the application of glutamate or its analogN-methyl-d-aspartate (NMDA) to the cortical surface causes dilation of pial arterioles (2, 15, 28) and increases in local cerebral blood flow (CBF) (3, 38) in a variety of species. Because glutamate is the major neurotransmitter in the cerebral cortex, investigations into the mechanism of glutamate-induced vasodilation could lead to a better understanding of the mechanisms underlying the coupling of CBF to increases in cortical activity. Whereas the weight of the evidence favors nitric oxide (NO) as the mediator of glutamate-induced vasoactivity in the brain (3, 15, 28, 38), recent studies (1, 3) suggest that products of the P-450 epoxygenase pathway are also involved in the CBF response to glutamate. Thus multiple mediators may be involved in the dilation response of cerebral arterioles to glutamate or NMDA.

By virtue of its link to oxidative metabolism and its potent dilating effects, the nucleoside adenosine (Ado) is an excellent candidate as a mediator of the increase in CBF during enhanced cortical activity (13, 25). Our laboratory (25, 29), in addition to others (11), has provided evidence supporting a role for Ado in the regulation of CBF during somatosensory stimulation. Consistent with these findings, the application of glutamate or its analogs leads to the release of endogenous Ado within various brain tissues, including cortical slices (9, 20,21), cultured cortical synaptosomes (19), the caudate nucleus (41), and the hippocampus (7). We therefore hypothesize that Ado contributes to glutamate-induced dilation of cerebral arterioles.

The purpose of the present study is to determine the effect of Ado receptor blockade on pial arteriolar dilation evoked by the topical application of glutamate and its analog NMDA. Cloning experiments have isolated four distinct G protein-coupled cell surface receptors that govern the activity of Ado: the A₁, A_2A, A_2B, and A₃ subtypes (13, 17, 18). Our laboratory has previously shown that Ado receptors, specifically those of the A_2A subtype, are involved in the dilation of pial arterioles during somatosensory stimulation (29). Our laboratory (33) has also demonstrated that Ado-induced dilation of intracerebral arterioles in vitro is mediated primarily by the A_2A receptor subtype. Therefore, in addition to using the nonselective Ado antagonist theophylline (Theo), we also evaluated the effect of the A_2A receptor-selective antagonist 4-{2-[7-amino-2-(2-furyl)(1,2,4)triazolo (2,3-a)(1,3,5)triazin-5-ylamino]ethyl}phenol (ZM-241385) on glutamate-induced dilation of pial arterioles.

METHODS

All experiments were conducted in accordance with standards established by the University of Washington Animal Welfare Committee. Male Sprague-Dawley rats (350–400 g; Charles River) were initially anesthetized with halothane (1.5–2.0%). The right femoral artery was cannulated to facilitate the continuous monitoring of mean arterial blood pressure (MABP) and the measurement of blood gas values. The right femoral vein was likewise cannulated to allow the delivery of intravenous drugs. The rat was tracheotomized, paralyzed with tubocurarine chloride (1 mg/kg iv) and was mechanically ventilated to maintain physiological blood gas values. Rectal temperature was maintained with a heating pad. Halothane anesthesia was replaced with α-chloralose and urethane (50 and 500 mg/kg ip). The skull was fixed in a stereotaxic frame and the pial vasculature was visualized through a closed cranial window implanted over the right somatosensory cortex, as described in detail by Morii et al. (32).

One pial arteriole (20–35 μm in diameter) per rat was chosen for study. Vessel diameter was measured continuously with a video dimension analyzer. Glutamate and other drugs were dissolved in artificial cerebrospinal fluid (CSF), and applied (0.25 ml/min) through the cranial window by means of a perfusion pump.

Initially, we examined the dose-dependent response of pial arterioles to topical glutamate application (n = 5). Glutamate (10 μM) was perfused under the window for a period of 10 min while stable (∼4 mmHg) intracranial pressure was maintained. The vascular response was measured for a period of 5 min, and the window was washed with CSF for 15 min. This process was repeated with increasing glutamate concentrations of 0.1 mM and 1 mM. In preliminary studies, we also examined the effect of repeated perfusion of glutamate on vascular tone and reactivity by perfusing the same dose (0.1 mM) of glutamate two times in each animal, each perfusion period being separated by a 15-min washout and 5-min equilibration period (n = 4).

