Location and Characteristics of Nitric Oxide Synthase in Sheep Spinal Cord and Its Interaction with α²-Adrenergic and Cholinergic Antinociception

After obtaining Animal Care and Use Committee approval, ten nonpregnant ewes of mixed Western breeds were anesthetized with 50 mg/kg intravenous pentobarbital and killed by intravenous injection of potassium chloride. The thoracic spinal cord was removed and immediately placed in ice-cold 50 mM Tris/HCl buffer, pH 7.4, containing 0.1 mM ethylenediamenetetraacetic acid, 0.5 mM dithiothreitol, 12 mM 2-mercaptoethanol, 2 micro Meter leupeptin, 1 micro Meter pepstatin A, and 1 mM phenylmethylsulfonic fluoride. Spinal meninges and vessels were removed by careful dissection and the spinal cord was frozen in liquid nitrogen and stored at -70 degrees C. In other experiments, six Sprague-Dawley rats were killed by decapitation under deep pentobarbital anesthesia and the cerebellum prepared and stored as described for spinal cord.

Homogenates (25% w/v) were prepared on ice with a tissue homogenizer and centrifuged at 1,000g for 10 min at 0-4 degrees C. The supernatants were passed through a 10DG desalting column (BioRad, Hercules, CA) to remove all substances of molecular mass less than 6,000 d. The eluents were centrifuged at 100,000g for 60 min at 4 degrees Celsius. This supernatant was considered to contain cytosolic nitric oxide synthase. The 100,000g pellet was washed three times and resuspended with an equal volume of the above Tris buffer, then recentrifuged at 100,000g for 60 min at 4 degrees C. The supernatant was discarded and the pellet was washed and resuspended with a sonicator (Cole Palmer, Chicago, IL) for 10 s x 3 with an equal volume of the above Tris buffer. This pellet was considered to contain membrane-bound nitric oxide synthase.

Nitric oxide synthase activity was determined by the conversion of [sup 14 Carbon]-L-arginine to [sup 14 Carbon]-citrulline, using a modification of a previously described method. [20]Briefly, 100-micro liter samples were added to 100 micro liter of reaction mixture containing 50 mM Tris/HCl, pH 7.4, containing 1.35 micro Meter [sup 14 Carbon]-L-arginine (final reaction mixture concentration 0.68 micro Meter; 200,000 dpm/tube, Dupont/NEN, Boston, MA) purified by AG1X-8 resin (OH-form, BioRad), 1 mM NADPH, 1.25 mM CaCl², micro gram/ml calmodulin, 2 micro Meter FAD (flavin adeninedinucleotide), 2 micro Meter FMN (flavin mononucleotide), and 10 micro Meter tetrahydrobiopterin (BH sub 4). The reaction tubes were shaken in a 32 degrees C water bath for 20 min and the reaction stopped by addition of 2 ml ice-cold buffer containing 30 mM hydroxyethylpiperazineethane sulfonic acid and 3 mM ethylenediamenetetraacetic acid. The reaction mixture was then applied to a chromatography column of 0.5 ml Dowex AG50WX-8 resin (Sodium+ form, pH 7.0, Sigma Chemical, St. Louis, Mo). [sup 14 Carbon]-Citrulline was eluted with 2 ml buffer and quantified by liquid scintillation with an efficiency of 95-97% (LKB, 1219, Rakebete, MD). Protein was measured as previously described [21]using bovine serum albumin as standard. Nitric oxide synthase activity was expressed as pmol [sup 14 Carbon]-citrulline sup -1 *symbol* mg protein sup -1 *symbol* min sup -1.

To validate separation of citrulline by the column, 10 micro liter of reaction mixture after Dowex AG50WX-8 resin separation was applied to silica gel 60 plates (20 x 20 aluminum, Alltech Associates, Los Altos, CA) for thin-layer chromatography, using a mobile system containing CHCl³/methanol/NH⁴OH 2:2:1 (vol/vol), and staining with ninhydrin spray. Each lane was cut into 10-mm sections and radioactivity was measured by scintillation counting. The retention factors with this system are 0.25 for arginine, 0.33 for ornithine, and 0.76 for citrulline.

