Neuronal Nitric Oxide Synthase Activation by Vasoactive Intestinal Peptide in Bovine Cerebral Arteries

Abstract

The participation of nitric oxide and vasoactive intestinal peptide (VIP) in the neurogenic regulation of bovine cerebral arteries was investigated. Nitrergic nerve fibers and ganglion-like groups of neurons were revealed by NADPH-diaphorase staining in the adventitial layer of bovine cerebral arteries. NADPH diaphorase also was present in endothelial cells but not in the smooth muscle layer. Double immunolabeling for neuronal nitric oxide synthase and VIP indicated that both molecules co-localized in the same nerve fibers in these vessels. Transmural nerve stimulation (200 mA, 0.2 milliseconds, 1 to 8 Hz) of endothelium-denuded bovine cerebral artery rings precontracted with prostaglandin F2α, produced tetrodotoxin-sensitive relaxations that were completely suppressed by N^G-nitro-l-arginine methyl ester (l-NAME) and by the guanylyl cyclase inhibitor 1H-[1,2,4]oxadiazolo[4,3-a]quinoxaline (ODQ), but were not affected by the adenylyl cyclase inhibitor 9-(tetrahydro-2-furanyl)-9H-purin-6-amine (SQ 22,536), nor by VIP tachyphylaxis induced by pretreatment with 1 μmol/L VIP. Transmural nerve stimulation also elicited increases in intracellular cyclic GMP concentration, which were prevented by l-NAME, and small decreases in intracellular cyclic AMP concentration. Addition of VIP to bovine cerebral artery rings without endothelium produced a concentration-dependent relaxation that was partially inhibited by l-NAME, ODQ, and SQ 22,536. The effects of l-NAME and SQ 22,536 were additive. VIP induced a transient increase in intracellular cyclic GMP concentration, which was maximal 1 minute after VIP addition, when the highest relaxation rate was observed, and which was blocked by l-NAME. It is concluded that nitric oxide produced by perivascular neurons and nerve fibers fully accounts for the experimental neurogenic relaxation of bovine cerebral arteries and that VIP, which also is present in the same perivascular fibers, acts as a neuromodulator by activating neuronal nitric oxide synthase.

Immunohistochemical and ultrastructural studies show that cerebral arteries are surrounded by a dense network of perivascular nerves, part of which contains vasodilator neuropeptides such as vasoactive intestinal peptide (VIP) (Larsson et al., 1976; Suzuki et al., 1988), substance P (Uddman et al., 1981; Suzuki et al., 1989), or calcitonin gene-related peptide (Suzuki et al., 1989). In addition, functional studies reveal that cerebral arteries relax when perivascular nerves are electrically stimulated. This effect occurs in the presence of adrenergic and cholinergic receptor blockade (Lee et al., 1978; Toda, 1981) and it is therefore considered to be a nonadrenergic noncholinergic (NANC) neurogenic relaxation. It has been recently reported that the neurogenic relaxation of cerebral arteries from several species is substantially reduced when nitric oxide (NO) synthesis is prevented by l-arginine analogues (Toda and Okamura, 1990; González and Estrada, 1991; Gaw et al., 1991; Ayajiki et al., 1993). The NO involved in neurogenic relaxation originates in perivascular nerves containing nitric oxide synthase (NOS) and NADPH-diaphorase activity (Bredt et al., 1990; Estrada et al., 1993; Nozaki et al., 1993).

