Determining sodium azide concentration in blood by ion chromatography.

We describe a simple method for measuring sodium azide concentrations in aliquots of blood and other tissues. Aliquots are acidified, converting azide to volatile hydrazoic acid (HN3) which is then trapped in sodium hydroxide. We analyze the resulting aliquots by ion chromatography, using a sodium tetraborate eluent and suppressed conductivity detection. The method is sensitive to at least 100 ng/mL.

Bioactivation of nitroprusside by porcine endothelial cells.

Sodium nitroprusside (Na2[(CN)5FeNO], SNP), which is stable, diamagnetic, and not detectable by electron paramagnetic resonance (EPR) spectroscopy, can be activated by one-electron reduction. The initial product, which retains the five cyanides and is here called penta, has a distinctive EPR signal. Penta spontaneously dissociates the trans-cyanide ligand resulting in a second paramagnetic species called tetra, which has a different and distinctive EPR signal. Tetra is able to transfer its NO ligand to a suitable acceptor, and all four equatorial cyanides subsequently dissociate. However, excess free cyanide shifts the tetra-penta equilibrium in the direction of penta and prevents NO release. This study was an attempt to extend the above results on SNP reduction, which were obtained in a model hemoglobin system, to intact porcine cells by characterizing all EPR-detectable intermediates. When porcine aortic endothelial or smooth muscle cells in culture were incubated under anaerobic conditions with SNP, an EPR spectrum was obtained, which could be resolved into the signal for penta and a signal previously described as a nonheme iron-nitrosyl-sulfur complex, Fe-NOSR. Tetra was not detected. This FeNOSR has some differences in its stability and location from that described by others in activated macrophages. When incubations were carried out under air, penta could not be detected, but a somewhat diminished signal for FeNOSR was still detectable. When incubations were carried out in the presence of excess free cyanide, conditions under which reduced SNP does not nitrosylate hemoglobin, the penta signal became stronger and the FeNOSR signal, though decreased, was still observed. Depletion (95%) of intracellular reduced glutathione in endothelial cells had no effect on the FeNOSR signal strength. We conclude that SNP is activated in porcine endothelial cells by a one-electron reduction to penta, which apparently dissociates its trans-cyanide to form tetra which then goes on to form FeNOSR upon reaction with a membrane-bound thiol. Glutathione is not involved in any of these reactions.

An impurity present in some samples of SIN-1 oxidizes it to nitric oxide in anaerobic solutions.

N-Morpholino-N-nitrosoaminoacetonitrile (SIN-1), a nitrovasodilator metabolite of the drug, molsidomine, is widely used in studies on the pharmacology and toxicology of nitric oxide (NO) because solutions of SIN-1 'spontaneously' release NO in a pathway involving molecular oxygen. Preliminary results, however, suggested that SIN-1 could react with hemoglobin in anaerobic solutions to release NO and form NO-hemoglobin. Electron paramagnetic resonance (EPR) studies showed that heme(III) of methemoglobin was not being reduced, thereby not serving as the oxidant in the reaction generating NO-hemoglobin. When anaerobic solutions of SIN-1 and hemoglobin kept in the light and in the dark were compared, substantially more NO-hemoglobin was eventually generated in the dark, indicating that SIN-1 did not undergo photochemical decomposition to NO under the conditions used. Solutions of NO-hemoglobin were equally stable under these same conditions of light and dark. The initial pH (7.0) of stirred, unbuffered solutions of SIN-1 decreased at nearly the same rate whether or not oxygen was present. Anaerobic and aerobic solutions plateaued at the same pH, namely 5.4. Anaerobic solutions of SIN-1 in phosphate buffer, pH 7.4, released NO to the gas phase, where it was identified by trapping it with hemoglobin on agarose beads and deriving the characteristic NO-hemoglobin EPR spectrum. High pressure liquid chromatography revealed the presence of an unknown species with a retention time between that of SIN-1 and molsidomine. Samples from two different lots of SIN-1 contained this impurity which appears to oxidize SIN-1 to products that release NO in the absence of oxygen. This unknown impurity may be unstable toward light.

Effects of temperature, oxygen, heme ligands and sulfhydryl alkylation on the reactions of nitroprusside and nitroglycerin with hemoglobin.