We examined the effect of the Ado antagonist Theo on glutamate- and NMDA-induced vasodilation. Glutamate (100 μM, n = 6) or NMDA (100 μM, n = 6) was superperfused for a period of 10 min. After a 5-min equilibration period, the baseline dilation to glutamate or NMDA was measured. The window was then flushed for 15 min with CSF. Theo (10 μM) was then perfused under the window for 10 min, followed by another 5-min equilibration period. This concentration of Theo has previously been shown to have no effect on resting vessel diameter but to completely abolish the dilation response to 10 μM Ado (36). Glutamate or NMDA was coapplied with Theo for 10 min, followed by a 5-min equilibration period. The effect of Theo on glutamate- or NMDA-induced vasodilation was measured. To determine whether Theo may have nonspecific inhibitory effects on vasodilation, we also evaluated the responses of pial arterioles to mild hypercapnia before and after Theo application. Mild hypercapnia (relative to baseline blood gas values while under mechanical ventilation) was induced by connecting the respirator to a gas mixture of 5% CO₂-40% O₂-55% N₂ for ∼2 min, after which baseline (arterial Pco₂: 32–36 mmHg) ventilation was restored. CO₂ reactivity was expressed as the percent increase in diameter per mmHg increase in arterial Pco₂.

We next determined the effect of the Ado A₁-specific antagonist 8-cyclopentyl-1,3-dipropylxanthine (CPX) on the pial arteriolar response to glutamate. CPX was applied topically at a concentration (0.1 μM, n = 6) that has been shown to potentiate the vascular response to sciatic nerve stimulation, presumably by antagonizing the neuroinhibitory effects of endogenous Ado (29). The experimental protocol was otherwise identical to that of Theo application.

We evaluated the effect of the A_2A receptor-selective antagonist ZM-241385 on glutamate-induced pial arteriolar dilation. ZM-241385 was applied topically (0.1 and 1 μM; n = 6; Ref. 6) or intravenously (1 mg/kg iv, n = 6). These concentrations have been previously shown to have no effect on vessel diameter but to completely abolish the dilation response to Ado A_2A agonists (33). The experimental protocol was otherwise identical to that of Theo application. We evaluated the response of these arterioles to mild hypercapnia induced by ventilation with 5% CO₂ before and after exposure to ZM-241385. Throughout all experiments, MABP, arterial Pco₂, Po₂, and pH were monitored and remained within physiological norms.

Glutamate and Theo were purchased from Research Biomedicals International (Natick, MA). CPX and ZM-241385 were purchased from Tocris (Baldwin, MO). Glutamate was initially dissolved in mock CSF to produce a 10 mM stock solution. The mock CSF was composed of (in meq/l) 156.5 Na⁺, 2.95 K⁺, 2.50 Ca²⁺, 1.33 Mg²⁺, 24.6 HCO${}_3^{-}$; and (in mg/dl) 66.5 dextrose and 40.2 urea. Theo was dissolved in mock CSF to produce 1 mM stock solution. The glutamate and Theo stock solutions were further diluted with CSF to obtain concentrations appropriate for experimental application. ZM-241385 was dissolved first in DMSO to produce a 10 mM stock solution, which was subsequently diluted with CSF to yield appropriate concentrations for this study. A 10 mM CPX stock solution was prepared in 50% ethanol-50% alkamus (Rhone Poulenc; Cranbury, NJ) and then diluted with CSF to yield experimental concentrations. Vehicle control studies for ZM-241385 and CPX were conducted by Meno et al. (29) and revealed no vehicle effect on resting arteriolar diameter or vasodilation induced by transient hypercapnia.

All listed values are expressed as means ± SE. Average responses to glutamate perfusion were expressed as percent change from baseline (before application) values. Because of variations in hypercapnic blood gas values from trial to trial, the hypercapnic dilation response (CO₂ reactivity) was normalized by dividing the percent increase in vessel diameter by the increase in arterial Pco₂ (in mmHg). Comparisons of data were made with a paired Student's t-test or with analysis of variance (for multiple groups), followed by a post hoc Bonferroni procedure. A value of P < 0.05 was considered significant.

RESULTS

Throughout all experiments, MABP and blood gas values were maintained stable and within physiological range (Table 1). Topical application of glutamate elicited a dose-dependent dilation of pial arterioles (Fig. 1). The lowest dose of glutamate (10 μM) did not significantly dilate pial arterioles (baseline diameter: 32 ± 3 μm), whereas higher concentrations of glutamate (100 μM and 1 mM) evoked increasingly greater vasodilation (17 ± 1% and 28 ± 3%, respectively). The glutamate analog NMDA (100 μM) caused a similar dilation response as that induced by glutamate of the same concentration (18 ± 4%). In an effort to minimize the possible excitotoxic effects of glutamate, the concentration of 100 μM was used in the adenosine antagonist studies.

