Nitric oxide scavenging by the cobalamin precursor cobinamide.

Nitric oxide (NO) is an important signaling molecule, and a number of NO synthesis inhibitors and scavengers have been developed to allow study of NO functions and to reduce excess NO levels in disease states. We showed previously that cobinamide, a cobalamin (vitamin B12) precursor, binds NO with high affinity, and we now evaluated the potential of cobinamide as a NO scavenger in biologic systems. We found that cobinamide reversed NO-stimulated fluid secretion in Drosophila Malpighian tubules, both when applied in the form of a NO donor and when produced intracellularly by nitricoxide synthase. Moreover, feeding flies cobinamide markedly attenuated subsequent NO-induced increases in tubular fluid secretion. Cobinamide was taken up efficiently by cultured rodent cells and prevented NO-induced phosphorylation of the vasodilator-stimulated phosphoprotein VASP both when NO was provided to the cells and when NO was generated intracellularly. Cobinamide appeared to act via scavenging NO because it reduced nitrite and nitrate concentrations in both the fly and mammalian cell systems, and it did not interfere with cGMP-induced phosphorylation of VASP. In rodent and human cells, cobinamide exhibited toxicity at concentrations > or =50 microM with toxicity completely prevented by providing equimolar amounts of cobalamin. Combining cobalamin with cobinamide had no effect on the ability of cobinamide to scavenge NO. Cobinamide did not inhibit the in vitro activity of either of the two mammalian cobalamin-dependent enzymes, methionine synthase or methylmalonyl-coenzyme A mutase; however, it did inhibit the in vivo activities of the enzymes in the absence, but not presence, of cobalamin, suggesting that cobinamide toxicity was secondary to interference with cobalamin metabolism. As part of these studies, we developed a facile method for producing and purifying cobinamide. We conclude that cobinamide is an effective intra- and extracellular NO scavenger whose modest toxicity can be eliminated by cobalamin.

Nitric oxide (NO) 1 has multiple cellular functions including regulation of cell growth, differentiation, and apoptosis, and many physiological roles including modulation of blood pressure, platelet aggregation, and synaptic plasticity (1)(2)(3). Studies of NO functions have been aided by pharmacologic agents that raise or lower NO levels (4). A number of disease states including sepsis and hepatic failure are characterized by abnormally high NO production, and removing the excess NO could have salutary effects (5)(6)(7).
One approach to lower NO concentrations is to reduce NO synthesis. Four nitric-oxide synthase (NOS) isoforms are present in mammals: neuronal NOS (nNOS or NOS I), inducible NOS (iNOS or NOS II), endothelial NOS (eNOS or NOS III), and a recently described mitochondrial NOS (mtNOS) (8,9). nNOS and eNOS are constitutively expressed in many tissues and produce pico to nanomolar concentrations of NO in response to increased intracellular calcium (10). iNOS is expressed in a variety of cell types, and its level can be induced many-fold by endotoxin (lipopolysaccharide, LPS) and tumor necrosis factor-␣, with iNOS producing almost 1,000 times higher NO, i.e. nano to micromolar, concentrations than nNOS and eNOS (7,(11)(12)(13). Less is known about mtNOS, but it accounts for 50% of cellular NO production in rat liver, and low micromolar NO concentrations have been found in rat heart mitochondria (9,14). A large number of NOS inhibitors have been generated, most of which are arginine analogs including isothiourea derivatives, and because of the multiple biochemical roles of arginine, these agents can have affects other than NOS inhibition (15)(16)(17).
Another approach to reduce NO levels is to use a NO scavenger. For example, the heme moiety of hemoglobin binds NO with great avidity, but heme and free extracellular hemoglobin can be highly toxic, particularly in whole animals (18). Thus, other NO scavengers have been considered including dithiocarbamate derivatives that chelate iron and thus bind NO, but these too can have adverse effects (19,20). Cobalamin (vitamin B 12 ) is structurally similar to heme and also binds NO, but with considerably less efficiency than heme (21). We found recently that the cobalamin precursor cobinamide, which lacks the dimethylbenzimidazole ribonucleotide tail of cobalamin ( Fig. 1), has a more than 100 times greater affinity for NO than cobalamin; moreover, each cobinamide molecule can potentially neutralize two NO molecules compared with only one NO molecule for cobalamin (22). We now present studies demonstrating the use of cobinamide as a NO scavenger, both in a Drosophila model and in cultured mammalian cells. Cobinamide became * 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.
‡ These authors contributed equally to this work. § Supported by Institutional National Research Service Award 5T32 CA81211. toxic to cells at concentrations considerably higher than would be needed to neutralize NO produced in vivo, and the toxicity was reversed fully by cobalamin; cobalamin did not affect NO scavenging by cobinamide, and thus the combination of these two corrinoids could be a very effective method to scavenge NO.

