A Conserved Val to Ile Switch near the Heme Pocket of Animal and Bacterial Nitric-oxide Synthases Helps Determine Their Distinct Catalytic Profiles

Nitric oxide (NO) release from nitric oxide synthases (NOSs) is largely dependent on the dissociation of an enzyme ferric heme-NO product complex (Fe(III)NO). Although the NOS-like protein from Bacillus subtilis (bsNOS) generates Fe(III)NO from the reaction intermediate N-hydroxy-l-arginine (NOHA), its NO dissociation is about 20-fold slower than in mammalian NOSs. Crystal structures suggest that a conserved Val to Ile switch near the heme pocket of bsNOS might determine its kinetic profile. To test this we generated complementary mutations in the mouse inducible NOS oxygenase domain (iNOSoxy, V346I) and in bsNOS (I224V) and characterized the kinetics and extent of their NO synthesis from NOHA and their NO-binding kinetics. The mutations did not greatly alter binding of Arg, (6R)-tetrahydrobiopterin, or alter the electronic properties of the heme or various heme-ligand complexes. Stopped-flow spectroscopy was used to study heme transitions during single turnover NOHA reactions. I224V bsNOS displayed three heme transitions involving four species as typically occurs in wild-type NOS, the beginning ferrous enzyme, a ferrous-dioxy (Fe(II)O(2)) intermediate, Fe(III)NO, and an ending ferric enzyme. The rate of each transition was increased relative to wild-type bsNOS, with Fe(III)NO dissociation being 3.6 times faster. In V346I iNOSoxy we consecutively observed the beginning ferrous, Fe(II)O(2), a mixture of Fe(III)NO and ferric heme species, and ending ferric enzyme. The rate of each transition was decreased relative to wild-type iNOSoxy, with the Fe(III)NO dissociation being 3 times slower. An independent measure of NO binding kinetics confirmed that V346I iNOSoxy has slower NO binding and dissociation than wild-type. Citrulline production by both mutants was only slightly lower than wild-type enzymes, indicating good coupling. Our data suggest that a greater shielding of the heme pocket caused by the Val/Ile switch slows down NO synthesis and NO release in NOS, and thus identifies a structural basis for regulating these kinetic variables.

Nitric oxide synthases (NOSs) 1 are flavoheme enzymes that generate nitric oxide (NO) from L-arginine (Arg) (1,2). The overall reaction consumes 1.5 NADPH and 2 O 2 and involves two steps, the first being Arg hydroxylation to form N-hydroxy-L-Arg (NOHA), and the second being NOHA oxidation to form citrulline and NO (see Scheme 1).
The NOS heme is located in a catalytic "oxygenase" domain that also contains the binding sites for Arg and the essential redox cofactor (6R)-tetrahydrobiopterin (H 4 B) (3,4). The heme is ligated to a cysteine thiolate (5-7) and catalyzes a reductive activation of molecular oxygen in conjunction with H 4 B in each of the two steps of NO synthesis (8 -11). The NOS heme also binds self-generated NO as an intrinsic feature of catalysis (12)(13)(14). Each molecule of NO generated by NOS has a high probability of binding to the heme before it is released from the enzyme. The resulting ferric heme-NO product complex (Fe III NO) has been observed to build up as a transient species during NOHA oxidation reactions catalyzed by NOS oxygenase domains (NOSoxy) when run in a stopped-flow spectrophotometer under single turnover conditions (9,13,14). These experiments also provided spectral and kinetic information regarding the formation of an initial ferrous heme-dioxy species (Fe II O 2 ), whose disappearance then coincides with k cat and formation of the transient Fe III NO product complex (see Scheme 2). Because dissociation of the Fe III NO product complex is required for NO to escape from the enzyme, it is an essential step for the biologic functions of animal NOSs. In fact, this dissociation rate is one of three key kinetic parameters that together determine the overall catalytic behavior of a given NOS (15,16). Slower rates of NO dissociation dispose the Fe III NO product complex toward its reduction during catalysis, which then places the ferrous heme-NO enzyme species into a futile cycle that does not release NO (Fig. 1). Interestingly, there is only a modest variation in Fe III NO dissociation rates among the three mammalian NOS isoforms, which range from 2 to 5 s Ϫ1 when measured in NOHA single turnover reactions at 10°C (16).
