Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Normal orofacial blood flow is important for the maintenance of healthy teeth and gums. The blood vessels and tissues of the oral cavity are continuously exposed to foods, spices, bacteria and endogenous substances (e.g., kallikreins and kinins) that stimulate the production of endogenous prostanoids (Greenberg and Palmer, 1978; Samuelsson, 1979). The evidence that oxygen-derived free radicals may play an important role in vascular reactivity was provided by experiments demonstrating that, in cerebral arterioles (Kontos, 1985), an acute increase in arterial blood pressure causes excessive activation of arachidonic acid via the cyclooxygenase pathway with subsequent increased production of O₂ (Okabe et al., 1985). O₂ may then escape into the extracellular space via an anion channel (Kontos, 1985) and, after dismutation to H₂O₂, form HO·. The HO· is a potent oxidant, and studies by our group demonstrated that formation of HO· causes prolonged vasodilation and oxidative vascular injury (Okabe et al., 1983; Marshall and Kontos, 1990).

Vascular responses to oxygen-derived free radicals are currently a very active area of investigation, because these agents have been implicated in several pathological conditions and in the vascular changes associated with ischemia-reperfusion (Frei, 1994) and aging (Feher, et al., 1987). It has been indicated that the dominant effect of oxygen-derived free radicals is vasodilation (Lamb and Webb, 1984); however, they can also evoke vasoconstriction (see Rubanyi, 1988). We have previously shown that HO·-induced injury of endothelial cell inhibits both the production and effect of endothelium-derived relaxing factor (Todoki et al., 1992). Thus, impairment of vasodilator function caused by excessive production of free radicals sets the stage for increased reactivity to vasoconstrictor stimuli; free radicals generated in the vascular wall may act directly on smooth cells, or interact with the production and/or biological activity of endogenous vasoactive mediators formed in endothelial cells.

The inhibitory mechanical responses of the rabbit lingual artery, which subserves blood flow to the tongue and gums by tributaries, consists of two components; one is atropine-sensitive but the other is a noncholinergic component (Brayden and Large, 1986). This type of mixed dilator response is not unusual; it has been noted in several different tissues including the cat salivary gland (Lundberg, 1981), nasal mucosa (Eccles and Wilson, 1974), tongue (Lundberg et al., 1982) and the dog hind limb (Brody and Shaffer, 1970). The identity of the noncholinergic inhibitory transmitter in these tissues is not known. CGRP, a 37-amino acid peptide, is synthesized via the alternative processing of the primary RNA-transcript of the calcitonin gene (Amara et al., 1982; Rosenfeld et al., 1983). CGRP has been show to be widely distributed in the central and peripheral nervous system (Goodman and Iverson, 1986), and it also exists in nerve fibers throughout the cardiovascular system (Mulderry et al., 1985). We reported recently that CGRP, but not NO, released from nonadrenergic, noncholinergic nerves in response to perivascular nerve stimulation produces relaxation of canine lingual artery that is mediated by activation of CGRP₁ receptors (Kobayashi et al., 1995). An important role for CGRP as a mediator of neurogenic vasodilation of the lingual artery has thus been established.

It is widely accepted that in some of the pathophysiological conditions, the first target of oxygen-derived free radicals is the vascular system (Rubanyi, 1988). However, the effects of oxygen-derived free radicals on neurotransmission in CGRP nerves remained unknown. We, therefore, tested the direct effects of oxygen-derived free radicals on CGRP-mediated neurogenic vasodilation in canine lingual artery. The present study addresses two questions: (1) Does exposure of the artery preparation to oxygen-derived free radicals (especially HO· generated from Fenton's reagent) impair CGRP-mediated neurogenic relaxation? (2) What is the mechanism of impaired relaxation at the cellular level?

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Mongrel dogs of either sex (10-15 kg) were used for all studies.

