Influence of Heme-Thiolate in Shaping the Catalytic Properties of a Bacterial Nitric-oxide Synthase*

Background: NOSs possess a highly conserved tryptophan residue, proximal to the heme-thiolate bond. Results: Replacement of this Trp by His or Phe in Bacillus subtilis NOS altered both thermodynamic and kinetic parameters and NO synthesis. Conclusion: B. subtilis NOS control catalysis by tuning the electron density of its heme-thiolate bond. Significance: This is the first study to investigate these relationships in a bacterial NOS. Nitric-oxide synthases (NOS) are heme-thiolate enzymes that generate nitric oxide (NO) from l-arginine. Mammalian and bacterial NOSs contain a conserved tryptophan (Trp) that hydrogen bonds with the heme-thiolate ligand. We mutated Trp66 to His and Phe (W66H, W66F) in B. subtilis NOS to investigate how heme-thiolate electronic properties control enzyme catalysis. The mutations had opposite effects on heme midpoint potential (−302, −361, and −427 mV for W66H, wild-type (WT), and W66F, respectively). These changes were associated with rank order (W66H < WT < W66F) changes in the rates of oxygen activation and product formation in Arg hydroxylation and N-hydroxyarginine (NOHA) oxidation single turnover reactions, and in the O2 reactivity of the ferrous heme-NO product complex. However, enzyme ferrous heme-O2 autoxidation showed an opposite rank order. Tetrahydrofolate supported NO synthesis by WT and the mutant NOS. All three proteins showed similar extents of product formation (l-Arg → NOHA or NOHA → citrulline) in single turnover studies, but the W66F mutant showed a 2.5 times lower activity when the reactions were supported by flavoproteins and NADPH. We conclude that Trp66 controls several catalytic parameters by tuning the electron density of the heme-thiolate bond. A greater electron density (as in W66F) improves oxygen activation and reactivity toward substrate, but decreases heme-dioxy stability and lowers the driving force for heme reduction. In the WT enzyme the Trp66 residue balances these opposing effects for optimal catalysis.

requires another round of oxygen activation, in which H 4 X is proposed to operate as the electron donor ( Fig. 1).
Both animal NOS and bacterial NOS possess a conserved tryptophan residue that forms stacking interactions with the porphyrin ring and hydrogen bonds with the heme-thiolate bond (Fig. 2). Replacement of this proximal Trp residue by histidine in murine iNOS (W188H) increased the midpoint potential of the heme group by ϩ88 mV and reduced the rate of NO synthesis compared with wild-type iNOS. In addition, the W188H mutation stabilized a heme intermediate that formed downstream of the FeO 2 species, which reacted slowly with L-Arg to form NOHA (35). Surprisingly, the W188F mutant of iNOS had a defective heme binding, thus hampering any further characterization (36). Substitution of Trp 409 by Phe in rat nNOS reduced the heme midpoint potential of the protein, led to a faster formation of the FeO 2 species and to greater rates of Fe(II)-NO oxidation compared with the wild-type protein (37)(38)(39). Thus, the proximal Trp residue may play a role in controlling the reactivity of the enzyme by tuning the properties and reactivity of its heme.
We set out to investigate these relationships in Bacillus subtilis NOS. Replacement of the conserved Trp residue (Trp 66 ) by His or Phe led to the expression of stable proteins, with characteristic spectroscopic features as described recently (34). We utilized these mutations to increase (His) or decrease (Phe) the H-bonding properties of the heme-thiolate bond of B. subtilis NOS and inquired on its effects on: heme midpoint potential, enzyme oligomeric state, substrate binding, FeO 2 stability, kinetics of heme transitions during catalysis, kinetics and extent of product formation in both steps, reactivity of the FeNO complex, and NO synthesis. This is the first comprehensive study to: (i) provide a side-by-side comparison of the effect of replacing the proximal Trp with an electron donating (His) versus an electron withdrawing (Phe) residue in catalysis by a bacterial NOS. (ii) Demonstrate that H 4 T enables the formation of a Fe(III)-NO species during NOHA-driven single turnover reactions and (iii) show that H 4 T supports NO synthesis by B. subtilis in an in vitro reconstitution system.