Nitrovasodilators react with hemoglobin (Hb) to form heme(III) and nitric oxide (NO)Hb. These reactions can be exploited as models for events that take place at the cellular level leading to the biological effects of the prodrugs. Sodium nitroprusside (SNP) is known to undergo a one-electron reduction in its reaction with heme(II), resulting in the labilization of the cyanide ligand trans to the NO ligand. This reduced form is here called "penta." Upon dissociation of the trans-cyanide, the resulting species is here called "tetra." Dissociation of the trans-cyanide is obligatory for transfer of the NO to a heme(II) group. NO release from penta is blocked by excess free cyanide in solution, which prevents the formation of tetra. As reported here, both penta and tetra had unique EPR signals when frozen at -196 degrees, but only tetra gave an EPR signal at 22 degrees. NOHb also has a unique EPR signal, but it could not be detected when SNP was incubated with Hb in air or 10 or 5% oxygen. NOHb was detected in similar incubations under 1% oxygen, but the levels were 3- to 10-fold lower than those found under 100% nitrogen. The concentration of tetra was also much lower under 1% oxygen and penta was not detectable, suggesting that oxygen may either shift the penta-tetra equilibrium towards tetra or that penta may be susceptible to oxidation by molecular oxygen. Nitroglycerin (GTN) also generated much less NOHb but more heme(III) under 1% oxygen than under nitrogen. Carbon monoxide (CO), which binds to heme(II), completely blocked the reactions of SNP and GTN with Hb, whereas N-ethylmaleimide (NEM) alkylation of globin sulfhydryl groups increased both NOHb and heme(III) formation. 13C NMR studies on uniformly 13C-labeled SNP suggested that oxygen had little effect on the concentrations of the NMR-detectable species in the reaction. In summary, the most oxygen-sensitive step in the nitrosylation of Hb by SNP was probably the transfer of NO to heme(II). However, the penta-tetra equilibrium was affected by oxygen, temperature and cyanide. No evidence was found for the involvement of the globin sulfhydryl groups in either the GTN or the SNP reaction with Hb.

Identification of increased nitric oxide biosynthesis during pregnancy in rats.

We reported previously that plasma levels, urinary excretion, and metabolic production of cyclic guanosine 3',5'-monophosphate (cGMP) are increased in gravid rats, and postulated that endogenous nitric oxide (NO), a potent vasodilator and immune modulator, may mediate this change. Four lines of evidence are now presented demonstrating increased biosynthesis of NO during pregnancy in rats: 1) Urinary excretion and plasma levels of the stable NO metabolite, nitrate, are elevated in pregnant rats; urinary excretion of nitrate is increased in pseudopregnant rats. 2) The urinary excretion of cGMP also increases during pregnancy and pseudopregnancy, paralleling the rise in urinary nitrate excretion. 3) Chronic treatment with the NO synthase inhibitor, NG-nitroarginine methyl ester (NAME), inhibits the increase in urinary nitrate excretion. 4) Nitric oxide hemoglobin is detected by electron paramagnetic resonance spectroscopy in the blood of pregnant, but not in nonpregnant, rats. The results show endogenous NO production is increased in gravid rats. This finding raises the possibility that NO may contribute to maternal vasodilation and uterine immune suppression of normal pregnancy.

Nitric oxide hemoglobin in patients receiving nitroglycerin as detected by electron paramagnetic resonance spectroscopy.

Blood specimens were obtained from 30 adult patients admitted to a coronary care unit after the decision to use nitroglycerin had been made by their physicians. The samples were drawn before nitroglycerin administration, within 1 hour after starting nitroglycerin, after several hours of therapy, and more than 4 hours after discontinuing therapy. One patient was admitted twice, accounting for 31 sets of blood specimens. A positive identification of nitric oxide-hemoglobin (NOHb), with electron paramagnetic resonance (EPR) spectroscopy could be made in the blood of 10 of the subjects after they had been receiving nitroglycerin for several hours (third blood sample). In seven subjects this third blood sample was not drawn, and they were dropped from the study. A final positive finding of NOHb was made in 10 of 24 patients. NOHb has not been identified previously in human subjects given nitroglycerin, and a significant dose-response relationship was observed between nitroglycerin and NOHb. We ascribe our inability to detect NOHb in all subjects before nitroglycerin (basal levels) and after nitroglycerin in 14 subjects to concentrations that were below the limits of detection of the technique as used. Subtraction of the EPR signal for plasma ceruloplasmin was necessary to detect the NOHb EPR signals. Thus we have shown EPR spectroscopy to be a highly specific and sensitive method for detecting and quantifying NOHb in human subjects. Further refinements in the technique to improve sensitivity are possible.(ABSTRACT TRUNCATED AT 250 WORDS)

Formation of nitric oxide hemoglobin in erythrocytes co-cultured with alveolar macrophages taken from bleomycin treated rats.