Given the above, we were surprised to find that a NOS-like enzyme expressed in Bacillus subtilis (bsNOS) generated NO from NOHA in a single turnover reaction, but the enzyme exhibited a dissociation rate for its Fe III NO product complex that was 10 -20-fold slower than its mammalian NOS counterparts (17). The crystal structure of bsNOS (18) revealed that it contains an amino acid switch (Val to Ile) that creates greater shielding of its heme pocket compared with that in the mammalian enzymes (Fig. 2). Interestingly, the Ile is conserved among other bacterial NOS-like enzymes, whereas the use of Val at this position is conserved among animal NOSs. We hypothesized that this substitution may help to regulate the dissociation rate of the Fe III NO product complex. We therefore generated complementary mutations in mouse inducible NOSoxy (iNOSoxy, V346I) and in bsNOS (I224V), 2 and characterized the kinetics and extent of their NO synthesis from NOHA and their NO release kinetics. Our findings suggest that the difference in heme pocket shielding caused by the Val/Ile switch creates kinetic differences that impact both NO synthesis and NO release in NOS.

EXPERIMENTAL PROCEDURES
Materials-All reagents and materials were obtained from Sigma, Aldrich, Alexis, or sources described previously (19).
Mutagenesis-Site-directed mutagenesis of mouse ⌬65iNOSoxy DNA in the pCWori expression plasmid (coding for amino acids 66 -498 plus a His 6 tag at the C terminus) and bsNOS DNA in the pET15b expression plasmid (coding for amino acids 1-370 plus a six-His tag at the N terminus) 2 were performed using the QuikChange site-directed mutagenesis kit from Stratagene. The mutation codon (bold and underlined) and a silent restriction site (italic, Nae1 and XhoI for V346I and I224V mutations, respectively) was incorporated into the primers as follows: V346I, 5Ј-GCA CTG CCG GCC ATA GCC AAC ATG CTA CTG-3Ј; V346IR, 5Ј-CAG TAG CAT GTT GGC TAT GGC CGG CAG TGC-3Ј; I224V, 5Ј-GGC GTG CCA ATT GTT TCT GAT ATG AAG CTC GAG GTC GGG GG-3Ј; I224VR, 5Ј-CCC CCG ACC TCG AGC TTC ATA TCA GAA ACA ATT GGC ACG CC-3Ј. The mutations were confirmed at the molecular biology core facility of the Cleveland Clinic by sequencing ϳ500 consecutive base pairs including the mutation sites. No other mutations were observed.
Protein Expression and Purification-Wild-type and mutant enzymes were overexpressed in Escherichia coli BL21 and purified using Ni 2ϩ -nitrilotriacetate affinity chromatography as reported previously (17,20). Concentrations of NOS enzymes were determined from the 444 nm absorbance of the ferrous-CO complex, using an extinction coefficient of 76 mM Ϫ1 cm Ϫ1 (21).
Imidazole and Arg Binding Affinity Measurement-Binding affinities were measured by perturbation difference spectroscopy as reported previously (20). In general, enzymes were incubated with 200 M H 4 B and then titrated first with imidazole. Spectra were recorded at room temperature after each addition. Double reciprocal plots of the peak to trough absorbance difference versus the imidazole concentration gave the apparent binding constant of imidazole (K d ). Binding affinity of Arg was then measured in the same way except that substrate Arg was added gradually to enzyme solutions that contained H 4 B and imidazole at either 1.6 mM (iNOSoxy enzymes) or 10 mM (bsNOS enzymes). The K s value of Arg was then calculated using equation 1, K s ϭ K obs /{1 ϩ [imidazole]/K d (imidazole)}, in which K obs is the apparent binding constant determined for Arg.
