J Biol Chem, Vol. 275, Issue 17, 12581-12589, April 28, 2000

Steady-state and Transient Kinetics of Escherichia coli Nitric-oxide Dioxygenase (Flavohemoglobin)
THE B10 TYROSINE HYDROXYL IS ESSENTIAL FOR DIOXYGEN BINDING AND CATALYSIS^*

Anne M. Gardner, Lori A. Martin, and Paul R. Gardner

From the Division of Critical Care Medicine, Children's Hospital Medical Center, Cincinnati, Ohio 45229

Yi Dou, and John S. Olson

From the Department of Biochemistry and Cell Biology and the W. M. Keck Center for Computational Biology, Rice University, Houston, Texas 77005

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Escherichia coli expresses an inducible flavohemoglobin possessing robust NO dioxygenase activity. At 37 °C, the enzyme shows a maximal turnover number (V_max) of 670 s¹ and K_m values for NADH, NO, and O₂ equal to 4.8, 0.28, and ~100 µM, respectively. Individual reduction, ligand binding, and NO dioxygenation reactions were examined at 20 °C, where V_max is ~94 s¹. Reduction by NADH occurs in two steps. NADH reduces bound FAD with a rate constant of ~15 µM¹ s¹, and heme iron is reduced by FADH₂ with a rate constant of 150 s¹. Dioxygen binds tightly to reduced flavohemoglobin, with association and dissociation rate constants equal to 38 µM¹ s¹ and 0.44 s¹, respectively, and the oxygenated flavohemoglobin dioxygenates NO to form nitrate. NO also binds reversibly to reduced flavohemoglobin in competition with O₂, dissociates slowly, and inhibits NO dioxygenase activity at [NO]/[O₂] ratios of 1:100. Replacement of the heme pocket B10 tyrosine with phenylalanine increases the O₂ dissociation rate constant ~80-fold and reduces NO dioxygenase activity ~30-fold, demonstrating the importance of the tyrosine hydroxyl for O₂ affinity and NO scavenging activity. At 37 °C, V_max/K_m(NO) is 2,400 µM¹ s¹, demonstrating that the enzyme is extremely efficient at converting toxic NO into nitrate under physiological conditions.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The flavoHbs¹ belong to a 1.8 billion-year-old family of globin molecules that includes O₂-binding Hbs and Mbs isolated from animals, plants, fungi, protozoa, bacteria, and worms (1-6). FlavoHbs have a unique two-domain structure containing linked Hb and reductase domains with extensive homology to the mammalian Hbs and metHb reductases (1, 7). Other Hbs appear to be co-expressed with associated metHb reductases (8). An O₂ transport or storage function, like that of the erythrocytic Hbs and muscle Mbs, has been suggested for some microbial and plant (flavo)Hbs (4, 9); however, other functions including the catalysis of oxidations have long seemed more likely (10-12).

Recently, we described an NO dioxygenase (NOD) produced by Escherichia coli that utilizes O₂ and NAD(P)H to convert NO to nitrate (Equation 1) and identified it as a flavoHb (13, 14). Subsequent studies have shown that related bacterial and yeast flavoHbs display a similar NOD activity.²

(Eq. 1)

A role for flavoHbs in NO detoxification is supported by the ability of flavoHbs to protect bacteria against NO or nitroso compounds (13, 14, 16-18) and by their induction in bacteria exposed to NO, nitrate, nitrite, or nitroso compounds (13, 14, 17, 19-22). However, the mechanism of NO detoxification, and thus the function of the flavoHbs, is obscured by the possibility of multiple reaction mechanisms involving NO. Other NO detoxifying activities for flavoHbs, including denitrosylation of nitrosothiols (17), NO reduction (17, 23), and NO sequestration (16, 23), have been offered to explain the protection flavoHbs afford to bacteria against NO and nitrosoglutathione. Thus, an understanding of the biological function(s) of the flavoHbs demands a greater knowledge of their various activities.

In this report, steady-state, reduction, and ligand binding kinetics of the E. coli NOD (flavoHb) were measured in order to define its function and the mechanism of NO dioxygenation. We also examined the effects of amino acid substitutions at the highly conserved Tyr(B10) position on NOD activity, reduction, and ligand binding kinetics (7, 24). Key differences between flavoHb and other Hbs are discussed in light of this specialized but perhaps ancient NO dioxygenation and detoxification function of hemoglobin.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cells and Reagents-- The flavoHb-deficient E. coli strain RB9060 (25) was generously provided by Alex Ninfa (University of Michigan). Plasmid pAlter containing the E. coli hmp gene was prepared as described previously (13). FAD, NADPH, and bovine hemin were purchased from Sigma. NADH, bovine liver catalase (260,000 units/ml), and deoxyribonuclease were obtained from Roche Molecular Biochemicals. Saturated NO was prepared as described previously (26). Saturated O₂ (1.14 mM) was prepared by vigorously scrubbing a solution of 50 mM potassium phosphate, pH 7.8, containing 0.1 mM EDTA (buffer A) at 25 °C and atmospheric pressure with 99.993% O₂ (Praxair, Bethlehem, PA) in a rubber septum-sealed glass vial vented with a syringe needle. Manganese-containing superoxide dismutase (2700 units/mg) (27) was purified to near homogeneity from E. coli strain DH5 containing the multicopy plasmid pD11c bearing the sodA gene (28).

