Introduction

Top Abstract Introduction Comparison of nos active... Role of h4b Electron transfer Conclusions References

The nitric oxide synthase (NOS)¹ isoforms consist of a heme domain linked to a flavoprotein by a CaM-binding peptide. The flavoprotein domain exhibits strong sequence and cofactor resemblance to cytochrome P450 (P450) reductase, but the heme domain has virtually no structural similarity to P450 other than the fact that a thiolate is coordinated to the heme iron atom. Nevertheless, the heme domain is similar to P450 in terms of spectroscopic, biochemical, and catalytic properties, and much of our understanding of the function of NOS is based on our comparatively advanced understanding of the structure and mechanism of P450. A discussion of NOS, a member of the heme-thiolate family of proteins that includes P450, is therefore appropriate within the context of a P450 symposium.

NOS catalyzes the oxidation of L-Arg to ^.NO and citrulline (Stuehr, 1997; Marletta, 1988; Knowles and Moncada, 1994). Three major NOS isoforms have been identified: NOS-I (nNOS), a form initially associated with the brain (Bredt and Snyder, 1990); NOS-II (iNOS), a form most closely associated with macrophages (Xie et al., 1992); and NOS-III (eNOS), an isoform that is localized in epithelial cells (Pollock et al., 1991). All three isoforms require heme, FAD, FMN, H4B, NADPH, O₂, and CaM to be functional (Stuehr, 1997; Marletta, 1988; Knowles and Moncada, 1994; Marletta, 1993; Masters, 1994). All three are also homodimers of a polypeptide in which the heme- and H4B-binding domain is linked via a consensus CaM-binding sequence to a flavoprotein with binding sites for one FAD, one FMN, and NADPH (McMillan et al., 1992; White and Marletta, 1992; Klatt et al., 1992; Bredt et al., 1991; Tayeh and Marletta, 1989; Kwon et al., 1989). The isoforms differ, however, in their tissue localization, regulation, and function. nNOS is longer than the other isoforms due to the presence of a PDZ region at the amino terminus that is involved in subcellular targeting of the protein (Brenman et al., 1995), and eNOS is distinguished by the presence of myristoylation and palmitoylation sites at the amino terminus that serve a similar function (Busconi and Michel, 1993; Garcia-Cardeña et al., 1996). Furthermore, nNOS and eNOS are constitutive enzymes that are regulated by the Ca²⁺-dependent binding of CaM to the CaM-binding sequence. Their activity is thus physiologically controlled by local changes in the Ca²⁺ concentration (Moncada and Higgs, 1993; Nathan and Xie, 1994). In contrast, iNOS binds CaM in essentially a Ca²⁺-independent, irreversible manner, and its activity is transcriptionally regulated by cytokines rather than by changes in the Ca²⁺ concentration (Moncada and Higgs, 1993; Nathan and Xie, 1994).

The three NOS isoforms catalyze the same two-step reaction sequence and appear to have the same catalytic mechanism. They oxidize L-Arg to the stable intermediate N-hydroxy-L-arginine, and subsequently oxidize this intermediate to NO and citrulline. Both steps in this sequence are NADPH- and O₂-dependent (Stuehr, 1997; Marletta, 1988; Knowles and Moncada, 1994). The electrons required for the reaction flow from NADPH to the FAD, then to the FMN, and finally to the heme iron atom. The flow of electrons from the FMN to the heme iron is gated by the binding of CaM to the CaM-binding sequence and is therefore the locus of the Ca²⁺-dependent regulation of the activities of nNOS and eNOS (Abou-Soud and Stuehr, 1993). Of course, the electron flow is permanently turned on in iNOS because CaM is bound to that isoform in an essentially irreversible manner (Cho et al., 1992).

