Mild Hypercapnia Induces Vasodilation via Adenosine Triphosphate-sensitive K⁺Channels in Parenchymal Microvessels of the Rat Cerebral Cortex

The institutional animal care and use committee approved this study. Male Wistar rats (300–400 g) were anesthetized with inhalation of halothane, 3%, in 100% oxygen (3 l/min). A midline thoracotomy was performed; 50 ml saline was infused intracardially into the left ventricle; simultaneously, a right atrial incision was made for blood drainage. The animals were then decapitated, and the brains were rapidly removed and rinsed with artificial cerebrospinal fluid (control solution) with the following composition (in mm): NaCl, 119; KCl, 4.7; CaCl², 2.5; MgSO⁴, 1.17; KH²PO⁴, 1.18; NaHCO³, 25; and glucose, 11. In general, brain slice preparation was done according to previously reported methods. 9,10Brains were cut freehand into blocks containing the neocortex, followed by immediate sectioning into slices (200 μm thick) with a mechanical tissue slicer (Vibratomes 1000; Ted Pella, Inc., Redding, CA). Throughout the slicing procedure, brain blocks were continuously bathed in the control solution bubbled with 93% O²+ 7% CO²at 4°C. An individual slice was then transferred to a recording chamber filled with control solution, which was mounted on an inverted microscope (IX70; Olympus, Tokyo, Japan).

The recording system consisted of a recording chamber (3 ml) and a tubing compartment (7 ml) including the perfusion chamber (fig. 1). The slices were continuously superfused with control solution at a flow rate of 1.5 ml/min bubbled with 93% O²+ 7% CO²(Pco²= 40 mmHg; pH = 7.4; 37°C) and 90% O²+ 10% CO²(Pco²=50 mmHg; pH = 7.3; 37°C) for normal and hypercapnic conditions, respectively. In some experiments, artificialcerebrospinal fluid was changed from the normocapnic control solution to the hypercapnic normal pH solution (Pco²= 50 mmHg; pH = 7.4; 37°C), which was prepared by modifying the composition of NaCl (109 mm) and NaHCO³(35 mm) bubbled with 90% O²+ 10% CO². An intraparenchymal arteriole (ID, 5–9.5 μm) was located within the neuronal tissue, and its ID was continuously monitored with live computerized videomicroscopy.

The videomicroscopy equipment consisted of an inverted microscope, a ×40 objective (Olympus), and a ×2.25 video projection lens (Olympus). Arterioles were microscopically identified and differentiated from venules by the presence of the layer of vascular smooth muscle cells, and these were confirmed by hematoxylin-eosin staining of the slice after each experiment (unpublished data, Hiroyuki Kinoshita, M.D., Ph.D., Wakayama, Japan [October 2001]). The image of a parenchymal arteriole was transmitted to a video camera (C6790–81; Olympus) and displayed on a computer via  a media converter (Physio-Tech, Tokyo, Japan). We defined the intraluminal diameter as the length between the internal margins of the arteriolar walls. Changes in the intraluminal diameter in cerebral microvessels were recorded on computer image files and then analyzed using image analysis software with a sensitivity to 0.01 μm (Physio-Tech).

Each slice was equilibrated for at least 30 min before the start of the experimental protocols. All experiments were performed during submaximal constriction in response to prostaglandin F^2α(5 × 10⁻⁷m). Glibenclamide (5 × 10⁻⁶m), iberiotoxin (10⁻⁷m), or N^G-nitro-l-arginine methyl ester (10⁻⁴m) was applied 10 min before addition of prostaglandin F^2α(5 × 10⁻⁷m). In some experiments, the condition of perfusion was changed from normal to hypercapnic once during constriction in response to prostaglandin F^2α. Concentration responses to acetylcholine (10⁻⁶to 10⁻⁴m), levcromakalim (3 × 10⁻⁸to 3 × 10⁻⁷m), and sodium nitroprusside (3 × 10⁻⁸to 3 × 10⁻⁶m) were obtained in the absence or the presence of glibenclamide, iberiotoxin, or N^G-nitro-l-arginine methyl ester (10⁻⁴m). Concentration responses were obtained in a cumulative fashion by adding vasoconstrictor or vasodilator substances into the bubbling chamber connected to, but separated from, the recording chamber. Only one concentration response was made for each slice. The experiment duration for each slice was within 3 h. In the preliminary experiments, we found that prostaglandin F^2α(5 × 10⁻⁷m) can produce submaximal vasoconstrictor effects (≈70% constriction compared with maximal contraction induced by prostaglandin F^2α[10⁻⁵m]) on the cerebral arterioles under our experimental conditions (unpublished data, Katsutoshi Nakahata, M.D., Wakayama Japan [October 2002]). The amount of dilation of the cerebral arteriole induced by a mild increase in levels of carbon dioxide (50 mmHg), acetylcholine, levcromakalim, or sodium nitroprusside was normalized by using the constriction produced by prostaglandin F^2α(5 × 10⁻⁷m) in each arteriole. Therefore, the percent dilation was calculated by the following equation:

. The percent constriction was calculated by the following equation:

D^control, D^PGF, and D^dilatorwere the arteriolar diameter during the control condition, the arteriolar diameter after administration of prostaglandin F^2α(5 × 10⁻⁷m), and that after addition of the vasodilator including carbon dioxide, respectively.