Vessel preparation, isometric tension recording and electrical stimulation. In accordance with our institutional Animal Care Committee guidelines, the animals were anesthetized with sodium pentobarbital (30 mg/kg i.v.) and then sacrificed by rapid exsanguination. The lingual artery, which is the largest collateral branch of the external carotid artery, was carefully isolated and dissected into several ring segments after immersion in ice-cold modified Krebs-Ringer solution (118.3 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 25.0 mM NaHCO₃, 0.026 mM calcium disodium ethylenediaminetetraacetic acid and 11.1 mM glucose, aerated with 95% O₂/5% CO₂; pH 7.2-7.3). The endothelium was removed from all rings by gently rubbing the luminal surface with a wooden implement.

6

5

5

6

6

6

HO·-generating system. For the HO·-generating system, 3 × 10⁴ M H₂O₂ plus 2 × 10⁴ M FeSO₄ were used. A simple mixture of H₂O₂ and an iron (II) salt forms the HO·, as was first observed by Fenton in 1894 (see Walling, 1982, for a review): Catalase and DMTU were used to scavenge H₂O₂ and HO·, respectively; deferoxamine, a powerful iron chelator, was also used. The timed sequence of reagent addition is described under "Results."

ESR analysis. The production of HO· by Fenton's reagent was verified by ESR spectroscopy. ESR detection of the spin adduct was performed by a JES-RE3X, X-band spectrometer (Jeol, Tokyo, Japan) at the following instrument settings: modulation amplitude, 0.2 mT (100 kHz); receiver gain, 2.5 × 10; recording range, 5 mT; recording time, 2 min; time constant, 0.03 sec; microwave power, 8 mW; and magnetic field, 335.6 ± 5 mT. DMPO (100 mM) was used as the spin trap. The desired reaction mixtures (200 µl) were prepared in the reaction bath (37°C) and transferred to a quartz ESR flat cell (130 µl), which was in turn placed in the cavity of the ESR spectrometer; ESR detection of the spin adduct was performed at room temperature. Sequential ESR scans were then started 30 sec after the addition of DMPO to the reaction mixture. To quantitate the DMPO spin adducts detected, the Mn⁺⁺ standard ESR spectrum (MnO) was obtained.

Immunohistochemistry. The intact and HO·-exposed endothelium-denuded ring preparations were fixed in 40% paraformaldehyde and picric acid for 24 to 72 hr at 4°C (Saito et al., 1986). The tissues were dehydrated with ethanol embedded in paraffin and sectioned at 5 µm. Anti-CGRP polyclonal antibody (RPN 1842; Amersham, Buckinghamshire, England) was applied at 1:300 dilution for 16 to 24 hr at 4°C, and sections were stained by the immunoperoxidase technique with a streptoavidin-biotin kit (Nichirei, Tokyo, Japan); then they were counter stained lightly with Mayer's hematoxylin. Negative control staining was also carried out by the substitution of the normal rabbit serum for the primary antiserum.

Reagents. The drugs and chemicals used were: acetylcholine chloride, catalase, capsaicin, glibenclamide, guanethidine monosulfate, norepinephrine hydrochloride, superoxide dismutase (from bovine blood, 2,800 U/mg of protein) (Sigma, St. Louis, MO); DMTU, H₂O₂ (Aldrich, Milwaukee, WI); nitroglycerin (Nihon Kayaku, Tokyo, Japan); deferoxamine mesylate (Ciba-Geigy); DMSO, FeSO₄, papaverine hydrochloride (Wako Chemicals, Osaka, Japan); DMPO (Labotec, Tokyo, Japan; 99-100% pure, GC assay by Dojindo Laboratories, Kumamoto, Japan); cromakalim (Pola Cosmetics, Yokohama, Japan); CGRP and human-8-37 CGRP (Peptide Institute, Inc., Osaka, Japan). All agents except for capsaicin, cromakalim and glibenclamide were dissolved in distilled water. Capsaicin, cromakalim or glibenclamide was dissolved in DMSO and diluted in Krebs-Ringer solution before being added to the tissue bath (final concentration of DMSO, 0.1 mM). All other reagents were of analytical grade.