EXPERIMENTAL PROCEDURES
Production and Analysis of Cobinamide-Approximately 200 mg of hydroxocobalamin (OH-Cbl, Sigma) was dissolved in 1 ml of concentrated HCl and heated at 65°C for 8 min to hydrolyze the phosphoester bond linking the dimethylbenzimidazole ribonucleotide moiety to the corrin ring ( Fig. 1 and Ref. 23). The solution was cooled on ice and applied to a l-ml C18 solid phase extraction column (Fisher Scientific). The HCl and unreacted OH-Cbl were removed by batch elution in water and 10% acetone, respectively, and the diaquocobinamide (cobinamide) product was eluted in 20% acetone and concentrated under reduced pressure in a SpeedVac (Savant Industries). Purity of the cobinamide was confirmed both spectrophotometrically by comparison with published spectra (22,24,25) and by high performance liquid chromatography (HPLC) using a C18 reversed phase column eluted isocratically in 100 mM NaH 2 PO 4 , pH 4.0, and 15% methanol (v/v) (25,26); the column effluent was monitored at multiple wavelengths by a diode array detector. The yield was generally Ͼ80%, or ϳ160 mg of cobinamide was produced per batch. When stored at Ϫ20°C, the cobinamide was spectroscopically and biologically stable for at least 1 month.
Assessment of Malpighian Tubule Secretion in Drosophila melanogaster-An elegant in vitro method has been devised to measure fluid transport by the Malpighian tubules of D. melanogaster; tubular secretion is stimulated markedly by NO donors and LPS, the latter via induction of the Drosophila NOS gene (27,28). Briefly, two pairs of Malpighian tubules, each with a ureter, were dissected from 10 wild type Oregon R adult flies anesthetized on ice. The tubules were mounted in liquid paraffin with the distal end of one tubule bathed in a 10-l droplet of Schneider's Insect medium. Fluid secretion rates were determined at room temperature by measuring the size of drops formed at the end of the ureter every 10 min. Basal fluid secretion rates were measured for three 10-min intervals prior to adding 10 M Deta-NONOate (Cayman Chemical Co.) or 1 M LPS (Sigma) to the droplet of Schneider's medium; after three more 10-min intervals, 10 M cobinamide, with or without 10 M OH-Cbl, was added to some of the tubules for a further 30 min. Each data point represents the mean from at least 20 pairs of tubules analyzed on three separate occasions.
In some experiments, flies were grown for 48 h on food supplemented with 250 M cobinamide prior to measuring the rates of tubular secretion. The supplemented food was generated by liquifying standard fly food paste by heating to ϳ40°C and after adding cobinamide, allowing the food to cool to room temperature. Malpighian tubules were dissected from the flies, and rates of tubular secretion were measured in the absence and presence of 1 M LPS as described above.
Measurement of Cobinamide Uptake into Mammalian Cells-To study cobinamide uptake into mammalian cells, we radioactively labeled cobinamide with 14 (29), but to be sure no [ 14 C]KCN remained, we lowered the pH to 6 to form HCN (pK a of KCN is 9.2) and then bubbled argon through the solution to remove any [ 14 C]HCN. The specific activity of the [ 14 C]dicyanocobinamide product was 104 mCi/mmol (using a molar extinction coefficient of 2.8 ϫ 10 4 at 348 nm) (25).
In the uptake studies, ϳ1 ϫ 10 6 baby hamster kidney (BHK) cells cultured in Dulbecco's modified Eagles' medium (DMEM) containing 10% fetal bovine serum (FBS) were grown to subconfluence in 6-well cluster dishes. The cells were incubated at 37°C for 5 min with [ 14 C]dicyanocobinamide at concentrations ranging from 100 nM to 100 M. At the end of the incubation, the cells were washed rapidly four times with ice-cold phosphate-buffered saline, harvested with a rubber policeman, and collected by centrifugation for 15 s at 10,000 ϫ g. The cell pellet was dissolved in 500 l of 0.1 N NaOH, and radioactivity in sample aliquots was measured by liquid scintillation counting. Cobinamide uptake was linear between 2 and 15 min of incubation and from 0.5 to 2 ϫ 10 6 cells; counts obtained in 0-time samples were Ͻ10% of those obtained in the 2-min samples.