The following pharmacologic agents were used: acetylcholine, dimethyl sulfoxide, glibenclamide, N^G-nitro-l-arginine methyl ester, prostaglandin F^2α, sodium nitroprusside (Sigma, St. Louis, MO), and levcromakalim (GlaxoSmithKline plc, Greenford, United Kingdom). Drugs were dissolved in distilled water such that volumes of less than 60 μl were added to the perfusion system. Stock solutions of levcromakalim (10⁻⁵m) and glibenclamide (10⁻⁵m) were prepared in dimethyl sulfoxide (3 × 10⁻⁴m). The drug concentrations are expressed as final molar concentrations.

The data are expressed as mean ± SD; n refers to the number of rats from whom brain slices were obtained. Statistical analysis was performed using repeated-measures ANOVA followed by the Sheffé F test. Differences were considered to be statistically significant at P < 0.05.

Figure 2shows a representative example of vasodilator responses induced by a mild increase in levels of carbon dioxide (50 mmHg; pH = 7.3). Prostaglandin F^2α(5 × 10⁻⁷m) produced vasoconstriction of parenchymal arterioles, and addition of carbon dioxide (Pco²= 50 mmHg) induced marked vasodilation that reached the maximum within 20 min. The cumulative data demonstrate significant vasodilation induced by hypercapnia in the cerebral parenchymal arterioles, whereas the selective ATP-sensitive K⁺channel antagonist glibenclamide (5 × 10⁻⁶m) completely abolished this vasodilation (fig. 3). In contrast, the selective Ca²⁺-dependent K⁺channel antagonist iberiotoxin (10⁻⁷m) did not alter the vasodilation induced by hypercapnia in arterioles contracted with prostaglandin F^2α(fig. 4).

Hypercapnia under normal pH (Pco²= 50 mmHg; pH = 7.4) produced no vasodilation in cerebral parenchymal arterioles (n = 5;fig. 5).

The nitric oxide synthase inhibitor N^G-nitro-l-arginine methyl ester (10⁻⁴m) did not affect the vasodilation induced by a mild increase in levels of carbon dioxide (50 mmHg) (fig. 6A), whereas this compound completely abolished the vasodilation induced by acetylcholine (10⁻⁶to 10⁻⁴m) (fig. 6B).

In arterioles contracted with prostaglandin F^2α(5 × 10⁻⁷m), the selective ATP-sensitive K⁺channel opener levcromakalim (3 × 10⁻⁸to 3 × 10⁻⁷m) induced vasodilation in a concentration-dependent fashion (fig. 7). The highest concentration of levcromakalim used in the current study produced almost full vasodilation. This vasodilator effect of levcromakalim was completely abolished by glibenclamide (5 × 10⁻⁶m) (fig. 7). Glibenclamide did not alter the vasodilator effect of the nitric oxide donor sodium nitroprusside (3 × 10⁻⁸to 3 × 10⁻⁶m) in cerebral parenchymal arterioles (fig. 8).

To our knowledge, this is the first study demonstrating that in rat cerebral parenchymal arterioles, a mild increase in levels of carbon dioxide (50 mmHg) induces vasodilation, which is completely abolished by glibenclamide but not by N^G-nitro-l-arginine methyl ester. These results suggest that in parenchymal microvessels of the rat cerebral cortex, ATP-sensitive K⁺channels, but not nitric oxide derived from endothelial as well as neuronal enzymes, play a major role in vasodilator responses produced by mild hypercapnia.

The specific effects of carbon dioxide on parenchymal cerebral arterioles have not been well studied, because penetrating microvessels are quite small and difficult to access when embedded within the neuronal tissues. However, Lee et al.  10introduced a new system, which is capable of evaluating microvessels in the brain parenchyma using live videomicroscopy. The advantage of this method is the capacity to directly evaluate vasodilation or vasoconstriction in smaller vessels, approximately greater than 20 μm in diameter. In addition to these beneficial points, we developed this microscopic system with computer image analysis. Indeed, the image analysis software used in the current study has a higher sensitivity of 0.01 μm (compared with previously reported systems), showing that much smaller arterioles with IDs of less than 10 μm can be examined. 10,11Although a previous study of rat brain slices demonstrated that cerebral microvessels are capable of responding to the alteration in carbon dioxide tension, 11the role of ATP-sensitive K⁺channels in intraparenchymal cerebral arterioles has not been studied.