	Comparison of NOS Active Site Structures

Top Abstract Introduction Comparison of nos active... Role of h4b Electron transfer Conclusions References

In order to carry out structural and mechanistic studies, we have developed systems for expression of the NOS isoforms in Escherichia coli and have expressed and purified the recombinant proteins (Gerber and Ortiz de Montellano, 1995; Rodriguez-Crespo et al., 1996; Gerber et al., 1997a and 1997b). All of these proteins, including rat nNOS, bovine and human eNOS, and murine and human iNOS, are obtained in a correctly folded, heme-bound state. Successful expression of iNOS requires co-expression of a gene coding for CaM, as the protein does not fold correctly and is inactive in the absence of CaM co-expression (Gerber et al., 1997b; Fossetta et al., 1996). CaM co-expression is not required for the expression of nNOS or eNOS, but the yield and quality, at least of eNOS, is higher when it is co-expressed with CaM (Rodriguez-Crespo and Ortiz de Montellano, 1996). The recombinant proteins are soluble and are obtained free of H4B because there is no H4B, and no myristoylation or palmitoylation, in E coli.

We have employed aryldiazenes as topological probes to explore the active sites of the NOS isoforms in much the same way that we previously used them to characterize the topologies of cytochrome P450 enzymes (Ortiz de Montellano, 1995). Aryldiazenes react with the heme group of hemoproteins to form stable -bonded aryl-iron complexes (Fe-Ar). The absorption maxima of these aryl-iron complexes when the iron is also thiolate ligated, as in the P450 and NOS enzymes, is at approximately 480 nm. Formation and decay of the aryl-iron complexes is therefore readily monitored by spectroscopic methods. Earlier studies with the P450 enzymes demonstrated that in situ oxidation of the aryl-iron complexes with ferricyanide causes the aryl group to migrate from the iron to one of the four nitrogens of the heme (Ortiz de Montellano, 1995). This migration only occurs within the intact active site if the iron is thiolate-ligated. The four possible N-phenylprotoporphyrin IX isomers thus formed can be demetallated and individually quantitated by high-performance liquid chromatography (Swanson and Ortiz de Montellano, 1991). Earlier studies with the crystalline P450 enzymes show that the N-arylporphyrin isomer ratio is primarily determined by the degree to which each of the four porphyrin nitrogens is sterically protected by active site residues. Topological information is thus provided by the N-aryl porphyrin regioisomer ratios, using different aryldiazene probes; by the rates of aryl-iron complex formation and decay; and by the changes caused by cofactors, substrates, and inhibitors in these rates and isomer ratios (Ortiz de Montellano, 1991).

The three NOS isoforms react with phenyldiazene (PhN==NH) to give phenyl-iron complexes (Gerber and Ortiz de Montellano, 1995; Gerber et al., 1997b). In general, the binding of CaM stimulates the phenyldiazene reaction, whereas the binding of both L-Arg and H4B inhibits it, as expected if the biopterin cofactor and the substrate obstruct the entry channel or the heme site itself. The phenyldiazene reaction is slower with eNOS than with the other two isoforms both in the presence and absence of L-Arg or H4B, which suggests that the eNOS active site is smaller than those of the other two isoforms. This conclusion is supported by the finding that 2-naphthyldiazene forms a 2-napthyl-iron complex with nNOS and iNOS, but not eNOS, and p-biphenyldiazene only forms the p-biphenyl-iron complex with nNOS. The minimum height above the iron required to erect the phenyl complex is ~5.8 Å; the 2-naphthyl complex, 7.1 Å; and the p-biphenyl complex, ~9.9 Å. The ceiling height directly over the iron is therefore between ~6 and 10Å, with eNOS closer to the lower and nNOS to the higher limit.

