N 5-(1-Imino-3-butenyl)-l-ornithine

Nitric oxide synthase (NOS) catalyzes the NADPH- and O2-dependent conversion ofl-arginine to nitric oxide (NO) and citrulline; three isoforms, the neuronal (nNOS), endothelial, and inducible, have been identified. Because overproduction of NO is known to contribute to several pathophysiological conditions, NOS inhibitors are of interest as potential therapeutic agents. Inhibitors that are potent, mechanism-based, and relatively selective for the NOS isoform causing pathology are of particular interest. In the present studies we report that vinyl-l-NIO (N 5-(1-imino-3-butenyl)-l-ornithine;l-VNIO) binds to and inhibits nNOS in competition with l-arginine (K i = 100 nm); binding is accompanied by a type I optical difference spectrum consistent with binding near the heme cofactor without interaction as a sixth axial heme ligand. Such binding is fully reversible. However, in the presence of NADPH and O2,l-VNIO irreversibly inactivates nNOS (k inact = 0.078 min−1;K I = 90 nm); inactivation is Ca2+/calmodulin-dependent. The cytochromec reduction activity of the enzyme is not affected by such treatment, but the l-arginine-independent NADPH oxidase activity of nNOS is lost in parallel with the overall activity. Spectral analyses establish that the nNOS heme cofactor is lost or modified by l-VNIO-mediated mechanism-based inactivation of the enzyme. The inducible isoform of NOS is not inactivated byl-VNIO, and the endothelial isoform requires 20-fold higher concentrations to attain ∼75% of the rate of inactivation seen with nNOS. Among the NOS inactivating l-arginine derivatives,l-VNIO is the most potent and nNOS-selective reported to date.

Nitric oxide synthase (NOS) 1 catalyzes the oxidation of Larginine to nitric oxide (NO) and citrulline; NADPH and O 2 are cosubstrates (1)(2)(3). Three major isoforms of NOS have been identified to date. The neuronal (nNOS) and endothelial (eNOS) isoforms are constitutively expressed and are regulated by Ca 2ϩ /calmodulin, whereas the activity of the inducible isoform (iNOS) is controlled transcriptionally and is not affected by changes in intracellular Ca 2ϩ . Although amino acid sequence homology among the isoforms is limited (ϳ50%) (3), all are comprised of a C-terminal reductase domain that binds NADPH and the cofactors FAD and FMN and a N-terminal oxygenase domain that binds L-arginine and the heme and tetrahydrobiopterin (BH 4 ) cofactors (1)(2)(3). The reductase domain is related in function and amino acid sequence to cytochrome P450 reductase (4), whereas the oxygenase domain is related in function (but not sequence) to the cytochromes P450. Binding of Ca 2ϩ /calmodulin to a region between the domains permits electron flow from the reductase domain to the oxygenase domain and also stimulates electron flow within the reductase domain (5). The resulting reduction of the heme cofactor allows O 2 to be activated, permitting the cytochrome P450like oxidation of L-arginine to N -hydroxy-L-arginine and the subsequent further oxidation of that tightly bound intermediate to citrulline and NO (1,2,6).
Nitric oxide synthase-derived NO is important in many physiological processes including blood pressure homeostasis (7)(8)(9), neurotransmission (10,11), and immune function (12), but overproduction of NO can have pathological consequences (13). For example, excess NO resulting from overexpression of iNOS in response to endotoxin or inflammatory cytokines is a major contributor to the vascular disregulation seen in septic shock (14,15) and in patients receiving interleukin-2-based immunotherapy (16,17). Inappropriate activation of nNOS is implicated in chronic visceral pain (18,19), in migraine headache (20), and in several neurodegenerative diseases (e.g. Parkinson's disease (21,22)), and is thought to contribute to postischemic reperfusion injury in stroke (23).