Although NO currently is considered as a peripheral neurotransmitter involved in NANC relaxation in many organs (Ny et al., 1995; Kim et al., 1995; Takahashi et al., 1995; Takahashi and Owyang, 1995), a controversy exists as to whether neuropeptides, such as VIP or calcitonin gene-related peptide, also contribute to the neurogenic relaxation, based on the existence of a relaxation component resistant to NOS inactivation in some tissues (Kim et al., 1995; Ny et al., 1995; Takahashi et al., 1995; Takahashi and Owyang, 1995). In fact, VIP, which colocalizes with NOS in peripheral nerves from different territories (Ekblad et al., 1994; Hayashida et al., 1996; Takahashi et al., 1995; Klimaschewski et al., 1994; Barroso et al., 1996), has been shown to contribute to NANC relaxation in some tissues, such as the rabbit corpus cavernosum (Kim et al., 1995), the cat lung (Takahashi et al., 1995), or the rat stomach (Takahashi and Owyang, 1995), but not in others, such as the cat esophagus (Ny et al., 1995) or the human trachea (Ward et al., 1995). In cerebral blood vessels, VIP has been proposed as a functional dilator neurotransmitter (Lee et al., 1984; Bevan et al., 1986; Gaw et al., 1991), but conclusive experimental evidence has never been provided.

Nitric oxide and VIP exert their relaxant effect on smooth muscle cells through two separate signal transduction mechanisms. Activation of guanylyl cyclase by NO results in the accumulation of cyclic GMP (cGMP) (Schmidt et al., 1993), whereas the interaction of VIP with specific membrane receptors leads to increases in cyclic AMP (cAMP) concentration on activation of adenylyl cyclase (Huang and Rorstad, 1984; Suzuki et al., 1985).

In the current report, we analyze the participation of these two pathways in the relaxation produced by stimulation of perivascular nerves in bovine cerebral arteries and their possible interaction. We show that neuronal NO fully accounts for the neurogenic relaxation and that VIP, which is present in the same nerve terminals, acts as a neuromodulator by regulating NO synthesis in the nitrergic nerves.

METHODS

Bovine brains were obtained from a local slaughterhouse and transported to the laboratory in phosphate-buffered saline solution (PBS) at 4°C. A first-order branch of the anterior cerebral artery was isolated for functional experiments and for biochemical determinations. The endothelium was removed by gently introducing a cotton thread through the vascular lumen, taking care to preserve the adventitial layer.

Isometric tension recordings

Arterial rings, 4 mm in length, were suspended on two intraluminal parallel wires, introduced in an organ bath containing a physiologic solution (in mmol/L: 115 NaCl, 4.6 KCl, 1.25 CaCl₂, 25 NaHCO₃, 1.2 KH₂PO₄, 1.2 MgSO₄, 0.01 EDTA, and 11 glucose), and connected to a Piodem strain gauge for isometric tension recording. All rings were equilibrated at a passive tension of 1 g for 90 minutes. Before starting the experiments, rings were contracted with prostaglandin F_2α and acetylcholine (1 to 10 μmol/L) was added to confirm functionally the absence of endothelium. Only arteries that did not dilate in the presence of acetylcholine were used. Sodium nitroprusside (10 μmol/L) was added at the end of each experiment, and the relaxation obtained was considered to be 100%. In experiments in which guanylyl cyclase was inhibited, the maximal relaxation was obtained with papaverine (100 μmol/L) in both control and treated vessels.

Transmural nerve stimulation (TNS; 200 mA, 0.2 milliseconds, 1 to 8 Hz) was applied to precontracted arterial rings using two parallel platinum electrodes, one at each side of the vessel, connected to a CS-20 stimulator (Cibertec, Madrid, Spain). The neurogenic nature of the stimuli was demonstrated, since TNS responses were blocked by tetrodotoxin (1 to 10 μmol/L). The interval between consecutive stimuli was at least 8 to 10 minutes. TNS always was performed in the presence of 5 μmol/L indomethacin, 1 μmol/L phentolamine, 1 μmol/L propranolol, and 0.1 μmol/L atropine to prevent any effect mediated by prostaglandins, noradrenaline, or acetylcholine. Two or three reproducible responses were obtained before starting the experiments. N^G-nitro-l-arginine methyl ester (l-NAME, 100 μmol/L), 1H-[1,2,4]oxadiazolo[4,3-a]quinoxaline (ODQ, 10 μmol/L), or 9-(tetrahydro-2-furanyl)-9H-purin-6-amine (SQ 22,536, 200 μmol/L) were added to the organ bath, and the effect of TNS was recorded 20 minutes later.