Shift of the phenyl group from the iron to the porphyrin nitrogens gives, with all three NOS isoforms, mixtures of the four N-phenylprotoporphyrin IX regioisomers in which the isomer with the N-phenyl on pyrrole ring D predominates (Gerber and Ortiz de Montellano, 1995; Gerber et al., 1997b). The regioisomer ratios for the CaM-bound proteins without H4B or L-Arg are (N_B:N_A:N_C:N_D, where the subscript indicates the N-phenyl pyrrole ring): nNOS, 11:12:03:74; eNOS, 04:04:11:81; and iNOS, 10:17:08:65 (Gerber et al., 1997b). Addition of H4B decreases the proportion of the ND regioisomer to 41%, 47%, and 40%, respectively, for nNOS, eNOS, and iNOS. These results indicate that, in all three isoforms, the most open region of the active site in the absence of L-Arg and H4B is above pyrrole ring D. Furthermore, they indicate that H4B binds close to pyrrole ring D and thereby decreases both the rate of the reaction with phenyldiazene and the extent to which the phenyl in the preformed complex shifts to the nitrogen of pyrrole ring D (fig. 1). These structural inferences are confirmed by the recent crystal structure of the L-Arg- and H4B-containing iNOS heme domain, which shows that H4B binds in the substrate access channel close to pyrrole ring D of the heme (Crane et al., 1998). A similar active site geometry is observed in the crystal structure of eNOS (T. Poulos, personal communication, 1998).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Schematic representation of the binding of H4B relative to the NOS heme group based on the N-aryl shift data. The pyrrole rings of the heme are labeled. The iNOS crystal structure coordinates indicate that H4B is bound in the indicated quadrant of the heme site but not quite as directly over pyrrole ring D.

	Role of H4B

Top Abstract Introduction Comparison of nos active... Role of h4b Electron transfer Conclusions References

The reason for the absolute catalytic requirement of NOS for H4B remains unclear. Many of the functions ascribed to H4B can be satisfied by dihydrobiopterin, a cofactor that binds to NOS without forming a catalytically active enzyme. The binding of H4B to H4B-free NOS causes a shift of the heme iron atom from the low- to the high-spin state (Rodriguez-Crespo et al., 1996), but a similar shift is observed with dihydrobiopterin (Presta et al., 1998). H4B promotes the dimerization of iNOS (Baek et al., 1993) and stabilizes the nNOS dimer (Klatt et al., 1995), but it is not required for the dimerization of eNOS (Rodriguez-Crespo et al., 1996; Rodriguez-Crespo and Ortiz de Montellano, 1996). Furthermore, dihydrobiopterin is able, at least for iNOS, to promote dimerization without conveying catalytic activity to the dimer (Presta et al., 1998). Dihydrobiopterin, like H4B, facilitates electron transfer from the flavoprotein domain to the iron to give the ferrous enzyme (Presta et al., 1998). Indeed, reduction of the ferric to the ferrous state can be observed in the complete absence of any biopterin cofactor (fig. 2). Thus neither the allosteric effect of H4B on the iron coordination state, nor its dimer-stabilizing properties, nor its effect on reduction of the ferric to the ferrous enzyme, accounts for the absolute catalytic requirement for H4B. These results are consistent with the view that H4B plays a critical redox role not met by dihydrobiopterin, but efforts to identify a redox role for H4B have not been successful. However, a recent low temperature study implies a redox role for the biopterin cofactor and provides a possible explanation for the previous failures to detect such a role (Bec et al., 1998). At 30^oC, Bec et al. observed a spectrum that they assigned to the nNOS ferrous dioxy complex. This spectrum, in the presence of H4B and L-Arg, was converted to a new species with an absorbance maximum at 428 nm. This reaction was much faster at 30^oC than oxidation of the flavin groups, which was only observed spectroscopically as the temperature was raised. Furthermore, this sequence of reaction steps resulted in the production of N-hydroxy-L-arginine. The reaction sequence is not observed when dihydrobiopterin is used instead of H4B. These results led the authors to propose that H4B provides the electron required to activate the ferrous dioxygen complex and that the biopterin radical thus formed is rapidly reduced, under normal turnover conditions, by electron transfer from the flavin groups (fig. 3). There are ambiguities in this study, notably the following: (a) the ferrous dioxy spectrum does not agree with that reported earlier by stopped-flow studies (Abu-Soud et al., 1997), (b) the spectroscopically detected intermediate is not observed with a heme domain dimer that lacks the flavin groups, and (c) artifacts can be introduced at low temperature that are not pertinent to turnover at physiological temperatures. Nevertheless, the low temperature studies provide a paradigm that may explain the unique requirement for H4B in the catalytic function of NOS.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Formation of the ferrous-CO complex with an absorption maximum at 450 nm in an incubation of H4B-free eNOS with NADPH under an atmosphere of CO. eNOS obtained by expression in E coli was used without any exogenously added biopterin. The broad spectrum is that of the protein prior to the addition of NADPH.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Reaction scheme proposed by Bec et al. (1998) invoking a role for H4B in the first hydroxylation of L-Arg by the NOS enzymes. The iron in brackets represents the NOS heme iron atom. The reductase represents the flavins in the enzyme