Appreciation of the pathological roles of NOS-derived NO has stimulated interest in the design and synthesis of NOS inhibitors for possible therapeutic use in disorders associated with overproduction of NO (24,25). To be pharmacologically useful, compounds should be well transported into the target tissue, strongly inhibitory, NOS isoform-selective, and chemically stable under biological conditions. In attempting to meet these goals, L-arginine analogs are particularly attractive inhibitor candidates because they are effectively transported by the ubiquitous system y ϩ amino acid transporter (26,27) and thus show good activity in vivo. N -Methyl-L-arginine (L-NMA), the prototypic NOS inhibitor (28), has been shown, for example, to reverse the hypotension of septic shock (14,15,29) and cytokine-induced shock (17,30,31) in animal studies and in early clinical trials. Unfortunately, NMA and several other L-arginine analogs including N -amino-L-arginine (32)(33)(34) and N -(1-iminoethyl)-L-ornithine (L-NIO) (35,36) show little isoform selectivity; their pharmacological use may thus cause undesirable inhibition of physiological processes controlled by nontargeted NOS isoforms. We (37,38) and others (39) have shown that the S-alkyl-L-thiocitrullines show modest selectivity (up to 50-fold) for nNOS over eNOS and iNOS and have proposed that these compounds may be of use in treating dis-* This work was supported in part by National Institutes of Health grant DK48423. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In the present studies, we have examined several novel Larginine antagonists and find that vinyl-L-NIO (N 5 -(1-imino-3butenyl)-L-ornithine, L-VNIO) (Fig. 1) is a potent, mechanismbased inhibitor that attacks the heme cofactor of NOS. It shows a marked selectivity for nNOS. Abstracts describing this work have been published (40,41).

Materials
Most biochemicals and reagents for organic syntheses were obtained from Sigma and Aldrich, respectively. N -Methyl-L-arginine and BH 4 were purchased from Chemical Dynamics (Plainfield, NJ) and Alexis (La Jolla, CA), respectively. L-[ 14 C]Arginine was from NEN Life Science Products. N -Alkyl-L-arginines (42) and L-NIO (43) were prepared by the general methods indicated. Rat nNOS was isolated from stably transfected kidney 293 cells (44) as described previously (45). Bovine eNOS (46) and mouse iNOS expressed in Escherichia coli 2 were generous gifts from Dr. Kirkwood Pritchard (Department of Pathology, Medical College of Wisconsin, Milwaukee, WI) and Dr. Bettie S. S. Masters (Department of Biochemistry, University of Texas Health Sciences Center, San Antonio, TX), respectively. 1 H and 13 C NMR spectra were obtained using a Bruker AC 300 MHz spectrometer. FAB mass spectral analyses were generously carried out by Dr. Frank Laib at the Department of Chemistry, University of Wisconsin, Milwaukee.

General Procedure for Preparation of N 5 -(1-Iminoalkyl)ornithines and N 5 -(1-Iminoalkenyl)ornithines
D-or L-Ornithine HCl (1.68 g, 10 mmol) and cupric acetate (0.5 g, 10 mmol) were dissolved in water (20 ml) and stirred for 10 min at room temperature. The solution was then filtered to remove minor impurities and cooled to 0°C, and the pH was adjusted to 9.5 by addition of cold 10% NaOH. Alkyl or alkenyl imidate (15 mmol), prepared separately from the corresponding nitrile and HCl (g) (43) was then added, and the mixture was allowed to stir at pH 9.0 -9.5 for 1 h at 0°C and for 2 h at room temperature. The pH was then adjusted to 7.4 with cold dilute HCl, and the mixture was stirred at room temperature overnight. Hydrogen sulfide gas (caution: toxic) was bubbled through the solution, and the resulting copper sulfide precipitate was removed by filtration through charcoal. The filtrate was passed through Chelex to remove any residual Cu 2ϩ , and the clear solution was evaporated to dryness by rotary evaporation under reduced pressure. The residue was washed with ethyl acetate, and the product was crystallized from ethanol to give pure N 5 -(1-iminoalkyl)ornithine or N 5 -(1-iminoalkenyl)ornithine.   