Data presentation. All data were expressed as mean ± S.E.M. The statistical tests of significance used were one-way analysis of variance and the Dunnett's multiple range test. Statistical software was obtained from Ricoh (Tokyo, Japan). A significance level of P < .05 was used to reject the null hypothesis.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

In preliminary experiments, oxygen-derived free radical species generated from H₂O₂ plus FeSO₄ under the same reaction conditions as those of free radical exposure studies was verified by using a highly sensitive ESR spectroscopy and the spin-trap DMPO in the absence of the artery ring preparations. The 1:2:2:1 quartet (the hyperfine splittings were An = A_H = 1.49 mT), characteristic of the HO·-DMPO spin adduct (Zweier, 1988), was observed (fig. 1). H₂O₂ or FeSO₄ alone did not produce the DMPO-HO· adduct (data not shown). The signal intensity of the second peak of the spectrum (normalized as a relative height against the standard signal intensity of the MnO marker) reached maximum value (15.3) immediately after the addition of H₂O₂/FeSO₄ into the bathing media in the presence of DMPO; the relative signal intensity was decreased, in a time-dependent manner, when DMPO was added 5 to 40 min after initiation of the Fenton reaction (0.8 at 40 min) (fig. 1). Thus, the exposure conditions to the ring preparations were chosen. We have used 40 min of exposure, which was repeated two times (to give 80 min of total exposure to HO·).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   ESR spectra of DMPO-HO· produced by Fenton's reagent (3 × 104 M H2O2 plus 2 × 104 M FeSO4) in the bathing media. H2O2/FeSO4 was added to the bathing media in the presence of spin-trap DMPO (0 min), or DMPO was added 5, 10, 20, 30 or 40 min after the addition of H2O2/FeSO4; and then after 30 sec, sequential ESR scans were started. The bathing medium was the same as that of the free radical exposure study except that canine lingual artery rings were omitted. Signals appearing at both sides of the ESR spectra correspond to Mn++ (MnO) installed in the ESR cavity as a reference. Experimental conditions for ESR detection are as described under "Materials and Methods."

We have previously confirmed that the endothelium-denuded canine lingual artery responds to electrical stimulation (4-16 Hz), and this response was abolished when guanethidine (5 × 10⁶ M) was added 20 min before the initiation of the electrical stimulation; even after guanethidine removal (washing), electrical stimulation elicited no response (Kobayashi et al., 1995). Therefore, in the presence of NE-induced active tone (NE, 10⁵ M), the preparations had no ability to contract in response to electrical stimulation (in the absence of NE) relaxed, in a frequency-dependent fashion, when neural elements in the arteries were stimulated electrically (fig. 2). Under the same experimental conditions (in guanethidine-treated ring preparations), the relaxation produced by electrical stimulation (4-16 Hz) at the steady-state (equilibrium) contraction induced by NE was nearly completely diminished by previous exposure to H₂O₂/FeSO₄ (fig. 2A). To determine whether it was HO· that were responsible for the observed inhibitory effect, the role of catalase, DMSO, DMTU and deferoxamine in protecting the relaxation of the ring preparations induced by electrical stimulation (4-16 Hz) from the H₂O₂/FeSO₄ exposure was tested. The results of this series of experiments are shown in table 1. Catalase, DMTU, DMSO, deferoxamine and a cocktail of catalase-deferoxamine significantly protected the relaxation against H₂O₂/FeSO₄ exposure, which suggested involvement of HO· generated from the Fenton reaction.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Effects of HO· exposure (A) and capsaicin (B) on the relaxant response in the presence of active tone induced by NE (105 M) of endothelium-denuded canine lingual artery rings to electrical stimulation (10V, 4-16 Hz, for 45 sec). The effect of guanethidine (5 × 106 M) on electrical stimulation in the absence (abolition of contraction) or presence (relaxation instead of the abolition of contraction) of active tone induced by NE was confirmed in all ring preparations before the start of the experiments; the denudation of the endothelium was also checked by the absence of relaxation induced by 105 M acetylcholine in all rings contracted to a stable plateau tension by the addition of NE. The ring preparation was exposed to HO· (generated from 3 × 104 M H2O2 plus 2 × 104 M FeSO4) for 40 min; the ring was serially washed and then exposed to HO· (A). Capsaicin (CAP, 106 M) treatment was performed for 30 min (B). Tracings are representative of six or more experiments.