The VIP concentration curves were performed in a cumulative manner in precontracted arteries. Only one curve was obtained for each vascular ring to avoid tachyphylaxis. Experiments in the presence and absence of l-NAME, ODQ, or SQ 22,536 were performed in pairs of vessels isolated from the same animal.

Cyclic GMP and cyclic AMP measurements

Arterial rings were mounted in the organ baths as for the functional experiments. The absence of relaxing responses to acetylcholine and the presence of TNS-induced relaxations were assessed in each segment. When l-NAME was added, TNS was applied 20 minutes later to confirm that NOS was inhibited. Arteries were contracted with prostaglandin F_2α and frozen with liquid nitrogen–embedded forceps without any further drug (controls), 60, 105, and 240 seconds after addition of 10 nmol/L VIP, and 12 and 40 seconds after initiation of TNS (8 Hz, 12 seconds) in the absence and presence of l-NAME. Frozen samples later were used for cGMP and cAMP determinations by radioimmunoassay using a kit from New England Nuclear (DuPont, Boston, MA, U.S.A.).

Histochemistry

The entire circle of Willis and its major branches were dissected out of the brain in the slaughterhouse, rinsed and perfused with PBS, and immediately introduced in fixative. The arteries used for NADPH-diaphorase staining were fixed for 2 hours in 4% paraformaldehyde in 0.1 mol/L phosphate buffer, pH 7.4, at 4°C, and then washed with phosphate buffer and crioprotected by incubation for 1 hour with increasing concentrations of sucrose (5% to 20%, weight/volume) and overnight with 30% (weight/volume) sucrose. Longitudinal sections (40 μm) were obtained with a freezing microtome, mounted on slides, and incubated for 70 minutes at 37°C in a mixture containing 1 mmol/L NADPH, 1 mmol/L nitro blue tetrazolium, and 0.3% Triton-X-100, in 0.1 mol/L phosphate buffer, pH 8.0. The sections were rinsed in phosphate buffer, dehydrated, and coverslips were fixed with Permount.

The vessels used for immunohistochemical study were fixed for 16 hours at 4°C in Zamboni's fixative (Steffanini et al., 1967) and washed in PBS containing 15% (weight/volume) sucrose and 0.01% (weight/volume) sodium azide. After pretreatment with 0.2% (volume/volume) Triton X-100 in PBS for 2 hours, and with Pontamine Sky blue (Cowen et al., 1985) for 30 minutes, the arterial segments were incubated overnight at room temperature with a rabbit antibody against NOS (Hammersmith Hospital/Wellcome, 1:400) and a guinea-pig antibody against VIP (Euro-Diagnostica, 1:300) diluted in PBS. Arteries then were washed in PBS and incubated with a mixture of tetramethylrhodamine isothiocyanate–conjugated mouse antiserum against rabbit IgG (Sigma, 1:100) and fluorescein isothiocyanate–conjugated mouse antiserum against guinea-pig IgG (Sigma, 1:10) for 1 hour at room temperature. The arteries were mounted and examined under a microscope equipped for epiillumination with filters selective for fluorescein and rhodamine fluorescence. In control experiments, no immunostaining was observed when primary antisera were omitted, replaced with nonimmune serum, or preadsorbed with their corresponding antigens (0.1 to 1 μmol/L) for 24 hours at 4°C. Labeled secondary antisera did not exhibit cross-reactivity with IgG from inappropriate species.

Statistical analysis

Data are expressed as arithmetic means plus/minus standard deviation, except for EC₅₀ values, which are presented as geometric means with confidence limits. Statistical comparisons between groups were made using Student's t test for paired or unpaired samples, except for EC₅₀ values, which were analyzed by the nonparametric Wilcoxon test. P < 0.05 was considered to be statistically significant.