Optical Difference Spectroscopy-Studies were carried out on a Shimadzu model 2501 dual beam UV-visible spectrophotometer using either nNOS as isolated (ϳ20% low spin heme) or nNOS pretreated with imidazole (100% low spin heme). In the former case, 0.5 ml of nNOS as isolated (432 g) in 50 mM Tris-HCl buffer, pH 7.5, 10% glycerol, and 0.1 mM EDTA was placed in the sample and reference cuvettes at 15°C, and the base-line spectrum was adjusted to zero. Sequential samples of buffer and L-VNIO in buffer were added to the reference and sample cuvettes, respectively, and optical difference spectra at increasing concentrations of L-VNIO were obtained. Similar studies using imidazole in place of L-VNIO were carried out to determine the K s for that ligand (K s Imid ϭ 86.2 M, data not shown). The effect of L-VNIO on imidazole liganded nNOS was then determined by initially adding 1.0 mM imidazole to the cuvettes, setting the base line to zero, and then adding sequential samples of buffer and L-VNIO in buffer to the reference and sample cuvettes, respectively, as described above. For these studies K s L-VNIO was calculated from the relationship K s(app) The buffer in the sample cuvette was then saturated with CO, and the difference spectrum between 400 and 500 nm was determined (48).

Inhibition of NOS Isoforms by L-VNIO and Related
Compounds-All NOS isoforms are inhibited by a variety of Larginine analogs that compete with L-arginine for the amino acid binding site; we have reported previously that N -alkyl-Larginines with n-alkyl substituents up to 4 carbons and the isosteric N 5 -(1-iminoalkyl)-L-ornithines are good to excellent inhibitors (24). Consistent with these findings, in initial rate studies L-VNIO was found to be a potent inhibitor of nNOS, and its binding was competitive with L-arginine (Fig. 2). A replot of the data shows that K i L-VNIO is ϳ100 nM (Fig. 2, inset); this value is substantially lower than the K m for L-arginine (1.4 M). Similar studies showed L-VNIO also inhibits eNOS and iNOS competitively with respect to L-arginine, but the K i values are much higher (i.e. 12.0 M for eNOS and 60 M for iNOS) ( Table  I). Note that the K i L-VNIO /K m L-Arg ratios for nNOS, eNOS, and iNOS are 0.07, 3.33 and 4.80, respectively, indicating that L-VNIO competes for the L-arginine binding site of nNOS particularly well (Table I). D-Arginine is neither a substrate nor an inhibitor of NOS (1), and none of the NOS isoforms was inhib-ited by D-VNIO when tested at 100 M in the presence of 20 M L-arginine (data not shown).
The isoform selectivity exhibited by L-VNIO is mirrored to a degree by structurally related L-arginine analogs (Fig. 1). As shown in Table I, methyl-L-NIO and ethyl-L-NIO (the saturated analog of L-VNIO) are also more potent inhibitors of nNOS than of eNOS or iNOS. However, expressed on a K i /K m basis, neither of these inhibitors shows biologically significant selectivity for nNOS over eNOS; they, in fact, show a very modest selectivity (3-5-fold) for iNOS over the constitutive isoforms. The prototypic N 5 -(1-iminoalkyl)-L-ornithine inhibitor, L-NIO, also shows biologically insignificant isoform selectivity (Table I).