	Electron Transfer

Top Abstract Introduction Comparison of nos active... Role of h4b Electron transfer Conclusions References

The maximum rates of NO-synthesis by the three CaM-bound NOS isoforms differ, the activities of nNOS and iNOS being comparable (500-1500 nmol·nmol¹·min¹) but severalfold higher than that for eNOS (100-200 nmol·nmol¹·min^-1) (table 1) (Nishida and Ortiz de Montellano, 1998). The intrinsic ability of the flavin domain to deliver electrons can be independently evaluated by measuring the rate at which it reduces cytochrome c, an alternative electron acceptor (Klatt et al., 1992). All three isoforms reduce cytochrome c, but in the absence of Ca²⁺/CaM the activities of the constitutive nNOS and eNOS isoforms are much lower than that of the permanently CaM-bound iNOS (table 1). When CaM binds to the constitutive isoforms, their ability to reduce cytochrome c is greatly increased (Klatt et al., 1992; Heinzel et al., 1992). However, whereas the enhanced cytochrome c reductase activity of CaM-bound nNOS is comparable to that of iNOS, that of CaM-bound eNOS remains tenfold lower (table 1). It is to be noted that the rates of cytochrome c reduction in all cases are much higher than the corresponding rates of NO synthesis.

To examine the link between the intrinsic reductase activity of the flavin domain and the overall activity of the NOS isoforms, we constructed and characterized chimeras in which the isoform flavin domains were interchanged. All six of the possible chimeras of this type have been assembled and examined, but the results have only been published for two of them: E/N, in which the eNOS heme and CaM domains are fused to the nNOS reductase domain, and I/N, in which the iNOS heme and CaM domains are fused to the neuronal reductase domain (Nishida and Ortiz de Montellano, 1998). Comparison of the NO-synthesizing and cytochrome c-reducing activities of the chimeras with those of the parent wild-type enzymes indicate that the intrinsic activity of the reductase is a major determinant of the overall activity of the enzyme. This is well-illustrated by the E/N chimera, which has a CaM-bound cytochrome c reductase activity of ~6000 min¹, a value similar to that of CaM-bound nNOS (~7000 min¹) and iNOS (~6000 min¹) but much higher than that of CaM-bound eNOS (~700 min¹) (table 1). This result indicates that the activity of the nNOS reductase domain is not attenuated when it is placed in the context of the eNOS heme and CaM-binding domains. More importantly, the NO-synthesizing activity of the E/N chimera is fourfold higher than that of wild-type eNOS and approaches the activities of wild-type nNOS and iNOS (table 1). One clear conclusion from this study is that the ability of the reductase domain to deliver electrons to the heme is a major limiting step in the overall activity of the enzyme. Thus the activity of the chimeras parallels that of the protein that provides the reductase domain (fig. 4).

View larger version (46K): [in this window] [in a new window]   Fig. 4.   The NO-synthesizing and cytochrome c-reducing activities of three wild-type NOS isoforms and three NOS chimeras. N, I, and E indicate, respectively, nNOS, iNOS, and eNOS. The open bars in the cytochrome c reductase panel indicate activities in the absence of added CaM. Closed and hatched bars indicate, respectively, the activities in the presence of CaM of the wild-type and chimeric proteins. Lines are shown that connect the forms of the enzyme having the same reductase domain. eNOS and the N/E chimera, in which the eNOS reductase replaces the reductase of nNOS, have relatively low activities in both assays. In contrast, nNOS, iNOS, and chimeras with the nNOS reductase domain, including E/N, have high activities.