FlavoHb Mutagenesis-- The E. coli hmp gene was mutagenized in the plasmid pAlter using the Altered Sites II mutagenesis system (Promega, Madison, WI) and the synthetic oligonucleotides 5'-CGCCCATTTCTTCGACCGTATGTT-3', 5'-GTTAACCGCCCATTTCCATGACCGTATGTTTACTC-3', 5'-GTTAACCGCCCATTTCGAGGACCGTATGTTTACTC-3' for Tyr²⁹(B10) replacements with Phe, His, and Glu, respectively. Site-directed mutations were identified by targeted DNA restriction analysis and were verified by automated DNA sequencing.

Purification of FlavoHbs-- FlavoHbs were expressed from the native hmp promoter in the multicopy plasmid pAlter plus hmp in strain RB9060 grown under nitrate-inducing conditions (14). One-liter cultures were grown anoxically in medium containing 5 g of yeast extract and 10 g of tryptone per liter of 100 mM potassium phosphate, pH 7.0, 10 mM NaNO₃, 20 mM glucose with 2 µM hemin and either tetracycline (12 µg/ml) or ampicillin (50 µg/ml). Cells were collected by centrifugation and were washed with chilled buffer A containing 5 mM NaN₃. Azide was included to inhibit the loss of heme. Cells were lysed by sonication in 2 volumes of buffer A with ~1 mg of deoxyribonuclease, and lysates were clarified by centrifugation. NOD was fractionated from extracts containing 12 mg/ml protein at 35-55% ammonium sulfate saturation at 4 °C. The precipitant was resuspended in 50 mM Tris-Cl, pH 8.0, 1 mM EDTA plus 5 mM NaN₃ (buffer B) and was dialyzed extensively at 4 °C against ~100 volumes of the same buffer. The dialysate was then separated on a 2.5 × 18-cm DEAE-Sepharose (Amersham Pharmacia Biotech) column in buffer B using a linear 50-400 mM NaCl gradient. Fractions having the highest A₄₁₀/A₂₈₀ ratios were pooled, concentrated, and separated on a 1 × 55-cm Superdex 75 column (Amersham Pharmacia Biotech) in buffer B containing 50 mM NaCl. For smaller preparations from 2-liter cultures, flavoHbs were further separated on 1-ml Hi-Trap Mono Q columns (Amersham Pharmacia Biotech) with a 50-200 mM NaCl gradient in buffer B. For larger preparations from 12-liter cultures, pooled Superdex 75 fractions were separated again on the DEAE-Sepharose as described above. FlavoHb fractions were concentrated and stored at 70 °C. FlavoHb yields were typically 2-5 mg/liter of culture. Sample purity was estimated from SDS-polyacrylamide gel electrophoresis. For the wild-type and Phe(B10) substitution, purity was judged to be 95%, and for the B10 Glu and His mutants, purity was estimated at 80%. Where indicated, reconstitution with heme was achieved by incubating flavoHb with stoichiometric excesses of heme, dithiothreitol, and dithionite, and reactants were separated from flavoHb on a Sephadex G25 column in N₂-purged buffer containing 50 mM Tris-Cl, pH 8.0, and 1 mM EDTA.

Assay of NOD and Other FlavoHb Activities-- NOD was measured amperometrically (13, 14) using an ISO-NO electrode (World Precision Instruments, Sarasota, FL) in a thermostatted 2-ml reaction mix containing buffer A, 1 µM FAD, and the indicated concentrations of NAD(P)H and NO. FlavoHbs were routinely diluted in buffer A containing 1 mM dithiothreitol and kept on ice. Dithiothreitol was included to stabilize flavoHb and had no effect on NO decomposition rates at the carry-over concentrations of <5 µM. Rates of whole cell NO consumption were measured with 1 µM NO at 37 °C unless specified otherwise (13, 14). O₂ concentration was varied by first scrubbing the reaction mixture with N₂ and then adding O₂ from a saturated O₂ solution. NO consumption rates were corrected for background rates of NO decomposition for each assay condition.

NADH oxidase activity was followed at 340 nm in a 0.5-ml reaction containing 100 µM NADH, and 1 µM FAD in buffer A at 37 °C. NO reductase activity was measured amperometrically in an anaerobic 2-ml reaction at 37 °C containing buffer A, 1 µM FAD, 10 mM glucose, 16 units of glucose oxidase, 260 units of catalase, 100 µM NADH, and 5 µM NO. NO and flavoHb were added following O₂ depletion, and NO decomposition rates were determined with 1 µM NO. FAD reductase activity was assayed at 450 nm in a 1-ml reaction as described for the measurement of NO reductase activity except that 20 µM FAD was included and NO was omitted.

Cofactor and Protein Assays-- Heme was measured using the alkaline pyridine method (29). FAD was assayed using fluorescence at 520 nm with excitation at 460 nm following a 3-min incubation of flavoHb at 100 °C (30). E = 6,220 M¹ cm¹ at 340 nm was used for the measurement of NAD(P)H. Protein was assayed using the dye-binding assay (31) with a bovine serum albumin standard.

FlavoHb Reduction Kinetic Measurements-- All reactions were carried out in 0.1 M sodium phosphate, 0.3 mM EDTA at pH 7.0, and 20 °C. FlavoHbs were exposed to air on ice to allow complete oxidation as monitored at 404 nm. Vials containing flavoHbs were sealed and flushed with N₂ for 10-20 min with gentle agitation to remove O₂. An aliquot was removed with a syringe and diluted into anaerobic buffer. The resultant sample (3-4 µM heme) was injected in a Gibson-Dionex stainless steel stopped-flow apparatus and rapidly mixed with anaerobic buffer containing various amounts of NADH. The system was presoaked with concentrated deoxyhemoglobin to remove any residual O₂. FAD reduction was followed at 460 nm, and heme iron reduction was monitored at 430 nm.