Effect of L-VNIO Binding on the Heme Spectrum of nNOS-A variety of studies indicate that the reactive guanidinium nitrogen of substrate L-arginine is bound near the iron of the NOS heme cofactor; N -hydroxy-L-arginine is then formed when that guanidinium nitrogen is oxidized by O 2 , which has been bound and activated as a sixth, axial heme iron ligand (1,2). Although L-arginine does not bind near enough to heme iron to act as a sixth axial ligand (49), other inhibitors such as Lthiocitrulline do covalently interact with heme iron (24,50,51). Such interactions can be revealed by optical difference spectroscopy in which perturbations of the heme spectrum caused by substrates or inhibitors are determined (49). As shown in Fig. 3A, increasing concentrations of L-VNIO, when added to solutions of native nNOS, cause a type I difference spectrum. This result, which is similar to that seen with L-arginine, indicates that L-VNIO does not interact covalently with heme iron but does bind sufficiently close to its sixth axial position to displace an endogenous heme ligand (the identity of the ligand displaced is presently unknown). The displacement of the endogenous ligand, which is present in ϳ20% of nNOS as isolated (49), is responsible for the spectral change shown in Fig. 3A. Fig. 3B is a replot of the data in Fig. 3A showing that the L-VNIO dissociation constant, K s L-VNIO , is 1.1 M; the previously reported value for K s L-Arg is 2.5 M (49). We also determined K s L-VNIO using imidazole-saturated nNOS. These studies confirmed that L-VNIO gives a type I optical difference spectrum and provide a potentially more accurate estimate of K s L-VNIO since the signal is larger (i.e. heme iron is initially 100% low spin) and the amounts of L-VNIO added are larger, making it unnecessary to correct for L-VNIO bound to nNOS. With imidazole saturated nNOS, K s L-VNIO , calculated as described under "Methods," was 1.4 M (data not shown). Note that K s L-VNIO is a simple binding constant measured in the absence of NADPH whereas K i L-VNIO (Table I) is determined under conditions of substrate turnover. The two values need not agree (52).
Inactivation of nNOS by L-VNIO-The rate measurements used to construct Fig. 2 were obtained immediately after initiation of the enzymatic reaction. If the reaction mixtures were  The K i and K m values shown are averages of at least duplicate determinations made under initial rate conditions using the oxyhemoglobin assay as described under "Methods;" duplicate measurements agreed within Ϯ5%. monitored for several minutes, the progress curves were clearly concave downward indicating occurrence of a progressive irreversible inhibition of nNOS. Such inactivation was examined directly by incubating nNOS with various concentrations of L-VNIO in the presence of NADPH and monitoring the reaction mixtures for residual nNOS activity at intervals (Fig. 4A). As shown, L-VNIO, but not D-VNIO, caused a first-order inactivation of nNOS. There was no evidence of nNOS reactivation in these studies; product formation was constant over the time of the assay. In separate studies, passage of reaction mixtures containing L-VNIO-inactivated nNOS through small gel-filtration columns did not restore activity (not shown). A replot of the data in Fig. 4A shows that the rate constant for inactivation, k inact , is 0.078 min Ϫ1 and K I , the equilibrium binding constant of L-VNIO to nNOS prior to inactivation, is 90 nM (Fig. 4B). The latter value is similar to the K i measured under turnover conditions. Inactivation of nNOS by L-VNIO is dependent on NADPH, O 2 , and Ca 2ϩ /calmodulin, but is independent of added flavins (Table II). As shown, loss of activity in the absence of NADPH, O 2 , or Ca 2ϩ /calmodulin is no greater than seen in the absence of L-VNIO. The moderate loss of activity (5-9%) seen in the absence of these substrates or cofactors reflects the instability of nNOS under the conditions of this experiment. The effects of BH 4 on inactivation are complex (Table II; experiments 9 -12). Omission of BH 4 from otherwise complete reaction mixtures containing L-VNIO results in an apparent increase in inhibition (73.4 versus 56.9%), but the difference is attributable mainly to the