Alveolar macrophages, taken from rats treated with a single intratracheal dose of bleomycin, release reactive nitrogen intermediates in the form of nitric oxide which are cytostatic to murine leukemia L1210 cells. When cultured in the presence of erythrocytes the cytostatic activity of alveolar macrophages was inhibited which corresponded with an increase in nitrosylated hemoglobin content when compared with erythrocytes cultured alone. These results suggest that erythrocytes inhibit alveolar macrophage cytostatic activity by preventing reactive nitrogen intermediates from reaching target cells because the hemoglobin serves as a sink for reactive nitrogen intermediates in the form of nitric oxide.

Acute neurotoxicity of sodium azide and nitric oxide.

Sodium azide is a chemical of rapidly growing commercial importance with a high acute toxicity and an unknown mechanism of action. Although it has some chemical properties and biological effects in common with cyanide, its lethality does not appear to be due to inhibition of cytochrome oxidase. Unlike cyanide it is a potent vasodilator and inhibitor of platelet aggregation presumably by virtue of its conversion to nitric oxide in vivo and in isolated preparations of blood vessels and thrombocytes. It is not clear whether the high toxicity of azide is due to nitric oxide or to the parent anion. Of a number of possible azide antagonists tested in intact mice only phenobarbital in both anesthetic and subanesthetic doses afforded statistically significant protection against death. Diazepam, phenytoin, and an anesthetic dose of a ketamine/xylazine combination had no effect. Major motor seizures are sometimes seen in human azide poisoning, and these are a regular feature of azide poisoning in laboratory rodents. Solutions of nitric oxide given systemically to mice produced no signs of toxicity, but doses 1,000-fold lower placed in the cerebroventricular system of rats produced brief but violent tonic convulsive episodes. A dose of 0.61 mmol/kg azide as given systemically regularly produced convulsions whereas a dose of 6 mumol/kg given icv produced seizures in rats. The icv convulsive dose of azide was 50-fold larger than the icv dose of nitric oxide. These results suggest that azide lethality is due to enhanced excitatory transmission in the central nervous system perhaps after its conversion to nitric oxide.(ABSTRACT TRUNCATED AT 250 WORDS)

Nitric oxide hemoglobin in mice and rats in endotoxic shock.

Mice given ip bacterial endotoxin (LPS) at 10 mg/kg showed a statistically significant decrease in plasma glucose and an increase in hematocrit at 2 h after injection. Glucose was still decreased at 4 h, but the hematocrit had returned to control values. Nitrosylated hemoglobin (HbNO) was detected at 3, but not at 2 h. By 4 h it had increased 5-fold. When N-monomethylarginine (NMMA) at 100 mg/kg, ip was given 2 h after LPS in mice, the HbNO concentration at 4 h was significantly reduced, but the hypoglycemia was worsened because NMMA itself produced a significant hypoglycemia. Rats given iv LPS, 20 mg/kg, showed a fleeting, transient rise in mean arterial pressure (MAP) lasting only a few min. Thereafter, the MAP tended to drift slowly downward over 4 h, but when the MAP at 30 min intervals was compared to the pre-LPS MAP, there were no significant differences. Plasma glucose in unanesthetized rats was significantly elevated at 1 h, back to control at 2 h, and significantly decreased at 3 h. HbNO was detected as early as 1 h after injection. By 2 h the HbNO concentrations exceeded the highest levels found in mice, and they were still increasing as late as 5 h after injection. Unanesthetized rats showed toxic signs and 3/12 rats died within 4 hours of LPS administration. These results are consistent with a model for endotoxic shock in which LPS stimulates an inducible pathway for NO synthesis.

Embryotoxic effects of sodium azide infusions in the Syrian hamster.