EXPERIMENTAL PROCEDURES
Reagents-H 4 B and H 4 T were purchased from Schircks Laboratories (Jona, Switzerland). CO gas was obtained from Praxair, Inc. (Danbury, CT). N 5 -CH 3 -H 4 T (Eprova, Switzerland) was a generous gift of Dr. Donald Jacobsen (Department of Cell Biology, Cleveland Clinic). N-Hydroxy-L-Arg (NOHA) and 14 C-labeled L-arginine (L-[ 14 C]Arg) were purchased from MP Biomedicals (Solon, OH). EPPS was purchased from Fisher Scientific (Pittsburgh, PA). DTT was purchased from RPI Corp. (Mount Prospect, IL). All other reagents were purchased from Sigma.
Protein Expression and Purification-Wild-type and mutant BsNOS proteins containing a His 6 tag attached to their N termini were overexpressed in E. coli strain BL21(DE3). BsNOS proteins were expressed and purified as described (34). Protein concentration was determined from the absorbance at 444 nm of the ferrous heme-CO complex, using an extinction coefficient of 76 mM Ϫ1 cm Ϫ1 (⌬⑀ 444 -500 nm ) (25). All proteins were purified to homogeneity (Ն95%) as assessed by SDS-PAGE  The spatial arrangement of L-Arg, heme, and its axial Cys, H 4 T, and Trp 66 are shown. The structure of the active site was created from the crystallographic data available for native BsNOS (PDB 1M7V (6)) using PyMol software. In the crystal structure utilized herein, the proximal Trp is in position 56 due to a truncation of the first 10 amino acids of the protein.
(supplemental Fig. S1A). The oligomeric state of wild-type and mutant B. subtilis NOSs reconstituted with H 4 T and L-Arg was examined by size-exclusion chromatography. Protein samples (ϳ150 M) in EPPS buffer (40 mM, pH 7.6, 150 mM NaCl) were incubated with 2 mM L-Arg, 400 M H 4 T, and 1.2 mM DTT for 15 min. Samples were injected on a Superdex 200 resin preequilibrated with EPPS buffer (40 mM, pH 7.6, 150 mM NaCl) supplemented with 100 M L-Arg, 40 M H 4 T, and 120 M DTT. Under these conditions, all proteins existed predominantly in the dimeric state (80 -97%) (supplemental Fig. S1B).
Imidazole and Arginine Binding-Imidazole and L-Arg binding affinities were studied at 25°C by perturbation difference spectroscopy according to methods described previously (27,28). NOS samples (around 5 M) in 40 mM EPPS buffer, pH 7.6, with 10% glycerol, 0.6 mM DTT, 0.2 mM H 4 T were titrated by stepwise addition of imidazole, to a final concentration of 10 mM. The K d of imidazole (K d (imid)) was calculated by fitting the data to a simple saturation binding equation. The K d of L-Arg (K d ) was determined under the same conditions, in the presence of 10 mM imidazole. The data were fit to a simple saturation binding equation, and K d was calculated according to the Equation 1.
Single Turnover Reactions-L-Arg hydroxylation and NOHA oxidation experiments were carried out in a Hi-Tech SF61-DX2 stopped-flow instrument (Hi-Tech Scientific, Salisbury, UK) coupled to a diode array detector, as previously described (40). An anaerobic solution of 20 M ferrous NOS, 2 mM L-Arg (or 1 mM NOHA), 0.2 mM H 4 T (or another pterin, where indicated), and 1 mM DTT in 40 mM EPPS, pH 7.6, containing 10% glycerol and 150 mM NaCl was mixed at 10°C with a syringe containing oxygen-saturated buffer, 2 mM L-Arg, 0.2 mM H 4 T, and 1 mM DTT. Sequential spectral data were fitted to an A 3 B 3 C model using the Specfit/32 global analysis software, version 3.0 (Spectrum Software Associates, Marlborough, MA), which calculates the spectra of the different enzyme species and their concentration change versus time. The reported rates are the average of 5-10 measurements. The error associated to these measurements was 2-5%. Ferrous protein was generated by addition of dithionite. The reactions were started by mixing with oxygen-saturated buffer and allowed to proceed at room temperature for 10 min. The reactions were quenched with a solution of 0.5 M HCl, 20% 2-propanol, 1 mM L-Arg, and 1 mM NOHA. Quenched samples were stored at Ϫ80°C for further HPLC analysis. Infinite time reactions with NOHA were performed as described for the reactions of L-Arg, except the concentrations of reagents were: 150 M NOS, 400 M NOHA, 1 mM H 4 T, and 3 mM DTT. Quenched samples were stored at Ϫ80°C and citrulline formation was determined using a published HPLC procedure (see below).
