Radiolabeled neuronal nitric oxide synthase inhibitors: synthesis, in vivo evaluation, and primate PET studies.

UNLABELLED: The objectives of this study were to synthesize neuronal nitric oxide synthase (NOS-I)-selective imaging agents based on the 2 potent, selective inhibitors AR-R 17443 [N-(4-((2-((phenylmethyl) (methyl)-amino)ethyl)phenyl)-2-thiophenecarboximidamide)] and AR-R 18512 [(N(2-methyl-1,2,3,4-tetrahydroisoquinoline-7-yl)-2-thiophenecarboxim idamide)] in positron-emitting form and to evaluate regional brain uptake in rodents and primates. METHODS: [11C]AR-R 17443 and [11C]AR-R 18512 were produced by N-alkylation of the corresponding desmethyl precursors using [11C]iodomethane. Regional brain uptake of [11C]AR-R 17443 and [11C]AR-R 18512 was assayed in rats and NOS-I knockout mice, and PET was performed in baboons. Tracer kinetic modeling used a 2-compartment plasma and brain tissue model. RESULTS: Yields of [11C]AR-R 17443 and [11C]AR-R 18512 ranged from 8% to 16% at the end of synthesis, with specific activities of 50-178 GBq/micromol (1,350-4,800 Ci/mmol) at the end of synthesis. In rat cerebellum and cortex at 30 min after injection, [11C]AR-R 17443 showed 1.01 +/- 0.01 and 1.63 +/- 0.12 percentage injected dose per gram (%ID/g) uptake, respectively, whereas [11C]AR-R 18512 showed 0.88 +/- 0.01 and 1.30 +/- 0.07 %ID/g uptake, respectively. Attempts to block tracer uptake by pretreatment with the NOS-I-selective inhibitor 7-nitroindazole or the corresponding unlabeled inhibitor (or desmethyl precursor to AR-R 17443 of similar potency) were unsuccessful. A small but significant (20%) decrease in cerebellar uptake of [11C]AR-R 18512 was present in NOS-I knockout mice compared with control mice. PET of [11C]AR-R 18512 in baboons with concurrent regional cerebral blood flow (rCBF) determination before and after administration of blocker showed dose-related decreases in cerebellar uptake that were greater than or equal to decreases in rCBF. Plasma metabolites accounted for 27% of total activity at 30 min after injection. Kinetic modeling of binding potentials revealed a distribution volume of 334 in cerebral blood that dropped 51% after blocker administration. CONCLUSION: Rodent studies for [11C]AR-R 17443 and [11C]AR-R 18512 showed little evidence of specific NOS-I binding. In baboons, we detected a higher uptake of [11C]AR-R 18512 in the cerebellum than in the cortex (approximately 5%, accounting for decreased rCBF because of blockade), indicating minimal specific binding. Analogs of higher affinity are likely required if this class of agents is to prove viable for PET.

Discovery and development of neuronal nitric oxide synthase inhibitors.

The role of neuronally derived nitric oxide (NO) in neurotransmission and neural injury remains an area of active investigation. NO generation has been postulated to be involved in the deleterious events surrounding ischemia/reperfusion injury either directly or via the production of more reactive oxidants such as peroxynitrite. In our search for novel therapeutics for the treatment of a variety of neurological diseases including stroke, we have discovered novel, potent, and selective inhibitors of the neuronal nitric oxide synthase (nNOS) isoform. These compounds have proven to be effective in models of ischemia/reperfusion supporting the role of nNOS in these processes. The effects of these compounds as well as additional aspects critical to their development will be presented.

N-nitro-L-arginine and N-monomethyl-L-arginine exhibit a different pattern of inactivation toward the three nitric oxide synthases.