FlavoHb Ligand Binding Kinetic Measurements-- O₂, CO, and NO association rate constants (k'_O2, k'_CO, and k'_NO) were determined by measuring the time courses for ligand rebinding after laser photolysis of the corresponding complexes using a 300-ns excitation pulse at 577 nm (32). Solutions of the ferrous complexes contained ~30-50 µM flavoHb and 300-600 µM NADH. Rebinding to flavoHb-Fe²⁺ was followed at either 430 nm (O₂) or 436 nm (NO and CO). NO rebinding to flavoHb-Fe³⁺ was monitored at 419 nm in the absence of any reducing agent. To prepare flavoHb-Fe²⁺-O₂ complexes, NADH was added to buffer equilibrated with various N₂/O₂ gas mixtures in a sealed 1-mm cuvette. FlavoHbs were injected into the cuvette, shaken, and immediately photolyzed. Time courses were recorded within 2 min of flavoHb injection, since the NADH oxidase activity of flavoHb rapidly depletes O₂ and NADH at such high µM flavoHb concentrations, and soon either flavoHb-Fe²⁺ or flavoHb-Fe³⁺ is generated, depending upon which substrate is present in excess. Rate constants for O₂ dissociation (k_O2) were determined by stopped-flow rapid mixing techniques using the CO displacement method (32). Solutions of flavoHb-Fe²⁺-O₂ were mixed with CO-saturated buffer, and the displacement reaction was followed at 422 nm. The observed displacement rate (r_obs) is equal to k_O2/(1 + k'_O2[O₂]/k'_CO[CO]). k_O2 was calculated from this expression using k'_O2 and k'_CO determined as described above. CO dissociation rate constants (k_CO) were determined by displacing bound CO with high concentrations of NO. Reduced flavoHb-Fe²⁺ incubated with 100 µM CO was mixed with buffer equilibrated with 1 atmosphere of NO (2 mM). The rate constants for NO dissociation (k_NO) from flavoHb-Fe²⁺ were measured by injecting flavoHb-Fe²⁺-NO complex into a solution containing 10 mM dithionite and 1 mM CO, and the time course of flavoHb-Fe²⁺-CO formation was monitored at 422 nm (33). Recombinant sperm whale Mb was used as the standard and control for all measurements.

Model for NOD Kinetics and Mechanism-- The following reactions and rate constants describe the transient and steady-state kinetics for NO dioxygenation by flavoHb.

(Eq. 2)

The enzyme species, E(n), are defined by n, the number of electrons added, and a prime to indicate when the heme iron atom is reduced and capable of binding O₂. Thus, the E(2) species represents flavoHb with FADH₂/Fe³⁺, E'(2) is FADH·/Fe²⁺, E(1) is FADH·/Fe³⁺, and E'(1) is FAD/Fe²⁺. P refers to the product nitrate. A more detailed and complex reaction scheme is shown in Fig. 10.

Initial hydride transfer from NADH to FAD is assumed to be a simple bimolecular process with an apparent rate equal to k'_H[NADH]. As described under "Results" (Fig. 8A), the initial two-electron reduction of FAD probably involves rapid NADH binding followed by hydride transfer to form FADH₂. However, most of the initial velocity measurements were carried out at high [NADH], where the reduction of FAD becomes effectively a simple one-step bimolecular process defined by k'_H.

Electron transfer from the flavin to the iron atom is modeled as a simple, irreversible process with a rate equal to k_ET. O₂ binding is assumed to be reversible with association and dissociation rate constants k'_O2 and k_O2, respectively. The oxidative reaction of NO with flavoHbO₂ is modeled as a bimolecular step with a rate equal to k'_ox[NO], based on the previous work of Eich et al. (34), and then nitrate release occurs by a first order process governed by k_p.

The NO dioxygenation process occurs again for the one-electron reduced enzyme. For the sake of simplicity, the rate constants for the second set of O₂ binding and NO oxidation steps are assumed to be identical to those for the E(2) to E(1) oxidation. As described under "Discussion," the rate of O₂ binding depends on the fraction of reduced iron present in the intermediate, which in turn almost certainly depends on the number of electrons present in the enzyme. Thus, k'_O2 may be different for E(2) versus E(1), and further experiments are needed to determine these differences.

The rate of product formation or NO consumption is given by the following,

(Eq. 3)

and can be analyzed at fixed [NO] and [O₂] in terms of a hyperbolic dependence on NADH concentration. Expressions for V_max(obs, NADH) and K_m(obs, NADH) can be derived by applying steady-state assumptions for the rates of change of the various enzyme intermediates. The resultant equations are as follows,

(Eq. 4)

(Eq. 5)

where V_max(obs, NADH) is expressed as a turnover number in units of NO heme¹ s¹ or simply s¹. These expressions and the rate constants measured in transient kinetic experiments further serve to guide our understanding of the NOD steady-state kinetics. For example, at saturating [NO] and [O₂], K_m(obs, NADH) = V_max/2k'_H, where V_max = k_pk_ET/(k_p + k_ET), and estimates of all three rate constants, k'_H, k_ET, and k_P can be obtained in rapid mixing experiments.

When initial velocity measurements are made at saturating [NADH], the steady-state expression reduces to V_max(obs, NADH) or as follows.

(Eq. 6)

At saturating [NADH] and [NO], K_m(obs, O₂) should approximate V_max/k'_O2. In practice it is difficult to achieve this situation due to the inhibition observed at NO/O₂ ratios of 1:100, since competition between O₂ and NO for intermediates E'(2) and E'(1) inhibits NOD activity.

At saturating levels of both O₂ and NADH, the initial velocity equation simplifies to the following.