Peroxide Assay-H 2 O 2 -dependent NOHA oxidation assays were performed as described previously (20). In short, enzymes were incubated at room temperature with NOHA, dithiothreitol, and different concentrations of H 4 B in 96-well plates. Reactions were initiated by adding 30 mM H 2 O 2 and stopped after 10 min by adding catalase. Griess reagent solution was then added to enable the detection of nitrite production as the absorbance change at 550 nm. Nitrite was quantitated based on NaNO 2 standard solutions.
Single Turnover NOHA Reactions-NOHA oxidation reactions were carried out in a Hi-Tech SF-61 stopped-flow apparatus equipped for anaerobic work and coupled to a Hi-Tech MG-6000 diode array detector, as reported previously (14). An anaerobic solution that contained the dithionite-reduced enzyme at concentrations indicated in the text, 40 mM Hepes, 0.5 or 0.2 mM NOHA, and 0.2 mM H 4 B was transferred into the stopped-flow instrument and rapidly mixed with air-saturated Hepes buffer at 10°C. Ninety-six spectral scans were obtained after each mixing. Sequential spectral data were fit to different reaction models using the Specfit global analysis program (provided by Hi-Tech Ltd.), which could calculate the number of different enzyme species, their spectra, and their concentrations versus time during the single turnover reactions. Data from six to eight sequential reactions were averaged to obtain the final traces.
Citrulline Analysis-Amino acids in aliquots taken from single turnover reactions were derivatized with o-phthalaldehyde and then SCHEME 1 SCHEME 2 FIG. 1. Kinetic model for NO synthesis by NOS. The enzyme molecules engage in a productive cycle that releases free NO and in a futile cycle that releases a higher oxide of nitrogen. Reduction of ferric enzyme to ferrous (k r ) enables the heme to bind O 2 and initiates both Arg hydroxylation (k cat 1) and NOHA oxidation (k cat 2). After NO is made, an immediate product of catalysis is the ferric heme-NO complex (Fe 3ϩ -NO), which can either release NO (k d ) or become reduced (k r Ј) to generate a ferrous heme-NO complex (Fe 2ϩ -NO). The ferrous heme-NO complex dissociates extremely slowly and instead regenerates the active ferric enzyme by reacting with O 2 (k ox ). Adapted from Refs. 15 and 16. analyzed by reverse-phase HPLC with fluorescence detection (19). Samples were filtered through an Amicon Centricon device (10,000 milliwatt cut-off) prior to derivatization. Samples were injected onto a Hewlett-Packard ODS-Hypersil column that was eluted with a gradient of buffer A (5 mM ammonium acetate, pH 6.0, 20% methanol) and buffer B (100% methanol). Retention times and concentrations of amino acids were calculated based on NOHA and citrulline standard solutions.
Kinetics of NO Binding to Ferric Enzymes-The procedure was as detailed previously (22). Anaerobic buffered solutions containing 2 M ferric enzyme, 400 M H 4 B, 1.2 mM dithiothreitol, and 400 M NOHA were rapidly mixed at 10°C in the stopped-flow instrument with a buffered solution containing different concentrations of NO. Six to eight sequential reactions were run at each NO concentration and then averaged to obtain the final traces. The NO solutions were made by diluting a chilled saturated NO solution in chilled anaerobic buffer and assuming a NO concentration of 2 mM for a saturated solution at 10°C.

