The Nitric Oxide Reductase Activity of Cytochrome c Nitrite Reductase from Escherichia coli*

Cytochrome c nitrite reductase (NrfA) from Escherichia coli has a well established role in the respiratory reduction of nitrite to ammonium. More recently the observation that anaerobically grown E. coli nrf mutants were more sensitive to NO· than the parent strain led to the proposal that NrfA might also participate in NO· detoxification. Here we describe protein film voltammetry that presents a quantitative description of NrfA NO· reductase activity. NO· reduction is initiated at similar potentials to NrfA-catalyzed reduction of nitrite and hydroxylamine. All three activities are strongly inhibited by cyanide. Together these results suggest a common site for reduction of all three substrates as axial ligands to the lysine-coordinated NrfA heme rather than nonspecific NO· reduction at one of the four His-His coordinated hemes also present in each NrfA subunit. NO· reduction by NrfA is described by a Km of the order of 300 μm. The predicted turnover number of ∼840 NO· s–1 is much higher than that of the dedicated respiratory NO· reductases of denitrification and the flavorubredoxin and flavohemoglobin of E. coli that are also proposed to play roles in NO· detoxification. In considering the manner by which anaerobically growing E. coli might detoxify exogenously generated NO· encountered during invasion of a human host it appears that the periplasmically located NrfA should be effective in maintaining low NO· levels such that any NO· reaching the cytoplasm is efficiently removed by flavorubredoxin (Km ∼ 0.4 μm).

Nitric oxide (nitrogen monoxide or NO⅐) 3 is functionally important throughout the biosphere (1). In humans it serves as a signaling molecule and vasodilator and in Bacteria and Archaea it can provide a substrate for anaerobic respiration. However, NO⅐ is also a potent cytotoxin that forms part of the innate response of hosts to the infective invasion of pathogens. For this purpose NO⅐ is produced from L-arginine by inducible NO⅐ synthase and from nitrite in response to intragastric acidity (2,3). As a consequence, enteric food-borne pathogens such as Escherichia coli have developed mechanisms for NO⅐ detoxification to ensure their survival in the range of oxic, micro-oxic, and anoxic environments encountered in their animal hosts. E. coli lacks genes homologous to those encoding the dedicated respiratory NO⅐ reductases found in bacteria such as Paracoccus denitrificans and Pseudomonas stutzeri (4,5). Alternative enzymes must be employed for NO⅐ management, and three in particular have been implicated in the removal of NO⅐, flavohemoglobin, flavorubredoxin, and cytochrome c nitrite reductase (NrfA) (6 -8).
Flavohemoglobin is a cytoplasmic protein expressed in the presence of nitrate during aerobic and anaerobic growth (9,10). In the presence of oxygen flavohemoglobin converts NO⅐ to nitrate at a rate that has been reported to range from 10 to 670 NO⅐ s Ϫ1 . In the absence of oxygen, flavohemoglobin converts NO⅐ to nitroxyl (NO Ϫ ). However, the rates of NO⅐ reduction are significantly lower than those of NO⅐ oxidation and may not be sufficient for effective NO⅐ detoxification. Under such anoxic conditions flavorubredoxin and NrfA could provide alternative catalysts for NO⅐ removal (6,11,12). Flavorubredoxin is a cytoplasmic protein that reduces ϳ15 NO⅐ s Ϫ1 with a Michaelis constant (K m ) of ϳ0.4 M. However, there is relatively little kinetic information available regarding NO⅐ removal by the periplasmically located NrfA that may represent the first line of defense toward exogenously generated NO⅐.
A possible role for NrfA in NO⅐ detoxification emerged only recently with the observation that an E. coli nrf strain had much greater sensitivity to NO⅐ than the parent strain under anaerobic conditions (6,13). The nrf gene products have a well established role coupling quinol oxidation to nitrite reduction during anoxic and micro-oxic growth in the presence of nitrate or nitrite (14). Quinol oxidation is catalyzed by NrfD and electrons are transferred, most likely via NrfC and NrfB, to NrfA, a homodimeric, deca-heme-containing cytochrome c nitrite reductase (15). NrfA homologs are present in a wide range of bacteria, and in vitro they have been shown to reduce not only nitrite but also NO⅐ and hydroxylamine to ammonium without releasing detectable intermediates (15)(16)(17)(18)(19). As a consequence, it was proposed that NO⅐ reduction by NrfA, rather than nonspecific reduction by another Nrf component, was responsible for providing E. coli with resistance toward NO⅐ under anaerobic conditions (6).