Assessment of VASP Phosphorylation-Vasodilator-stimulated phosphoprotein (VASP) is expressed in a wide variety of cells (30); it is phosphorylated in response to NO and other vasodilators such as cGMP, with phosphorylation conveniently quantified by Western blotting because it retards the gel mobility of the protein (31). Approximately 1 ϫ 10 6 rat C6 glioma or CS-54 vascular smooth muscle cells in 12-well cluster dishes were transfected with 50 ng of an expression vector for VSV epitope-tagged VASP as described previously; C6 cells additionally received 25 ng of cGMP-dependent protein kinase (G-kinase) I␣ expression vector (31,32). The cells were cultured for 36 h in DMEM supplemented with 10% FBS and were treated during the last 30 min with the indicated concentrations of the NO donor PAPA-NONOate (Cayman Chemical Co.) or the membrane-permeable cGMP analog 8-pCPT-cGMP (Biolog, Inc.) in the case of C6 cells or the calcium ionophore A23187 (Calbiochem) in the case of CS-54 cells. Some of the cultures received cobinamide or human hemoglobin, prepared as described previously (33), simultaneously with the PAPA-NONOate. The cells were extracted in situ in a gel sample buffer containing 1% SDS, and lysates were subjected to PAGE and Western blotting. VASP was detected using a mouse anti-VSV monoclonal antibody (Sigma) as described previously (31). The blots shown were reproduced at least three times.
Measurement of Nitrite and Nitrate-NO has a very short half-life and is oxidized rapidly to nitrite and nitrate under physiological conditions. Hence, one of the most common methods for assessing NO production is to measure nitrite and nitrate concentrations, which is generally done using the Griess reagent (33,34). We measured nitrite and nitrate concentrations in both the Drosophila Malpighian tubule secretion system and in the C6 and CS-54 cells. In the Drosophila system, 20 tubules were incubated in 100 l of Schneider's medium, and the tubules were stimulated for 60 min with either 10 M Deta-NONOate or 10 M LPS, in the absence or presence of 10 M cobinamide. The C6 and CS-54 cells were incubated for 30 min with 15 M PAPA-NONOate and 300 nM A23187, respectively, in the absence or presence of 15 M cobinamide. In all cases, the medium was collected at the end of the incubation period, and nitrite and nitrate in the medium were measured using a nitric oxide quantitation kit from Active Motif, which is an enhanced Griess reagent-based method. Because Deta-NONOate and PAPA-NONOate will continue to release NO even after the end of the incubation with the tubules and cells, all subsequent steps were performed at room temperature immediately after harvesting the medium.
Assessment of Cobinamide Cytotoxicity-The effect of varying concentrations of cobinamide on the growth of BHK, CS-54, and C6 cells, and human foreskin fibroblasts and human umbilical venous endothelial cells was assessed by counting the number of cells daily for 3 days using a model ZM Coulter Counter (Coulter Electronics). All of the cells were grown in DMEM containing 10% FBS, except the umbilical endothelial cells, which were grown in M199 medium containing endothelial cell growth supplement and 20% FBS (35). Cells were plated in 6-well culture dishes at initial densities of 1.5-3 ϫ 10 5 cells/well.
Measurement of the Activity of Methionine Synthase and Methylmalonyl-CoA Mutase in Vitro-BHK cells were extracted as described previously at a density of ϳ50 ϫ 10 6 /ml in a buffer containing 100 mM Tris-HCl, pH 7.4, 5 mM dithiothreitol, 1 mM EDTA, and a protease inhibitor mixture (34). Methionine synthase and methylmalonyl-CoA mutase activities were measured in the extracts in the absence and presence of 1-200 M cobinamide. Methionine synthase activity was measured at 37°C according to the method of Weissbach et al. (36), except the [ 14 C]methyltetrahydrofolate substrate was separated from the [ 14 C]methionine product by thin layer chromatography on cellulose acetate plates developed in butanol:acetic acid:water (4:1:5); the R F values for substrate and product were 0.26 and 0.44, respectively.
For methylmalonyl-CoA mutase, the extracts were preincubated for 10 min at 37°C with 5 M deoxyadenosylcobalamin to convert apoenzyme to holoenzyme, followed by a 10-fold dilution in extract buffer. Enzyme activity was measured at 30°C according to the method of Kikuchi et al. (37), with the methylmalonyl-CoA substrate separated from the succinyl-CoA product by HPLC on a C18 reverse phase column eluted in 100 mM sodium phosphate, pH 4.0, containing 15% methanol; amounts of substrate and product were determined by comparison with known standards. Both assays were linear with time from 5 to 15 min and with a protein concentration from 0.1 to 0.5 mg/ml.