Dependence of [sup 14 Carbon]-citrulline production on the following cofactors was tested: (1) Calcium²+: Deletion of calcium from the reaction mixture and addition of 0.9 mM ethylenediamenetetraacetic acid, and 2.5 mM MgCl². (2) Calmodulin: Addition of calmodulin (10 micro gram/ml) to the reaction mixture, or addition of the calmodulin antagonist, calmidazolium, 0.1 mM. (3) NADPH: Deletion of NADPH from the reaction mixture. (4) BH⁴: Addition of BH sub 4, 10 micro Meter to the reaction mixture. (5) FAD/FMN: Addition of 2 micro Meter FAD, and 2 micro Meter FMN to the reaction mixture.

[sup 14 Carbon]-Citrulline production was determined in the presence of L-NAME, and aminoguanidine (10 sup -8 to 10 sup -3 M). N^G-nitro-D-arginine methyl ester (D-NAME), 10 sup -5 M, was used as a control. To determine the effect of substrate concentration, [sup 14 Carbon]-citrulline production was determined in the presence of unlabeled L- and D-arginine (10 sup -5 M).

For kinetic analysis, nitric oxide synthase activity was determined by adding increasing concentrations of unlabelled L-arginine (2.5, 5, 10, 15, 20 micro Meter). A Line-weaver-Burke plot was constructed and the maximal velocity (V^max) and Michaelis Menton constant (K^m) were determined by linear regression.

Nitric oxide synthase activity was determined in the 100,000g supernatant from tissue dissected from the dorsal horn, the intermediolateral cell column (IML), and the ventral horn, as previously described in sheep. [22]In addition, freshly removed spinal cord was fixed in 4% paraformaldehyde, frozen, and cut into 40-micro meter sections using a cryostat. Sections were stained for NADPH diaphorase with nitrotetrazolium blue as previously described. [23]NADPH diaphorase was indicated in this method by intense dark blue staining.

After approval from the Animal Care and Use Committee, 14 ewes of mixed Western breeds weighing between 45 and 50 kg were studied. After a 48 h fast, anesthesia was induced with 20-30 mg/kg intramuscular ketamine and maintained with 1.5-2% inhalational halothane. A polyvinyl catheter was inserted percutaneously into a jugular vein, and 80 mg gentamicin and 1.6 mg atropine were administered intravenously. The ewe was positioned prone and a midline incision was made over the cervical vertebral bodies to expose the dura mater at the C1-C2 intervertebral space. A small nick was made in the dura mater through which a polyvinyl 21-G catheter was inserted and advanced 8 cm caudad. Intrathecal location of the catheter was confirmed by free flow of cerebrospinal fluid via the catheter before surgical closure. The catheter was secured with skin sutures and taped to the neck. Each sheep received 800,000 units aqueous penicillin and 1 g dihydrostreptomycin intramuscular for 3 days after surgery. Prophylactic analgesics were not administered, although the nonsteroidal anti-inflammatory drug, flunixin (1.1 mg/kg intravenous) was available for behavioral evidence of pain. In all cases, animals were standing and eating within 1 h of completion of surgery, and no animal demonstrated behavioral evidence of pain requiring treatment. Studies were begun 5 days after surgery.

Antinociception Testing. All ewes were placed in stainless steel metabolic carts with their heads restrained during the period of testing to prevent excessive movement. The wool covering both front legs was sheared just above the carpus. Antinociception was quantified using a pressure stimulus to the shaft of the radius as previously described. [24]Mechanical thresholds were measured using a pneumatically driven device, which gradually pushed a blunt metal pin against the skin over the lower end of the radius. The force applied to the foreleg is simultaneously displayed on a digital meter in newtons and charted on a standard strip chart recorder. A predetermined cutoff of 20 N was used to prevent tissue damage. Increasing pressure resulted in a well-defined withdrawal of the leg in response to the noxious stimulus, at which time pressure was immediately released and the force was recorded. Each sheep was trained for a minimum of 2 or 3 days before any study. The stability of baseline withdrawal threshold was tested in pilot studies.

Drug Protocol. For each experiment, baseline withdrawal thresholds were determined, followed by intrathecal injection of the test drug and retesting 15, 30, 45, and 60 min after injection. Animals then received incremental intrathecal injections of clonidine to achieve cumulative doses of 10, 30, and 100 micro gram. Clonidine was administered at 30-min intervals to coincide with onset to peak effect. [15].