nNOS destabilizing effect of removing BH 4 from the reaction mixtures. Thus nNOS loses 9.2 and 37.7% of its activity when incubated without L-VNIO in the presence and absence of BH 4 , respectively. The loss attributable to L-VNIO is ϳ48% in the presence of BH 4 (i.e. 56.9 -9.2%) and ϳ35% in its absence (i.e. 73.4 Ϫ 37.9%); omitting BH 4 presumably causes a greater autoinactivation of nNOS resulting in less of the activity loss being attributable to L-VNIO. Inclusion of higher levels of BH 4 in the reaction mixtures (250 M versus 50 M; experiment 11) decreased L-VNIO-mediated inactivation significantly to ϳ21% (i.e. 30.3 Ϫ 9.2%), suggesting that BH 4 also offers some direct protection from inactivation. That even high levels of BH 4 can not fully protect the enzyme is evident from experiment 12, which shows that a 5-fold higher level of L-VNIO (1.  (24,35,36). In Fig. 5, inactivation of nNOS by these compounds is compared with that seen with L-VNIO. As shown, inactivation by 0.5 M L-VNIO is comparable to that seen with 2.0 M L-NMA and greater than that seen with 10 M L-NIO. As was seen with L-VNIO, inhibition by L-NIO is NADPHdependent; nNOS inactivation by L-NMA has previously been shown to be NADPH-dependent (53,54). Interestingly, the next higher homolog of L-NIO, methyl-L-NIO, is a much weaker inactivator of nNOS, requiring a 10-fold higher concentration (100 M) to duplicate the inactivation seen with L-NIO. Further extension of the iminoalkyl group to form the saturated analog of L-VNIO (i.e. ethyl-L-NIO) results in a compound that does not inactivate nNOS under the conditions examined (Fig. 5). Note that ethyl-L-NIO does bind competitively with L-arginine and thereby inhibits nNOS (Table I); such inhibition does not, however, progress to inactivation.

28), and by L-NIO
Effect of L-VNIO on Other Reactions Catalyzed by nNOS-In addition to NO synthesis, nNOS catalyzes the NADPH-dependent reduction of cytochrome c, an activity attributed to the reductase domain (1,2). It also catalyzes the O 2 -and Ca 2ϩ / calmodulin-dependent, L-arginine-independent oxidation of NADPH, an activity requiring both the reductase and the oxygenase nNOS domains and resulting in the formation of superoxide (1,2). As shown in Table III (experiments 1 and 2), L-VNIO has no effect on cytochrome c reduction but potently inhibits NADPH oxidation. That cytochrome c reduction is due to direct transfer of electrons to it from the reductase domain and not to formation of superoxide by the oxygenase domain is evident from experiments 3 and 4 showing that superoxide dismutase has no effect on the rate of cytochrome c reduction in the presence or absence of L-VNIO. In the absence of Ca 2ϩ / calmodulin the rate of electron flow within the reductase domain is greatly diminished and electrons do not flow from the reductase domain to the heme cofactor of the oxygenase domain (5). As shown in experiments 5-8, omission of Ca 2ϩ /calmodulin from the reaction mixtures decreases cytochrome c reduction ϳ 96% and nearly eliminates NADPH oxidation. There is again no effect of L-VNIO and superoxide dismutase on cytochrome c reduction, confirming that this reductase domain specific reaction is not affected. The minimal residual NADPH oxidase activity seen in the absence of Ca 2ϩ /calmodulin (experiment 5) is apparently unrelated to that seen in the presence of Ca 2ϩ / calmodulin (experiment 1) since it is not inhibited by L-VNIO; it may reflect a very slow, normally undetectable formation of superoxide by the flavins of the reductase domain.

Inactivation of nNOS by L-VNIO Results in Heme
Loss-Brief treatment of nNOS with dithionite or with NADPH in the presence of Ca 2ϩ /calmodulin reduces the heme cofactor from the Fe 3ϩ to the Fe 2ϩ state. With heme reduced, nNOS binds CO with a characteristic increase in absorption at 443 nm that is useful in quantitating bound, functional heme (1, 2). As shown in Table IV, treatment of nNOS with L-VNIO, but not D-VNIO, causes loss of the heme cofactor as judged by CO binding studies. Loss of heme closely parallels the loss of overall NOS activity indicating that loss of heme functionality or heme binding to the enzyme can account for virtually all mechanismbased inactivation.