RESULTS
Because the complementary Val and Ile mutations are located near the NOS heme, we first examined whether they altered the heme environment or the binding properties of the heme or the oxygenase domain. Table I summarizes some spectral properties and binding affinities of the purified mutant and wild-type enzymes. The maximal absorbance values for the Soret peak that we observed in the absence of Arg and H 4 B indicate that the mutants each mimicked their wild-type counterpart in that the V346I iNOSoxy had its ferric heme poised in a predominantly low spin state, whereas the I224V bsNOS had its heme poised in a predominantly high spin state. Their Soret peak positions became similar to each respective wild-type enzyme in the presence of Arg and/or H 4 B, or when dithiothre-itol, CO, or imidazole was bound to the heme as a sixth ligand. V346I iNOSoxy displayed a similar binding affinity toward imidazole as it did the wild-type iNOSoxy but had poorer affinity toward Arg. This is consistent with results obtained for an analogous mutant of neuronal NOS (V567L) that had an altered substrate recognition profile (39). The I224V bsNOS displayed a lower affinity toward imidazole but an increased affinity toward Arg. In an H 2 O 2 -driven NOHA oxidation assay the H 4 B concentration dependence of V346I iNOSoxy was somewhat enhanced. We can conclude that the mutations did not greatly alter NOS binding of Arg, H 4 B, or small heme ligands, or greatly alter the electronic properties of the heme or various heme-ligand complexes. These results are consistent with wild-type iNOSoxy and bsNOS enzymes also being mostly similar in these regards despite their containing either a Val or Ile at the same position.
We next determined how the amino acid substitutions might influence the kinetics and extent of catalysis. We utilized stopped-flow spectroscopy to study the heme transitions that were associated with catalysis of NOHA oxidation by the mutants in a single turnover reaction. Solutions of ferrous enzymes containing NOHA and H 4 B were rapidly mixed with O 2 -containing buffer in a stopped-flow spectrophotometer equipped with a rapid-scanning diode array detector, and the collected spectral data were subject to global analysis using software provided by the instrument manufacturer. We have done this type of analysis previously (14,17) for the NOHA reactions of wild-type iNOSoxy and bsNOS. In those cases the spectral data best fit to an A3 B3 C3 D model with three  consecutive monophasic transitions that together discern four spectrally distinct species. These are in order of appearance: the beginning ferrous enzyme, a ferrous-dioxy (Fe II O 2 ) intermediate, a ferric-NO (Fe III NO) intermediate, and the ending ferric enzyme.
In the case of I224V bsNOS, we observed the same three transitions involving the same four species (Fig. 3A). The Soret maxima for the Fe II O 2 and Fe III NO intermediates as calculated by global analysis were 427 and 439 nm, respectively, as compared with Soret values of 429 and 440 nm as taken from the collected absorbance traces. Fig. 3B depicts the calculated concentrations of each species during the first 150 ms of the single turnover reaction, whereas Fig. 3C depicts the calculated concentrations of the Fe III NO and ferric heme species over a longer reaction time period to indicate their complete transition. The maximal concentrations indicated for the Fe II O 2 and Fe III NO species during the reaction (Fig. 3B) imply that practically all of the mutant enzyme molecules participated in a productive reaction to generate NO. Indeed, the reaction generated 0.7 citrulline/heme (Table II), which was slightly lower than the product yield for wild-type bsNOS.
The I224V mutation altered the kinetic profile of the single turnover reaction. The calculated rates for the three heme transitions were each faster in I224V bsNOS relative to those of wild-type bsNOS (Table III). The Fe II O 2 formation rate in I224V bsNOS even exceeded that found in wild-type iNOSoxy. Thus, for I224V bsNOS the general progression and product yield of the reaction was normal, but the kinetics of each transition became faster and thus were more like iNOSoxy.