Chemicals

The following compounds were used: prostaglandin F_2α, acetylcholine chloride, sodium nitroprusside, tetrodotoxin, indomethacin, phentolamine hydrocloride, dl-propranolol, atropine methyl bromide, N_W-nitro-l-arginine methyl ester, hemoglobin (bovine), VIP (porcine), NADPH, and nitro blue tetrazolium, all of them purchased from Sigma Chemical Co. (St. Louis, MO, U.S.A.). SQ 22,536 was from Research Biochemicals International (Natick, MA, U.S.A.), and ODQ was from Tocris (Bristol, U.K.). Stock solutions (10 to 30 mmol/L) were made in water except for indomethacin, prostaglandin F_2α, and ODQ, which were dissolved in 5% NaHCO₃, ethanol, and dimethyl sulfoxide, respectively. All subsequent dilutions were made in 0.9% NaCl.

RESULTS

Histochemical staining for NADPH-diaphorase activity in longitudinal sections of bovine pial arteries revealed the presence of large bundles of positive-staining fibers running longitudinally in the adventitial layer (Fig. 1A). These bundles branched off in different directions, leading to single varicose nerve fibers (Fig. 1B). In some areas, heavily stained neurons presenting clear nuclei and grouped in a ganglion-like manner were observed (Figs. 1C and 1D). NADPH-diaphorase staining also was present in endothelial cells but not in the smooth muscle layer (Fig. 1C). Double immunolabeling for neuronal NOS and VIP in whole-mounted arterial segments showed a co-localization of both markers in the same perivascular nerve fibers (Figs. 1E and 1F).

Fig. 1.

(A and B) Longitudinal sections of bovine pial arteries through the adventitial layer, stained for NADPH diaphorase. Positive nerve bundles with a branching pattern and containing varicosities are observed (×200). (C) Longitudinal section of a bovine pial artery across the vessel wall, stained for NADPH diaphorase. Cross-sectioned fibers (arrowhead) and ganglion-like structures (arrows) in the adventitial layer (ADV) showed positive staining. A portion of labeled endothelium can be seen at the bottom (END). No staining was observed in the smooth muscle cell layer (SM). (×290). (D) Higher magnification (×600) of the ganglion-like structures shown in Fig. 1C. Two groups of positive neuronal cell bodies (arrows) with clear nuclei (arrowheads) are visualized. (E and F) Whole-mount preparation of a bovine anterior cerebral artery double immunostained for NOS (E) and VIP (F). Both VIP and NOS immunoreactivities coexist in the same nerve varicosities (arrows) (×250).

To determine the contribution of NO and VIP to the relaxation induced by depolarization of perivascular nerves, TNS was applied to endothelium-denuded precontracted arterial rings before and after blockade of NOS, guanylyl cyclase, or adenylyl cyclase. TNS induced a transient relaxation that was abolished by the NOS inhibitor l-NAME (Figs. 2A and B), but not by the inactive enantiomer d-NAME (Fig. 2A). The effect of l-NAME was partially reversed by the simultaneous administration of l-arginine (Fig. 2A). In the presence of the specific guanylyl cyclase inhibitor ODQ (Garthwaite et al., 1995; Moro et al., 1996), arterial rings did not relax on TNS (Figs. 2A and B); however, the neurogenic response was not modified by the adenylyl cyclase inhibitor SQ 22,536 (Fig. 2B). Simultaneous application of l-NAME and SQ 22,536 produced the same effect as l-NAME alone (Fig. 2B). In addition, TNS response was not altered when applied after 30 minutes' preincubation of the vessels in the presence of 1 μmol/L VIP (Fig. 2B), conditions in which VIP response was clearly diminished because of tachyphylaxis (Fig. 2C).

Fig. 2.

(A) Representative recordings of the relaxation induced by transmural nerve stimulation (8 Hz, 12 seconds) in two endothelium-denuded bovine cerebral arteries, precontracted with prostaglandin F_2α, before and after treatment with 100 μmol/L d-NAME, 10 μmol/L l-NAME, and 10 μmol/L l-NAME plus 3 mmol/L l-arginine (left panel), or with 10 μmol/L ODQ (right panel). (B) Relaxation of endothelium-denuded bovine cerebral arteries induced by TNS, in the absence (control) and presence of 100 μmol/L l-NAME (NAME), 10 μmol/L ODQ, 200 μmol/L SQ 22,536 (SQ), 100 μmol/L l-NAME plus 200 μmol/L SQ 22,536 (SQ + NAME) or after incubation for 30 minutes with 1 μmol/L VIP. (C) Relaxation induced by 2 nmol/L VIP in control conditions and after incubation for 30 minutes with 1 μmol/L VIP showing tachyphylaxis. Values represent the mean ± SD of the results obtained in arterial segments from five to eight animals. *P < 0.05.