(Eq. 7)

This situation can be achieved experimentally without inhibition, since the ratio [NO]/[O₂] can be kept at 1/100. Under these conditions, K_m(obs, NO) equals V_max/k'_ox, and the rate constant for the bimolecular reaction of NO with bound O₂ can be calculated from the steady-state data.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

FAD Is Required for the NOD Activity of FlavoHb-- The time course for flavoHb-catalyzed NO decomposition was measured with or without FAD added as a cofactor, using purified flavoHb (~0.1 nM) containing 0.28 mol fractions of FAD. With 10 µM NADH as reducing substrate, a linear initial rate is observed with 1 µM FAD (Fig. 1A, line 1), but in its absence, a slower initial rate is measured that rapidly declines to zero after 40 s (line 2). A 2-min preincubation of flavoHb (~0.1 nM) at 37 °C with 10 µM NADH eliminates the initial rate in the absence, but not in the presence, of 1 µM FAD (data not shown). The FAD dependence of NOD activity following the initial ~40-s interval of decline (Fig. 1B) yields an apparent K_d for FAD of 40 nM (inset). With 200 µM NADPH as reducing substrate, an initial rate is not detectable without FAD (13, 14), and a similar K_d for FAD is estimated.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   FAD requirement of NOD activity. A, NO decomposition by purified flavoHb (8.8 ng; 0.20 pmol) was measured at intervals with (line 1) or without (line 2) 1 µM FAD. B, the effect of [FAD] on NOD activity was measured after >45 s of reaction time at 1 µM NO. Double reciprocal plots of the FAD dependence of the rate (inset) were used to determine an apparent Kd for FAD. Reactions were initiated with 2 µM NO in a 2-ml reaction containing 200 µM O2 and 10 µM NADH at 37 °C as described under "Materials and Methods." FlavoHb contained 0.28- and 0.42 mol fractions of FAD and heme, respectively.

The data indicate a dependence of NOD activity on added FAD. Together, these results and the substoichiometric FAD content of various E. coli flavoHb preparations (13, 30) warranted the inclusion of FAD in flavoHb activity assays.

Dependence of NOD Activity on NAD(P)H-- Double reciprocal plots of initial velocity versus [NADH] at various O₂ concentrations show roughly parallel lines (Fig. 2A) as predicted by the reaction mechanism described by Equation 3. At saturating levels of both NO and O₂, the K_m values for NADH are 4.8 and 3.2 µM at 37 and 20 °C, respectively. The V_max values are 500 and 83 NO heme¹ s¹ at 37 and 20 °C, respectively (Fig. 2B). In comparison, the K_m for NADPH at 20 °C is ~60-fold greater, 180 µM, but the V_max value is the same as that for NADH (Fig. 2C).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   NAD(P)H dependence of NOD activity. The NOD activity of purified flavoHb was measured with varied concentrations of NADH (A and B) or NADPH (C). A, NOD activity was assayed at 37 °C with 0.1 µM NO, 670 µM O2 (line 1), 100 µM O2 (line 2), 25 µM O2 (line 3), 12.5 µM O2 (line 4). B, NOD activity was assayed with 1.8 µM NO and 670 µM O2 at 37 °C (line 1) or 260 µM O2 at 20 °C (line 2). C, NOD activity was measured at 20 °C with 1.8 µM NO and 260 µM O2.

Role of O₂ in NOD Activity-- The contribution of free O₂ to NO decomposition was examined, since NO reacts rapidly with free O₂ (k₂ ~ 5,000 µM¹ s¹ (35, 36)), and this reaction may interfere with measurements of NOD activity. When added in competitive excess relative to NO, superoxide dismutase (1 mg/ml) does not significantly affect the rate of NO consumption measured with 100 µM NADH and 1 µM NO. With 200 µM NADPH, a small (20 ± 8%; n = 3 ± S.D.) inhibition by superoxide dismutase is observed. The results indicate a minor or insignificant role for O₂ in NO decomposition by purified flavoHb and agree with previous results obtained under less optimal conditions (13, 17).

Dependence of NOD Activity on O₂-- The apparent K_m for O₂ is ~100 µM when NOD is assayed with saturating NO (1 µM) and saturating levels of NADH or NADPH at 37 °C (Fig. 3A, lines 1 and 2). When the concentration of NO was lowered to 0.1 µM, both V_max and K_m(O₂) values decreased by roughly the same amount so that parallel lines are observed in the double reciprocal plot (lines 1 and 3). Under these more physiological conditions, the K_m for O₂ is ~35 µM at 37 °C. Similar behavior was observed at 20 °C, and the K_m for O₂ is ~27 µM at high [NO] and ~13 µM at low [NO] (Fig. 3, lines 4 and 5). In E. coli incubated at 37 °C, half-maximal NOD activity is observed at ~60 µM O₂ (Fig. 3B), a value within range of the K_m(O₂) values determined for the isolated flavoHb.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   O2 dependence of NOD activity. A, NOD activity of purified flavoHb was measured at 37 °C with varying [O2], 1 µM NO, and 100 µM NADH (line 1, circles) or 600 µM NADPH (line 2, triangles). Activity was measured as described for line 1 except that the temperature was 20 °C (line 3). NOD activity was assayed with 0.1 µM NO and 100 µM NADH at 37 °C (line 4) or at 20 °C (line 5). B, NOD activity of flavoHb was measured at 37 °C with 1 µM NO and 200 µM O2 in E. coli strain RB9060 bearing pAlter plus hmp and overproducing flavoHb.

In the absence of O₂, the NO reductase activity (23) of the flavoHb is 0.14 NO heme¹ s¹. This NO reductase activity corresponds to ~0.02% of the NOD activity measured with saturating O₂.