Assessment of the Activity of Methionine Synthase and Methylmalonyl-CoA Mutase in Vivo-The activities of methionine synthase and methylmalonyl-CoA mutase were assessed in intact BHK cells by following incorporation of [ 14 C]formate into purine nucleotides and [ 14 C]propionic acid into protein, respectively; both of these assays have been used previously as surrogate measurements of the in vivo activities of the enzymes (33,38). Briefly, about 1 ϫ 10 6 BHK cells were incubated in 6-well cluster dishes with 10 Ci of [ 14 C]formate for 90 min or 20 Ci of [ 14 C]propionic acid for 16 h; cobinamide, at concentrations of 1-200 M, was added 6.5 h before the formate label (for a total incubation time of 8 h) and simultaneously with the propionate label. At the end of the incubation, the cells incubated with [ 14 C]formate were extracted in 0.4 N perchloric acid, heated to 100°C for 70 min to convert purine nucleotides to bases, and applied to Dowex 50 cation exchange columns to separate the purine bases from unincorporated [ 14 C]formate (39). The cells incubated with [ 14 C]propionic acid were extracted in ice-cold 10% trichloroacetic acid, heated to 80°C for 30 min to solubilize precipitated nucleic acids, and after recooling to 4°C, precipitated protein was collected on glass microfiber filters (39). Both assays were linear with time and cell number from 0.5 to 2 ϫ 10 6 cells/ml.

RESULTS
Production and Analysis of Cobinamide-The standard method for producing diaquocobinamide starts with dicyanocobinamide, removing the cyanide by acid treatment, and exposure to strong light (23). As mentioned earlier, cobinamide has a very high binding affinity for cyanide, and hence it is difficult to remove cyanide completely from the cobinamide preparations. Moreover, exposure to light over a prolonged period can potentially alter the corrin ring. Because we were interested in producing cyanide-free cobinamide for use in biological systems, we started with OH-Cbl as the initial substrate. The dimethylbenzimidazole ribonucleotide tail was removed by brief acid treatment, and the diaquocobinamide was purified by batch elution over a small sample preparation column as described under "Experimental Procedures." Beginning with about 200 mg of OH-Cbl, we obtained ϳ150 -170 mg of high purity cobinamide. Fig. 2 shows the absorbance spectrum of a typical cobinamide preparation at pH 3 having a major peak of 348 nm and smaller relatively equal peaks at 494 and 520 nm (23,24); in preparations containing contaminants, the 494 and 520 nm peaks tend to either merge together into one broad peak, or the 520 peak became predominant, and a broad band at 455 nm became evident (25). Further evidence for high purity of the preparations was that at pH 12 the A 344 /A 356 ratio was 1.06, well within the range of 1.05-1.11 reported previously for pure dihydroxocobinamide (25), and HPLC analyses of the cobinamide product yielded a single peak when monitored at multiple wavelengths between 300 and 600 nm (40).
Efficacy of Cobinamide as a NO Scavenger in a Drosophila Fluid Secretion Model-It has become clear that Drosophila is an excellent model for human disease and drug discovery (41,42). The Malpighian tubules of D. melanogaster are the organs for fluid transport and osmoregulation in the insect, corresponding to vertebrate kidneys. Rates of tubular secretion can be measured in vitro after extracting tubules from flies; NO stimulates secretion via activation of soluble guanylate cyclase, thereby increasing the intracellular cGMP concentration and activating the cGMP/G-kinase transduction pathway (27,28). We studied the effect of cobinamide on tubular fluid secretion stimulated both by a NO donor and by LPS, an inducer of the Drosophila NOS gene. In addition, to simulate conditions in a whole animal, we administered cobinamide to the flies via their food and then measured the effect of LPS on rates of tubular fluid secretion.
In the experiments with a NO donor, we treated tubules with 10 M Deta-NONOate, which caused a rapid and sustained increase in the rate of fluid secretion ( Fig. 3A; Deta-NONOate was added to all tubules after a 30-min basal period, as indicated by the arrowhead). Adding 10 M cobinamide to Deta-NONOate-treated tubules reduced the rate of fluid secretion significantly, almost returning to the basal unstimulated level ( Fig. 3A; cobinamide was added at 60 min to some of the tubules as indicated by the arrow and the filled circles). Adding cobinamide alone to the tubules was without effect (data not shown). To increase endogenous NO production by the tubules, we used LPS and observed a marked stimulation of tubular secretion ( Fig. 3B; LPS was added at 30 min as indicated by the arrowhead). As in the experiments with Deta-NONOate, cobinamide rapidly reduced tubular secretion, returning rates near to the basal state ( Fig. 3B; cobinamide was added at 60 min to some of the tubules as indicated by the arrow and the filled circles). Thus, cobinamide scavenges both extracellularly administered and intracellularly produced NO in a Drosophila whole organ system.