Determination of Reaction Products by HPLC-L-[ 14 C]Arg and [ 14 C]NOHA were extracted from the reaction mixtures (100 l each) with 5 l of 2-propanol 40 mM HCl, and were vortexed for 45 min at room temperature. These solutions were then centrifuged at 10,000 ϫ g for 10 min to eliminate any precipitated materials. 50 l was injected on a Nucleosil C18-HPLC column (particle size 5 m, 250 ϫ 4.6 mm), and the amino acids were separated using an isocratic method with elution buffer (50 mM sodium acetate, pH 6.50). Under these conditions L-[ 14 C]NOHA and L-[ 14 C]Arg eluted at ϳ13 and 18 min, respectively. The radioactivity of each sample was determined using a scintillation counter. Citrulline formation from NOHA was determined by derivatization of the amino acid products with naphthalene-2,3-dicarboxyaldehyde in the presence of cyanide followed by HPLC analysis, according to a published procedure (41).
Midpoint Potential Measurements-Spectroelectrochemical titrations were carried out in a glove-box (Belle Technology, Dorset, UK) under N 2 atmosphere, as previously described (35,42). Briefly, NOS proteins were made anaerobic by gel filtration in a Sephadex G-25 column (PD 10, GE Healthcare) equilibrated with anaerobic buffer (100 mM phosphate buffer, pH 7.0, 125 mM NaCl). Protein samples were diluted to a 3.5-ml final volume (final concentration Ϸ10 M) and L-Arg (2 mM) and H 4 T (100 M) were added. The following electron mediator dyes (0.5-1 M) were used: phenosafranine (E m ϭ Ϫ252 mV), benzyl viologen (E m ϭ Ϫ358 mV), methyl viologen (Ϫ450 mV), and anthraquinone-2-sulfonate (E m ϭ Ϫ225 mV). The titration was carried out at 15°C by bolus additions of a sodium dithionite solution. Absorption spectra were recorded with a Cary 50 spectrophotometer equipped with a dip-probe detector, and the potentials were measured with an Accumet AB15 pH meter (Fisher Scientific) using a silver/silver chloride microelectrode saturated with 4 M KCl.
Ferrous Heme-NO Complex Oxidation (k ox )-Wild-type, W66H, and W66F BsNOS proteins (ϳ5 M) in 100 mM EPPS (pH 7.6, 150 mM NaCl, 10% glycerol) containing 2 mM L-Arg, 0.2 mM H 4 T, and 1 mM DTT were reduced with dithionite in an anaerobic cuvette. The Fe(II)-NO complex of W66F was unstable at neutral pH, hence these reactions were carried out under the conditions described above using CHES buffer (100 mM, pH 9.5, 150 mM NaCl, 10% glycerol). The Fe(II)-NO complexes were generated by adding successive aliquots of an anaerobic NO-saturated buffer. The samples were then transferred to an anaerobic stopped-flow instrument using a gas tight syringe, and the reactions were initiated by mixing the anaerobic Fe(II)-NO protein samples with air-saturated buffer at 10°C. Spectra were taken during the course of the reactions and the data were fit to a single exponential model A 3 B using Specfit global analysis software.