Analysis of the effects of L-Arg and H4B on the rate of NADPH consumption by the three wild-type isoforms and the two chimeras establishes, furthermore, that modulation of the enzyme activity by the substrate and cofactor are exclusively mediated through interactions with the heme domain. Thus the consumption of NADPH is greatly stimulated by H4B and L-Arg in iNOS but not nNOS or eNOS (fig. 5). This is also observed with the I/N but not E/N chimera. The consumption of NADPH by H4B- and L-Arg-free eNOS is slightly stimulated by the addition of L-Arg and slightly depressed by the addition of H4B, in a manner similar to the activity observed when both the cofactor and substrate are present (Nishida and Ortiz de Montellano, 1998). This same pattern, which differs from those observed with nNOS and iNOS, is observed for the E/N chimera. Analysis of the dimerization properties of the chimeras shows that this property also resembles that of the parent heme domain. The effects of the substrate and cofactor on NADPH consumption are a composite of their effects on the rate of electron transfer to the heme, the rate of autooxidation of the ferrous dioxy complex, and the amount of NADPH consumed by uncoupled reduction of molecular oxygen.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Heme-domain specificity of the effects of L-Arg and H4B on the catalytic activity of NOS.

A third important feature that is affected by exchanging the reductase domains is the insensitivity of the activity of iNOS to the Ca²⁺-concentration. The activity of wild-type iNOS remains essentially constant as the concentration of the Ca²⁺-chelating agent EGTA is increased from 0 to 2.5 mM. However, the activity of the I/N chimera is no longer insensitive to the presence of the chelating agent and exhibits a decreased NO synthesizing activity in the presence of 100 mM EGTA, although it is still active at the highest concentration of EGTA examined (fig. 6) (Nishida and Ortiz de Montellano, 1998). Thus the protein contacts that govern the tight association of CaM to iNOS and the apparent independence of CaM-bound iNOS to the Ca²⁺-concentration involve, at least partially, residues of the flavin domain. This finding agrees with the conclusions of Ruan and coworkers, based on studies of nNOS and iNOS chimeras, that sequences of iNOS in addition to those in the consensus CaM-binding sequence are required in order to produce the Ca²⁺-independence of the enzyme (Ruan et al., 1996).

View larger version (16K): [in this window] [in a new window]   Fig. 6.   EDTA-dependence of the activity of wild-type iNOS and the I/N chimera.

We have further explored very recently the role of flavin domain residues in controlling the Ca²⁺/CaM-dependent activation of eNOS. The peptide insert identified by Salerno et al. (Salerno et al., 1997) in the eNOS FMN domain has been deleted from the cDNA and the protein has been expressed and purified. A comparison of the wild-type and insert-deleted proteins shows that removal of the insert (a) decreases the concentration of Ca²⁺ required to activate the protein in the presence of CaM and (b) elevates the total activity of the reductase domain as measured by its ability to reduce cytochrome c (C. Nishida and P. Ortiz de Montellano, unpublished results, 1998).

	Conclusions

Top Abstract Introduction Comparison of nos active... Role of h4b Electron transfer Conclusions References

Electron transfer, and control of the rate of electron transfer, are critical in determining the activities of the NOS isoforms. The NOS isoforms resemble P450 enzymes in this, as in many other, respects because electron transfer to the ferrous dioxy complex is also a key rate-determining step in the P450 systems. Indeed, H4B may be essential for the formation of NO because it functions as a rapidbut transientsource of electrons in the activation of oxygen by the NOS enzymes. Although the absolute rate of catalytic turnover is determined by the ability of the reductase domain to provide electrons, the effects of H4B and L-Arg on electron transfer are determined exclusively by interactions of the substrate and cofactor with the heme domain of the protein. Control of electron transfer from the flavin to the heme domain, and within the flavin and heme domains, is a complex but critical aspect of the function of the NOS enzymes.