We performed an identical stopped-flow analysis of the NOHA reaction catalyzed by the complementary V346I iNOSoxy mutant. As shown in Fig. 4A, the calculated spectra were typical for the ferrous, Fe II O 2 (Soret maxima at 429 nm), and ferric heme species. However, the calculated spectrum of the Fe III NO intermediate was different. It had two Soret peaks with maxima at 412 and 440 nm. A closer examination of the visible region (Fig. 4A, inset) showed that there were absorbance peaks at 545, 584, and 645 nm in the spectrum of this species, which strongly suggest that it is a mixture of Fe III NO and ferric heme species. An inspection of the actual spectral traces that were recorded during this transition confirmed that there was no buildup of a pure Fe III NO species during the reaction (Fig. 4B). The recorded spectrum contains two Soret peaks at 413 and 440 nm, which match well with the calculated spectrum in Fig. 4A and confirm that there was a concurrent formation of a Fe III NO and a ferric heme species during this period of the NOHA reaction catalyzed by V346I iNOSoxy. A replica reaction that contained twice the concentration of NOHA gave identical spectral and kinetic results (data not shown) indicating that incomplete NOHA   binding was not a factor. Indeed, we found that 0.78 citrulline/heme was generated in the NOHA reaction (Table II). This yield is 80% of that of the wild-type reaction and indicates that catalysis was relatively well coupled to product formation in V346I iNOSoxy. Fig. 4C shows the calculated concentrations of the heme species during the NOHA reaction catalyzed by V346I iNOSoxy. In the case of the third species, as noted above, its concentration trace actually represents a mixture of Fe III and Fe III NO species. The calculated rates for the three heme transitions were all slower relative to wild-type iNOSoxy, and the rate of Fe II O 2 formation was slower than that in bsNOS (Table III). Thus, our analysis of the V346I iNOSoxy reaction suggested that the product yield was good, the general progression was altered somewhat with respect to a smaller buildup of the Fe III NO intermediate, and the kinetics of each transition became slower and more like bsNOS.
To obtain an independent estimate for the dissociation rates of the Fe III NO product complexes, we focused on the absorbance change at 650 nm versus time in each of the NOHA reactions. The absorbance gain at 650 nm during the final transition represents the buildup of the ferric enzyme species and provides an independent estimate of the dissociation rate of the Fe III NO product complex. Absorbance traces recorded during the appropriate time periods from each of the four reactions are shown in Fig. 5. They all fit well to a single exponential equation, which gave estimated dissociation rates of 0.34 Ϯ 0.01, 0.60 Ϯ 0.02, 2.09 Ϯ 0.06, and 0.49 Ϯ 0.02 s Ϫ1 , respectively, for bsNOS, I224V bsNOS, iNOSoxy, and V346I iNOSoxy. These values are of the same rank order and quan- titatively similar to the rates that were determined by global analysis (see Table III) and provide independent confirmation that the I224V mutation speeds dissociation of the Fe III NO product complex in bsNOS, whereas the V346I mutation slows dissociation of this product complex in iNOSoxy.
We next studied binding kinetics of extrinsic NO to V346I iNOSoxy alone and in comparison with wild-type iNOSoxy. Anaerobic solutions that contained a ferric enzyme, NOHA, and H 4 B were mixed in the stopped-flow spectrophotometer with anaerobic solutions that contained different concentrations of NO. Buildup of the Fe III NO product complex was monitored at 438 nm. The absorbance gain was monophasic and fit well to a single exponential function in all cases. Fig. 6 contains plots of the apparent rate constants that we observed at each NO concentration for wild-type iNOSoxy and V346I iNOSoxy. When fit to a linear function both plots gave a positive intercept consistent with NO binding being reversible. The estimated k on and k off values derived from the graphs are given in Table IV. The estimated k on and k off values that we obtained for wild-type iNOSoxy that contained NOHA plus H 4 B are in general agreement with values we derived in identical experiments for ferric iNOSoxy that contained bound Arg and H 4 B (22). The data in Table IV indicate that ferric V346I iNOSoxy has an ϳ8-fold slower k on for extrinsic NO and an ϳ11-fold slower k off for NO compared with wild-type iNOSoxy. The slower k off for NO relative to wild-type as we estimated graphically in Fig. 6 qualitatively confirms our other measures of NO release rates by V346I iNOSoxy versus wild-type. The slower k on toward extrinsic NO is also consistent with our observation that the rate of Fe II O 2 formation in V346I iNOSoxy was slower than in wild-type iNOSoxy during the NOHA single turnover reactions (see Table III).