View this table: [in this window] [in a new window]   TABLE 1 Effect of early exposure to H2O2/FeSO4 reaction on the relaxant response of endothelium-denuded canine lingual artery rings to electrical stimulation (10 V, 4-16 Hz, for 45 sec) and effects of various reactive oxygen species scavengers The endothelium-denuded rings were previously treated with guanethidine (5 × 106 M), and the effect was confirmed in all rings before the start of the experiments; the denudation of the endothelium was also checked (see fig. 2, legend). Exposure to H2O2/FeSO4 reaction of the ring preparations was performed under the same conditions as those of figure 2 except that the scavengers used were added 1 min before the addition of H2O2/FeSO4 to the bathing medium. Mean values ± S.E.M. (n = 5-7) are shown; n refers to the number of dogs from which the lingual artery was taken. When added, doses were: 3 × 104 M H2O2, 2 × 104 M FeSO4, 100 U/ml catalase, 1 mM DMTU, 100 mM DMSO and 1 mM deferoxamine.

The abolition afforded by treatment with capsaicin of the electrical stimulation-induced relaxation of lingual artery ring preparations demonstrated in figure 2B is similar to that seen in HO·-exposed ring preparations (fig. 2A) and may be a result of similar mechanisms. Therefore, it is possible that HO· may damage the capsaicin-sensitive nerves and, by so doing, affect CGRP-mediated relaxation of the lingual artery preparations. An immunohistochemical test of this hypothesis is presented in figure 3. In control endothelium-denuded ring preparation, numerous CGRP-like immunoreactive fibers were found to be present in the adventitia; the CGRP-like immunoreactive fibers showed typical varicose profiles (fig. 3A). After the ring preparation was treated with capsaicin (fig. 3B) or exposed to H₂O₂/FeSO₄ (fig. 3C), the CGRP-like immunoreactivity was markedly diminished, which suggested that the relaxant response to electrical stimulation disappeared when endogenous CGRP was depleted by HO· like by capsaicin (Buck and Burks, 1986).

View larger version (102K): [in this window] [in a new window]   Fig. 3.   CGRP-like immunoreactivity in canine lingual artery. There were many immunoreactive fibers in the control tissue (A); immunoreactive material was observed in the adventitial layer. After exposure to capsaicin (B) or HO· (C), the intensity of CGRP-like immunoreactivity was reduced markedly. Exposure to capsaicin (106 M) or to HO· (generated from 3 × 104 M H2O2 plus 2 × 104 M FeSO4) of the artery preparations was performed under the same conditions as those of figure 2.

To determine whether the relaxation elicited by CGRP and cromakalim, a ATP-sensitive K⁺ channel opener, is altered by HO·, the influence of H₂O₂/FeSO₄ exposure on the effects of exogenously added CGRP and cromakalim was assessed. Figure 4 shows that previous exposure to HO· attenuated the human CGRP-(8-37)-inhibitable relaxation produced by exogenous CGRP. HO· also diminished the relaxation elicited by cromakalim (fig. 5A); the relaxant response to cromakalim was glibenclamide-sensitive (fig. 5B). The inhibitory effect of HO· on the relaxation of the ring preparations induced by CGRP and cromakalim was significantly protected by catalase, DMTU, DMSO, deferoxamine and a cocktail of catalase-deferoxamine (table 2); the protective effects of scavengers used are similar to the inhibitory effect on HO·-induced attenuation of the relaxation produced by electrical stimulation (see table 1).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Effects of HO· exposure (A) and human CGRP-(8-37) (B) on the relaxant response to exogenously added CGRP in the lingual artery rings precontracted with NE. The endothelium-denuded rings were previously treated with guanethidine (5 × 106 M), and the effect was confirmed in all rings before the start of the experiments (see fig. 2 legend). Exposure to HO· of the ring preparations was performed under the same conditions as those of figure 2. When added, doses were: 108 M CGRP, 3 × 104 M H2O2, 2 × 104 M FeSO4 and 2 × 10 8 M human CGRP-(8-37). Human CGRP-(8-37) was added 30 min before the addition of NE (105 M) to the bathing media. Tracings are representative of six experiments.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Effects of HO· exposure (A) and glibenclamide (B) on the relaxant response to cromakalim in the lingual artery rings precontracted with NE. The endothelium-denuded rings were treated previously with guanethidine (5 × 106 M), and the effect was confirmed in all rings before the start of the experiments (see fig. 2, legend). Exposure to HO· of the ring preparations was performed under the same conditions as those of figure 2. When added, doses were: 106 M cromakalim, 3 × 104 M H2O2, 2 × 104 M FeSO4 and 105 M glibenclamide. Glibenclamide was added 30 min before the addition of NE (105 M) to the bathing medium. Tracings are representative of six experiments.