The intracellular concentrations of the cyclic nucleotides cGMP and cAMP were measured before and during TNS-induced relaxation. In agreement with the functional results described earlier, significant increases of cGMP concentration were elicited by TNS (Fig. 3A), whereas a concomitant small decrease in cAMP concentration was produced (Fig. 3B). The TNS-induced increment in cGMP concentration was caused by guanylyl cyclase activation by NO because it was abolished in the presence of l-NAME (Fig. 3A). Basal concentrations of cGMP in unstimulated endothelium-denuded arteries were significantly lower than those measured in intact vessels (60 ± 8 fmol/mg protein, n = 15, versus 200 ± 40 fmol/mg protein, n = 8, respectively), and similar to those detected in intact vessels when NOS was inhibited by l-NAME (60 ± 15 fmol/mg protein, n = 4). In the absence of endothelium, l-NAME produced no further decrease in basal cGMP concentration (Fig. 3A).

Fig. 3.

(A) Intracellular cyclic GMP (cGMP) concentrations in bovine cerebral arteries in control conditions and during application of transmural nerve stimulation (TNS), in the absence and presence of 100 μmol/L l-NAME. (B) Intracellular cyclic AMP (cAMP) concentrations in bovine cerebral arteries in control conditions and during application of TNS. Values represent the mean ± SD of the results obtained in three to eight arterial segments. *P < 0.05 as compared with control values.

To analyze the signal transduction mechanisms of VIP action in bovine cerebral arteries, concentration–response curves were obtained when the NO/cGMP pathway was blocked by either l-NAME or ODQ, when the cAMP pathway was inhibited by SQ 22,536, and when transduction through both pathways was prevented by simultaneous addition of l-NAME and SQ 22,536. Addition of VIP to arterial rings without endothelium produced a sustained concentration-dependent relaxation (Fig. 4). The relaxation induced by low concentrations of the peptide was inhibited by l-NAME (Fig. 4A), resulting in a significant increase in the EC₅₀ values without modification of the E_max (Table 1). In the presence of ODQ, a reduced response also was observed (Fig. 4B), except at the highest concentration used. Analysis of the data revealed a small but significant increase in EC₅₀ values without modification of the E_max (Table 1), an effect similar to that of l-NAME. SQ 22,536 also inhibited VIP-induced relaxations in bovine cerebral arteries, although its effects were more evident at higher peptide concentrations (Fig. 4C), leading to a significant decrease in the E_max values without modification of the EC₅₀ (Table 1). When both l-NAME and SQ 22,536 were added simultaneously, the relaxation produced by VIP was decreased throughout the whole range of the concentration–response curve, indicating that both effects were additive (Fig. 4D and Table 1).

Fig. 4.

Relaxation induced by cumulative concentrations of VIP in bovine cerebral arteries without endothelium in the absence (open circles) and in the presence (filled circles) of 100 μmol/L l-NAME (A), 10 μmol/L ODQ (B), 200 μmol/L SQ 22,536 (C), or 100 μmol/L l-NAME plus 200 μmol/L SQ 22,536 (D). Values represent the mean ± SD of the results obtained in paired arterial segments from five to eight animals. *P < 0.05.

Table 1.

EC₅₀ and E_max values for VIP-induced relaxation in bovine cerebral arteries in the absence and presence of L-NAME, SQ 22,536 or both

	EC50*	Emax†
Control (7)	1.5(0.02−6.1)	97.2 ± 10
L-NAME (7)	3.7 (0.8−11.9)‡	96.8 ± 10
Control (8)	4.88 (−0.9−19.1)	95.8 ± 8.5
ODQ (8)	10.3 (−7.3−47.1)‡	84.4 ± 18.3
Control (7)	3.9 (2.2−6.9)	89.9 ± 15
SQ 22,536 (7)	3.7(1.8−11.6)	68.1 ± 25‡
Control (5)	4.0 (2.4−6.4)	85.8 ± 15
SQ 22,536 + L-NAME (5)	3.9 (−4.6−20.7)	52.9 ± 22‡

*

Values (nmol/L) are expressed as geometric means with confidence limits.