Dependence of NOD Activity on NO-- Double reciprocal plots of NOD activity versus [NO] show normal hyperbolic kinetics at 200 µM O₂ (Fig. 4A, line 1). At 50 and 25 µM O₂, inhibition of NOD activity occurs at high [NO] when the NO/O₂ ratio exceeds ~1:100 (compare lines 1-3). Under noninhibiting conditions, roughly parallel lines are observed for 1/v_i versus 1/[NO] at varying concentrations of O₂. The dependence of the observed V_max values on [O₂] gives a K_m for O₂ equal to ~80 µM in close agreement with the results in Fig. 3. With saturating O₂ (670 µM) and NADH at 37 °C, the K_m value for NO is 0.28 µM (Fig. 4B, line 1). At 20 °C with saturating O₂ (260 µM) and NADH, the K_m value for NO is 0.11 µM (Fig. 4B, line 2). The corresponding NOD turnover numbers (V_max) at 37 and 20 °C are 670 and 94 NO heme¹ s¹, respectively. In E. coli incubated at 37 °C under normoxic conditions (200 µM O₂), the apparent K_m for NO is 0.40 µM (Fig. 4C).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   NO dependence of NOD activity. A, NOD activity of purified flavoHb was measured at different NO concentrations with 100 µM NADH and 200 µM O2 (line 1), 50 µM O2 (line 2), or 25 µM O2 (line 3). B, NOD activity of purified flavoHb was measured at 37 °C with varying concentrations of NO, 100 µM NADH, and a saturating 670 µM O2 (line 1) or at 20 °C with a saturating 260 µM O2 (line 2). C, NOD activity of flavoHb expressed in cells was measured with varying concentrations of NO and 200 µM O2 as described under "Materials and Methods." FlavoHb contained a 0.42 mole fraction of heme. Strain RB9060 bearing pAlter plus hmp and overexpressing flavoHb was used for measurements of whole cell NO consumption.

Effects of Tyrosine B10 Substitutions on the NOD Activity of FlavoHb-- The role of the highly conserved flavoHb Tyr(B10) residue (7, 24) in regulating NOD activity was examined by site-directed mutagenesis. Phe was incorporated to eliminate potential hydrogen bonding to the flavoHb-Fe²⁺-O₂ complex via the Tyr hydroxyl group, and replacements with His and Glu were chosen for their potential stabilizing or destabilizing polar effects, respectively.

All of the Tyr(B10) substitutions reduce NOD activity significantly, from 15- to 35-fold, both in cell extracts (data not shown) and for the purified flavoHbs (Fig. 5, compare lines 1-4). The effect is clearly due to a decrease in the observed value of V_max and not an increase in the K_m for NO. At 50 nM NO, the NOD activity of the mutants decreases only 2.5-5-fold relative to the wild-type enzyme, whereas at 1 µM NO the drop in activity is 30-fold. Inhibition by NO is apparent at ~100 nM NO for the Phe replacement (Fig. 5, compare lines 1 and 2), whereas excess substrate inhibition of the wild-type enzyme only occurs at [NO] > 2 µM (line 5).

View larger version (19K): [in this window] [in a new window]   Fig. 5.   NOD activity of Tyr(B10) mutants. The NOD activity of flavoHb Tyr(B10) (lines 1 and 5) and mutants with Phe (line 2), Glu (line 3), and His (line 4) were measured with 100 µM NADH, 200 µM O2, and 1 µM FAD at various NO concentrations. Inset, magnification of the NO dependence of mutant NOD activities. The heme contents of Tyr, Phe, Glu, and His(B10) flavoHbs were 0.42, 0.28, 0.17, and 0.03 mole fractions, respectively.

The NADH oxidase activities of Tyr(B10), Phe, His, and Glu are 0.18, 0.48, 1.29, and 0.34 NADH heme¹ s¹, respectively, suggesting a decreased stability of the flavoHb-Fe²⁺-O₂ complex in the mutants. The NO reductase activities of the Tyr(B10) and Phe mutants are 0.14 and 0.07 NO heme¹ s¹, respectively. Clearly, Tyr(B10) plays a key role in NOD activity, while the other minor activities of E. coli flavoHb are not substantially altered (Table I).

Spectral Characterization of FlavoHb-- Spectra of flavoHb-Fe³⁺, unliganded flavoHb-Fe²⁺, flavoHb-Fe²⁺-O₂, flavoHb-Fe²⁺-CO, flavoHb-Fe²⁺-NO, and flavoHb-Fe³⁺-NO are shown in Fig. 6. The absorbance maxima of these complexes (Table II) are similar to those previously reported for this and other flavoHbs (5, 23, 30, 37-39).²

View larger version (31K): [in this window] [in a new window]   Fig. 6.   Absorption spectra of flavoHb ligand and redox forms. FlavoHb ligand and redox forms were prepared and measured at 5 µM heme as described under "Materials and Methods." The flavoHb forms are Fe2+ (deoxy) (line 1), Fe2+-O2 (oxy) (line 2), Fe2+-CO (line 3), Fe3+H2O (met) (line 4), Fe2+-NO (line 5), and Fe3+-NO (line 6). flavoHb was reconstituted with hemin as described under "Materials and Methods" and contained 0.95 and 0.41 mol fractions of heme and FAD, respectively.