View this table: [in this window] [in a new window]   TABLE 2 Effect of early exposure to H2O2/FeSO4 reaction on the relaxant response of endothelium-denuded canine lingual artery rings to CGRP and cromakalim and effects of various reactive oxygen species scavengers The endothelium-denuded rings were previously treated with guanethidine (5 × 106 M), and the effect was confirmed in all rings before the start of the experiments; the denudation of the endothelium was also checked (see fig. 2, legend). Exposure to H2O2/FeSO4 reaction of the ring preparations was performed under the same conditions as those of figures 4A and 5A, except that the scavengers used were added 1 min before the addition of H2O2/FeSO4 to the bathing medium. Mean values ± S.E.M. (n = 5-6) are shown; n refers to the number of dogs from which the lingual artery was taken. When added, doses were: 3 × 104 M H2O2, 2 × 104 M FeSO4, 100 U/ml catalase, 1 mM DMTU, 100 mM DMSO, 1 mM deferoxamine, 108 M CGRP and 106 M cromakalim.

The nitrovasodilators act on sensory fibers to release CGRP, which then diffuses to the vascular smooth muscle where it activates soluble guanylate cyclase to cause vasodilation in feline cerebral arterioles (Wei et al., 1992). A test of this in canine lingual artery is presented in figure 6. The relaxant response to nitroglycerin was unaffected when CGRP stores had been diminished by previous HO· exposure, which suggested that HO· did not have an effect on guanylate cyclase-dependent mechanisms in the canine lingual artery.

View larger version (9K): [in this window] [in a new window]   Fig. 6.   Effect of HO· exposure on the relaxant response to nitroglycerin in the endothelium-denuded lingual artery ring precontracted with NE (105 M). The ring was previously treated with guanethidine (5 × 106 M), and the effect was confirmed before the start of the experiment (see fig. 2, legend). Exposure to HO· of the ring preparation was performed under the same conditions as those of figure 2. When added, doses were: 105 M nitroglycerin, 3 × 104 M H2O2 and 2 × 104 M FeSO4. Tracings are representative of five experiments.

We have examined the possibility that electrolysis of a physiological buffer serves as a source of oxygen-derived free radicals, and may generate O₂ and/or closely related species of free radicals, thereby reducing the relaxation induced by electrical stimulation. In our system, relaxation response to electrical stimulation (4-16 Hz) of the ring preparations was unaffected by the treatment with superoxide dismutase (56 and 110 U/ml), catalase (100 U/ml), DMSO (100 mM) or DMTU (1 mM) (data not shown). Thus, it seems that the interaction between free radicals generated from electrolysis of a physiological buffer used and relaxant response to electrical stimulation is negligible under the experimental conditions used in the present study.

Figure 7 shows that production of the HO·-DMPO spin adduct by H₂O₂/FeSO₄ was effectively blunted by HO· scavengers (DMTU and DMSO, fig. 7A). Upon the addition of DMSO but not DMTU to the system, another spectrum appeared. The hyperfine coupling constants of this signal are An = 1.64 mT and A_H = 2.24 mT. These values coincide with the values reported for DMPO-CH₃ (Buettner, 1987); DMSO reacts selectively with the HO· and equivalent amount of methyl radical (·CH₃) is generated. When compared with the effects afforded by DMTU and DMSO, CGRP itself in higher concentrations (10⁶ -10⁵ M) slightly decreased HO·-DMPO spin adduct produced by H₂O₂/FeSO₄ (fig. 7B) under the same reaction conditions as those of the exposure experiments of the ring preparations.