E. coli NrfA has been reported to reduce ϳ450 NO⅐ s Ϫ1 whereas the enzyme from Desulfovibrio desulfuricans reduces only 30 NO⅐ s Ϫ1 under comparable conditions (16). In addition, it has been noted that D. desulfuricans NrfA exhibits similar rates of NO⅐ and nitrite reduction whereas the enzyme from Sulfurospirilum deleyianum reduces NO⅐ at one hundredth the rate of nitrite reduction (16,17). Given these diverse observations and the proposed role for E. coli NrfA in NO⅐ detoxification, we have set about providing a more complete and quantitative description of the NO⅐ reductase activity of this enzyme. Our previous studies employed protein film voltammetry (PFV) to quantitate the nitrite and hydroxylamine reductase activities of E. coli NrfA (19,(21)(22)(23)(24)(25)(26). In this approach graphite electrodes substituted for NrfB to deliver electrons directly to NrfA that had been adsorbed on the electrode surface. Michaelis-Menten parameters were deduced for each catalytic cycle from the variation of catalytic current (rate) with substrate concentration. In addition, enzyme activities were resolved across the electrochemical potential domain, revealing multiple modulations of the catalytic rate in response to the application of an increased driving force for the reaction being catalyzed. Because these modulations of activity were independent of the nature of the electrode surface, they were proposed to reflect intrinsic properties of NrfA nitrite and hydroxylamine reduction. Thus, to facilitate comparison with the other reductase activities of this enzyme we have chosen to define the NO⅐ reductase activity of E. coli NrfA by PFV.

EXPERIMENTAL PROCEDURES
Protein Purification-E. coli cytochrome c nitrite reductase NrfA was purified and quantitated as described previously (15). The enzyme had a specific activity of 1500 mol nitrite consumed min Ϫ1 mg Ϫ1 measured by enzyme-dependent oxidation of dithionite-reduced methyl viologen (1 mM) in 1 mM nitrite, 2 mM CaCl 2 , 50 mM Hepes, pH 7.0, at 20°C. Samples of the enzyme in 50 mM Hepes, pH 7.0, were stored as aliquots frozen in liquid nitrogen (15).
Reagent Preparation and NO⅐ Quantification-All solutions were prepared with "Trace-SELECT-Ultra" water (Sigma-Aldrich). The buffer-electrolyte was 2 mM CaCl 2 , 50 mM Hepes, pH 7.4, or 25 mM each of Hepes, Mes, Taps, and acetate with 2 mM CaCl 2 for studies from pH 4 to 9. Stock solutions for each substrate were prepared daily. For nitrite (50 mM) the appropriate mass of NaNO 2 was dissolved in ice-cold buffer-electrolyte and the pH confirmed to be that desired. For hydroxylamine (2 M) the appropriate mass of NH 2 OH.HCl was dissolved in icecold buffer-electrolyte and brought to the desired pH with aliquots of 10 M NaOH.
Buffer-electrolyte solutions saturated with NO⅐ were prepared from NO⅐ gas (98.5%; Aldrich) that had been bubbled through an anaerobic aqueous solution of 100 mM NaOH. The NO⅐ concentration in the resultant solution was determined by titration against a defined quantity of horse heart myoglobin. Here ferric myoglobin, ⑀ 410 nm ϭ 186 mM Ϫ1 cm Ϫ1 (27), was reduced with sodium ascorbate and phenazine ethosulfate (final concentrations 1 mM and 5 M, respectively) in a sealed anaerobic cuvette. Addition of aliquots of NO⅐ caused a decrease in the peak intensity at 434 nm with concomitant appearance of a peak at 421 nm that was indicative of formation of the ferrous myoglobin-NO⅐ complex (28). The concentrations of NO⅐ and myoglobin in the cuvette were considered equal when NO⅐ addition failed to perturb the spectrum. Stock solutions prepared and quantitated as described above contained between 1.7 and 2 mM NO⅐.