DISCUSSION
Factors that control heme-NO dissociation are particularly important for NOS enzymes. This is because their natural product is NO, and newly generated NO molecules coordinate to the ferric heme at the end of each catalytic cycle before leaving the enzyme (12-14) (see Fig. 1). In the three mammalian NOS, the rates of Fe III NO dissociation and reduction are set so that much of the NO can escape from the enzyme before the Fe III NO product complex is reduced to ferrous, which then dooms it to an oxidative reaction that forms a higher oxide of nitrogen in place of NO (15,16). The Fe III NO dissociation rates of mammalian NOS enzymes are within a range that is typical for other ferric heme proteins, whereas in bsNOS this rate is near the lower end of the range (Table V). In contrast, the rates of ferric heme reduction in mammalian NOS enzymes range between 0.1 and 4 s Ϫ1 at 10°C, which are much slower than in other heme enzymes that contain attached flavin domains like cytochrome P450BM3 (23) or flavohemoglobin (24). We suspect that this circumstance evolved in the mammalian NOS to enable their NO release from the heme. As discussed previously (17,18), the slower NO dissociation in bsNOS may predispose it to release less NO and thereby utilize NO in ways that may be distinct from its mammalian counterparts. Indeed, NO release from intact Bacillus cells has yet to be demonstrated.
Clearly, the amino acid residues that define heme pockets or active-site channels can impede NO release from most hemeproteins (Table V). Flash photolysis experiments have been done with the neuronal NOS and endothelial NOS Fe III NO product complexes to study NO rebinding kinetics within the heme pocket (25,26). These studies report that most of the photolysed NO (Ͼ80%) undergoes very rapid recombination with the ferric heme (within picoseconds) with much of the remaining NO binding to the heme within nanoseconds. This amounts to a highly efficient geminate or  near geminate recombination within the pocket. Thus, binding NO within the heme pocket occurs much faster than the bimolecular process that takes place when the enzyme is exposed to extrinsic NO. The reported bimolecular association rate constants for NO are in the range of 10 5 -10 7 M Ϫ1 s Ϫ1 at 10 or 20°C (22,25,26). The k off values for mammalian NOS ferric-NO complexes are reported to range between ϳ2 and 50 s Ϫ1 at 10 or 20°C. The faster k off values have typically been derived from graphic analysis of observed K on values determined in either laser-flash or stopped-flow NO-binding experiments, as we have done here in Fig. 6. The slower k off values have typically been derived from equilibrium NO titration experiments (25,26,40), in ligand displacement studies (16,25,26) or in single turnover catalytic studies (13,14,22,25,26) that directly follow NO dissociation from the ferric heme as we have done here in Figs. 3-5. Thus, a newly formed NO molecule is likely to undergo multiple ferric heme binding and dissociation events within the NOS heme pocket before it escapes from the enzyme. Under such circumstances, it is not surprising that residues like iNOSoxy Val-346 and bsNOS Ile-224, which help to define the size of the NOS heme pocket exit, would also influence the macroscopic association and dissociation rates of NO for the ferric heme.
Although the addition of an extra methyl group caused by the Val to Ile substitution removes only 10 Å 3 of the heme cavity, it extends the roof at the entrance of the heme pocket ϳ1.5 Å and into contact with the carboxylate side chain of heme pyrrole ring D, thereby completely sequestering heme ligands from solvent (Fig. 2). Structural superpositions of the conserved heme pockets of bsNOS (Protein Data Bank code: 1M7V), iNOSoxy (1DWX), and eNOSoxy (1FOP) bound to Arg and NO indicate that the terminal Ile methyl will generate a van der Waals interaction with heme-bound NO. Although protein dynamics allow access to the heme iron whether the Val or Ile is present, it is not surprising that even the conservative residue substitutions at a position intimate with the heme coordination site have significant influence over the rates of NO binding and release. Our finding that the complementary Val to Ile mutations had opposite affects on NO-binding rates among the iNOSoxy and bsNOS enzymes argue strongly that the described changes in side chain volume at this particular location help to regulate NO release from NOS. In this way, our results provide a structural basis that links heme pocket geometry to the 10 -20-fold difference in NO release rate that is observed between the animal and bacterial NOS enzymes.