NO Synthesis-NO synthesis by BsNOS and mutant proteins was assessed via a reconstitution assay as described previously (20). Briefly

Spectrocoscopic Properties and Substrate
Binding-We first examined the spectroscopic properties and the ability of the two mutant proteins to bind imidazole, the substrate L-Arg, and the stable reaction intermediate NOHA. UV-visible data for BsNOS, W66H, and W66F reconstituted with H 4 T are given under supplemental Table S1. First, the UV-visual spectral features of BsNOS reconstituted with H 4 T are almost identical to those observed for the protein reconstituted with H 4 B (11, 34). The UV-visible spectra of the Fe(III), Fe(II), and Fe(II)-CO complexes of BsNOS, W66H, and W66F reconstituted with H 4 T are given in Fig. 3. The spectra of the Fe(II), Fe(II)-CO, and Fe(III)-imidazole complexes of W66H and W66F are similar but not identical to that of wild-type BsNOS (supplemental Table S1). In the presence of L-Arg and H 4 T all three proteins exist in the typical Fe(III) high-spin configuration. Reduction with dithionite and exposure to CO resulted in the formation of the characteristic Fe(II)-CO complex with a Soret band appearing at 445-449 nm. The Fe(II)-CO species was unstable in the case of W66F (Fig. 3C), slowly converting to form an uncharacterized species with an absorption maximum at 420 nm. This could be a hexa-coordinated Fe(II)-CO complex, in which the axial Cys is protonated (44), however, the exact nature of this species in NOS is currently unknown. This was also observed for W66F reconstituted with H 4 B (34), as well as in the corresponding Trp 3 Phe mutants of Staphylococcus aureus NOS and eNOS (45), suggesting that the hexa-coordinated ferrous state in this mutant NOS is unstable compared with wild-type BsNOS. Binding dissociation constants, K d , for imidazole, L-Arg, and NOHA in the presence of H 4 T were measured spectrophotometrically (supplemental Fig. S2). The affinity constants of BsNOS wild-type for imidazole and L-Arg in the presence of H 4 T were 160 Ϯ 9 M and 1.20 Ϯ 0.06 M, respectively. K d values for imidazole and L-Arg binding to BsNOS in the presence of H 4 B have been reported previously as 384 and 4.8 M (11), respectively. This suggests a slightly increased affinity of BsNOS for its substrate L-Arg and the ligand imidazole in the presence of H 4 T. A similar, yet more pronounced effect of H 4 T on substrate binding affinity has been reported for Drosophila melanogaster iNOS (28). Binding affinities of BsNOS and W66H for imidazole were very similar (160 Ϯ 9 and 122 Ϯ 7 M, respectively), and higher (ϳ4-fold) than that observed in W66F (523 Ϯ 29 M). Both wild-type and mutant BsNOS proteins displayed competent binding for L-Arg, with K d values following the order:  Fig. S2). The lower affinity for both L-Arg and NOHA in W66H compared with wild-type BsNOS resembles the previously reported effects of the analogous mutation in iNOSoxy (W188H) (35) and P450 BM3 (F393H) (46).
Redox Potentiometry-A possible consequence of replacing the proximal Trp residue by His or Phe is an alteration of the heme midpoint potential. Redox titrations in the presence of L-Arg and H 4 T were performed for BsNOS, W66H, and W66F and the results are shown in Fig. 4. The calculated midpoint potentials were: Ϫ361, Ϫ302, and Ϫ427 mV for BsNOS, W66H, and W66F, respectively. Thus, substitution of Trp 66 by His increased the midpoint potential by ϩ59 mV with respect to the wild-type protein. A similar effect was observed for the corresponding mutations in iNOSoxy (W188H) (35) and cytochrome P450 BM3 (F393H) (46 -48). In contrast, substitution of Trp 66 with Phe resulted in a substantial decrease of the heme midpoint potential (Ϫ66 mV). Thus, changes in hydrogen bonding of the heme-thiolate in BsNOS have a direct effect on the redox properties of the heme.  Table S1. NOVEMBER 11, 2011 • VOLUME 286 • NUMBER 45

JOURNAL OF BIOLOGICAL CHEMISTRY 39227
Stopped-flow Analysis of a Single Turnover L-Arg Hydroxylation Reaction in the Presence of H 4 T -We next investigated the heme transitions that occur in BsNOS, W66H, and W66F during catalysis in the presence of H 4 T and L-Arg. Anerobic samples of each protein in the presence of H 4 T and L-Arg were reduced with dithionite and mixed with air-saturated buffer containing H 4 T and L-Arg in a stopped-flow instrument. A minimum of 100 spectra were collected during the course of the reaction and subjected to global analysis. We found that L-Arg hydroxylation reactions of both wild-type and mutant proteins could be best fit to a two-exponential model A 3 B 3 C, with Fe(II), Fe(II)-O 2 , and Fe(III) as the only detectable species (supplemental Fig. S3). A summary of the observed rate