†

Percentage of maximum relaxation obtained with sodium nitroprusside or papaverine in each segment. Values are expressed as mean ± SD.

‡

P < 0.05 as compared with control values.

To assess further the participation of the NO–cGMP pathway in the relaxation induced by VIP in bovine cerebral arteries, cGMP levels were measured at different times after addition of the peptide. Cyclic GMP increased 1 minute after addition of VIP to the organ bath, remained elevated for 2 minutes, and returned to control levels 4 minutes after addition of the peptide (Fig. 5A). In the presence of l-NAME, no increase in cGMP was observed on VIP administration (Fig. 5B).

Fig. 5.

(A) Intracellular cyclic GMP (cGMP) concentration in bovine cerebral arteries, in control conditions and at different times after addition of 10 nmol/L VIP. (B) Intracellular cyclic GMP concentrations in bovine cerebral arteries in control conditions and 1 minute after addition of 10 nmol/L VIP, in the absence and in the presence of 100 μmol/L l-NAME. Values represent the mean ± SD of the results obtained in 8 to 14 arterial segments. *P < 0.05, as compared with control and with NAME + VIP.

DISCUSSION

We investigated the involvement of NO and VIP in the regulation of bovine cerebral artery function by perivascular nerves. Our results suggest that there is an interplay between these two messengers, which are contained in the same nerve population, with NO being the peripheral neurotransmitter directly relaxing smooth muscle cells and VIP acting as an autocrine neuromodulator.

The presence of perivascular nitrergic nerves, shown here by NADPH-diaphorase staining, used as a marker for NOS (Hope et al., 1991; Dawson et al., 1991), and by NOS immunohistochemical study, is in agreement with previous results in the cerebral arteries of rats, cats, guinea-pigs, and humans (Barroso et al., 1996; Bredt et al., 1990, Estrada et al., 1993, Nozaki et al., 1993). However, it is interesting to notice the detection in bovine vessels of stained neuronal cell bodies grouped in ganglion-like structures, with morphologic features similar to those previously described in cats (Estrada et al., 1993; Lee, 1994). Although nitrergic neurons in the sphenopalatine ganglion have been shown to project to cerebral arteries in the rat (Nozaki et al., 1993; Minami et al., 1994; Suzuki et al., 1994), the presence of intrinsic NOS neurons in the cerebral vessel wall in cats and cows suggests that some of the nitrergic nerves have a local origin, at least in these species.

The coexistence of NOS and VIP demonstrated here for bovine cerebral arteries seems to be a common feature of NANC nerves in the peripheral nervous system. Partial or total co-localization of these two transmitters has been reported in different locations, including the gastrointestinal tract of several species (Ekblad et al., 1994), the canine corpus cavernosum (Hayashida et al., 1996), rat nasal mucosa (Takahashi et al., 1995), guinea-pig heart (Klimaschewski et al., 1994), and cerebral blood vessels (Nozaki et al., 1993; Barroso et al., 1996). This frequent co-localization of VIP and NOS in peripheral nerve terminals suggests a possible functional interaction between these two relaxing pathways.

Our results show that VIP-mediated activation of adenylyl cyclase did not occur during neurogenic stimulation of bovine cerebral blood vessels, because of the following: (1) there was no intracellular accumulation of cAMP, (2) the relaxation was not impaired by inhibition of adenylyl cyclase with SQ 22,536, and (3) VIP-induced tachyphylaxis did not modify the vascular relaxation to TNS. On the contrary, the guanylyl cyclase pathway must have been activated, since (1) significant increases in intracellular cGMP concentrations were detected during TNS, and (2) the relaxing response to TNS was completely blocked by the adenylyl cyclase inhibitor ODQ. The disappearance of any significant relaxation or cGMP accumulation in the presence of the NOS inhibitor l-NAME points to the conclusion that the NANC relaxation of bovine cerebral arteries was entirely caused by activation of the NO–cGMP pathway.