Ability of L-VNIO to Inactivate eNOS and iNOS-As shown in Fig. 6, L-VNIO is a relatively weak inactivator of eNOS and it does not inactivate iNOS. Inactivation of eNOS is NADPHdependent and follows first order kinetics, but even with a 20-fold higher concentration of L-VNIO the rate of inactivation of eNOS does not match that seen with nNOS (compare the rate of eNOS inactivation using 10 M L-VNIO to that seen with nNOS using 0.5 M L-VNIO). Detectable inactivation of iNOS did not occur even with 100 M L-VNIO.

DISCUSSION
Studies from this laboratory and others establish that all NOS isoforms bind a variety of L-arginine and L-homoarginine analogs. Initial binding is in all cases competitive with L-arginine, and all of the analogs inhibit all NOS isoforms if added in sufficiently high concentration (1,24). Three general mechanisms of inhibition and/or inactivation have been distinguished. Some analogs such as L-canavanine (55), N -cyclopropyl-L-arginine (56), N -nitro-L-arginine with iNOS (57) and N 5 -(1-iminobutyl)-L-ornithine with nNOS (this work) bind to and dissociate from NOS rapidly but do not react covalently; they are classic competitive inhibitors and do not cause inactivation. L-Thiocitrulline is also in this class, although it reversibly interacts to a limited degree (ϳ20%) as a covalent sixth axial ligand to the heme cofactor (50). "Slow on-slow off" inhibitors comprise the second type of NOS inhibitors and are exemplified by the slow binding and very slow dissociation of N -nitro-L-arginine from nNOS and eNOS (58,59). These inhibitors, which include the S-alkyl-L-thiocitrullines (38,39), are not altered by the enzyme, but their binding is apparently accompanied by a conformational change in NOS that is only slowly reversible.
Mechanism-based inhibitors (k cat inhibitors) constitute the third type of NOS-inhibiting amino acids. Of these, the best characterized is L-NMA, which has been shown to be hydroxylated by nNOS and iNOS as a pseudosubstrate; that product, N -hydroxy-N -methyl-L-arginine, undergoes further NOSmediated reaction to form CH 3 NO ϩ , formaldehyde, NO, and citrulline by complex mechanisms (53,54). A presently unidentified intermediate in this latter transformation apparently causes damage to and loss of the NOS heme cofactor, but the product formed is not yet identified (60). N -Allyl-L-arginine (56) and N -amino-L-arginine (34) have also been previously shown to cause mechanism-based, irreversible inactivation of NOS, and there is a brief report that L-NIO (35) (this work) causes similar inactivation. Inactivation by these compounds has not yet been mechanistically characterized. Although Namino-L-arginine causes somewhat more rapid inactivation of nNOS (k inact(max) ϭ 0.35 min Ϫ1 ) (34), neither it nor the other previously reported mechanism-based arginine analog inactivators shows significant NOS isoform selectivity.
The present studies establish that L-VNIO is a potent, mechanism-based NOS inactivator with substantial nNOS selectivity. Thus, initial binding of L-VNIO to nNOS is followed by NADPH-and O 2 -dependent mechanism-based inactivation of the overall reaction and of the oxygenase domain-dependent NADPH oxidase reactions. The reductase domain-specific cytochrome c reduction activity in the presence or absence of Ca 2ϩ /calmodulin is not lost. Consistent with damage to the oxygenase domain, inactivation correlates with loss of the heme cofactor as judged by CO difference spectroscopy. Isoform selectivity derives both from differences in initial binding (i.e. the K i values for eNOS and iNOS are 120-and 600-fold higher than that for nNOS) and from differences in subsequent enzyme-mediated activation of L-VNIO to a reactive derivative (i.e. iNOS is not subject to mechanism-based inactivation).