constants is given in Table 1 (Table 1). An enhanced conversion rate of the Fe(II)-O 2 species in the presence of the methylated cofactor at position N 5 has been reported for the reactions of iNOS with N 5 -CH 3 -H 4 B (49). It was proposed that a faster electron transfer from N 5 -CH 3 -H 4 T to the heme-oxy species occurs via an increased H 4 B . ϩ stability afforded by the presence of the N 5 -methyl group (49). Although this effect on Fe(II)O 2 stability was not observed in B. subtilis NOSs, W66F displayed the fastest rates of Fe(II)O 2 conversion with respect to BsNOS and W66H, regardless of the pterin used. Thus, replacement of Trp by Phe enhanced the processing of the heme-oxy intermediate in B. subtilis NOS, similar to the reported features of the corresponding mutant in rat nNOSoxy (W409F) (37)(38)(39). This implies that the properties of the Fe(II)O 2 species in BsNOS are affected by the electronic structure of its heme, as observed in mammalian NOS. A comparison of single turnover reactions with L-Arg and H 4 B in BsNOS proteins versus the iNOSoxy and W188Hoxy mammalian counterparts showed that in general, the heme transition rates in the bacterial NOSs are slightly slower than those seen in the mammalian proteins (   H 2 T, respectively. Formation of the Fe(II)O 2 species was faster in the presence of the oxidized pterins, whereas its conversion to ferric was significantly slower than that observed in the reactions performed with reduced biopterin or folate (supplemental Tables S2 and S3). This suggests that the reduced pterins may play a role in L-Arg oxidation via electron transfer, as it occurs in mammalian NOSs. A graphic representation of Fe(II)O 2 conversion rates for selected hemeproteins is provided in Fig. 6. A comparison with other heme proteins shows that the reactivity of the Fe(II)O 2 species in B. subtilis NOS falls between the fast transition rates observed in mammalian NOSs and the extremely slow rates of Fe(II)O 2 disappearance reported for most members of the cytochrome P450 family (supplemental Table S4). In B. subtilis NOSs the rates of conversion of the Fe(II)O 2 species correlated well with the heme midpoint potentials. The W66H complex was the most long-lived in the presence of L-Arg and H 4 B, H 4 T, or N 5 -CH 3 -H 4 T, whereas the opposite was true for W66F (Table 1).

Stopped-flow Analysis of a Single Turnover NOHA Oxidation
Reaction in the Presence of H 4 T-We next examined NOHA oxidation reactions under single turnover conditions. Earlier studies by our laboratory demonstrated that a Fe(III)-NO complex builds up in BsNOS during single turnover reactions with H 4 B and NOHA (3). Given the plasticity in pterin utilization proposed to be at play in bacterial NOSs (30), it was interesting to investigate whether H 4 T could support the formation of a Fe-NO species in B. subtilis NOS. Analysis of the reaction profiles of BsNOS and the W66 mutants shows that H 4 T also supports formation of a Fe(III)-NO complex (Fig. 7). Analysis of single turnover reactions in the presence of NOHA and H 4 T showed that formation of Fe(II)-O 2 was followed by fast formation of a Fe(III)-NO complex, which reacted further to form Fe(III). Fit of the data to a three exponential sequential model (A 3 B 3 C 3 D) yielded the rates provided in Table 1 Fig. S4).
Due to the elusive nature of the Fe(II)O 2 species in W66F, the reaction was modeled to a double-exponential sequential model, A 3 B 3 C. Formation of the Fe(III)-NO complex was fast (apparent k obs ϳ 26 s Ϫ1 , however, its rate of conversion to Fe(III) (0.30 s Ϫ1 ) did not differ markedly (ϳ1.5-fold) from that of BsNOS WT (0.19 s Ϫ1 ) or W66H (0.20 s Ϫ1 ) ( Table 1).
Extent and Kinetics of L-Arg Hydroxylation-To determine whether substitution of Trp for His or Phe had an impact in the total yield and kinetics of L-Arg hydroxylation we measured the time course of NOHA formation in the presence of H 4 T by rapid-quench and HPLC, using L-[ 14 C]Arg as the substrate (Fig.  8). The observed rates of [ 14 C]NOHA formation were 4.7 Ϯ 1.1, 2.9 Ϯ 0.2, and 17 Ϯ 2 s Ϫ1 , for BsNOS, W66H, and W66F, respectively. These rates follow the trend W66F Ͼ BsNOS Ͼ W66H, which are consistent with the rates for the heme transitions observed by rapid UV-visible scan during single turnover reactions. NOHA formation from L-Arg, and citrulline formation from NOHA were also determined, for infinite incubation times (15 min) (supplemental Table S5). Production of NOHA from L-Arg was characterized by notably low yields in both BsNOS and W66 mutants. In contrast, production of citrulline from NOHA was a relatively efficient process; the extent of product formation was ϳ0.6 citrulline per heme (supplemental Table S5).