Because of the volatility of NO⅐ its concentration in the electrochemical cell was measured immediately after each experiment. Calibration against myoglobin proved too time consuming to allow effective experimentation, so a modified version of the acidified Griess reaction was adopted to allow rapid analysis in triplicate (29). Briefly, aerobic cuvettes containing 1.9 ml of 17 mM sulfanilic acid, 0.4 mM N-(1-naphthyl) ethylenediamine dihydrochloride in 2 mM CaCl 2 , 50 mM Hepes, pH 7.4, were sealed and taken into the anaerobic chamber where voltammetry was performed. Each cuvette was injected with 50 l of the solution to be assayed, shaken for 1 min, and removed from the anaerobic chamber where 50 l of phosphoric acid (85%) was added. The cuvettes were left for 15 min and the absorbance measured at 540 nm. NO⅐ concentrations were calculated using ⑀ 540 nm ϭ 42.2 (mM NO⅐) Ϫ1 cm Ϫ1 determined by calibration of the modified Griess reaction with NO⅐ solutions whose concentrations have been determined by titration against ferrous myoglobin.
Protein Film Voltammetry-Experiments employed a threeelectrode cell configuration placed inside a Faraday cage housed in a N 2 -filled chamber with atmospheric O 2 Ͻ 2 ppm as described previously (30). The jacketed sample chamber was incubated at the desired temperature by connection to a thermostatted water bath. The cell employed an Ag/AgCl (saturated KCl) reference electrode, and potentials are reported with respect to the Standard Hydrogen Electrode following addition of 197 mV to the measured potential.
Pyrolytic graphite edge working electrodes of 3-mm diameter were polished immediately prior to use with an aqueous slurry of 0.3 m Al 2 O 3 , sonicated, rinsed, and dried with a tissue. The freshly polished electrode was then taken into the anaerobic chamber together with a frozen aliquot of NrfA (0.17 M). Immediately the NrfA sample had thawed, 3 l was placed on the electrode surface and after ϳ20 s excess solution was removed from the electrode with a tissue. The electrode was then rinsed with buffer-electrolyte and placed into the electrochemical cell. During experiments with NO⅐ a rubber O-ring was placed around the shaft of the working electrode to form a light seal to the electrochemical cell.
To define the variation of catalytic rate with nitrite or hydroxylamine concentration, sequential additions of substrate were made to the electrochemical cell while cyclic voltammetry was performed. Each titration employed a freshly prepared NrfA film with electrode rotation at 3000 rpm where the response suffered no limitation from substrate mass transport (19,25). To define the catalytic current due to NrfA-specific substrate reduction, the current recorded with a "bare" electrode was subtracted from that recorded with the NrfA film in place. Prior to determination of K m and i max the catalytic currents were adjusted for a first order loss of signal magnitude (t1 ⁄2 ϳ 70 min) that occurred during the course of each experiment.
For direct comparison of nitrite and NO⅐ reduction rates a freshly prepared NrfA film was placed in nitrite and two voltammograms were measured with electrode rotation at 3000 rpm. The film was then transferred to a solution of approximately equimolar NO⅐. Two voltammograms were measured with electrode rotation at 100 rpm, and aliquots of the NO⅐ solution were immediately analyzed by the Griess reaction. A rotation rate of 100 rpm gave steady state voltammetry without rapid loss of NO⅐ from solution as occurred at higher rotation rates. Catalytic currents for nitrite reduction were calculated as above. Those for NO⅐ reduction were obtained after subtraction of the response from a bare electrode in NO⅐-containing buffer-electrolyte and adjustment for a reproducible 10 (Ϯ2) % loss of magnitude due to film transfer that was identified in control experiments.