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Effects of HO· scavengers (DMTU and DMSO; A) and CGRP (B) on the ESR spectrum of DMPO-HO· produced by Fenton's reagent in the bathing media. Reaction conditions were identical with those of figure 1 except that H2O2/FeSO4 was added to the bathing media containing spin-trap DMPO in the presence of DMTU (1 mM), DMSO (100 mM) or CGRP (108 to 105 M). Signals appearing at both sides of ESR spectra correspond to Mn++ (MnO) installed in ESR cavity as a reference. Experimental conditions for the ESR detection are as described under "Materials and Methods."

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

We have previously suggested that particular experimental procedures with canine lingual artery preparations may prove to be a valuable model for the study of CGRP- but not NO-mediated vascular response (Kobayashi et al., 1995); this model (but endothelium was removed from all ring preparations to completely exclude the endothelium-derived relaxing factor-dependent mechanism) was used in the present study. The results of the present study show that HO· affect pre- and postjunctional sites of CGRP nerve fibers. This postulate is inferred from the following significant observations: 1) oxygen-derived free radical species generated from H₂O₂/FeSO₄ reaction was confirmed to be HO· by ESR study (fig. 1); 2) the CGRP-like immunoreactivity of perivascular nerves diminished markedly after exposure of the ring preparation to H₂O₂/FeSO₄ (fig. 3), and CGRP-mediated relaxation evoked by the electrical stimulation was almost completely inhibited (fig. 2), which suggests that HO· can deplete endogenous CGRP localized prejunctionally; 3) early exposure to H₂O₂/FeSO₄ attenuated human CGRP-(8-37)-abolishable relaxation induced by exogenous CGRP (fig. 4) and glibenclamide-abolishable relaxation induced by cromakalim (fig. 5), which suggests that HO· damage CGRP-mediated relaxation that is caused by activation of ATP-sensitive K⁺ channels at postjunctional site; and 4) the observed protective effects afforded by radical scavengers against inhibition of electrical stimulation-induced relaxation caused by H₂O₂/FeSO₄ exposure (table 1) were reproduced on the inhibition of exogenous CGRP- or cromakalim-induced relaxation (table 2), suggesting a result of similar mechanisms.

The observation that the DMPO-HO· signal produced by the H₂O₂/FeSO₄ reaction was reduced by competition with the addition of the HO· scavengers, DMTU and DMSO (fig. 7A), suggests that the signal arises from trapping HO·. Because DMTU might also scavenge H₂O₂ (Curtis et al., 1988), H₂O₂ may be considered as a precursor for DMPO-HO· adduct formation, and the observed effect of early exposure to H₂O₂/FeSO₄ may be caused by H₂O₂. However, selective HO· scavenger DMSO and deferoxamine, a powerful iron chelator, effectively protected the effect afforded by H₂O₂/FeSO₄ exposure (tables 1 and 2); thus HO·, rather than H₂O₂, is the active agent.

A COOH-terminal fragment of human CGRP, CGRP-(8-37), was introduced as a putative CGRP-receptor antagonist (Chiba et al., 1989). The existence of at least two classes of CGRP receptors has been suggested, one being sensitive to CGRP-(8-37) and the other being insensitive (Dennis et al., 1990). Receptors sensitive to CGRP-(8-37) are designated as CGRP₁ receptors (Dennis et al., 1989). Kobayashi et al. (1995) found that CGRP-(8-37) inhibited reversibly the relaxation of canine lingual artery preparations induced by periarterial nerve stimulation; the relaxation elicited by exogenously added CGRP was also inhibited by CGRP-(8-37) (fig. 4). Consistent with the data reported herein, the receptors present on canine lingual artery appear to be CGRP₁ receptors and HO· may produce CGRP₁-receptor dysfunction.