The precise mechanism by which L-VNIO is transformed by nNOS (and presumably by eNOS, albeit more slowly) into an inactivating intermediate is not yet known. Previous studies establish that NOS-mediated metabolism of L-arginine and L-NMA proceeds via cytochrome P450-like hydroxylation and oxidation reactions (1,2); in the case of L-NMA, carbon-centered free-radical intermediates are formed (54). We designed L-VNIO with the intent of developing an arginine analog that would place an allyl moiety, known to be highly susceptible to free radical activation, in close proximity to the NOS heme cofactor. The observations that initial binding is competitive with L-arginine and that steady-state K i values are lower than or comparable to the K m for L-arginine support the view that L-VNIO binds to nNOS as an L-arginine analog. The observation that a type I difference spectrum accompanies L-VNIO binding to imidazole liganded or native nNOS indicates that a portion of the inhibitor binds closely enough to the heme cofactor to displace (but not replace) imidazole or the endogenous sixth axial heme ligand present in nNOS as isolated. Because the binding site for the non-reactive guanidinium nitrogen of  Portions (25 l) were removed immediately after addition of nNOS and also after 10 min of incubation at 25°C; residual nNOS activity was determined using the oxyhemoglobin assay as described under "Methods." Percent inactivation was calculated by comparison of activity observed after 10 min to the average activity of all samples (except experiment 4; see footnote a) removed immediately (0.63 Ϯ 0.05 nmol/min). All values shown reflect means Ϯ S.D. for triplicate determinations. Control experiments showed that the small amount of L-VNIO carried over into the assay mixtures used to determine residual activity did not cause measurable inhibition. Expt  L-arginine (i.e. the nitrogen not near the heme iron and thus not hydroxylated or converted to NO) does not accomodate groups larger than ϭNH or -NH 2 (1), 3 we know further that it is the allyl moiety of L-VNIO that must be near heme. Several specific mechanisms for nNOS-mediated L-VNIO activation thus seem plausible. Activated oxygen bound to heme may epoxidize the allyl moiety, forming N 5 -(1-imino-3,4-epoxybutyl)-L-ornithine; this species could then react with protein or heme nucleophiles in a manner that destroys heme functionality or binding. Alternatively, by a process similar to that described for L-NMA (54), the allyl moiety of L-VNIO may lose H ⅐ to form a reactive allyl radical that damages or derivatizes the heme cofactor. Finally, we have considered the possibility that -electrons of the allyl moiety attack the highly electrophilic perferryl heme species that normally hydroxylates L-arginine to form N -hydroxy-L-arginine (1, 2); this reaction would generate a highly reactive carbocation at the terminus of L-VNIO. Although additional studies are necessary to confirm which of these or related mechanisms is(are) operative, we note that all of the proposed mechanisms depend critically on the unsaturation of the L-VNIO side chain. The mechanistic proposals are thus consistent with our observation that the saturated analog, ethyl-L-NIO, does not inactivate nNOS.
In summary, the present studies establish L-VNIO as the first neuronal isoform selective, mechanism-based amino acid inactivator of NOS. Our studies show further that selectivity can be achieved both in initial binding and in subsequent enzyme-mediated activation. The latter observation is of particular interest. Whereas all NOS isoforms catalyze the same overall reaction for L-arginine, our results indicate the isoforms can differ qualitatively as well as quantitatively in their ability to metabolize and thereby activate L-arginine analogs. There is thus the possibility of designing mechanism-based inhibitors that will have extremely high levels of isoform selectivity.  ϪCa 2ϩ /calmodulin ϩ L-VNIO 64.6 Ϯ 3 1.15 Ϯ 0. 34 7 ϪCa 2ϩ /calmodulin ϩ superoxide dismutase 65.5 Ϯ 0.3 ND 8 ϪCa 2ϩ /calmodulin ϩ L-VNIO ϩ superoxide dismutase 66.2 Ϯ 1.4 ND a ND, not determined.