As part of these studies, we measured the amount of NO released by Deta-NONOate or produced by LPS-treated tubules by incubating tubules for 60 min in Schneider's medium and measuring the sum of nitrite and nitrate in the medium. In the absence of the two drugs, no nitrite or nitrate could be FIG. 2. Spectral analysis of cobinamide. Cobinamide was produced from OH-Cbl as described under "Experimental Procedures" and was purified on a C18 solid phase extraction column. After concentration, the sample was brought to pH 3.0, and the spectrum of the diaquocobinamide product was recorded between 300 and 700 nm. detected in the medium, implying that the basal level of NO production by the tubules was very low. Treating the tubules with 1, 10, or 100 M Deta-NONOate increased the nitrite and nitrate concentrations in the medium to levels above the starting concentrations of the drug, which was likely because Deta-NONOate releases 2 mol of NO/mol of parent compound (Fig.  3C shows data for 10 M Deta-NONOate). The tubules were probably exposed to somewhat lower NO concentrations, because as mentioned previously, Deta-NONOate will continue to release NO during the measurement of nitrite and nitrate. Treating the tubules with 10 M LPS to induce NOS levels also increased medium nitrite and nitrate concentrations, but the levels were less than in the tubules treated with 10 M Deta-NONOate (Fig. 3C). In both the Deta-NONOate-and LPStreated tubules, cobinamide reduced the amount of combined nitrite and nitrate by about 50% (Fig. 3C), which is consistent with our previous work showing that each cobinamide molecule can neutralize two NO molecules, converting one to nitrite and binding the second one (22); thus, a maximal 50% reduction in nitrite and nitrate could be expected.
To determine whether systemically administered cobinamide would affect rates of tubular fluid secretion, we fed the flies food containing 250 M cobinamide for 2 days; at this concentration, cobinamide had no apparent detrimental effect on the flies. LPS-stimulated secretion rates of Malpighian tubules were compared between cobinamide-fed flies and flies fed normal control food. We found that tubular secretion rates were reduced by about 50% in the cobinamide-fed flies compared with control flies (Fig. 4). In addition to providing further evidence for the effectiveness of cobinamide as a NO scavenger, these data indicate that cobinamide was absorbed through the gut of the flies and was transported to the Malpighian tubules.
Cobinamide Uptake by Mammalian Cells-Some NO scavengers, such as hemoglobin, are solely extracellular and neutralize only that NO that diffuses out of cells. The studies in Drosophila indicated that cobinamide was likely taken up by the cells, but insect and mammalian transport systems may differ. To determine whether cobinamide has access to intracellular NO in mammalian cells, we produced [ 14 C]dicyanocobinamide and studied its uptake in BHK cells. We found a progressive curve-linear increase in radioactivity in BHK cells over a 1,000-fold concentration range of dicyanocobinamide from 100 nM to 100 M, suggesting that both passive diffusion and active transport of the dicyanocobinamide were occurring. Because of the progressive increase in cellular radioactivity, a K m could not be calculated. Although it is possible that the cobinamide was not actually taken up by the cells, but rather just bound to a surface receptor, this seems unlikely because of the continued increased in radioactivity over a three log scale of dicyanocobinamide concentrations. Thus, these data suggest that cobinamide has the potential to serve as both an intra-and extracellular NO scavenger in mammalian systems.
Efficacy of Cobinamide as a NO Scavenger in Mammalian Cells-To determine whether cobinamide can serve as a NO scavenger in mammalian cells, we examined NO stimulation of VASP phosphorylation in two different types of cultured cells. VASP is an important regulator of actin dynamics and thus of cellular processes such as cell adhesion and motility (31). Its function is regulated by phosphorylation, and NO, via activation of the cGMP/G-kinase transduction pathway, is a major inducer of VASP phosphorylation. To study the effect of cobinamide on NO-induced phosphorylation of VASP, we chose rat C6 glioma cells and CS-54 vascular smooth muscle cells, both of which have an active NO/cGMP transduction pathway (34,43). In C6 cells we studied the effect of cobinamide on exogenously generated NO, and in CS-54 cells we studied the effect of cobinamide on endogenously produced NO.
When C6 cells were treated for 30 min with 15-60 M PAPA-NONOate, VASP phosphorylation was induced, as evidenced by generation of a VASP form with reduced electrophoretic mobility (Fig. 5A, compare first lane, no PAPA-NONOate, to third, fifth, and seventh lanes, showing 15, 30, and 60 M PAPA-NONOate, respectively). At all PAPA-NONOate concentrations, adding 100 M cobinamide to the culture medium prevented the increase in VASP phosphorylation (Fig. 5A, compare third, fifth, and seventh lanes with fourth, sixth, and eighth lanes showing cells treated with cobinamide and PAPA-NONOate). Human hemoglobin (100 M) yielded results similar to those with cobinamide, indicating that under the experimental conditions, cobinamide and hemoglobin were equally effective NO scavengers. Cobinamide also prevented increases in VASP phosphorylation when used at an equimolar concentration as PAPA-NONOate (Fig. 5B shows 30 M each PAPA-NONOate and cobinamide, but similar results were also found at 15 and 60 M). As discussed further below, PAPA-NONOate releases 2 mol of NO and thus as in the Drosophila Malpighian tubules, each cobinamide molecule appeared to neutralize two NO molecules.