Our evidence suggests there was incomplete Fe III NO formation in V346I iNOSoxy despite its catalyzing an efficient NOHA single turnover reaction (0.78 NO/heme). A simple explanation is that not all of the NO is captured by the ferric heme. This would be consistent with crystallographic data showing that the Ile-224 methyl group has a steric interaction with hemebound NO in bsNOS. 3 Perhaps the same steric interaction is magnified in the iNOSoxy V346I mutant to the point of its destabilizing NO binding in a subpopulation of the enzyme molecules. This phenomenon might enable a fraction of NO to quickly escape from the heme pocket even while escape of the heme-bound NO is retarded. We can now test this interesting possibility.
The Ile/Val mutations affected other aspects of catalysis in addition to altering NO binding and the Fe III NO dissociation rates. These include changing the rates of the first two heme transitions in the NOHA reaction, namely the rate of Fe II O 2 formation and k cat . As was observed for Fe III NO dissociation, the Ile substitution slowed these two kinetic parameters, whereas the Val substitution sped them up. At this point, we can assume that all of the kinetic effects are because of the mutations either increasing or decreasing heme pocket access. Thus, one can rationalize that Ile supports slower rates of Fe II O 2 formation and extrinsic NO binding, because O 2 and NO access to the heme are similarly influenced by protein structural features in NOS (22,27,28) and in other hemeproteins (29). In contrast, the mutational effects on k cat were unexpected and cannot be so easily explained. The only documented way to slow k cat in iNOSoxy is to slow the H 4 B reduction of the Fe II O 2 intermediate during the Arg hydroxylation reaction (19), but it is not at all clear how the Val/Ile substitutions might influence that process. Perhaps the substitutions favor NOHA/heme-oxy interactions that are either more or less optimal for catalysis. Indeed, crystal structure data predict that there is a very tight fit between NOHA and O 2 in the iNOSoxy heme pocket (30) and also suggest a steric interaction between bound O 2 and Ile-224 in bsNOS. 3 In any case, the relative inability of the Ile/Val substitutions to measurably alter the spectral properties of the Fe II O 2 intermediate or the Fe III NO product complex suggests that their influence on k cat involves relatively subtle effects. This issue can be addressed in future studies.
Given the special constraint that Fe III NO formation puts on NOS catalysis, why has the enzyme not evolved to support a faster NO dissociation? This would minimize the danger of ferric heme reduction becoming too fast and would even allow NOS to support a faster rate of NO synthesis in the steady state. Although many parameters are likely to determine NO dissociation from a hemeprotein, enlarging heme pocket access is certainly one way to speed NO dissociation. But therein lies a problem, because the heme pocket has multiple functions that must remain in harmony with one another. For example, any positive effect of widening the entrance regarding NO release might be counteracted by an increase in active-site solvation, a less optimal shielding of heme-oxy catalytic intermediates, or by issues related to substrate binding within the active site. Indeed, related heme-thiolate oxygenase enzymes like cytochrome P450 are thought to require a relatively shielded distal pocket to perform their oxygen activation chemistry and catalysis (31,32). On the other hand, the available sequence data indicate that closing down the heme pocket (relative to mammalian NOS) must confer some selective advantage to the bacterial NOS-like enzymes. In fact, they contain another conserved residue switch (Ser to His-134 in bsNOS) that helps to further close down their heme pocket (18). But a problem associated with minimizing NO release in this way is that one must still accommodate entry of Arg and O 2 into the heme pocket. In fact, our data indicate that the Val to Ile substitution in iNOSoxy presents a kinetic barrier for NO and O 2 to access the heme, apparently mimicking what occurs for O 2 binding in bsNOS. This brings up related concerns about the biological O 2 tension under which each NOS has evolved to operate. There likely is a range of useful heme entryway sizes, and we suspect that those in the mammalian and bacterial NOSs are set according to their required functions and the environment under which they must operate. Beyond the structure-function insights, our mutants suggest a means to examine how changing the NO release rate of a given NOS might impact its biological function in the host organism.