Animals were studied 4 or 5 times, with experiments separated by a minimum of 2 days and performed in random order. All 14 sheep were studied with intrathecal saline and with intrathecal neostigmine pretreatment before clonidine, whereas the other experimental groups consisted of seven animals to reduce the number of exposures of each animal to the test compounds. Previous studies have not demonstrated tolerance to the effect of intrathecal clonidine in sheep using this treatment schedule, [15]and a separate analysis revealed no effect of order of experiment on result in any group.

To test the nitric oxide-dependence of clonidine's antinociceptive effect, animals were pretreated with 1 mg L-arginine, 1 mg D-arginine, 1 mg NMLA, or the combination of L-arginine and NMLA. To test the nitric oxide-dependence of neostigmine's enhancement of clonidine's effect, animals were pretreated with neostigmine plus 1 mg L-arginine, neostigmine plus 1 mg D-arginine, and neostigmine plus 1 mg NMLA. Doses of neostigmine and NMLA were chosen based on previous studies in sheep [8,15,25]and the doses of L- or D-arginine were determined in pilot experiments.

BH⁴was purchased from Research Biochemicals International (Natick, MA). All other drugs were obtained from Sigma Chemical. Aminoguanidine was dissolved at a stock concentration of 40 mM in 0.065 N HCl in water and adjusted to pH 7.4. Calmidazolium was prepared in dimethylsulphoxide. For antinociception studies, all drugs were dissolved in preservative-free normal saline and injected in a 1.0-ml volume followed by 0.5 ml normal saline flush (2x catheter dead space).

Results are expressed as mean+/-SEM. For studies of nitric oxide synthase activity, comparisons between means were made by Student's unpaired t test, or by analysis of variance followed by Newman-Keuls test. All values for IC⁵⁰were obtained from sigmoid logistic curves using SigmaPlot (Jandel Scientific, San Rafael, CA).

Antinociception responses were converted to the percent maximum possible effect (% MPE) according to the following formula: Equation 1.

Individual dose-response curves were compared using a two-way analysis of variance for repeated measures. The dose producing 50% maximal effect (ED⁵⁰) was calculated by standard methods, [26]and these values were compared among groups by one-way analysis of variance. P < 0.05 was considered significant.

Thin-layer chromatography revealed that the [sup 14 Carbon]-labeled product eluted from the Dowex AG50WX-8 column was purely [sup 14 Carbon]-citrulline (data not shown).

Nitric oxide synthase activity in supernatant was less in sheep spinal cord (23+/-1.3 pmol sup -1 *symbol* mg protein sup -1 *symbol* min sup -1) than in rat cerebellum (160+/-3.5 pmol sup -1 *symbol* mg protein sup -1 *symbol* min sup -1; P < 0.01).

Both the cytosolic (supernatant) and membrane-bound particulate fraction (pellet) of sheep spinal cord caused [sup 14 Carbon]-citrulline formation. However, the relative amounts of activity and dependence on cofactors differed between the supernatant and the pellet. Approximately 85% of the total [sup 14 Carbon]-citrulline formation was in the supernatant. Nitric oxide synthase activity in the supernatant was nearly abolished by removal of Calcium²+ or NADPH, or by addition of the calmodulin antagonist, calmidazolium, in the reaction mixture, whereas these manipulations had less effect on [sup 14 Carbon]-citrulline production by the pellet (Figure 1). Similarly, [sup 14 Carbon]-citrulline formation was increased significantly in the supernatant, but not the pellet, by additions of BH⁴, calmodulin, FAD, and FMN (Figure 2).

[sup 14 Carbon]-Citrulline production was linear over 30 min of incubation. Kinetic analysis yielded a K^mfor L-arginine of 2.45 micro Meter and V^maxof 91 pmol sup -1 *symbol* mg protein sup -1 *symbol* min sup -1.

L-NAME was more potent at inhibiting [sup 14 Carbon]-citrulline production than was aminoguanidine (IC⁵⁰= 1.36 x 10 sup -7 M for L-NAME versus 3.4 x 10 sup -5 M for aminoguanidine; Figure 3). D-NAME caused no inhibition of nitric oxide synthase activity (Figure 4). Similarly, isotope dilution by addition of excess unlabeled L-arginine reduced production of [sup 14 Carbon]-citrulline from the [sup 14 Carbon]-L-arginine tracer, whereas addition of unlabeled D-arginine had no such effect (Figure 4).