Because it was possible that cobinamide could be inhibiting the NO stimulation of VASP phosphorylation through some mechanism other than NO scavenging, we studied the effects of cobinamide on cGMP stimulation of VASP phosphorylation. We used the membrane-permeable cGMP analog 8-pCPT-cGMP at a concentration of 30 M to induce VASP phosphorylation and found no effect of cobinamide at concentrations from 30 to 100 M (Fig. 5C, compare second lane, CPT-cGMP alone, with third lane, CPT-cGMP plus cobinamide). Thus, cobinamide did not interfere with the activation of G-kinase or the phosphorylation of VASP, and thus, it appeared to be acting via NO scavenging.
Because calcium activates types I and III NOS, we treated CS-54 cells with the calcium ionophore A23187 to increase endogenous NO production. We found that A23187 increased VASP phosphorylation (Fig. 5D, compare fourth lane with first lane) and that cobinamide significantly attenuated this effect (Fig. 5D, compare third with fourth lane). Thus, cobinamide is an effective intra-and extracellular NO scavenger in mammalian cells.
As in the studies of the Malpighian tubules, we measured nitrite and nitrate concentrations in the culture medium of the C6 and CS-54 cells. As mentioned above, PAPA-NONOate, like Deta-NONOate, releases two NO molecules/mol of parent compound, and we found that treating C6 cells for 30  The medium was harvested, and the sum of nitrite plus nitrate was measured as described in the legend to Fig. 3. The data are the mean Ϯ S.D. of at least two independent experiments performed in triplicate; the y-axis on the left is for the PAPA-NONOate-treated C6 cells, and the y-axis on the right is for the A23187-treated CS-54 cells. Low nitrite and/or nitrate concentrations in the medium of untreated cells were subtracted from that of the treated cells. sequent measurement of nitrite and nitrate. Treating CS-54 cells with 300 nM A23187 increased nitrite and nitrate concentrations in the medium to about 12 M (Fig. 5E). When 15 M cobinamide was added to the PAPA-NONOate-treated C6 cells or the A23187-treated CS-54 cells, there was a significant decrease in the nitrite plus nitrate concentrations, with a larger effect in the CS-54 cells than in the C6 cells (Fig. 5E). Thus, the cobinamide-induced decrease in VASP phosphorylation in the two cell types was reflected by a reduction in nitrite and nitrate, providing further evidence that cobinamide was functioning as a NO scavenger in reducing VASP phosphorylation.
Studies of Cobinamide Cytotoxicity-As a cobalamin analog, cobinamide could be cytotoxic by interfering with cobalamin metabolism or function. Moreover, as a NO scavenger, cobinamide could interfere with the proproliferative and antiapototic actions of NO in some cell types (44,45). To study the potential toxicity of cobinamide in mammalian cells, we performed several sets of experiments. First, we assessed the effect of cobinamide on the growth of BHK, C6, and CS-54 cells, as well as two primary human cell lines, i.e. foreskin fibroblasts and human umbilical venous endothelial cells; the latter cells were included to determine whether there was differential toxicity of cobinamide to primary cells than to established cell lines. We found that at concentrations between 1 and 50 M, cobinamide had no effect on cell growth, but at concentrations Ն50 M, cobinamide inhibited the growth of all five cell types, albeit by a minimal amount at 50 M (Fig. 6, data are shown at cobinamide concentrations of 50, 100, and 200 M in BHK cells (main figure) and C6 and CS-54 cells (inset); similar results were found in the two primary cell lines). Growth inhibition was reversed completely by an equimolar concentration of cobalamin at all three cobinamide concentrations in all five cell types ( Fig. 6; data are shown only for 200 M cobinamide/cobalamin in BHK cells (main figure)). These latter data suggest that the mechanism of toxicity was through competitive interference of cobalamin metabolism or function rather than through NO scavenging. Thus, cobinamide was toxic to cells at only relatively high concentrations, and its toxicity could be reversed fully by cobalamin.
The next set of experiments was directed at determining whether cobinamide inhibited the activities of either methionine synthase or methylmalonyl-CoA mutase, the two vitamin B 12 -dependent enzymes in mammalian cells. At concentrations as high as 200 M, we found no effect of cobinamide on the activities of either enzyme as measured in BHK cell extracts. Methionine synthase was assumed to be in the holoenzyme form (46), and methylmalonyl-CoA mutase apoenzyme was converted to holoenzyme by incubation with 5 M deoxyadenosylcobalamin followed by a 10-fold dilution of the extract as described under "Experimental Procedures." Thus, at the highest concentration of cobinamide tested, the cobinamide concentration was 400 times that of the deoxyadenosylcobalamin concentration. The lack of inhibition of either enzyme by cobinamide is consistent with studies reported by others on various cobalamin analogs including cobinamide (47,48).