One interesting finding is that cGMP levels were low in endothelium-denuded arteries and that they were not further decreased by l-NAME. This indicates that there is no basal NO production in nitrergic terminals in unstimulated arteries. On the contrary, endothelial NOS is tonically activated, since, in resting conditions, arteries with intact endothelium presented higher cGMP concentrations, and these decreased to the levels measured in the absence of endothelium when treated with l-NAME.

Although VIP does not behave as a neurotransmitter in bovine cerebral vessels, the possibility exists that it acts as a neuromodulator by modifying the release of the coexisting neurotransmitter, as is the case with other neuropeptides. We show here that VIP causes relaxation of bovine cerebral arteries by two independent pathways. First, and in accordance with the well known interaction of VIP with adenylyl cyclase–coupled receptors (Huang and Rorstad, 1984; Suzuki et al., 1985), VIP relaxes the vessel by adenylyl cyclase activation, since the relaxation decreased in the presence of an inhibitor of this enzyme. Second, it activates the NO–cGMP pathway, as demonstrated by the increase in cGMP concentration and the partial inhibition of the VIP-induced relaxation by l-NAME and ODQ. The NO–cGMP pathway was predominant at low VIP concentrations, whereas the adenylyl cyclase inhibitor was more effective at high VIP concentrations. Furthermore, the cGMP increases induced by VIP were maximal 1 minute after addition of the peptide, the time at which the highest relaxation rate was reached, and returned to control levels 3 minutes later, when the vessel still was relaxed, indicating that the NO–cGMP pathway is important in the initial stages of VIP-induced relaxation. These findings suggest that in bovine cerebral arteries, VIP activates NOS by acting on receptors that differ from those coupled to adenylyl cyclase in both the affinity and the time course of the signal transduction. In relation to this possibility, it has recently been reported that, in dispersed gastric muscle cells, VIP interacts with a newly defined class of specific receptors, not recognized by the peptide histidine–isoleucine, which induce a G protein–dependent influx of Ca²⁺ and a subsequent activation of NOS (Murthy et al., 1993).

Shortly before the identification of perivascular nitrergic plexi in the cerebral vasculature, Gaw and others (1991) described a decrease in the relaxation produced by VIP in sheep cerebral arteries on incubation with hemoglobin or the NOS inhibitor N^G-monomethyl-l-arginine, and concluded that the neurogenic vasodilation was largely mediated by VIP through a direct action on smooth muscle by a mechanism involving synthesis of NO. However, it is unlikely that endogenous VIP released by TNS acts on NOS-related smooth muscle receptors, because an interaction with smooth muscle cells also would result in a concomitant participation of the cAMP-mediated pathway. Moreover, in contrast to what happens in the gastrointestinal smooth muscle, where a new isoform of Ca²⁺/calmoduline–dependent membrane-bound NOS has been described (Berezin et al., 1994; Murthy and Makhlouf, 1994), this enzyme is not present in cerebral artery smooth muscle cells, as demonstrated by the absence of NADPH-diaphorase staining both in arterial sections (current results) and in isolated smooth muscle cells derived from bovine cerebral arteries and maintained in culture (Estrada et al., 1995). The removal of the endothelial layer and the absence of NADPH diaphorase in the smooth muscle cells in our arterial segments point to the perivascular nitrergic fibers and neurons as the only source of NOS in these vessels. Because of the co-localization of NOS and VIP, we propose that this neuropeptide acts as a neuromodulator, increasing the synthesis and release of the coexisting neurotransmitter NO by acting on regulatory autoreceptors coupled to NOS activation.

Acknowledgments: The authors thank Professor Julia Polak for providing the antiserum to NOS, and Ms. Carmen Gómez and Ms. Angeles Bravo for their valuable technical assistance.