NO Synthesis-NO synthesis by BsNOS and mutant proteins was assessed using a three-component reconstitution assay (FLDR, YkuN, and NOS) as described previously (20). Our results indicate that both W66H and W66F are less proficient than wildtype BsNOS for NO synthesis (Fig. 9). In addition, H 4 B supported NO synthesis to a greater extent compared with H 4 T (Fig. 9). However, the relative NO synthesis efficiency of each Trp 66 mutant compared with wild-type remained unchanged. This suggests that the diminished NO synthesis observed in the Trp 66 background is likely caused by alterations in the heme-thiolate environment per se, rather than it being an isolated effect inherent to pterin preference or substrate affinity. The total yield of NO per heme indicates that under these experimental conditions B. subtilis NOS performs ϳ20 turnovers, and suggest that in vivo, H 4 B might be more competent than H 4 T to drive NO synthesis.
Oxidation Rate of the Ferrous-NO complex (k ox )-During catalysis, the Fe(III)-NO forms as an immediate product complex and can be reduced by NOS reductase (or the corresponding reductase partner in bacterial NOSs) at rates comparable with that of Fe(III)-NO dissociation (50,51). To return to the catalytic cycle, the ferrous-NO complex must react with O 2 . We went on to determine the observed rate constants for reaction of the Fe(II)-NO complex with O 2 in BsNOS, W66H, and W66F at half-air saturation ([O 2 ] ϳ 120 M). The oxidation of anaerobic, pre-formed heme Fe(II)-NO complexes by O 2 was investigated by stoppedflow spectroscopy. Spectral data are given in Fig. 10. In all cases, only two species were observed, namely Fe(II)-NO and Fe(III), and therefore the oxidation of Fe(II)-NO by O 2 could be best fit to a single-step reaction (Fig. 10). We observed that at neutral pH the Fe(II)-NO complex in W66F (Soret maximum at 439 nm) converted quickly to form a species with a Soret maximum at 415 nm and a marked shoulder at 439 nm. This presumably corresponds to a Fe(II)-NO complex in which the axial Cys has been protonated. This phenomenon is thought to arise by weakening of the Fe-S via disruption of the H-bonding interaction brought about by replacement of the proximal Trp with Phe (52). A similar behavior was observed in the W409F mutant of rat nNOSoxy. 4 To minimize this proton-mediated effect, the reactions of W66F were performed at  Table S4. pH 9.5 (note: previous studies in our laboratory indicated that pH has a negligible effect on the oxidation rates of Fe(II)-NO by O 2 ). 4 k ox values were found to be 0.092, 0.063, and 1.17 s Ϫ1 for BsNOS, W66H, and W66F, respectively. Thus, the oxidation rates of Fe(II)-NO in B. subtilis NOS are slow in the range of those reported for nNOSoxy and W409F, respectively (supplemental Table S6).

DISCUSSION
The effect of H-bonding on midpoint potentials and metalloprotein properties is not unprecedented. Chang and Traylor (54) were among the first groups to show that the basicity of the proximal nitrogen in myoglobin (His) influences the affinity of its heme for oxygen and carbon monoxide. A few years later,  ϳ416 and 437 nm), and Fe(III) resting enzyme (Soret peak ϳ392 nm, high-spin). The reaction of Fe(II) with dioxygen to form FeO 2 is very fast in W66F, thus the reaction could be best fit to a two-exponential model: A 3 B 3 C, where species A is likely a mixture of Fe(II) and FeO 2 .
Jensen and co-workers (55) provided evidence that the NH-S hydrogen bond in ferredoxin and rubredoxin could be important to modulate the midpoint potentials of the Fe-S clusters of the enzyme. The elucidation of the first crystal structures of heme-and iron-sulfur proteins provided strong evidence that a relationship exists between changes in oxidation state of the iron atom and hydrogen bond geometry (56,57). These early studies indicated that this proximal hydrogen bond could be the link between changes in midpoint potentials and the reactivity of the iron atom (56,57). Later on, model studies conducted by Ueyama et al. (58) suggested that the NH-S hydrogen bond of ferredoxin model complexes contributed to the redox potential of the Fe-S center. More recent approaches utilizing site-directed mutagenesis confirmed that these relationships hold true for a number of enzyme systems, including peroxidases (59) and cytochrome P450 (46, 60 -62).