Vessels that were repeatedly exposed to 100 μM glutamate maintained vascular tone and reactivity (as indicated by the recovery of resting vessel diameter and dilation responses to glutamate). The first application of glutamate elicited a 19 ± 6% change in vessel diameter from a baseline of 25 ± 4 μm. The response to the second perfusion of glutamate did not significantly differ from the first, with 16 ± 5% dilation from a baseline of 27 ± 1 μm. These experiments also serve as time controls for the antagonist studies. To ensure the viability and responsivity of the vessels, all studies were limited to no more than two glutamate applications.

The nonselective Ado antagonist Theo markedly attenuated the pial arteriolar response to glutamate perfusion (Fig. 2, n = 6). Similarly, Theo significantly diminished NMDA-induced vasodilation (Fig. 3, n = 6). Superperfusion of Theo had no appreciable effect either on baseline vessel diameter (Table1) or on the dilation response of pial arterioles to mild hypercapnia (2-min inhalation of 5% CO₂) (Figs. 2 and 3). Tables 1 and2 summarize blood pressure and blood gas values before and during mild hypercapnia, respectively. In contrast, topical coapplication of the Ado A₁-selective antagonist CPX (0.1 μM) failed to attenuate glutamate-induced pial arteriolar dilation. CPX also did not affect resting vessel diameter (Table 1).

Topical coapplication of the A_2A-specific antagonist ZM-241385 with 100 μM glutamate significantly attenuated glutamate-induced vasodilation (Fig. 4) both at 0.1 μM (n = 6) and 1 μM (n= 6). These attenuations were of comparable magnitude. Systemic administration of ZM-241385 (1 mg/kg iv) similarly reduced glutamate-evoked dilation (n = 6). Neither topical nor intravenous administration of ZM-241385 affected resting vessel diameter (Table 1) or CO₂ reactivity (Fig. 4).

DISCUSSION

The present study is the first to demonstrate that Ado contributes to glutamate-evoked dilation of pial arterioles within the cerebral cortex. This conclusion is based on the findings that the Ado antagonists Theo and ZM-241385 significantly attenuated glutamate-induced vasodilation without affecting either baseline vessel diameter or dilation responses to mild hypercapnia. The inhibitory effect of the A_2A receptor-selective antagonist ZM-241385 in the absence of an effect by the A₁ blocker CPX further suggests that glutamate dilates pial arterioles primarily through the activation of Ado A_2A receptors.

The role of Ado in linking neuronal activity to CBF has been predominantly examined through functional activation studies. These experiments utilized the stimulation of somatosensory pathways, coupled with the measurement of arteriolar dilation or CBF in the activated region of the somatosensory cortex. Two studies from our laboratory (25, 29) independently demonstrated that Ado antagonists inhibited the dilation response of pial arterioles to sympathetic nervous system (SNS) in the rat. This observation was corroborated by Dirnagl et al. (11) with the use of a whisker deflection model, thus supporting the role of Ado as a mediator of neurovascular coupling.

Meno et al. (29) further investigated the Ado receptor subtype involved in the dilation response of pial arterioles to cortical activation. In their study, blockade of the A₁receptor subtype by CPX resulted in the potentiation of the dilation response, presumably through the “disinhibition” of cortical activity and a consequent rise in vasoactive metabolites. The A_2A-selective antagonist ZM-241385 significantly attenuated pial arteriole dilation evoked by SNS, suggesting the involvement of the A_2A receptor subtype. Moreover, ZM-241385 blocked dilation to the A_2A-specific agonist 2-p-(2-carboxylethyl)-phenethylamino-5′-N-ethylcarboxy-amidoadenosine, but it did not affect vasodilation induced by mild hypercapnia. Consistent with a previous in vitro vessel study (33), these results demonstrate that ZM-241385 is both an effective and specific antagonist for the A_2A receptor in rat cerebral arterioles.