Kinetics of FlavoHb Reduction-- When flavoHb-Fe³⁺ is reduced anaerobically with NADH, complex time courses are observed. There is a very fast decrease at 460 nm (Fig. 7, left panels), indicating rapid reduction of FAD and initial production of FADH₂. There is also a rapid increase at 430 nm (right panels), indicating reduction of the heme iron atom. The rate constant for the fast phase at 430 nm was significantly smaller than that observed at 460 nm (Fig. 8A, open versus closed symbols). At very high [NADH], the initial bleaching of FAD followed at 460 nm is too fast to measure, whereas the reduction of iron is a simple exponential process with a limiting rate of 150 s¹ at 20 °C (Fig. 7, middle panels). At low [NADH], the rates for the fast phases for FAD and iron reduction are similar, and the time course for iron reduction shows a distinct lag. Identical behavior was observed for the Phe(B10) mutant, showing that this replacement has no effect on the reductive half-reaction (Fig. 7, bottom panels, and Fig. 8, circles versus triangles).

View larger version (39K): [in this window] [in a new window]   Fig. 7.   Kinetics of flavoHb reduction. Time courses for anaerobic flavin and heme reduction with NADH were followed at 460 nm (left panels) and 430 nm (right panels), respectively, for both the wild-type (WT, top and middle panels) and the Phe(B10) mutant flavoHb (bottom panels).

View larger version (17K): [in this window] [in a new window]   Fig. 8.   Dependence of the fitted rate constants of the fast reduction phases, kobs, on [NADH]. A, dependence on [NADH] for rate constants determined at both 460 nm (closed symbols) and 430 nm (open symbols). Data for two independent preparations of wild-type flavoHb (circles) and for a single preparation of the Phe(B10) mutant (triangles) are shown. The solid line represents a linear fit to all the data at measured at 460 nm with a slope of 10.6 µM1 s1 and a y intercept of 136 s1. The dashed line represents a hyperbolic fit to all of the data at 430 nm using the parameters from the fit in B. B, double reciprocal plot of 1/kobs at 430 nm versus 1/[NADH]. A linear fit gives a slope of 0.068 µM s1 and a y intercept equal to 0.0067 s1.

These data demonstrate the first two reductive steps in the mechanism shown by line 1 in Equation 2. First FAD is reduced to FADH₂ by a very fast bimolecular process, and then the heme iron atom is reduced by an intramolecular electron transfer process to produce a FADH^·/Fe²⁺ intermediate. The observed rate constant for the initial hydride transfer shows a linear dependence on [NADH] with an apparent bimolecular rate constant, k'_H in Equations 4 and 5, equal to 10-15 µM¹ s¹ (Fig. 8A, solid and dashed lines). A linear fit indicates an intercept of ~140 s¹ and a slope of 11 µM¹ s¹. These data suggest that the first reductive process also involves two steps, simple equilibrium binding of NADH to the enzyme with a dissociation rate constant of ~140 s¹ followed by a very rapid first order hydride transfer to form FADH₂. The rate of hydride transfer must be >600 s¹, and the K_d for NADH must be >50 µM because the observed rate of FAD disappearance at 460 nm does not saturate with increasing [NADH] (Fig. 8A). Since our steady-state measurements were designed to examine the dependences on O₂ and NO at high [NADH], we have chosen to model the reductive half reaction (Equation 2, line 1) as a simple bimolecular hydride transfer step followed by electron transfer from the flavin to the iron atom.

As shown in Fig. 8, A and B, and predicted by Equations 4 and 5, the apparent rate of iron reduction measured at 430 nm shows a hyperbolic dependence on [NADH] and a lag in the time course at low [NADH]. The limiting value, 150 s¹, represents the rate of electron transfer from reduced flavin to Fe³⁺, k_ET. The apparent K_m for NADH for iron reduction is 13 µM, which is approximately equal to k_ET/k'_H (10-15 µM).

The transfer of an electron from FADH₂ to Fe³⁺ must result in the formation of a flavin semiquinone, FADH^·. This semiquinone does not appear to be "blue," since only slow absorbance decreases were observed at wavelengths of 600 nm. As shown in Fig. 7, top panels, slow heme reduction phases exhibit a rate, ~12 s¹, that is independent of [NADH] and the wavelength of observation. This phase is variable from preparation to preparation and appears to involve further reduction of heme groups in protein lacking flavin. The FADH^· radical on the intact enzymes appears to donate an electron to the deflavo proteins, causing the formation of FAD and an increase in absorbance at 460 nm. The initial rapid phases for reduction of the flavoHb with NADH are little affected by introduction of 100 µM O₂, but the character and extent of the slow phases are markedly altered by the NADH oxidase activity of the enzyme.

Kinetics of O₂, NO, and CO Binding to FlavoHb-- Time courses for O₂ binding to flavoHb-Fe²⁺ reduced with NADH are shown in Fig. 9A. The observed rate for the flavoHb is roughly twice that of sperm whale Mb (compare lines 1 and 3) and depends linearly on [O₂], demonstrating that the reaction is bimolecular (compare lines 1 and 2). Mutation of Tyr(B10) to Phe has only a small effect on the rate of O₂ binding to flavoHb-Fe²⁺, with k'_O2 increasing from 38 µM¹ s¹ to 50 µM¹ s¹ (compare lines 2 and 4).

View larger version (22K): [in this window] [in a new window]   Fig. 9.   O2 binding and release from flavoHb. A, O2 association rates for wild-type Tyr(B10) flavoHb-Fe2+ (lines 1 and 2), sperm whale Mb-Fe2+ (line 3) and mutant Phe(B10) flavoHb-Fe2+ (line 4) were measured following flash photolysis by monitoring the absorbance of deoxy form at 430 nm as described under "Materials and Methods." Reactions were in buffer containing 260 µM O2 (lines 1 and 3) or 1250 µM O2 (lines 2 and 4). B, O2 dissociation from Tyr(B10) flavoHb-Fe2+ (lines 1 and 3), Mb-Fe2+ (line 4), and Phe(B10) flavoHb-Fe2+ (line 5) was measured by CO displacement in buffer containing 260 µM O2 (lines 1 and 4), 0 µM O2 (lines 2 and 5), or 1250 µM O2 (line 3) as described under "Materials and Methods." Tyr(B10) and Phe(B10) flavoHbs were reconstituted with hemin and contained 0.95/0.41 and 0.90/0.24 mole fractions of heme/FAD, respectively.