In the final set of experiments, we assessed in BHK cells the effect of cobinamide on the in vivo activities of methionine synthase and methylmalonyl-CoA mutase by following the incorporation of [ 14 C]formate into purine nucleotides and [ 14 C]propionic acid into protein; the former assay is a measure of carbon flux through the folate pathway and is dependent on methionine synthase activity, whereas the latter assay is dependent on methylmalonyl-CoA mutase activity. In both assays, we found that 100 M cobinamide decreased incorporation of the radioactive label by about 50% (Fig. 7; filled bars are data for [ 14 C]formate incorporation, and open bars are data for [ 14 C]propionic acid incorporation). The inhibition of enzyme activities was time-dependent, and in the case of methionine synthase, minimal inhibition was observed after 90 min of cobinamide exposure, the shortest time point that could be measured; inhibition increased progressively with time, and the data in Fig. 7 are for a total of 8 h of cobinamide exposure. Similar time dependence of enzyme inhibition was observed for methylmalonyl-CoA mutase, but over a longer time scale. As in the growth studies, the toxic effect of cobinamide was prevented completely by an equimolar concentration of cobalamin (Fig. 7). These latter data indicate that the mechanism of cobinamide toxicity is likely through interference with some aspect of cobalamin metabolism (as considered further under "Discussion"). Moreover, because methionine synthase and methylmalonyl-CoA mutase are the only two mammalian cobalamin-dependent enzymes, the cobalamin reversal data suggest that the [ 14 C]formate and [ 14 C]propionic acid incorporation assays accurately reflect the in vivo activity of these two enzymes.
Efficacy of Cobinamide as a NO Scavenger in the Presence of Cobalamin-Although cobinamide is likely to be used at concentrations where it is not toxic, it would be useful to know whether cobinamide and cobalamin can be used together when scavenging NO. Cobalamin itself is a weak NO scavenger (21) and would not be expected to interfere with the cobinamide scavenging of NO. We found that OH-Cbl had no effect on NO scavenging by cobinamide, either on tubular fluid secretion in Drosophila Malpighian tubules (Fig. 3, A and B, compare filled circles, cobinamide alone, with filled triangles, cobinamide plus cobalamin) or on VASP phosphorylation in mammalian cells (Fig. 5B, compare fourth lane, cobinamide alone, with fifth lane, cobinamide plus cobalamin). The combination of cobinamide plus cobalamin, like cobinamide alone, had no effect on VASP phosphorylation induced by 8-pCPT-cGMP (Fig. 5C, compare third and fourth lanes).

DISCUSSION
Since the original description of NO as endothelium-derived relaxation factor in 1978, it has become clear that NO has many physiological roles (1)(2)(3). It also has become clear that NO contributes to the pathophysiology of several disease states including septic and hemorrhagic shock, hepatic encephalopathy, hepatorenal syndrome, hemodialysis-related hypotension, and ischemia-reperfusion injury (5)(6)(7). In most of these diseases, abnormally high NO production induces profound vasodilation and vasopressor-refractory hypotension. A marked increase in iNOS underlies the elevated NO production in sepsis (49), and nonselective NOS inhibitors increase blood pressure in animal models of sepsis, but have had a mixed effect on sepsis-associated mortality in animals and humans (18,50); this may relate, in part, to inhibition of eNOS leading to microvascular vasoconstriction and decreased tissue and organ perfusion (51)(52)(53)(54)(55). Selective iNOS inhibitors should avoid some of the problems encountered with nonspecific inhibitors, but iNOS expression in some tissues may be functionally important as has been found for myocyte iNOS (11, 56 -59). Thus, even selective iNOS inhibition may lead to dysfunctional changes in cells and organs, and several groups of workers have urged caution in using any type of NOS inhibitor in sepsis (50,60,61).
There is a clear need, therefore, for agents that can lower NO levels without the attendant toxicity of NOS inhibitors. NO scavengers have the theoretical advantage of neutralizing only the portion of presynthesized NO to which the agent is exposed, without directly interfering with NOS function (18,62). Several NO scavengers have been identified, but not all are taken up by cells, and some, particularly free hemoglobin, exhibit unacceptable toxicity (63). Thus, there would appear to be the need for a nontoxic NO scavenger that could be used both clinically and in the laboratory.