In parenchymal arterioles of the rat cerebral cortex, a mild increase in levels of carbon dioxide produced marked vasodilation, which was completely abolished by glibenclamide but not by iberiotoxin. Glibenclamide and iberiotoxin are selective ATP-sensitive and Ca²⁺-dependent K⁺channel antagonists, respectively, suggesting that vasodilation induced by carbon dioxide in this preparation is mediated by ATP-sensitive K⁺channels. 8,12,13The selectivity of glibenclamide as an ATP-sensitive K⁺channel antagonist is supported by the following results. In the current study, glibenclamide abolished vasodilation in response to an ATP-sensitive K⁺channel opener, levcromakalim. In addition, the concentration of glibenclamide used in the current study did not affect vasodilation of cerebral parenchymal arterioles induced by the nitric oxide donor sodium nitroprusside. Therefore, it appears that, at least in the rat cerebral parenchymal arterioles, mild hypercapnia-induced vasodilation is mediated by the relatively selective activation of ATP-sensitive K⁺channels.

The current study was designed to evaluate the role of ATP-sensitive K⁺channels in cerebral microvessels in mild hypercapnia rather than severe hypercapnia, because mild hypercapnia appears to be more clinically relevant than the severe type. A previous study of larger cerebral arterioles demonstrated that mild hypercapnia, similar to that studied in the current study, induced vasodilation only partly via  ATP-sensitive K⁺channels. 7In humans, administration of glibenclamide did not alter the blood flow of the middle cerebral artery in response to mild hypercapnia, suggesting that the activity of ATP-sensitive K⁺channels may not contribute to hypercapnia-induced vasodilation in the large cerebral artery. 14In contrast, in cerebral parenchymal arterioles, mild hypercapnia produced marked vasodilation, which was completely abolished by a selective antagonist of ATP-sensitive K⁺channels, suggesting that these channels expressed in microvessels of cerebral cortex are very sensitive to the mild increase in levels of carbon dioxide. Therefore, ATP-sensitive K⁺channels may play a major role in cerebral vasodilation in mild hypercapnia only in cerebral microvessels. However, we cannot completely rule out the possibility that methodologic as well as species differences may contribute to these differential results.

In the cerebral microvessels, N^G-nitro-l-arginine methyl ester did not affect the vasodilation induced by a mild increase in levels of carbon dioxide. Because N^G-nitro-l-arginine methyl ester is a well-known nitric oxide synthase inhibitor, these results suggest that nitric oxide derived from endothelial as well as neuronal enzymes may not mediate vasodilation induced by mild hypercapnia in the cerebral parenchymal arteriole. 15In larger cerebral arteries as well as arterioles, nitric oxide, in the diverse degree, appears to play a role in cerebral vasodilation in response to hypercapnia. 7,16–19In the human cerebral circulation, it is contradictory whether the activity of nitric oxide synthase contributes to this vasodilation. 19,20Therefore, differential species as well as methodologic differences among these studies are the most likely explanation for these differential findings obtained for variable species including humans.

In our preparation, vasodilation in response to acetylcholine was completely abolished by treatment with N^G-nitro-l-arginine methyl ester, indicating the intact function of nitric oxide synthases in our experimental model. Previous studies of the isolated pressurized artery demonstrated that extraluminally applied acetylcholine can induce the activation of the endothelial enzyme concurrently with increased levels of endothelial Ca²⁺, resulting in production of nitric oxide, which suggests the possible role of endothelial nitric oxide synthase in vasodilation produced by acetylcholine in our preparation. 21However, we cannot rule out the possibility that in our preparation vasodilator substances induced the activation of neuronal nitric oxide synthase, because parenchymal arterioles in the brain slice are embedded in the neuronal tissues.