Time courses for the displacement of bound O₂ by CO are shown in Fig. 9B. The rate of O₂ dissociation from flavoHb-Fe²⁺-O₂ is much slower than that from sperm whale Mb (compare lines 1 and 4). The calculated values of k_O2 for wild-type flavoHb-Fe²⁺-O₂ and Mb-Fe²⁺-O₂ are 0.44 and 15 s¹, respectively. In this case, mutation of the Tyr(B10) to Phe causes an increase in the rate of O₂ dissociation (compare lines 2 and 5). The time courses for O₂ displacement from the mutant are multiphasic with two major components exhibiting apparent k_O2 values equal to 34 and 3.3 s¹ (Fig. 9B; Table III). Both of these rate constants are considerably larger than that of the wild-type flavoHb, which exhibits a monophasic time course. These results coupled with the O₂ association rate constants suggest that the affinity of E. coli flavoHb-Fe²⁺ for O₂ may decrease by as much as 60-fold when the Tyr(B10) is replaced with Phe (i.e. the K_d increases from 0.012 to 0.67 µM; Table III).

Rate constants for CO binding to flavoHb-Fe²⁺ are also listed in Table III, and in this case, biphasic time courses are observed for both the association and dissociation reactions. Similar biphasic time courses have also been observed for Vitreoscilla sp. Hb (24), and A. eutrophus and S. cerevisiae flavoHbs.² In the wild-type protein, the largest kinetic phase reveals an association rate constant for CO that is similar to that for O₂ binding. In addition, the apparent K_d value for CO binding is only 6-fold less than that for O₂. Replacement of Tyr(B10) with Phe causes much smaller increases in the CO dissociation rate and equilibrium constants than those for O₂ and as a result, the strong discrimination in favor of O₂ and against CO binding is lost.

The association rate constants for NO binding to the ferrous or ferric forms of both the wild-type and Phe(B10) flavoHb are approximately the same and nearly equal to those measured for O₂ and CO binding, 20-50 µM¹ s¹. The similarity of the ligand association rate constants is unusual and implies a rate-limiting, diffusive step for the binding of all three ligands (40). The rate constant for NO dissociation from flavoHb-Fe²⁺ is very small, k_NO = 0.0002 s¹, and little affected by the Tyr(B10) to Phe mutation. Thus, the affinity of the reduced enzyme for NO is still ~1,000 times greater than that for O₂. It should be noted that the ratio of NO/O₂ affinities in mammalian Mbs is much larger, ~200,000, and is similar to what is observed for the Tyr(B10) to Phe mutant (Table III) (41).

The binding of NO to wild-type flavoHb-Fe³⁺ is complex, but a simple interpretation of equilibrium titrations suggests a K_d of ~100 µM. Using the larger association rate constant in Table III, this K_d predicts a rate constant of ~4,000 s¹ for NO dissociation. This larger value is consistent with our inability to observe NO binding to flavoHb-Fe³⁺ in the stopped-flow apparatus, which has a dead time of ~0.003 s¹.

Kinetics of NO Dioxygenation by FlavoHb-Fe²⁺-O₂-- Measurements of the rate constants for the oxygenation of NO by flavoHb-Fe²⁺-O₂ (k'_ox) were attempted in the stopped-flow apparatus. Using human HbO₂ and MbO₂ as controls, the oxy complexes were reacted with 1, 2, 5, 10, and 20 µM NO. As described by Eich et al. (34), Mb-Fe²⁺-O₂ and Hb-Fe²⁺-O₂ show rapid bimolecular increases in absorbance at 406 and 409 nm, the positions of the peaks for Hb-Fe³⁺ and Mb-Fe³⁺, respectively. In contrast, no rapid increase at 404 nm indicative of formation of Fe³⁺ heme is observed when NADH-reduced flavoHbO₂ is mixed with NO. Instead, a very small absorbance decrease is observed at 404 nm, and the observed rate constant, 200 s¹, is independent of [NO]. Herold (42) described a similar first order decrease in absorbance at 407 nm for the release of nitrate following the reaction of NO with HbO₂. Given the likelihood of a similar reaction with flavoHbO₂, we estimate that the rate of nitrate release from flavoHb, k_p, is ~200 s¹.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

NOD Activity and the Function of FlavoHb-- It is clear from the data summarized in Table I that the E. coli flavoHb efficiently removes and detoxifies NO by dioxygenation. Under saturating conditions at 37 °C, the maximum turnover number (V_max) for NOD activity is 670 s¹, and the K_m for NO is 0.28 µM. Under these conditions, the apparent bimolecular rate constant for NO-induced oxidation of flavoHb-Fe²⁺-O₂ (k'_ox) determined from V_max/K_m(NO) is ~2,400 µM¹ s¹, indicating a very efficient reaction mechanism. This reaction rate constant is 40-70 times greater than the values of 35 and 60 µM¹ s¹ for mammalian MbO₂ and HbO₂ (34) and is only ~2 times lower than the diffusion-limited rate constant for the reaction of NO with free O₂ (35, 36). The high in vivo (60 µM) and in vitro (80-100 µM) values for the K_m(O₂) at 37 °C (Fig. 3) indicate that O₂ concentration will limit NOD function under physiological conditions. The strong dependence of NOD activity on [O₂] may explain the benefit for flavo(Hb) up-regulation in microbes in response to lower O₂ availability (9, 20, 22). Higher levels of (flavo)Hbs will be required to compensate for decreased NOD activity at a low [O₂].