A great advantage of studying these effects in NOS is that this hydrogen bond is provided by a Trp residue proximal to the hemethiolate, hence mutagenesis of this Trp by residues with different H-bonding capacities are generally feasible (an exception is the Trp 188 3 Phe mutant of murine iNOS, which has a defective heme binding (36)). The expression of the stable, dimeric mutants W66H and W66F of B. subtilis NOS permitted a side-by-side characterization of the role of the H-bonding properties of the hemethiolate in catalysis by a bacterial NOS. We found that replace-ment of Trp by His or Phe had profound effects on substrate binding affinity, the formation and disappearance of catalytically relevant heme intermediates, the heme midpoint potential, and the proficiency of the enzyme to drive NO synthesis.
Imidazole and Substrate Binding-Replacement of Trp by His or Phe did not disrupt Fe-Cys ligation in the mutant proteins, but had a measurable impact on the electronic properties of the heme and the binding affinity for the substrate L-Arg, and the stable intermediate, NOHA. These findings are in line with a recent spectroscopic study that showed that the proximal H-bond network modulates a number of electronic and structural properties in B. subtilis NOS (34).
Midpoint Potentials-The midpoint potential of wild-type BsNOS (Ϫ361 mV) is ϳ100 mV lower than that of wild-type mammalian NOSs (Ϫ250 to Ϫ270 mV) (42). Replacement of the proximal Trp by His increased the heme midpoint potential by 57 mV, whereas a 66-mV decrease was achieved via replacement of Trp with the electron-withdrawing residue Phe. This has important implications for catalysis; lower midpoint potentials pose a thermodynamic challenge for bacterial NOSs, making their heme more difficult to reduce. Consequently, lower heme midpoint potentials limit the repertoire of reductases capable of furnishing electrons into the bacterial NOSs.
Stability of FeO 2 during a Single Turnover L-Arg Hydroxylation Reaction in the Presence of H 4 T-Overall, it appears that the Fe(II)O 2 species is destabilized in proteins with lower redox potentials, and vice versa. An elevated heme midpoint potential decreases the driving force for reduction of the Fe(II)-O 2 complex, leading to a Fe(II)-O 2 that is more stable than the alternate Fe(III)O 2 Ϫ (ferric superoxy) species. Therefore, oxygen activation is disfavored in W66H, due to stabilization of the Fe(II)-O 2 complex, which in turn decreases the overall rate of catalysis. This general phenomenon has been referred as to a "thermodynamic trap" by Ost et al. (46). The observation that the Fe(II)O 2 species has a prolonged half-life in W66H and a faster disappearance in W66F compared with wild-type BsNOS concurs with the notion that modulation of the heme midpoint potential is crucial for tuning NOS reactivity. Indeed, other hemethiolate proteins featuring very negative midpoint potentials display an enhanced rate of oxygen activation and reactivity with their substrates (62,63). Conversion of the FeO 2 intermediate was accelerated in reactions driven by reduced pterins versus those performed in the presence of H 2 B or H 2 T, suggesting that the pterins may be employed as electron donors for oxygen activation, as it occurs in mammalian NOSs.    (6,17), our results suggest the pterin could operate as a source of electrons in oxygen activation reactions. Buildup of the FeNO complex was faster in W66F compared with WT BsNOS and W66H, and so was its transition to form Fe(III) enzyme. These kinetic features are overall unfavorable for NO synthesis, because a rapid formation and a low rate of dissociation of the Fe(III)-NO complex may lead to a larger partitioning of the available NO into the futile NObound form (39,50), thus enhancing the NO dioxygenase mode of the enzyme. A similar phenomenon has been observed in the corresponding mutant of nNOS (W409F) (38,39) and in nNOS harboring Fe-mesoporphyrin IX (42), both of which possess lower heme midpoint potentials compared with their respective native counterparts.