Catalytic current magnitudes were calculated at two potentials. At Ϫ0.3 V NrfA NO⅐ reduction was detected with little or no interference from electrodic NO⅐ reduction. At Ϫ0.4 V the current due to NrfA NO⅐ reduction was subject to greater error due to its superposition on a contribution from electrodic NO⅐ reduction but closer to the catalytic rate that would be achieved when the response becomes independent of further increase of driving force, i.e. at sufficiently negative electrode potentials. For all substrates the values of K m and maximum catalytic current magnitude (i max ) determined from Hanes and Lineweaver-Burk plots were within 15% of those determined by direct fit to the Michaelis-Menten equation. Activation enthalpies were determined from plots of ln i max versus T Ϫ1 by fit to a modification of the Eyring equation as shown in Equation 1 where ⌬S ‡ is the activation entropy and ⌬H ‡ the activation enthalpy, k b the Boltzmann constant, h the Planck constant, R the gas constant, and T the temperature. Fitting was performed with MicroCal Origin.

Detection of NO⅐ Reduction by NrfA
Using PFV-Previous studies have established that electrocatalytically active NrfA films are formed when freshly polished pyrolytic graphite edge electrodes are exposed to ice-cold NrfA solutions (19,(21)(22)(23). Featureless cyclic voltammograms were observed when films of NrfA prepared in this way were placed in 2 mM CaCl 2, 50 mM Hepes, pH 7.4, at 4°C with electrode rotation at 100 rpm (Fig.  1A, triangles). However, the voltammogram changed after the addition of NO⅐-saturated buffer-electrolyte to give 20 M NO⅐ in the electrochemical cell (Fig. 1A, heavy solid line). Reductive (negative) catalytic currents were detected below approximately Ϫ0.1 V. Control experiments performed with bare electrodes placed in 20 M NO⅐ showed evidence of catalytic reduction at potentials below approximately Ϫ0.4 V (Fig. 1A, circles). Thus, the sigmoidal catalytic response initiated just below Ϫ0.1 V can be attributed to NrfA-specific NO⅐ reduction.
Confirmation of the voltammetric detection of NrfA NO⅐ reduction was provided by experiments in the presence of cyanide. Cyanide is a potent inhibitor of NrfA nitrite and hydrox-ylamine reductase activities that is expected to inhibit NrfAspecific NO⅐ reduction (Fig. 1B) (21,22). Indeed, the sigmoidal catalytic feature just below Ϫ0.1 V arising from a NrfA film in 20 M NO⅐ disappeared on addition of cyanide whereas catalysis below Ϫ0.4 V, due predominantly to electrodic NO reduction, persisted with only a small drop in magnitude (Fig. 1A, light solid line).
A comparison of the PFV from NrfA in 20 M NO⅐ with that in 20 M nitrite showed reduction of both substrates occurred over a similar potential range (Fig. 1). NO⅐ reduction was repeatedly described by much smaller catalytic currents than those due to nitrite reduction. The distinct rotation rate dependences of the catalytic responses for nitrite and NO⅐ reduction confirmed that trace levels of nitrite were not responsible for the catalytic response detected in the NO⅐ experiments (Fig. 2). Voltammetry performed at 100 rpm in nitrite solutions showed a peak of current on scans to more negative potentials that was absent on the return to more positive potentials (Fig. 2). This form of response is indicative of substrate, i.e. nitrite, depletion in the vicinity of the adsorbed enzyme, and for this reason quantitative studies of nitrite reduction are performed at 3000 rpm where the response of the forward and reverse scans overlay (e.g. Fig. 1B) (25). By contrast, the voltammetry in NO⅐ solutions was essentially independent of electrode rotation rate above 90 rpm with no indication of substrate depletion at the electrode surface (Fig.  2). Below Ϫ0.2 V the catalytic current had slightly different magnitudes for the scans to positive and negative potentials but control experiments showed this arose from electrodic NO⅐ reduction. Importantly, the response just below Ϫ0.1 V that most clearly defined NrfA NO⅐ reduction was in good agreement for both scan directions and so indicative of a steady-state catalytic response. Having ruled out the possibility that detectable levels of nitrite were present in the NO⅐ experiments, it was concluded that NrfA is a NO⅐ reductase with distinct kinetic parameters for the reduction of NO⅐ and nitrite.