We found that cobinamide was an effective NO scavenger, both in a Drosophila tubular secretion model and in two types of cultured mammalian cells. In both systems, each cobinamide molecule appeared to neutralize more than one NO molecule, which is in agreement with our previous in vitro work (22). Although the previous work was performed in aqueous buffers, the current work was performed in serum-containing medium, and we have found recently that cobinamide binds tightly to serum albumin, both bovine and human. 2 Thus, the binding of cobinamide to serum albumin does not appear to interfere with its ability to scavenge NO, and the interaction of cobinamide with NO in the presence of serum is currently under study.
During bacterial biosynthesis of cobalamin, the dimethylbenzimidazole ribonucleotide tail is added last, and thus cobinamide is the penultimate precursor in cobalamin biosynthesis. Because of its place in cobalamin synthesis, cobinamide has been shown to contaminate bacterial vitamin B 12 preparations and to be present in animal tissues and human serum (64 -66). In a study of the enterohepatic circulation of corrinoids in humans, it was found that cobalamin analogs, which appeared to be mostly cobinamide, constituted 45% of total bile corrinoids (67). A detailed study of the pharmacokinetics of cobinamide and cobalamin after parenteral administration to rabbits found that cobinamide was retained more by the liver than cobalamin but that the urinary and fecal excretion of the two corrinoids was similar (68). Cobinamide binds poorly to transcobalamin II, the major cobalamin-binding protein in blood, but it binds tightly to haptocorrin, another important cobalamin-binding protein previously referred to as R binder (69 -71). Thus, the presence of cobinamide in serum and tissues, and its pharmacokinetic profile, are likely attributable to its high binding affinity for haptocorrin and possibly also albumin.
Previous studies have found little or no toxicity of cobinamide to mammalian cells, but these studies were done at low micromolar or submicromolar concentrations of the drug (48,72,73). We similarly found no toxicity of cobinamide to mammalian cells at low micromolar concentrations but began to observe toxicity at a concentration of about 50 M. Because each cobinamide molecule can potentially neutralize two NO molecules (22), cobinamide may be able to neutralize NO concentrations up to 100 M before exhibiting significant toxicity. Other than during pharmacological administration of NO, it is unlikely that NO concentrations ever exceed 10 M under physiological conditions (7,10,11). Moreover, we found that cobalamin completely prevented cobinamide toxicity and that cobalamin did not interfere with NO scavenging by cobinamide, either in the Drosophila Malpighian tubule secretion model or in mammalian cells; as mentioned previously, cobalamin itself is a NO scavenger (21).
At the relatively high concentrations where cobinamide began to exhibit toxicity, it appeared to interfere with cobalamin metabolism or function. This conclusion is based on the findings that cobinamide had no effect on the in vitro activities of methionine synthase and methylmalonyl-CoA mutase, the two mammalian cobalamin-dependent enzymes, whereas it inhibited in vivo assays of the enzymes in a time-dependent and cobalamin-reversible fashion. In the in vitro assays, the enzymes were in the cobalamin-containing holoenzyme form, indicating that cobinamide cannot compete with the bound cofactor. Other workers have also found no inhibition in vitro of the holoenzyme form of methionine synthase by cobalamin analogs, and in fact, cobinamide can restore the methionine synthase apoenzyme to full activity; but whether this occurs in vivo is not known (47). The effect of cobalamin analogs on methylmalonyl-CoA mutase activity has apparently not been studied, but 5Ј-deoxyadenosylcobinamide is inactive in restoring function to the apoenzyme of methylmalonyl-CoA mutase 2 V. Singh, V. S. Sharma, and G. R. Boss, unpublished observations. (74). The time-dependent inhibition of enzyme activity in the in vivo assays occurred over hours, suggesting that newly synthesized enzyme was inhibited; the half-life of methionine synthase is about 12 h and that of methylmalonyl-CoA mutase appears to be ϳ30 h (46,75). Thus, cobinamide could interfere with incorporation of methylcobalamin or deoxyadenosylcobalamin into newly synthesized methionine synthase and methylmalonyl-CoA mutase, respectively, or perhaps more likely, with the cellular transport of OH-Cbl or its conversion to the two coenzyme forms. Similar conclusions about the mechanism of toxicity of cobalamin analogs were reached previously (47). Whatever the precise mechanism(s) of cobinamide cytotoxicity is, it can be completely prevented by coadministration of cobalamin.
Because of its high binding affinity for cyanide, cobinamide could also be expected to be an excellent cyanide scavenger and could potentially be used in clinical states of cyanide toxicity, e.g. smoke inhalation and nitroprusside toxicity. Cobalamin, which has a binding affinity for cyanide several orders of magnitude less than that of cobinamide, has already been shown to be effective in cyanide toxicity and is used for this purpose in France (76 -78).
In conclusion, cobinamide is an effective NO scavenger in Drosophila Malpighian tubules and cultured mammalian cells and is likely to be useful in animal studies. It may also be beneficial in clinical states of excess NO and in cyanide toxicity and could be combined with cobalamin.