Product Yield and NO Synthesis-An examination of NOHA formation from L-Arg and citrulline formation from NOHA indicated that the efficiency of product formation is very similar between wild-type BsNOS and the W66 variants (ϳ0.3 NOHA per heme and ϳ0.6 citrulline per heme). This suggests that the mutations per se did not increase the extent of uncoupled O 2 reduction compared with wild-type BsNOS. However, the relatively low yields of NOHA production from L-Arg compared with that observed in mammalian NOSs suggests that the first catalytic step of NO synthesis is somewhat less efficient in bacterial NOSs. Both H 4 B and H 4 T supported NO synthesis by B. subtilis NOSs in a reconstitution system comprised of FLDR and YkuN. Under our experimental conditions, H 4 B proved to be a better supporter of NO synthesis compared with H 4 T. The causes of this effect are presently unknown.
Oxidation of FeNO by Dioxygen-The slow dissociation rates of the Fe(III)-NO complex in B. subtilis NOS increase the tendency of the enzyme to engage into a nonproductive cycle where the Fe(III)-NO is reduced by the reductase partner before it can be released. To avoid this, the Fe(II)-NO complex must react with O 2 (k ox ) to return to the catalytic cycle. We uncovered that oxidation of Fe(II)-NO by O 2 is remarkably slow in B. subtilis NOSs compared with the native mammalian NOSs (Table 1). Two other NOSs display comparable k ox values, namely D. melanogaster NOS (53) and Geobacillus stearothermophilus NOS (18). An immediate consequence of a slow k ox is that a significant proportion of the enzyme may accumulate as Fe(II)-NO during steady-state NO synthesis. The higher k ox observed in the W66F mutant is advantageous in that it diminishes buildup of the Fe(II)-NO species during catalysis, perhaps at the expense of a very negative heme midpoint potential.
A summary of the relevant thermodynamic and kinetic parameters for WT and W66 variants is given in Table 2. Despite the fact that the W66F mutant performed certain catalytic transitions (oxygen activation, Fe(III)-NO dissociation (k off ), and Fe(II)-NO oxidation (k ox )) faster than wild-type BsNOS and the W66H variant, its yield of NO synthesis was the lowest in the reconstitution system with FLDR and YkuN. Nonetheless, wild-type and both Trp 66 NOS variants displayed equivalent yields of product formation under single turnover conditions. It is therefore plausible that a negative midpoint potential has dual, counterproductive effects; whereas it is beneficial for the NO synthesis reactions occurring at the heme, it becomes a thermodynamic obstacle by compromising heme reduction by the reductase partner. A similar relationship has been observed for Fe-mesoporphyrin IX-substituted nNOS whose heme midpoint potential is more negative than that of native nNOS (42). In bacterial NOSs, this thermodynamic barrier could be overcome if NOS proteins partnered with the ubiquitous ferredoxins, which are capable of reducing very lowmidpoint potential redox centers.
This scenario comprising a low heme midpoint potential (favorable for oxygen activation but thermodynamically uphill toward heme reduction), slow Fe(III)-NO dissociation, and slow Fe(II)-NO oxidation compromises the NO synthesis activity of B. subtilis NOS, which may partly explain earlier and present observations regarding the poor NO synthesis capacity of B. subtilis NOS compared with mammalian NOSs. Furthermore, because most bacterial NOSs lack a covalently attached reductase domain, it is conceivable that the steady-state NO synthesis reaction is largely uncoupled compared with mammalian NOSs.
On the grounds of our observations of how modifying the heme-thiolate environment of B. subtilis NOS affected its catalytic behavior, we propose that low midpoint potentials, slow Fe(III)-NO dissociation, and slow Fe(II)-NO oxidation are key factors underlying the poor NO synthesis output of certain bacterial NOSs. Whether this is an evolutionary trait developed to satisfy perhaps slimmer NO needs in microbial metabolism remains to be investigated. Despite an increasing body of evidence suggesting that bacterial NOSs are well fit for NO synthesis, other functions for these proteins in vivo are also plausible. The remarkable flexibility regarding cofactor usage and their obligated promiscuity in terms of reductase partners may be ultimately, an unusual advantage.
Acknowledgments-We thank Dr. Donald W. Jacobsen for providing N 5 -CH 3 -H 4 T and technical support with HPLC experiments. We also thank Dr. Zhihao Yu for assistance with size exclusion chromatography and the generation of Fig. 2, and members of the Stuehr laboratory for general technical support.