The present study extends the above observations to glutamate-evoked dilation. We report that both Theo and ZM-241385 markedly suppressed the dilation response to glutamate, suggesting that the Ado A_2A receptor is involved in this response. There are differences between our findings and those of Meno et al. (29). First, in contrast to SNS-induced vasodilation, 1 μM ZM-241385 was no more effective than 0.1 μM in blocking the vascular response to glutamate. Because 0.1 μM ZM-241385 is highly A_2A receptor selective, with minimal functional effects on the A_2B receptor in cerebral arterioles (33), whereas 1 μM ZM-241385 may potentially antagonize both A_2A and A_2B receptors, Meno et al. (29) suggested that A_2B receptors may be involved in SNS-induced vasodilation. In contrast, our results suggest that glutamate-induced dilation is primarily mediated by the A_2A receptor. Second, Meno et al. reported that systemic administration of ZM-241385 significantly attenuated the dilation response to sciatic nerve stimulation, whereas topical application had no effect. In our hands, both topically and systemically applied ZM-241385 were highly effective in attenuating the dilator effects of topically applied glutamate. It is possible that topically applied ZM-241385 may not penetrate the cortex in sufficient concentrations to prevent Ado-induced dilation during somatosensory stimulation but that it could effectively antagonize the more superficial Ado receptors that are activated by glutamate superperfusion. Third, the A₁-specific antagonist CPX potentiated SNS-induced vasodilation (29) but had no effect on glutamate-induced vasodilation. We surmise that the differential activation of diverse cortical circuitries by the two experimental paradigms may explain such a discrepancy.

At odds with the findings of the present study, Bari et al. (2) reported that the intravenous administration (20 mg/kg) of Theo had no effect on NMDA-evoked dilation of pial arterioles in the newborn piglet. We speculate that differences in species or developmental stage of the experimental animal may have contributed to the discrepancy in results between the two studies.

A wealth of research supports the notion that Ado mediates the vascular response to glutamate in the cerebral cortex. Sciotti et al. (41) reported that within the caudate nucleus, Ado mediates changes in blood flow evoked by stimulation with the glutamate analog kainate. More specific to the cortex, several studies (9,19-21) have reported that glutamate evokes the release of Ado from cortical tissue. This release appears to be mediated by glutamate receptors of both the NMDA and non-NMDA subtypes (20).

There are several possible mechanisms by which glutamate might cause Ado release in the cortex. Activation of ionotropic glutamate receptors presumably would dissipate ionic gradients, the reestablishment of which would consume energy (14) and lead to the breakdown of intracellular ATP to Ado, which is then extruded via nucleoside transporters (9, 18). Alternately, glutamate may cause the release of ATP, which may subsequently be converted by ectonucleotidases to Ado (9, 42). Yet another potential mechanism of glutamate-evoked Ado release has been demonstrated in rat cortical synaptosomal cultures (19). Via a receptor-independent mechanism, glutamate uptake through cotransport with Na⁺ is capable of stimulating energy metabolism in both neurons and astrocytes (27).

Although cortical slice studies (9, 20, 21) have demonstrated the release of Ado by glutamate perfusion, the resulting concentration of Ado in the cortex is unknown. Thus the question remains as to whether vasoactive levels of Ado were reached under these conditions. Interstitial Ado concentrations have, however, been estimated by in vivo microdialysis in the rat hippocampus during NMDA stimulation (7) as well as in the caudate nucleus during kainate application (41). When stimulated with 1 mM NMDA or 100 μM kainate, interstitial Ado concentration reached the μM range within the respective brain regions. This Ado concentration, based on previous Ado dose-response studies (23, 33), is capable of eliciting the magnitude of dilation responses observed in the present study (Fig. 1).

In the present study, glutamate concentrations of 100 μM and 1 mM were necessary to induce appreciable vasodilation. There is concern that such concentrations may be excitotoxic. Under normal physiological conditions, glutamate levels are known to transiently reach the millimolar range only within the synaptic cleft where glutamate transport proteins within the surrounding glia then rapidly degrade glutamate levels (8). On a larger scale, this “glutamate sink” would similarly affect exogenously applied glutamate. Although 1 mM is near the upper bound of physiological glutamate concentrations, diffusion into the cortex coupled with its rapid uptake would reduce the effective glutamate levels to well below those that are excitotoxic (8). It should also be noted that excitotoxic effects have not been observed in previous studies (2, 15, 28) on glutamate- and NMDA-induced vasodilation (with superfusate concentrations ranging between 10 μM and 1 mM).

Other investigators have shown that NO synthase (NOS) (3, 15, 28,38) and P-450 epoxygenase (3) inhibitors also block the cerebral vascular response to glutamate/NMDA application. These findings, along with those of the present study, suggest that multiple mediators, including epoxyeicosatrienoic acids (EETs), NO, and Ado, may all play a role in neurovascular coupling. Although the specific nature and extent of interaction between these substances within this process remains to be firmly established, some insight into this increasingly complicated issue may be gleaned from the literature.