Previous studies demonstrated that ATP-sensitive K⁺channels can be activated by increased intracellular levels of adenosine diphosphate as well as decreased ATP levels, suggesting that these K⁺channels are closely related to intracellular energy status. 22In addition to these nucleotides, intracellular levels of protons reportedly regulate ATP-sensitive K⁺channels. 23Indeed, a recent study of cloned mouse ATP-sensitive K⁺channels expressed in smooth muscle cells demonstrated that hypercapnia as well as intracellular acidosis is capable of augmenting these currents, suggesting that the cloned ATP-sensitive K⁺channels are regulated by intracellular protons. 24In contrast, studies on isolated rat cerebral arteries indicated that carbon dioxide acts on ATP-sensitive K⁺channels from the outer side of the vascular smooth muscle cell membrane. 25Therefore, it has been unclear whether the activation of these channels induced by hypercapnia is mediated by the changes of extracellular carbon dioxide levels themselves or by those of intracellular proton levels indirectly produced by extracellular carbon dioxide. Under the condition of normal pH, applied carbon dioxide produced no vasodilation of the parenchymal arterioles in our preparation, indicating that in the cerebral arteriole the extracellular change of proton levels, but not carbon dioxide molecule levels, is responsible for vasodilation produced by hypercapnia. This conclusion is supported by many previous studies demonstrating that decreased extracellular pH plays an important role in cerebral vasodilation in response to hypercapnic acidosis. 8,18,25However, we cannot completely rule out the possibility that species and methodologic differences between these previous studies may contribute to the differential results regarding the mechanisms of hypercapnia action on ATP-sensitive K⁺channels expressed in vascular smooth muscle cells.

In the rat cerebral parenchymal microvessels, 3 × 10⁻⁷m levcromakalim produced almost maximal vasodilation, whereas this concentration of the compound can induce only 40% dilation of isolated aorta and carotid artery in the rat. 26,27A previous study of rats also demonstrated that in middle cerebral arteries with diameters of 200 to 300 μm, this concentration of levcromakalim induced up to 70% vasodilation. 28These results indicate that ATP-sensitive K⁺channel openers can produce more potent vasodilator responses in cerebral arteries than conduit arteries and that, even in the cerebral arteries, the arterioles that peripherally exist in brain parenchyma are likely to respond better to these openers than are relatively proximal arteries. In the current study of cerebral parenchymal arterioles, both the endothelium-dependent agonist acetylcholine and the nitric oxide donor sodium nitroprusside had somewhat less potent effects than the effects seen in larger conduit arteries like the aorta. 29,30These results suggest that the vasodilator response via  nitric oxide may not be a major regulatory mechanism in peripheral cerebral circulation, compared with that in conduit arteries.

In the current study, we used prostaglandin F^2αas a vasoconstrictor substance to produce resting vascular tone to mimic the in vivo  condition, because previous studies already demonstrated that this compound is capable of producing strong constriction of cerebral parenchymal as well as pial arterioles in the rat, indicating that these prostanoids play an important role in the regulation of cerebral parenchymal circulation. 31,32Previous studies on rat brain parenchymal arterioles also documented that vasoconstrictor substances other than prostaglandin F^2α, including potassium chloride or endothelin-1, can induce strong vasoconstriction. 10These results suggest that the above-mentioned substances are similarly effective as vasoconstrictors in cerebral parenchymal arterioles.

One may argue that the parenchymal arterioles in the current study were not perfused in vitro  and thus their reactivity to the vasodilator as well as vasoconstrictor substances may be altered as compared with the vasomotor reactivity in vivo . It is important to note that endothelial as well as nonendothelial control of vascular tone produced by shear stress is one of the determinants of vascular tone, especially in the resistance blood vessels including cerebral arterioles. 33,34Therefore, it may be difficult to extrapolate directly our results to in vivo  situations because we cannot rule out the possibility that the absence of intraluminal flow in cerebral arterioles may alter the behavior of our preparation. However, parenchymal arterioles in our preparation, similar to pressurized cerebral arterioles, responded well to vasoactive substances. Because the parenchymal arterioles of the brain slice are embedded in the intact neuronal tissues composed of both neuronal and glial cells attached to the arterioles, the resting vascular tone of these arterioles may be close to that seen under in vivo  conditions. In addition, even under in vivo  conditions, the intraluminal blood flow as well as the pressure of arterioles that exist in precapillary levels should be fairly low. These results appear to support the conclusion that the nonpressurized situation itself may not be a major problem during the study of arterioles embedded in the cerebral parenchyma.

A hypercapnia-induced increase in cerebral blood flow appears to be one of the mechanisms that modulates the acid-base balance in neuronal cells, resulting in maintenance of normal neuronal activity. Indeed, ATP-sensitive K⁺channels can be activated in ischemic situations including hypoxia, hypercapnia, and acidosis, indicating that these channels are involved in the regulation of cerebral blood flow during critical situations for neuronal cells. 7,8,35Our study using cerebral parenchymal arterioles may provide a hint to improve cerebral microcirculation during anesthesia.

The authors thank Zvonimir S. Katusic, M.D., Ph.D. (Professor of Anesthesia Research, Mayo Clinic, Rochester, MN), for his review of the manuscript.