Low NO reductase (0.013-0.24 NO heme¹ s¹), NADH oxidase (0.03-0.1 NADH heme¹ s¹), and FAD reductase activities of E. coli flavoHb have been previously reported (23, 37, 38, 43, 44). The rates that we measured for the anaerobic NO reductase, NADH oxidase, and anaerobic FAD reductase activities are more than 1000-fold lower than the NOD activity (Table I). Given the low NO reductase activity, flavoHb is unlikely to be involved in NO detoxification under anaerobic conditions. Moreover, there is evidence for a separate NO-inducible NO reductase activity in E. coli that performs this function (14). Other activities of flavoHb have been described (45-47), but their biological significance remains to be determined.

Mechanism of NOD Catalysis-- E. coli flavoHb appears to catalyze NO dioxygenation by a conventional flavoprotein mechanism in which two electron hydride transfer occurs separately from the two oxidation steps (Equation 2). Similar binary complex mechanisms have been proposed for metHb/cytochrome b₅ reductase (48). The resultant steady-state equations (Equations 4-6) predict parallel lines for plots of 1/v_i versus 1/[NADH] and plots of 1/v_i versus 1/[NO] at varying [O₂]. The observed data in Figs. 2A and 4A confirm this prediction. A summary of the relevant kinetic parameters for this reaction mechanism is given in Table IV.

View this table: [in this window] [in a new window]   Table IV Summary of kinetic parameters for the NOD activity of E. coli flavoHb Vmax and Km values were determined from initial velocities in Figs. 2-4. k'H, k'OX and k'O2 were calculated from initial velocities using the expressions Vmax/2 · Km(NADH), Vmax/Km(NO), and Vmax/Km(O2), respectively, as described in Equations 4-7. k'H and k'O2 were measured directly in Figs. 6-8. Vmax was calculated using the expression kpkET/(kp + kET), where kET = 150 s1 in Fig. 8, and kp was estimated to be ~200 s1. k'OX was estimated to be  600 µM 1s1, based on the dead time of the stopped flow apparatus and the inability to measure flavoHbO2 oxidation by NO. Theoretical Km values for NADH, NO, and O2 were calculated from the transient state parameters using the expressions Vmax/2k'H, Vmax/k'OX, and Vmax/k'O2.

The results in Fig. 8A suggest that the initial reduction of FAD may be a two-step process. However, we were not able to demonstrate saturation with increasing [NADH] due to the large rate of hydride transfer (>600 s¹) and the dead time limitations of the rapid mixing apparatus. Nevertheless, the apparent bimolecular rate constant determined for hydride transfer (k'_H) from the data in Figs. 7 and 8 correlates well with the value calculated from the steady-state parameters at 20 °C (Table IV). Similarly, the product release (k_P) and electron transfer rates (k_ET) measured in rapid mixing experiments at 20 °C predict a V_max value very close to that observed directly at saturating levels of all substrates, ~94 s¹. The correspondence between the transient and initial velocity measurements for these parameters adds confidence to the value of k'_ox calculated from V_max/K_m(NO).

There is up to an 11-fold difference between the measured value of K_m for O₂ at 20 °C and the value calculated using k'_O2 measured by laser photolysis. The latter rate constant predicts a low K_m for O₂, 2.4 µM, whereas a value of ~27 µM is observed in steady-state experiments (Table IV). There are several possible explanations for this discrepancy. Inhibition at high levels of NO will cause an apparent increase in the K_m for O₂; however, the K_m for O₂ is still very high at 0.1 µM NO, where no inhibition is observed (Fig. 3). Alternatively, the rate of O₂ dissociation, k_O2, could be much larger during the catalytic cycle than is seen in rapid mixing experiments for the fully reduced flavoHbO₂ complex. This interpretation is ruled out by the observation of roughly parallel lines in the 1/v versus 1/[NO] plots at various [O₂]. A high value of k_O2 predicts converging lines (Equation 6). The remaining explanation is that the bimolecular association rate constant for O₂, k'_O2, is ~11 times smaller during catalysis than is observed by laser photolysis (Table IV). The mechanistic cause of this difference is puzzling and could be due to a low level of iron reduction in the two- and one-electron reduced intermediates (i.e. at equilibrium the FADH₂/Fe³⁺ species is favored over FADH^·/Fe²⁺ and FADH^·/Fe³⁺ over FAD/Fe²⁺), or conformational changes in the distal portion of the heme pocket during catalysis slow the rate of ligand binding.

The discrepancy between the steady-state and transient values of k'_O2 and the inhibition seen at high [NO] indicates that the mechanism given in Equation 2 is an oversimplification. A more complete reaction scheme describing the NOD activity of flavoHb is shown in Fig. 10 and takes into account 1) inactivation by NO binding to deoxyheme groups; 2) the formation of a three-electron reduced flavoHb (i.e. FADH₂/Fe²⁺); and 3) the existence of multiple intermediates, rates, and sites for both reduction by NADH and oxidation by NO. The fact that Equations 4-7 can be used to interpret both the initial velocity and transient state reduction kinetics suggests that the rate of hydride transfer from NADH to FAD is probably independent of the reduction state of the iron atom. The latter idea is supported by the lack of effect of the Phe(B10) mutation on the kinetics of anaerobic reduction.

View larger version (24K): [in this window]