Pairs of osmotic minipumps containing 400 mg/ml (6.15 M) sodium azide in distilled water were subcutaneously implanted in timed pregnancy Syrian golden hamsters. The total delivered dose was calculated as 6 X 10(-2) mmol kg-1 hr-1 at the maximal pumping rate. Most dams exhibited obvious signs of toxicity during the period of pump implantation which was Days 7 through 9 of gestation. After removal of the pumps the dams were euthanized on Day 13 of gestation, and the uteri were removed for counting of the number of living, malformed, and resorbed fetuses. This dose rate resulted in a significantly increased incidence of resorptions of embryos over that in a control group implanted with pumps delivering only distilled water. The incidence of gross malformations exclusively in the form of encephaloceles was not different between control and azide-infused groups. The extent of nitrosylation of circulating hemoglobin was followed with time and found to involve only about 0.1% of the total blood pigment. Thus, this commercially important and widely distributed chemical with high acute toxicity is not considered to be teratogenic in hamsters, and it produces embryotoxicity only at dose rates that result in toxic signs in the dams.

Effect of cyanide on the reaction of nitroprusside with hemoglobin: relevance to cyanide interference with the biological activity of nitroprusside.

The reaction of sodium nitroprusside (SNP) with deoxyhemoglobin (Hb) results in two distinct EPR-detectable species, the one-electron-reduced nitroprusside ion [(CN)5FeNO]3- and nitrosylhemoglobin (HbNO). In the presence of excess cyanide (CN-) only the signal for [(CN)5FeNO]3- is observed. Thus, while free CN- does not interfere with Hb reduction of SNP, it prevents transfer of the NO moiety to Hb. Electrolytic reduction of SNP under similar conditions, however, leads to [(CN)5FeNO]3- and a small amount of [(CN)4FeNO]2- resulting from loss of the CN- trans to the NO. Excess free CN- shifts the equilibrium between these two species toward [(CN)5FeNO]3-, thereby reducing the concentration of [(CN)4FeNO]2-. Thus, [(CN)4FeNO]2- appears to be responsible for the transfer of NO to Hb. Consistent with this mechanism, both [(CN)5FeNO]3- and [(CN)4FeNO]2- are observed when SNP is added to erythrocyte lysates. Under these conditions HbNO is formed more rapidly due to the higher concentration of the latter species with the labile NO. This observation suggests that red blood cell constituents capable of binding CN- shift the equilibrium between the reduced SNP ions toward [(CN)4FeNO]2-. In the reaction of reduced glutathione (GSH) with SNP, [(CN)5FeNO]3- is formed as well as low concentrations of an EPR-detectable GSH-SNP adduct. Excess free CN- introduces a lag in the appearance of these signals, suggesting that GSH mediates SNP reduction by a different mechanism from that of Hb, although it too is inhibited by CN-.(ABSTRACT TRUNCATED AT 250 WORDS)

Decomposition and specific exchange of the trans-cyanide ligand on nitroprusside is facilitated by hemoglobin.

Most investigators agree that sodium nitroprusside (SNP) undergoes biotransformation in vivo to release free cyanide (CN-). Despite a demonstrated reactivity of SNP toward oxyhemoglobin (HbO2) in vitro, that reaction is probably not the major cause for CN- release in vivo. An unknown reaction in various vascular beds inactivates SNP more rapidly than the reaction in blood, but both result in the release of free CN-. Recently it has been claimed that SNP is stable in blood and that the apparent CN- release is an artifact due to photodecomposition of SNP during CN- analyses. In this study the release of free CN- from SNP was followed over 3 hr of incubation in plain buffer, in suspensions of red blood cells (RBC), in lysates, and in solutions of HbO2 purified by isoelectric focusing and shown to be free of methemoglobin (MetHb), valency hybrid species, and reduced glutathione (GSH). In each case the mixtures containing HbO2 released significantly more CN- than plain buffer where CN- release was at the limit of sensitivity of the method. Moreover, in the order of RBC, purified HbO2 solutions and lysates, each preparation released significantly more CN- in 1 hr than the preceding one. In lysates the reaction had gone to completion by 3 hr. Using 13C nuclear magnetic resonance (NMR) spectroscopy, SNP was shown to be inert to CN- exchange in solutions of Na13CN, but when GSH or MetHb was added, an exchange reaction occurred between the trans-CN- of SNP and 13CN- in solution. The same reaction proceeded even more rapidly in the presence of HbO2. The results of this study show that in the presence of GSH, MetHb and particularly in the presence of HbO2, SNP readily exchanges its trans-CN- ligand with excess free CN-. This reaction is believed to represent an obligatory precursor step in the total decomposition of SNP which occurs in the absence of free CN-, but in the presence of RBC, solutions of HbO2, and lysates. It appears that excess free CN- halts total decomposition at the stage of trans-CN- labilization.