Cytosolic nitric oxide synthase activity was concentrated in tissue dissected from the dorsal horn, followed by tissue from the IML (Figure 5). Tissue from the ventral horn contained approximately one third the nitric oxide synthase activity as that from the dorsal horn. Similarly, NADPH diaphorase histochemistry revealed dense staining in the dorsal horn and IML, with minimal cellular staining in the ventral horn (Figure 6).

The overall mean mechanical withdrawal threshold before drug injection was 5.8+/-1.0 N (mean+/-SD). None of the pretreatments affected mechanical withdrawal thresholds before clonidine injection (data not shown). Intrathecal clonidine produced dose-dependent antinociception in all experiments. Clonidine-induced antinociception was unaffected by NMLA, but was enhanced by L-(but not D-)arginine, and this enhancement by L-arginine was antagonized by NMLA (Figure 7, left). Neostigmine enhanced clonidine-induced antinociception, and this enhancement was not further increased by L-arginine, but was antagonized by NMLA and by D-arginine (Figure 7, right). Similarly, ED⁵⁰analysis revealed a twofold or threefold reduction in the apparent ED⁵⁰of clonidine by L-arginine and by neostigmine, and this reduction in ED sub 50 by L-arginine or neostigmine was blocked by NMLA (Table 1).

These data in sheep provide confirmation of work in other species concerning the location, identity, and function of nitric oxide synthase in the spinal cord, but also provide new insights into the role of nitric oxide in nociceptive processing.

Although it is well recognized that nitric oxide synthase, demonstrated by NADPH histochemistry or immunocytochemistry, is localized in the dorsal horn and the IML in mammalian spinal cord, as confirmed in this study, there has been little evaluation of the nature of this enzymatic activity. Nitric oxide synthase isoforms have been divided by tissue and subcellular location, dependence on cofactors, sensitivity to inhibition by differing compounds, quantity of nitric oxide synthesized, and ability of experimental conditions to induce increased activity. [10]Definition by these pharmacologic conditions has been recently supported by molecular biologic techniques, which demonstrate unique proteins, under the synthetic control of different genes, for these isoforms. [13].

The results of the current study suggest that nitric oxide synthase in the sheep spinal cord is constitutive, cytosolic, and consistent with the brain isoform. Increased sensitivity to L-NAME compared to aminoguanidine suggests a constitute nitric oxide synthase, [16,17](but see Laszlo et al. [27]) such as that present in endothelium, glia, and neurons. [28]Cytosolic nitric oxide synthase activity in the current study is Calcium²+ -, calmodulin-, and NADPH-dependent, as would be expected with neuronal nitric oxide synthase. [20,29]Similarly, the K^mfor L-arginine observed in the current study is similar to that previously observed for neuronal nitric oxide synthase. [20,30]The quantitative comparison of sheep spinal cord nitric oxide synthase activity with that of rat cerebellum, the brain area of highest such enzyme activity and most well characterized, must be qualified because optimal substrate conditions were not rigorously defined in rat tissue in the current experiment. Nonetheless, the results suggest a lower amount of such activity in sheep spinal cord, which is consistent with immunocytochemical studies.

Although contradictory data regarding the dependence of neuronal nitric oxide synthase on BH⁴, FMN, and FAD have been reported, it is now well accepted that to obtain maximal enzyme activity these cofactors need to be added. Binding domains for the flavins FMN and FAD are found in neuronal nitric oxide synthase, and these cofactors bind to neuronal nitric oxide synthase in stoichiometric amounts. [31]Highly purified neuronal nitric oxide synthase has an absolute requirement for these cofactors. [32]The results of the current study are in keeping with the reported dependence of rat cerebellar nitric oxide synthase for these cofactors, and suggest partial loss of cofactors during sample preparation. Inhibition of product formation from [sup 14 Carbon]-L-arginine by unlabeled L-arginine in the current study is far less than that predicted by simple isotopic dilution. This result is inconsistent with either simple competitive competition at nitric oxide synthase, or with an isotopic rate effect and requires an alternative explanation.