Hyperpolarizations and part of the vasorelaxations to CGRP are blocked by glibenclamide and not affected by blockers of Ca⁺⁺-activated K⁺ channels (Nelson et al., 1990). CGRP, when applied to the bath solution, activates glibenclamide-sensitive K⁺ channels in on-cell patches of smooth muscle cells from mesenteric arteries (Nelson et al., 1990). Glibenclamide appears to be selective for ATP-sensitive K⁺ channels and does not block a variety of other ion channels, including Ca⁺⁺ channels, inward rectifying and delayed rectifying K⁺ channels, and Ca⁺⁺-activated K⁺ channels (Ashcroft and Ashcroft, 1990; Langton et al., 1991). Therefore, relaxations to CGRP appear to involve activation of CGRP receptors on the vascular smooth muscle cells of the lingual artery and subsequent second-messenger activation of ATP-sensitive K⁺ channels. Cromakalim causes relaxation of vascular smooth muscle by opening ATP-sensitive K⁺ channels, leading to membrane hyperpolarization (Hamilton and Weston, 1989). In our system, the finding that early exposure to HO· markedly attenuated cromakalim-induced relaxation of the lingual artery preparations (fig. 5) suggests a mechanism dependent on ATP-sensitive K⁺ channel inactivation.

Standen et al. (1989) have isolated a ATP-sensitive K⁺ channel from the rabbit mesenteric artery that was activated by 10⁶ M cromakalim and blocked by 10⁵ M glibenclamide. However, the synthetic K⁺ channel opener, cromakalim, has been reported to activate aortic Ca⁺⁺-activated K⁺ channels incorporated into planar lipid bilayers (Gelband et al., 1989). Glibenclamide (10⁵ M), at a concentration which showed no effect alone on probability of opening (P_o) of Ca⁺⁺-activated K⁺ channels, reversed the pinacidil- or cromakalim-stimulated increase in P_o of Ca⁺⁺-activated K⁺ channels (Gelband and McCullough, 1993), which suggested that the vasorelaxant property of cromakalim and CGRP-mediated relaxation evoked by electrical stimulation observed in the present study may be due to increased P_o of Ca⁺⁺-activated K⁺ channels and that this channel modulation may be altered by HO· exposure. However, we ruled out this possibility because we previously confirmed that Ca⁺⁺-activated K⁺ channel blocker tetraethylammonium had without any effect on the vasorelaxant properties of cromakalim and the electrical stimulation in our system (Kobayashi, et al., 1995).

The inhibitory effect of HO· radical exposure and protection afforded by free radical scavengers in the relaxation elicited by exogenous CGRP were similar to those in the relaxation produced by cromakalim (tables 1 and 2). The mechanism by which HO· simultaneously produce CGRP-receptor dysfunction and ATP-sensitive K⁺ channel inactivation is not clear. Arora and Hess (1985) found that exposure of canine cardiac sarcolemmal vesicles to oxygen-derived free radicals generated from a xanthine-xanthine oxidase system resulted in the significant depression of B_max of [³H]quinuclidlinyl benzilate (a selective ligand of muscarinic receptors) binding without affecting the affinity constants of dissociation (K_D) and suggested that the decreased [³H]quinuclidlinyl benzilate binding mediated by oxygen-derived free radicals would be caused by a change in the number of receptors, possibly an alteration of receptor protein structures, rather than in the affinity of the receptors. These findings have led to the assumption that a target of HO· might be the functional and structural integrity of protein (e.g., receptor and/or ion channel). Katz and Messineo (1981) indicated that the functioning of the cell is affected profoundly by alteration in the lipid microenvironment of the cell membranes. Furthermore, the polyunsaturated fatty acid moiety of membrane lipids is readily oxidized by oxygen-derived free radicals. Therefore it is possible that HO· affect the following: 1) membrane lipids, the peroxidation of which may secondarily modify the membrane protein (Lee and Okabe, 1995); 2) exclusively the lipids, which alter the microenvironment of the protein (Katz and Messineo, 1981); or 3) both protein and membrane lipids independently and concurrently (Fligiel et al., 1984).