Generation of valency hybrids and nitrosylated species of hemoglobin in mice by nitric oxide vasodilators.

The hemoglobin fraction of blood samples from C57BL/6 male mice was purified by column chromatography and subjected to isolectric focusing (IEF) across a pH gradient. Densitometric scanning of the IEF gel showed the presence of a single peak corresponding to the fully reduced tetramer, H, or (alpha 2+ beta 2+)2. Twenty minutes after an ip injection of 1.1 mmol/kg NaNO2 blood samples treated the same way showed four peaks corresponding to the species: H = 43%, X or (alpha 2+ beta 3+)2 = 10%, Y or (alpha 3+ beta 2+)2 = 33%, and M or (alpha 3+ beta 3+)2 = 14%. In contrast blood samples from control CD-1 male mice showed the presence of three IEF distinct peaks which were all believed to be H valency forms, and six distinct peaks were seen after treatment in vivo with NaNO2. Thus, the C57BL/6 mice yield patterns similar to those observed after in vitro treatment of human red cells with NaNO2 (H. Kruszyna, R. Kruszyna, R. P. Smith, and D. E. Wilcox, 1987b, Toxicol. Appl. Pharmacol. 91, 429-438), and the CD-1 mice are a much less satisfactory model. The appearance and disappearance of the species X, Y, and M over time after ip injection of 1.1 mmol/kg NaNO2 or hydroxylamine HCl were followed in C57BL/6 mice by the technique of IEF. In each case the patterns were consistent with previously established patterns for the respective methemoglobinemias as determined by absorption spectrophotometry, and they were consistent with the suggestion that two pathways exist for the oxidation of H and for the reduction of M which proceed through X and Y, respectively. By using electron paramagnetic resonance (EPR) spectroscopy, we were also able to follow with time the concentration of nitrosylated heme (NO-heme) on reduced subunits in both mouse strains. The peak for the NO-heme coincided in time with the peak methemoglobinemia as determined by either IEF or absorption spectrophotometry. EPR was also used to determine NO-heme in CD-1 mice after injection of a series of NO-vasodilators with and without methylene blue (MB). Low, but clearly detectable amounts of NO-heme were found in the blood of animals given all xenobiotics tested including NaNO2, hydroxylamine HCl, glyceryl trinitrate, hydralazine, sodium nitroprusside, and sodium azide. MB has little effect on the response, and no NO-heme could be detected in control mice.(ABSTRACT TRUNCATED AT 400 WORDS)

Red blood cells generate nitric oxide from directly acting, nitrogenous vasodilators.

Human red blood cells (RBC) incubated under nitrogen with methylene blue and glucose at physiological temperature and pH can be used to test for the biotransformation of nitrogenous vasodilators to nitric oxide (NO). The NO generated was trapped as nitrosylated heme by reduced subunits (hemeII) on various hemoglobin valency species and quantified by electron paramagnetic resonance spectroscopy. It was possible to separate the various valency species of hemoglobin present in the mixture as (alpha 2 + beta 2)2, (alpha 2 + beta 3+)2, (alpha 3 + beta 2+)2, or (alpha 3 + beta 3+)2 by isoelectric focusing (IEF) unless cyanide (from nitroprusside) or azide was present in the mixture. These anions bind tenaciously to oxidized subunits (hemeIII) and prevent the separation of the various species by IEF. The fully oxidized tetramer, (alpha 3 + beta 3+)2, does not bind NO, but the other three species have hemeII units which can be nitrosylated. In the absence of cyanide or azide the valency species could be separated by IEF, and it was possible to quantify the degree of nitrosylation on each individual species. The various agents tested (nitrite, glyceryl trinitrate, hydroxylamine, hydralazine, nitroprusside, and azide) produced different patterns of valency species and degrees of nitrosylation of hemeII. When hemeIII ligands were present or in cases of very low yields, it was still possible to quantify the total concentration of NO-hemeII in the mixture. Thus, the method could still be used to test for NO formation. All of the so-called NO vasodilators tested yielded detectable amounts of NO in the system.