Quantitation of Steady-state NrfA Reduction of NO⅐, Nitrite, and Hydroxylamine-Initial experiments to define the relationship between catalytic rate and NO⅐ concentration involved titration of NO⅐ into the electrochemical cell. Cyclic voltammetry showed that the catalytic rate increased as the NO⅐ concentration increased (Fig. 3A). However, colorimetric quantitation of NO⅐ at the end of the experiment showed its concentration was only 86% of that expected. Similar behavior was observed in a num-ber of titrations, but the final NO⅐ level ranged from 40 to 86% of that expected. Because this prevented reliable assessment of the NO⅐ concentration at intermediate points in the titrations, an alternative experimental approach was adopted.
After NO⅐ addition to the cell a reproducible (3%) drop in concentration was found to have occurred after measuring two cyclic voltammograms. Thus, film transfer experiments were conducted with this restriction to assess the relative rates of nitrite and NO⅐ reduction at a number of substrate concentrations. The relationship between nitrite concentration and reduction rate was defined under comparable conditions by a nitrite titration. The two sets of results were combined and the  resultant plots fit to the Michaelis-Menten equation (Fig. 4 Here we note that two steps were taken to facilitate quantitation of the NrfA NO⅐ response as distinct from electrodic NO⅐ reduction. The first step was to perform experiments at 4°C rather than 20°C as used for previous PFV of NrfA because lower temperatures were found to improve the ratio of enzymatic to electrodic currents (not shown). The second step was to subtract the response of a freshly polished bare electrode recorded in a solution of given NO⅐ concentration from that of the NrfA-coated electrode prior to analysis (Fig. 3B). Quantitative experiments were still restricted to NO⅐ concentrations below 350 M because the decreased ratio of enzymic to electrodic current detected at higher NO⅐ concentrations introduced unacceptable uncertainty into quantitation of the NrfAspecific currents.
The Michaelis constants (K m ) and maximum catalytic currents (i max ) describing NO⅐ and nitrite reduction are compared in Table 1 where, for completeness, values for hydroxylamine reduction under comparable conditions are included. The K m values increased in the order nitrite ϽNO⅐ Ͻ Ͻhydroxylamine and differed by at least an order of magnitude. The values obtained here for nitrite and hydroxylamine reduction are comparable with those of ϳ25 and 127,000 M, respectively, at 20°C, pH 7.0, (19). We considered that the fits to the NO⅐ data might be less reliable because the experimentally attained rates approached only 50% of i max whereas those for nitrite and hydroxylamine reduction approached 90% of their respective maxima. Analysis of the nitrite reduction data for rates less than 50% of i max yielded K m values twice as large and i max values within 30% of those from the full data set. However, the kinetic parameters for NrfA substrate reduction are such that this possible error does not alter the major conclusions drawn from Table 1, namely that at 4°C and pH 7 all three substrates are reduced with similar maximal velocities but with significantly different K m values. The higher K m for NO⅐ than nitrite reduction was confirmed experimentally when it was found that the mass transport limited response from reduction of 2 M nitrite observed at 100 rpm was only lost on addition of ϳ10-fold more NO⅐ to the solution.
Catalytic waves for NrfA reduction of NO⅐, nitrite, and hydroxylamine are compared in Fig. 5. To facilitate their comparison these waves are presented at substrate concentrations corresponding to ϳ100, 75, 25, and 5% of their respective K m values at Ϫ0.4 V because this encompassed the range of NO⅐ concentrations from which data could be confidently extracted. At these concentrations the activity of NrfA toward all three substrates is initiated just below Ϫ0.1 V but the wave shapes are clearly substrate-dependent. Hydroxylamine reduction gives rise to essentially sigmoidal responses whereas the NO⅐ reduction waves show a peak of activity. By contrast, the nitrite reduction waves changed from peak-shaped to reflect a low potential activity "boost" as the nitrite concentration increased.
pH and Temperature Dependence of NrfA NO⅐ Reduction Rates-To assess the pH dependence of NrfA NO⅐ reductase activity, films were transferred between solutions of distinct pH, each containing 50 M NO⅐ (Fig. 6). Exposure to more acidic pH results in a sigmoidal increase of activity that is accompanied by displacement of the catalytic wave toward more positive potentials. These changes were independent of the order of exposure to pH and so fully reversible.