The nature of second messenger system of CGRP-mediated vasorelaxation is not known. Wei et al. (1992) recently suggested that cyclic GMP-dependent mechanism is involved in CGRP-induced dilation of feline cerebral arterioles. The importance of guanylate cyclase as a mediator of CGRP's action in vascular smooth muscle cells is of interest. The main mechanism of action of nitrodilators is that they act directly on vascular smooth muscle to generate NO, either spontaneously or through interaction with tissue components (see Ignarro, 1990). This agent then activates soluble guanylate cyclase either directly or through the formation of nitrosothiol intermediates. The result is an increase in cyclic GMP and cyclic GMP-dependent protein kinase with resultant smooth muscle relaxation. If the view that the mechanism of impairment afforded by HO· exposure of the CGRP-induced relaxation of lingual artery preparations involves decreased formation of cyclic GMP via guanylate cyclase inhibition is correct, it should be possible to attenuate nitroglycerin-induced relaxation of the preparations by HO· exposure. Early exposure to HO·, however, had no effect on this system (fig. 6). This points out that its action may not be at site(s) between the receptor site of interaction and the smooth muscle production of cyclic GMP; the susceptibility of the processes of production of cyclic GMP mediated by nitroglycerin in smooth muscle cells to HO· (or lipid peroxidation) may differ greatly from that of relaxation mediated by CGRP₁-receptor activation and opening ATP-sensitive K⁺ channels in canine lingual arteries. Thus it is likely that HO· may damage the postjunctional smooth muscle cell membrane site of CGRP nerve fibers, but not, however, the second messenger system linked to cyclic GMP formation in the smooth muscle cells. The remaining vasorelaxation may be mediated by other mechanisms, perhaps cyclic AMP-dependent mechanisms. In support of this is the finding that cyclic AMP activates AMP-sensitive K⁺ channels in myocytes (Notsu et al., 1992). In cultured rat aortic smooth muscle cells, CGRP stimulates the formation of cyclic AMP but not cyclic GMP (Kubota et al., 1985). Although these results suggest an important role for cyclic AMP in the response of vascular smooth muscle to CGRP, they do not establish that cyclic AMP mediates the effects of CGRP. The effect of oxygen-derived free radicals on the contribution of cyclic AMP to the actions of CGRP in the vascular smooth muscle cells merits further study.

There are numerous mechanisms whereby oxygen-derived free radicals contribute to the vascular (Sasaki and Okabe, 1993), skeletal muscle (Ishibashi et al., 1996) and myocardial dysfunctions (Okabe et al., 1989, 1991). Free radicals may interfere with normal control of vascular tone via chemical destruction of endogenous vasoactive agents (catecholamines, NO) (Wolin, 1991); very reactive HO· are particularly important mediators of tissue injury, and small unmyelinated C fibers are activated by real or threatened tissue injury, thereby releasing vasoactive agents, possibly CGRP. During 80 min of early exposure to HO·, endogenous CGRP-localized prejunctional site of CGRP nerve fibers in the lingual artery may be destroyed or released continuously, thus being depleted. Whether endogenous CGRP localized prejunctionally is destroyed or run out by HO· exposure remains to be determined. Irrespective of the exact mechanism involved, the findings (figs. 2 and 3) implicate HO·, like the pungent chemical capsaicin, as the species toxic to CGRP nerve fibers. Although CGRP (10⁶-10⁵ M) slightly reduced DMPO-HO· signal produced by H₂O₂/FeSO₄ (but 10⁸ M CGRP used in the relaxation experiments was without effect) (fig. 7B), our studies do not test the direct inactivation of CGRP by HO·, because exogenous CGRP or electrical stimulation that mediates release of CGRP was given after Fenton's reagent was removed from the tissue bath.

This is the first study to demonstrate that HO· can deplete endogenous CGRP localized prejunctionally and also damages CGRP-induced relaxation of canine lingual artery preparations that is caused by activation of ATP-sensitive K⁺ channels at postjunctional sites; the second messenger system of the relaxation mediated, at least, by cyclic GMP may be less susceptible to HO·.