Nitrite conversion to nitric oxide in red cells and its stabilization as a nitrosylated valency hybrid of hemoglobin.

The authors describe a method for the preparation, isolation and purification from human red blood cells of a stable form of hemoglobin containing both oxidized and reduced subunits. After isoelectric focusing across a pH gradient, it occupies the same position as the synthetically reconstituted or chemically generated species, (alpha 2+ beta 3+)2 (I). Unlike its synthetic or chemically generated counterpart, however, it (HbX) does not bind oxygen. A different method for the preparation in situ of the (alpha 3+ beta 2+)2 valency hybrid results in a pigment with chemical and spectral properties identical with those ascribed to its synthetically reconstituted counterpart, and both bind oxygen. HbX is generated in highest yield when red cells are incubated under N2 in the presence of glucose, methylene blue and nitrite for several hours. Its visible absorption spectrum differs from that reported for I, and it reacted very slowly with ferricyanide. Exposure to CO did not result in spectral shifts over that for HbX in air, but spectral shifts were produced with CO exposure of (alpha 3+ beta 2+)2. HbX had an inositol hexaphosphate difference-binding spectrum quite unlike that described for I. HbX also had a distinctive electron paramagnetic resonance spectrum that shifted after inositol hexaphosphate addition with a new three-line hyperfine pattern. Both spectra were characteristic for an unpaired electron in an iron-centered orbital with a hyperfine coupling to the nitrogen nuclear spin of a nitrosyl ligand on heme iron. The authors conclude that HbX is a form of I, but it has NO bound to its reduced subunits.

Cyanide and sulfide interact with nitrogenous compounds to influence the relaxation of various smooth muscles.

Sodium nitroprusside relaxed guinea pig ileum after the segment had been submaximally contracted by either histamine or acetylcholine, intact isolated rabbit gall bladder after submaximal contraction by either acetylcholine or cholecystokinin octapeptide, and rat pulmonary artery helical strips after submaximal contraction with norepinephrine. In each of these cases the relaxation produced by nitroprusside was at least partially reversed by the subsequent addition of excess sodium cyanide. Cyanide, however, in nontoxic concentrations did not reverse the spasmolytic effects of hydroxylamine hydrochloride, sodium azide, nitroglycerin, sodium nitrite, or nitric oxide hemoglobin on guinea pig ileum, nor did cyanide alone in the same concentrations have any effect. The similar interaction between nitroprusside and cyanide on rabbit aortic strips is not dependent on the presence of an intact endothelial cell layer. Also, on rabbit aortic strips and like cyanide, sodium sulfide reversed the spasmolytic effects of azide and hydroxylamine, but it had little or no effect on the relaxation induced by papaverine. Unlike cyanide, however, sulfide augmented the relaxation induced by nitroprusside, and it reversed the effects of nitric oxide hemoglobin, nitroglycerin, and nitrite. A direct chemical reaction between sulfide and nitroprusside may account for the difference between it and cyanide. Although evidence was obtained also for a direct chemical reaction between sulfide and norepinephrine, that reaction does not seem to have played a role in these results. These observations suggest the existence of at least three distinct subclasses of so-called nitric oxide vasodilators. At least in some cases cyanide and sulfide cannot be acting by the same mechanism in their modifications of the responses to the agonists.

Hypotensive effects of hydroxylamine in intact anesthetized dogs and cats.

When given IV in bolus doses to intact anesthetized dogs or cats, hydroxylamine hydrochloride produced transient but precipitous falls in the mean arterial blood pressure in a dose-related manner, as well as a significant methemoglobinemia. The half-time for recovery of the mean arterial pressure was also dose-related. These effects were very similar to those elicited by comparable doses of sodium nitrite, except that the half-recovery time for return to normal blood pressure was somewhat longer with nitrite. Although hydroxylamine has long been known to relax vascular smooth muscle in vitro, we are not aware of previous demonstrations of hypotensive effects in vivo. Acute poisoning by either nitrite or hydroxylamine is apt to result in both an anemic hypoxia due to methemoglobinemia and a stagnat (hypokinetic) hypoxia due to direct vasodilation. Hydroxylamine, but not nitrite, also appeared to stimulate respiration possibly through an effect on the chemoreceptors of the carotid body.