Protective Role of ATP-Sensitive Potassium Channels in Hypoxia-Induced Generalized Seizure

See allHide authors and affiliations

Science  25 May 2001:
Vol. 292, Issue 5521, pp. 1543-1546
DOI: 10.1126/science.1059829


Adenosine triphosphate (ATP)–sensitive potassium (KATP) channels are activated by various metabolic stresses, including hypoxia. The substantia nigra pars reticulata (SNr), the area with the highest expression of KATPchannels in the brain, plays a pivotal role in the control of seizures. Mutant mice lacking the Kir6.2 subunit of KATP channels [knockout (KO) mice] were susceptible to generalized seizures after brief hypoxia. In normal mice, SNr neuron activity was inactivated during hypoxia by the opening of the postsynaptic KATPchannels, whereas in KO mice, the activity of these neurons was enhanced. KATP channels exert a depressant effect on SNr neuronal activity during hypoxia and may be involved in the nigral protection mechanism against generalized seizures.

KATP channels (1) couple the intracellular metabolic state to electrical activity at the plasma membrane (2). We have previously reported the molecular structure of KATP channels (3, 4) comprising the inwardly rectifying K+ channel Kir6.2 and a sulfonylurea receptor with high affinity (SUR1 in pancreatic β cells) or low affinity (SUR2A in the heart) for sulfonylureas. High-affinity binding of [3H]glibenclamide in the brain is strongest in the SNr, suggesting high expression of the β cell type KATPchannel in this nucleus (5, 6). Because the SNr acts as a central gating system in the propagation of seizure (7–9) and generalized seizures can be evoked by metabolic stresses such as hypoxia and hypoglycemia (10), these KATP channels could well be involved in the development of seizure during ATP-depleted conditions.

Kir6.2 KO mice (11) were used to evaluate this possibility. Daily behavior and basal physiological values of KO mice were not significantly different from those of wild-type mice (12,13). However, responses to brief (150 s) hypoxia caused by oxygen deprivation (n = 19, 5.4 ± 0.2% O2) differed in KO and wild-type mice (14). The wild-type mice (10/10) all remained sedated during the challenge and revived normally. In contrast, the KO mice all responded with a myoclonic jerk (latency = 8.9 ± 1.1 s, n = 9) followed by severe tonic-chronic convulsion and death (survival time = 21.8 ± 5.2 s,n = 9) (Table 1). Under more severe hypoxic conditions (n = 6, 4.3 ± 0.2% O2, P < 0.0001 compared with 5.4 ± 0.2% O2), four of the six wild-type mice exhibited a generalized convulsion (latency = 25.8 ± 2.7 s) (15). Electroencephalogram (EEG) and electromyogram (EMG) (14) revealed a sequence of seizure patterns in conscious KO mice (n = 5) challenged with 5.4 ± 0.1% O2 (Fig. 1A). First, very low-voltage EEG for about 3 s indicated loss of consciousness. Then fast waves after an abrupt, sharp deflection lasted for several seconds in the EMG traces, corresponding to the tonic convulsion and myoclonus, after which bilateral, high-voltage sharp wave bursts were observed in the EEG traces. In wild-type mice under the same conditions, a medium- to low-voltage EEG predominated during the hypoxic period (Fig. 1B). This suggests that KATP channels participate in determining the seizure threshold during hypoxia. It is unlikely that the seizures observed in KO mice were produced by rapid cardiac arrest, because heartbeats continued after the seizure (16).

Figure 1

A generalized convulsive seizure in KO mice. (A) Representative EEG and neck-muscle EMG traces in a KO mouse just before application of N2 (in normoxia, 20.9% O2, horizontal bar) and for the initial 27 s of hypoxia (5.4% O2, total duration = 150 s). The myoclonus (asterisk) followed by the bilateral, convulsive seizure is represented. L and R, EEG traces from left and right cortices. Open and solid triangles, the onset and end of N2 flow delivery, respectively. (B) Data from a wild-type (WT) mouse.

Table 1

KO mice are susceptible to seizure by hypoxia. The mice were subjected to brief (150 s) hypoxia. Inspired oxygen concentrations are indicated. The ambient temperature was 23.0° to 25.6°C. Values are expressed as means ± SD. *,P < 0.0001 versus 5.4 ± 0.2 (unpairedt test).

View this table:

To investigate the role of SNr neuron activity, we recorded single unit activities from the SNr in acute slice preparations (17). In control, the firing rate of SNr neurons was not significantly different in wild-type and KO mice [25.2 ± 1.8 Hz (n = 60) and 22.6 ± 2.1 Hz (n= 47) for wild-type and KO mice, respectively]. However, in brief hypoxia (90 s) (18), wild-type neurons showed a marked decrease in firing rate to about one-third that before hypoxia (from 28.4 ± 2.0 Hz to 10.2 ± 3.2 Hz, n = 9,P = 0.0003; Fig. 2, A and C), whereas the firing rate of the KO neurons increased about 1.8-fold (from 30.6 ± 1.6 Hz to 53.8 ± 3.8 Hz, n = 8, P = 0.0002; Fig. 2, B and C). These results suggest that the KATP channel–mediated suppressive effect on SNr activity is sufficient to reverse the facilitation during hypoxia in these conditions.

Figure 2

Effect of hypoxia on firing rate of SNr neurons in acute slice preparations. Brief (90 s) hypoxia (solid bar) produced a marked decrease in firing rate of SNr neuron of wild-type mouse (A) but increased it in KO mouse (B). Insets represent traces of unit activities before, during, and after hypoxia (arrows). Spike amplitude increased in wild-type but decreased in KO mouse during hypoxia. (C) Hypoxia-induced changes in firing rate of wild-type (open circles, n = 9 from nine mice) and KO (solid triangles, n = 8 from eight mice) mice. Data points represent means ± SE.

