P. denitrificans NorBC can use quinols as alternate electron donors. Recombinant NOR expressed in E. coli inner membranes does not support NADH- or succinate-dependent NO-respiration. Since E. coli does not possess a bc₁ complex this clearly demonstrates that the recombinant NOR is unable to accept electrons directly from the ubiquinol pool. Moreover purified NOR is unable to catalyse NO reduction in the presence of either dithiothreitol plus ubiquinone or dithiothreitol plus menaquinone as an electron-donating system.

E. coli does not possess a membrane-bound respiratory NOR, instead it uses a number of soluble enzymes including NrfA, a pentahaem nitrite reductase [47], NorVW, a flavorubedoxin [48] and flavohaemoglobin [49] to detoxify NO. As already noted, E. coli does not possess a bc₁-dependent respiratory chain and the terminal respiratory complexes derive electrons directly from the quinone pool. Aerobically grown E. coli expresses two quinol-dependent terminal oxidases, cytochrome bo₃ and cytochrome bd, and although cytochrome bo₃ can reduce NO very slowly [50], it is unable to accept electrons from pseudoazurin. Consequently recombinant NOR expressed in E. coli membranes can be specifically assayed with ascorbate/pseudoazurin without the need of additional inhibitors.

Two previous studies have shown that NOR is not electrogenic and that the two substrate protons required for NO reduction are taken up from the periplasm [17,42]. In addition it has recently been reported that in the reaction of the fully reduced enzyme with molecular oxygen, electron transfer to the active site of NOR is coupled to proton uptake [28]. The combination of the conservation of both Glu¹²⁵ and Glu¹²² amongst NorB sequences and their predicted projection into the periplasm on a loop region between transmembrane helices III and IV, suggests a possible role in accepting protons at the entrance to a channel that conducts substrate protons to the dinuclear centre. To investigate this proposition further we constructed variants in which each of these two residues was replaced by an aspartate residue (E122D and E125D) or the equivalent amide (E122Q and E125Q).

The E122Q variant has a level of residual activity similar to the E122A variant. In contrast the E122D variant exhibits a high level of activity. This pattern of results is reminiscent of those obtained from variants of cytochrome c oxidase from R. sphaerioides in which the critical aspartate (Asp¹³²) at the entrance to the D-channel has been mutated. Substitution of this aspartate residue with anything else, other than a glutamate residue, leads to very low levels of oxidase activity, whilst the glutamate mutant has levels of activity similar to the wild-type enzyme [25].

In contrast, substitution of either an aspartate or glutamine residue for the glutamate residue at position 125 in NorB yields an enzyme that has less than 15% of the activity of the wild-type enzyme. In this context it is interesting to note that the glutamate residue (Glu¹⁰¹ in subunit II) at the mouth of the K-channel in cytochrome c oxidases will not tolerate substitution with any other residue [51]. In the absence of a structure it is not possible to say which of the two glutamate residues (Glu¹²² or Glu¹²⁵) in NorB is the primary proton acceptor, but we speculate that it is Glu¹²² and that the role of Glu¹²⁵ in the proposed channel might be structural. This would account for the observation that the length of the amino acid side-chain (e.g. E125Q) seems to be a more important determinant of activity than retaining a carboxylate in that position.

In conclusion this study demonstrates the usefulness of using pseudoazurin as a specific electron donor in experiments involving NorBC. Using recombinant pseudoazurin we measured changes in the steady-state activity of recombinant NOR expressed in E. coli membranes brought about by substitutions to the glutamate residues at positions 122 and 125 of NorB. The results are consistent with these residues forming the entrance to a proton-conducting channel, which we refer to as the ‘E-channel’, whose proposed function is to conduct protons from the periplasm to the active site of the enzyme during turnover.