Film transfer experiments with each substrate at a concentration equal to its K m at 4°C demonstrated that NrfA activity toward all three substrates increased when the temperature was raised from 4 to 25°C. The rates increased linearly with tem-  Table 1. Experimental conditions were as in Fig. 1. perature, and the increase was significantly greater for reduction of hydroxylamine than for nitrite and NO⅐ (Table 2). With the assumption that the K m for NO⅐ reduction, like those for nitrite and hydroxylamine reduction, is independent of temperature across this range, the variations in rates reflect the variation of i max with temperature. Fits of the data to the Eyring equation provided activation enthalpies (⌬H ‡ ) of ϳ25 kJ mol Ϫ1 for nitrite and NO⅐ reduction and ϳ65 kJ mol Ϫ1 for hydroxylamine reduction.

DISCUSSION
The work presented here has shown that E. coli NrfA is able to serve as a direct NO⅐ reductase. This demonstration can explain the increased sensitivity of anaerobically grown E. coli nrf mutants to NO⅐ when compared with the parent strain in addition to the strongly attenuated ability of the mutants to remove NO⅐ from anoxic cultures (6,13). Similar maximum catalytic rates are predicted for NO⅐ and nitrite reduction. Because we do not know the amount of electrocatalytically active enzyme adsorbed on the electrode surface, we cannot directly convert maximum catalytic current to a turnover number (k cat ). However, in solution assays at room temperature with reduced methyl viologen as the electron donor, NrfA nitrite reduction occurs with a k cat of ϳ700 NO 2 Ϫ s Ϫ1 that corresponds to an electron flux of ϳ4200 s Ϫ1 (13,15). If the enzyme, as the PFV suggests, can reach similar electron flux during NO⅐ reduction this corresponds to a k cat of ϳ840 s Ϫ1 . This is a value much higher than those so far reported for flavohemoglobin and flavorubredoxin from E. coli and the dedicated respiratory NO⅐ reductases of P. denitrificans and Ps. stutzeri (4,5,(7)(8)(9)(10)(11)(12).
During colonization of a human host, E. coli will experience predominantly micro-oxic and anoxic environments along with concentrations of nitrate and nitrite that should result in production of NrfA and flavorubredoxin, both of which may play a role in NO⅐ detoxification (11,31,32). At pH 7, the K m of ϳ300 M for NO⅐ removal by NrfA is clearly much larger than that of 0.4 M displayed by flavorubredoxin (11,12). However, the much higher turnover number predicted for NrfA results in a specificity constant (k cat /K m ) of ϳ2 ϫ 10 6 M Ϫ1 s Ϫ1 that lies within an order of magnitude of that of ϳ40 ϫ 10 6 M Ϫ1 s Ϫ1 for the cytoplasmically located flavorubredoxin. We estimate that NrfA will reduce 27 NO⅐ s Ϫ1 at the 10 M NO⅐ levels produced extracellularly by activated macrophages (33). Thus, in considering the manner by which E. coli detoxifies exogenous NO⅐ encountered during invasion of a human host it may be that NrfA maintains low NO⅐ levels in the periplasm so that any NO⅐ that does reach the cytoplasm is efficiently removed by flavorubredoxin.
The ability of NrfA to reduce NO⅐ in addition to nitrite offers E. coli the opportunity for respiratory flexibility and NO detoxification in a single enzyme. The 10-fold greater K m for NrfA reduction of NO⅐ than nitrite implies that the NO⅐ reductase activity will come into its own when NO⅐ concentrations are elevated significantly above those of nitrite. Given that activated macrophages can produce extracellular NO⅐ concentrations on the order of 10 M, such elevated NO⅐ levels may be encountered when E. coli is phagocytosed by resident macrophages (33). Conditions that could favor NrfA NO⅐ reduction may also arise in the rapidly changing environments of the gastrointestinal tract and are perhaps most likely in the acidic conditions of the stomach where nitrite disproportionates to a range of species, including NO⅐ (3). As a periplasmic protein NrfA will be subject to the wide range of pH exploited by the gastrointestinal tract (34), and our demonstration that NrfA NO⅐ reductase activity persists over a pH range of at least six units from pH 3 to 9 is consistent with retention of NrfA activity on passage through the gastrointestinal tract.