The ability of NG-nitro-L-arginine (NNA) and NG-methyl-L-arginine (NMMA) to inactivate native neuronal, endothelial cell, and macrophage nitric oxide synthases (nNOS, eNOS, and iNOS, respectively) was investigated. Each NOS isozyme (plus cofactors) was preincubated with either NNA or NMMA and then assayed for remaining activity by measuring the conversion of labeled L-arginine to labeled L-citrulline. Consistent with previous reports (Olken, N. M., et al., Biochem. Biophys. Res. Commun. 177, 828-833, 1991), NMMA was a mechanism-based irreversible inhibitor of iNOS, exhibiting time- and concentration-dependent inactivation of iNOS with a KI equal to 2.6 microM and a kinact equal to 0.042 min-1. When assayed without a preincubation period, NMMA exhibited typical reversible inhibition of iNOS (Ki = 3.9 microM). NMMA also reversibly inhibited nNOS and the eNOS with Ki equal to 0.65 and 0.7 microM, respectively. However, NMMA did not inactivate eNOS at concentrations up to 10 microM. In the presence, but not the absence, of 4 microM tetrahydrobiopterin, NMMA inactivated nNOS with a kinact equal to 0.022 min-1 and a KI equal to 2.0 microM. Since NNA did not inactivate iNOS at concentrations up to 25 microM, NNA is strictly a reversible inhibitor of iNOS (Ki = 8.1 microM). Neuronal NOS and eNOS, however, were rapidly inactivated by NNA with kintact equal to 0.083 and 0.047 min-1 and KI equal to 0.09 and 0.02 microM, respectively, when preincubated with NNA. Tetrahydrobiopterin did not affect the rate of inactivation of nNOS by NNA. In all cases, L-arginine protected against inactivation, suggesting that inactivation occurs at or near the active site. Thus, inactivation of the three NOS isozymes with NMMA and NNA reveals active-site differences between the isoforms.

Delayed production of nitric oxide contributes to NMDA-mediated neuronal damage.

Exposure of primary murine cortical neuron cultures to N-methyl-D-aspartate (NMDA) resulted in neuronal death as evidenced by release of lactate dehydrogenase (LDH) into the media. The addition of N-nitro-L-arginine (N-Arg) protected the neurons from death in a concentration-dependent manner when added after the NMDA, but not when the N-Arg was present with the NMDA. Protection by N-Arg was lost if L-arginine containing media was added to the cultures prior to the addition of the N-Arg. Treatment of the neurons with kainate prior to NMDA reduced subsequent NMDA-induced damage which was not prevented with N-Arg. These results suggest that delayed production of nitric oxide (NO) contributes to NMDA-induced neuronal damage in culture.

In vitro loading of apoferritin.

This study compared the effect of loading apoferritin either with ferrous ammonium sulfate in various buffers or with ceruloplasmin and chelated ferrous iron. It was shown that loading of apoferritin with ferrous ammonium sulfate was dependent on buffer and pH, and was directly related to the rate of iron autoxidation. The ceruloplasmin-dependent loading of apoferritin, however, was unaffected by these factors. Isoelectric focusing and amino acid analysis of the differently loaded ferritins showed that ferrous ammonium sulfate loading of apoferritin resulted in the depletion of the basic amino acids, lysine and histidine, probably as a result of protein oxidation. No significant differences in amino acid composition was noted for ceruloplasmin-loaded ferritin. Furthermore, ferritin loaded with ferrous ammonium sulfate released more iron than either native or ceruloplasmin-loaded ferritin when either paraquat or EDTA was used as an iron mobilizing agent. We suggest that the loading of apoferritin with ferrous ammonium sulfate occurred as a result of iron autoxidation and may result in oxidation of amino acids and loss of integrity of the protein, and that ceruloplasmin may act as a catalyst for the incorporation of iron into apoferritin in a manner more closely related to that occurring in vivo.

Ferritin as a source of iron for oxidative damage.

The generation of deleterious activated oxygen species capable of damaging DNA, lipids, and proteins requires a catalyst such as iron. Once released, ferritin iron is capable of catalyzing these reactions. Thus, agents that promote iron release may lead to increased oxidative damage. The superoxide anion formed enzymatically, radiolytically, via metal-catalyzed oxidations, or by redox cycling xenobiotics reductively mobilizes ferritin iron and promotes oxidative damage. In addition, a growing list of compounds capable of undergoing single electron oxidation/reduction reactions exemplified by paraquat, adriamycin, and alloxan have been reported to release iron from ferritin. Because the rapid removal of iron from ferritin requires reduction of the iron core, it is not surprising that the reduction potential of a compound is a primary factor that determines whether a compound will mobilize ferritin iron. The reduction potential does not, however, predict the rate of iron release. Therefore, ferritin-dependent oxidative damage may be involved in the pathogenesis of diseases where increased superoxide formation occurs and the toxicity of chemicals that increase superoxide production or have an adequate reduction potential to mobilize ferritin iron.