NO is widely believed to be the primary mediator of neurovascular coupling within the cerebral cortex. NOS inhibitors such asN^G-nitro-l-arginine attenuate the CBF response to functional activation (10, 24,37) as well as to direct glutamate or NMDA application in vivo (2, 15, 28, 38). Furthermore, the application of glutamate analogs evokes NO release in vivo (4, 5). These findings suggest that glutamate receptor stimulation causes the activation of neuronal NOS, NO formation, and NO diffusion to vascular smooth muscle, where it activates soluble guanylate cyclase, resulting in vasorelaxation.

However, the notion that NO directly mediates neurovascular coupling was recently challenged by a study (26) that demonstrated that the CBF response to whisker deflection, suppressed by continual NOS inhibition, was restored with the addition of either NO donors or cGMP. In contrast, addition of the NO-independent vasodilator papaverin failed to reverse the effects of NOS inhibition. Thus NO appears to play a “permissive” role, providing a basal level of cGMP required for the manifestation of the CBF response, but does not directly mediate the response itself. NO may also act as a neuromodulator by maintaining a level of neuronal activation necessary for the production of other vasoactive substances. NOS inhibition may thus prevent NMDA-evoked glutamate release (30) and lead to a profound decline in evoked activity in the cortex (35). By maintaining a basal level of cGMP and neuronal activity within the cortex, NO may “permit” the full manifestation of vasodilation evoked by other factors such as EETs or Ado.

It seems reasonable to surmise that Ado may also play a similar permissive role by providing a basal level of A_2A receptor stimulation in the dilation response to glutamate. There are, however, important differences between the effects of NOS inhibition and Ado receptor blockade in this study. First, unlike NOS inhibitors, which significantly decrease baseline CBF and vessel diameter (16), the Ado antagonists Theo and ZM-241385 had no effect on baseline diameter in this study (29, 31). These observations suggest that baseline A_2A receptor activation is extremely small. Thus it seems unlikely that Ado would act permissively, by providing a significant baseline level of stimulation requisite for the full manifestation of the vascular response to other vasoactive stimuli. Second, NOS inhibition abolished the CBF response to hypercapnia (22), potassium (12), and adrenoreceptor stimulation (6). In contrast, Ado receptor blockade specifically inhibited dilation responses to Ado and its analogs, but it did not affect the dilation responses to other vasodilator stimuli, such as sodium nitroprusside (11), ATP (33), acidic pH (33), or CO₂inhalation (present observations, see also Ref. 31). Third, NOS inhibition profoundly depressed somatosensory evoked potentials, suggesting that NO may modulate synaptic activity in the cerebral cortex (34, 35), whereas A_2A Ado receptor antagonists did not affect evoked activity in the brain (29). Taken together, these observations provide support for a mediatory rather than a permissive role by Ado in glutamate-induced dilation.

Recently, EETs have also been implicated as potential mediators of neurovascular coupling in the cerebral cortex (1, 3). The cytochrome P-450 epoxygenase inhibitors miconazole andN-methylsulfonyl-6-(2-propargyloxyphenyl)hexanamide blocked the CBF response to both direct NMDA application (3) and functional activation (39). These studies lead to the hypothesis that influx of calcium into glia triggers the production of EETs derived from arachidonic acid within the astrocytic phospholipid pool, which then diffuse to vascular smooth muscle, causing hyperpolarization and vasodilation. Because glia possess no known functional NMDA receptors, this mechanism necessitates a path of communication between activated neurons and the surrounding glia. Consistent with the findings of the present study, Ado released from activated neurons could evoke glial calcium influx (40), providing a potential mode of interaction between Ado and the P-450 epoxygenase pathway in neurovascular coupling.

In summary, the present study demonstrates that Ado, through its action at the A_2A receptor subtype, may be involved in glutamate-induced vasodilation in the rat cerebral cortex. These findings shed further light on the physiological role of Ado in the regulation of CBF during neuronal activity.

FOOTNOTES

• Address for reprint requests and other correspondence: A. C. Ngai, Dept. of Neurological Surgery, Harborview Medical Center, Box 359914, 325 Ninth Ave., Seattle, WA 98104-2499 (E-mail: [email protected]washington.edu).

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

• First published January 23, 2003;10.1152/ajpheart.00909.2002