There is little information available on the levels of hydroxylamine that may be encountered by E. coli and that could allow the hydroxylamine reductase activity of NrfA to placed in a physiological context (35). However, comparison of NrfA activity toward hydroxylamine in addition to NO⅐ and nitrite is valuable when considering the biochemical implications of our results. Each NrfA subunit contains five c-type hemes. Four of these hemes have bis-histidine ligation and the fifth has lysine and water (hydroxide) as axial ligands (15). Crystal structures of Wolinella succinogenes NrfA complexed with nitrite and hydroxylamine show that these substrates displace water (hydroxide) to become axial ligands to the lysine-coordinated heme and in so doing they define the site for binding and reduction of these substrates (20). Here we make several observations that point toward the site of NO⅐ reduction overlapping with that for nitrite and hydroxylamine reduction. Specifically, nitrite competes with NO⅐ for reduction by NrfA, reduction of all three NrfA substrates is strongly inhibited by cyanide (21,22), and reduction of all three substrates is initiated at similar potentials that we have proposed to reflect in part the reduction potential of the lysine-coordinated heme (19). In addition, the nitrite and NO⅐ reductase activities of NrfA show similar variation in rate and operating potentials with pH, suggesting that the same ionizable group(s) may modulate both activities. Thus, we propose that the lysine-coordinated heme forms the site of NO⅐ binding and reduction in NrfA.
The detailed mechanism(s) of NrfA substrate reduction have yet to be conclusively demonstrated. Nitrite, NO⅐, and hydroxylamine are each reduced stoichiometrically to ammonium such that no intermediates are detectable (15)(16)(17)(18)(19). Because all three substrates form the same product it has been suggested that NO⅐ and hydroxylamine are intermediates during nitrite reduction. A computational study and comparison to the chemistry of model complexes have supported this proposal, and in this context some discussion of the information from this study is warranted (20). The relative order of the K m values, with that for nitrite Ͻ NO⅐ Ͻ Ͻ hydroxylamine, is not immediately reconciled with the failure to detect reaction intermediates. The active site is buried deep in the interior of the enzyme with two channels providing access to the exterior (15). One channel is positively charged whereas the other is negatively charged, making it an attractive proposition that the first channel allows nitrite to access the active site while the second is for ammonium egress. We cannot know through which channel NO⅐, or indeed hydroxylamine, enter the active site, but in either case the channel is unlikely to have a high affinity for substrate and this may then be reflected in the measured K m .
NrfA also reduces these three substrates at distinct rates. The maximum rate of hydroxylamine reduction is greater than those for NO⅐ and nitrite reduction at physiological temperatures with the rate-limiting step lying between NO⅐ and hydroxylamine. The variation of predicted maximal catalytic rates with temperature yields a higher activation enthalpy for hydroxylamine reduction than those for reduction of NO⅐ and nitrite. Higher catalytic rates must reflect a smaller activation free energy and so the activation entropy for hydroxylamine reduction must be larger than those for NO⅐ and nitrite. Differences in NrfA processing of the substrates are also manifest in the catalytic wave shapes observed at lower substrate concentrations. The peaked responses, more prominent for nitrite reduction at higher temperatures than the 4°C used in this study, were previously proposed to arise from activation of the enzyme on reduction of active site heme followed by attenuation of activity on reaching potentials low enough to reduce one or more of the bis-histidine-coordinated hemes lying near the dimer interface (19). The mechanism by which heme reduction attenuates activity is not clear. Because the phenomenon is not apparent during hydroxylamine reduction, one possibility was that it reflected a change in coordination capabilities within the active site that had a greater effect for nitrite, with two N-O bonds, than hydroxylamine. However, it now appears that it is unlikely because NO⅐ reduction gives rise to peaked wave shapes. Alternative mechanisms for modulation of catalytic rate include redox-linked modulations of intra-molecular electron and proton transfer.