Effects of ceruloplasmin on superoxide-dependent iron release from ferritin and lipid peroxidation.

Ceruloplasmin (CP) effectively inhibited superoxide and ferritin-dependent peroxidation of phospholipid liposomes, using xanthine oxidase or gamma irradiation of water as sources of superoxide. In addition, CP inhibited superoxide-dependent mobilization of iron from ferritin, suggesting that CP inhibited lipid peroxidation by decreasing the availability of iron from ferritin. CP also exhibited some superoxide scavenging activity as evidenced by its inhibition of superoxide-dependent cytochrome c reduction. However, superoxide scavenging by CP did not quantitatively account for its inhibitory effects on iron release. The effects of CP on iron-catalyzed lipid peroxidation in systems containing exogenously added ferrous iron was also investigated. CP exhibited prooxidant and antioxidant effects; CP stimulated at lower concentrations, reached a maximum, and inhibited at higher concentrations. However, the addition of apoferritin inhibited CP and Fe(II)-catalyzed lipid peroxidation at all concentrations of CP. In addition, CP catalyzed the incorporation of Fe(II) into apoferritin. Collectively these data suggest that CP inhibits superoxide and ferritin-dependent lipid peroxidation via its ability to incorporate reductively-mobilized iron into ferritin.

Nitric oxide mediates iron release from ferritin.

Nitric oxide (NO) synthesis by cytotoxic activated macrophages has been postulated to result in a progressive loss of iron from tumor target cells as well as inhibition of mitochondrial respiration and DNA synthesis. In the present study, the addition of an NO-generating agent, sodium nitroprusside, to the iron storage protein ferritin resulted in the release of iron from ferritin and the released iron-catalyzed lipid peroxidation. Hemoglobin, which binds NO, and superoxide anion, which reacts with NO, inhibited nitroprusside-dependent iron release from ferritin, thereby providing evidence that NO can mobilize iron from ferritin. These results suggest that NO generation in vivo could lead to the mobilization of iron from ferritin disrupting intracellular iron homeostasis and increasing the level of reactive oxygen species.

Vanadate-dependent NAD(P)H oxidation by microsomal enzymes.

Vanadate-dependent NAD(P)H oxidation, catalyzed by rat liver microsomes and microsomal NADPH-cytochrome P450 reductase (P450 reductase) and NADH-cytochrome b5 reductase (b5 reductase), was investigated. These enzymes and intact microsomes catalyzed NAD(P)H oxidation in the presence of either ortho- or polyvanadate. Antibody to P450 reductase inhibited orthovanadate-dependent NADPH oxidation catalyzed by either purified P450 reductase or rat liver microsomes and had no effect on the rates of NADH oxidation catalyzed by b5 reductase. NADPH-cytochrome P450 reductase catalyzed orthovanadate-dependent NADPH oxidation five times faster than NADH-cytochrome b5 reductase catalyzed NADH oxidation. Orthovanadate-dependent oxidation of either NADPH or NADH, catalyzed by purified reductases or rat liver microsomes, occurred in an anaerobic system, which indicated that superoxide is not an obligate intermediate in this process. Superoxide dismutase (SOD) inhibited orthovanadate, but not polyvanadate-mediated, enzyme-dependent NAD(P)H oxidation. SOD also inhibited when pyridine nucleotide oxidation was conducted anaerobically, suggesting that SOD inhibits vanadate-dependent NAD(P)H oxidation by a mechanism independent of scavenging of O2-.

Alloxan- and glutathione-dependent ferritin iron release and lipid peroxidation.