A small amount of activity remained in the pellet after high speed centrifugation. Similarly, some nitric oxide synthase activity is present in the particulate fraction of the rat cerebellum homogenates, [11]and it was speculated to be of endothelial origin. In contrast to that report, the small amount of residual activity in the particulate preparation in the current study was minimally dependent on BH⁴, FAD, FMN, and calmodulin. Because [sup 14 Carbon]-citrulline production by the pellet was also NADPH-independent, this activity is unlikely to represent true nitric oxide synthase.

Both cytosolic [sup 14 Carbon]-citrulline production and NADPH diaphorase staining localized nitric oxide synthase activity to dorsal horn and IML in the sheep spinal cord. This agrees with the findings of previous functional studies in sheep, which demonstrated an L-NAME reversible effect of cholinergic ligands on spinal sympathetic outflow and spinal analgesia. [25]Similarly, these results agree with studies in rodents, which demonstrate a concordance between NADPH diaphorase histochemistry and nitric oxide synthase immunocytochemistry in the spinal cord, [18,19,33]and localization of both to the dorsal horn and IML.

The discrepancy between the relatively large amount of nitric oxide synthase activity in ventral horn tissue by [sup 14 Carbon]-citrulline production and small amount of NADPH diaphorase staining is somewhat puzzling. Although rat ventral horn motoneurons are capable of expressing nitric oxide synthase after spinal cord trauma, [34]they are not demonstrable by NADPH diaphorase or nitric oxide immunocytochemical staining in normal animals, similar to our results in sheep. Whether nitric oxide synthase is present, but inhibited in these cells in vivo, but is disinhibited in the in vitro conditions of the [sup 14 Carbon]-citrulline assay used in the current study is uncertain.

As in vascular endothelium, where acetylcholine is the prototypical stimulator for nitric oxide synthesis, there is mounting evidence that neuronally released acetylcholine in the spinal cord produces antinociception in part by stimulating nitric oxide synthesis. For example, behavioral analgesia from intrathecal injection of muscarinic agonists in rats is inhibited by nitric oxide synthase blockers. [7]Similarly, there is tonic spinal cholinergic activity in rats, and the antinociceptive result of this activity is antagonized by nitric oxide synthase blockers. [35]Except for the first day after laminectomy surgery, sheep do not exhibit tonic spinal cholinergic activity as evidenced by antinociception from intrathecal neostigmine alone, [36]precluding us from examining the nitric oxide dependence of antinociception from this agent alone in this species.

Antinociception from spinal alpha²-adrenoceptor stimulation is at least partially caused by activation of cholinergic neurons. In sheep, antinociception from intrathecal clonidine or dexmedetomidine is enhanced by neostigmine, but is not inhibited by the muscarinic antagonist, methylatropine. [8,15,25]We have hypothesized that a relatively small proportion of antinociception from spinal alpha sub 2 -adrenoceptor stimulation depends on acetylcholine release. [15]As such, blockade of this small contribution by methylatropine may reduce the potency of alpha²-adrenergic agonists by a small amount, but severalfold magnification of this cholinergic contribution by inhibition of acetylcholinesterase can have a greater effect on the potency of alpha sub 2 -adrenergic agonists.

The results of the current study are consistent with this hypothesis as well as with the nitric oxide-dependence of cholinergic agonists for spinal antinociception. Thus, L-arginine enhanced, whereas NMLA did not affect antinociception from intrathecal clonidine, similar to the interactions with cholinesterase inhibitors and cholinergic antagonists. Lack of enhancement by D-arginine suggests the effect of L-arginine was not a nonspecific action on clonidine distribution or bio-availability. Also, blockade of neostigmine-induced enhancement of intrathecal clonidine antinociception by NMLA is consistent with a nitric oxide-dependence of spinal cholinergic stimulation. [7]Whether the blockade of neostigmine enhancement of clonidine antinociception by D-arginine is owing to reduction in nitric oxide synthase substrate availability is not clear from the current experiments.

In summary, these data confirm location of nitric oxide synthase activity to the dorsal horn and IML of sheep, and enzyme kinetics, dependence on cofactors, and sensitivity to inhibitors suggest it is the neuronal nitric oxide synthase isoform. Antinociception from intrathecal clonidine is partially dependent on nitric oxide synthesis, and neostigmine-induced enhancement of intrathecal clonidine antinociception is totally dependent on nitric oxide synthesis.