The diabetogenic action of alloxan is believed to involve oxygen free radicals and iron. Incubation of glutathione (GSH) and alloxan with rat liver ferritin resulted in release of ferrous iron as assayed by spectrophotometric detection of ferrous-bathophenanthroline complex formation. Neither GSH nor alloxan alone mediated iron release from ferritin. Superoxide dismutase (SOD) and catalase did not affect initial rates of iron release whereas ceruloplasmin was an effective inhibitor of iron release. The reaction of GSH with alloxan resulted in the formation of the alloxan radical which was detected by ESR spectroscopy and by following the increase in absorbance at 310nm. In both instances, the addition of ferritin resulted in diminished alloxan radical detection. Incubation of GSH, alloxan, and ferritin with phospholipid liposomes also resulted in lipid peroxidation. Lipid peroxidation did not occur in the absence of ferritin. The rates of lipid peroxidation were not affected by the addition of SOD or catalase, but were inhibited by ceruloplasmin. These results suggest that the alloxan radical releases iron from ferritin and indicates that ferritin iron may be involved in alloxan-promoted lipid peroxidation.

Inhibition of superoxide and ferritin-dependent lipid peroxidation by ceruloplasmin.

Ceruloplasmin (CP) was found to inhibit xanthine oxidase and ferritin-dependent peroxidation of phospholipid liposomes, as evidenced by decreased malondialdehyde formation. Ceruloplasmin was also shown to inhibit superoxide-mediated mobilization of iron from ferritin, in a concentration-dependent manner, as measured spectrophotometrically using the iron(II) chelator bathophenanthroline sulfonate. Ceruloplasmin failed to function as a peroxyl radical-scavenging antioxidant as evidenced by its inability to inhibit free radical-initiated peroxidation of linoleic acid, suggesting that CP inhibited lipid peroxidation by affecting the availability of ferritin-derived iron. In addition, CP scavenged xanthine oxidase-derived superoxide as measured spectrophotometrically via its effect on cytochrome c reduction. However, the extent of the superoxide scavenging of CP did not quantitatively account for its effects on iron release, suggesting that CP inhibits superoxide-dependent mobilization of ferritin iron independently of its ability to scavenge superoxide. The effects of CP and apoferritin on iron-catalyzed lipid peroxidation in systems containing exogenously added ferrous iron was also investigated. In the absence of apoferritin, CP exhibited a concentration-dependent prooxidant effect. However, CP-dependent, iron-catalyzed lipid peroxidation was inhibited by the addition of apoferritin. Apoferritin did not function as a peroxyl radical-scavenging antioxidant but was shown to incorporate iron in the presence of CP. These data suggest that CP inhibits superoxide and ferritin-dependent lipid peroxidation largely via its ability to reincorporate reductively mobilized iron back into ferritin.

Iron release from ferritin and lipid peroxidation by radiolytically generated reducing radicals.

Iron is involved in the formation of oxidants capable of damaging membranes, protein, and DNA. Using 137Cs gamma radiation, we investigated the release of iron from ferritin and concomitant lipid peroxidation by radiolytically generated reducing radicals, superoxide and the carbon dioxide anion radical. Both radicals released iron from ferritin with similar efficiencies and iron mobilization from ferritin required an iron chelator. Radiolytically generated superoxide anion resulted in peroxidation of phospholipid liposomes as measured by malondialdehyde formation only when ferritin was included as an iron source and the released iron was found to be chelated by the phospholipid liposomes.

Effect of diquat on the distribution of iron in rat liver.

Diquat toxicity is proposed to be mediated through the generation of active oxygen species; however, the exact role of active oxygen in toxicity is not known. The generation of damaging oxygen radicals requires transition metals such as iron. In vitro studies have shown that redox cycling of diquat results in the release of iron from ferritin, thus, increasing the potential for active oxygen species generation. We sought to determine if diquat administration to male Sprague-Dawley rats would result in the release of iron from ferritin in vivo. Rats were treated with diquat dibromide (20 mg/kg body weight) and the effect on the iron distribution in liver was determined. The results show that diquat-treated animals had increased levels of hepatic low molecular weight chelatable iron (LMWC-Fe) and decreased levels of hepatic ferritin iron when compared to saline-treated animals. These results suggest that diquat toxicity may be associated with the release of iron from ferritin in vivo and that iron release from ferritin may be a process common to other free radical mediated toxicities.