To investigate the ionic mechanism of the suppression of SNr activity, we recorded from acutely dissociated neurons. γ-Aminobutyric acid–ergic (GABAergic) neurons, which constitute the majority of SNr neurons (19), were identified by their membrane electrical properties (20). With inside-out patch recordings, KATP channel currents were characterized in the SNr GABAergic neurons of wild-type mice (21). ATP decreased the open-state probability of the channels in a dose-dependent manner (half-maximal concentration = 12.0 μM, n = 4 to 8). It also was decreased significantly by tolbutamide (0.5 mM in the presence of 1 μM ATP) (n = 11, P < 0.005) and increased by the selective β cell KATP channel opener diazoxide (0.5 mM in the presence of 100 μM ATP) (n = 7, P < 0.005). The current-voltage relation showed inward rectification (conductance = 77.4 ± 0.3 pS; 140 mM K+ on both sides of the membrane, n = 7; 1 μM ATP). No such channels were observed in the neurons of KO mice (n = 32). These results indicate that pancreatic β cell type KATPchannels are functionally expressed in the SNr neurons of wild-type mice but not in KO mice. This confirms cell-attached patch (22) and reverse transcriptase polymerase chain reaction experiments showing that SUR1 and Kir6.2 are expressed in SNr neurons, whereas no SUR2A, SUR2B, or Kir6.1 is expressed (23).

The responses to hypoxia of the SNr neurons and the involvement of the KATP channels were further investigated by perforated patches (24). Under control normoxic condition (oxygenated with 100% O2), SNr neurons of both wild-type and KO mice exhibited similar tonic, high-frequency, spontaneous firings [10.8 ± 1.1 Hz (n = 38) and 13.0 ± 1.4 Hz (n = 26), respectively], and there was no significant difference in membrane properties (20). However, the membrane potentials of the SNr neurons were shifted in the opposite direction when these neurons were perfused with hypoxic solution (25). Wild-type SNr neurons were hyperpolarized, and the hyperpolarization was reversed by tolbutamide (0.1 mM,n = 10) (Fig. 3A). Diazoxide (0.3 mM) also produced hyperpolarization of −5.6 ± 0.8 mV (P < 0.0001, n = 22) (Fig. 3A). In contrast, KO neurons showed no such hyperpolarization but were depolarized (Fig. 3B). During the 20-min perfusion with hypoxic solution, wild-type and KO SNr neurons elicited the maximal hyperpolarization of −9.6 ± 1.0 mV (P < 0.0001,n = 13) and maximal depolarization of 7.9 ± 1.0 mV (P < 0.0001, n = 15), respectively (Fig. 3C). The hyperpolarizing effect of hypoxia on wild-type SNr neurons was mimicked by the mitochondrial metabolic inhibitor 2,4-dinitrophenol (30 μM) (16).

Figure 3

Hypoxia produced contrary effects on membrane potential of wild-type and KO SNr neurons. (A) Acutely dissociated wild-type SNr neuron is hyperpolarized with cessation of firings by diazoxide (0.3 mM, hatched bar) and by hypoxia (pO2 < 1 torr, solid line). Tolbutamide (0.1 mM, open bar) had no effect at resting potential but reversed hypoxia-induced hyperpolarization. (B) KO SNr neurons were depolarized during hypoxia. Both diazoxide and tolbutamide had no effect on membrane potential. The thick baseline at the end of the hypoxic period consists of spikes of very low amplitude. Recordings were performed by nystatin patch in the current clamp mode. The membrane potential gap observed at the onset and end of hypoxia is an artifact of solution switching. (C) Effect of hypoxia on membrane potentials of wild-type (n = 13) and KO (n = 15) mice. Solid and open columns represent membrane potentials in the normoxic condition and maximal hypoxic response, respectively. Values are means ± SE.

The contribution of postsynaptic KATP channels to the response to hypoxia of the GABAergic SNr neurons also was examined in acute slice preparations. When both excitatory and inhibitory fast neurotransmissions were blocked by a combination of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (20 μM),d,l-2-amino-5-phosphonovaleric acid (dl-APV) (50 μM), and bicuculline (20 μM), the firing rate of both wild-type and KO SNr neurons increased about 35% during the normoxic condition (36.4 ± 6.3%, n = 7,P = 0.0009 and 33.7 ± 4.4%, n = 6, P = 0.0011, respectively) (Fig. 4, A and B). However, hypoxia produced striking contrasts in the firing of wild-type and KO neurons (Fig. 4, A, B, and C). In addition, tolbutamide (200 μM) reversed the hypoxia-induced decrease in the firings of wild-type neurons. This increase was seen in the absence (P = 0.0042,n = 6; Fig. 4, D and E) or presence (P= 0.018, n = 4; Fig. 4F) of CNQX, APV, and bicuculline. Tolbutamide had no effect on firings in normoxia (n = 8, Fig. 4D). These results suggest that postsynaptic KATPchannels are critical in the hypoxia-induced inactivation of the SNr neurons seen in wild-type mice, although presynaptic modulatory effects on firings have been described (26, 27).

Figure 4

Effect of synaptic transmission blockade and tolbutamide on hypoxia-induced change in firing rate of SNr neurons. The combination of 20 μM CNQX, 50 μM dl-APV (APV), and 20 μM bicuculline (BIC) had no effect on hypoxia-induced change in firing rate of wild-type (A) and KO (B) neurons, although it increased the firing rate in the normoxic condition in both groups. (C) Changes in firing rate of wild-type (open circles, n = 4 from four mice) and KO (solid triangles, n = 3 from three mice) mice due to hypoxia in the presence of three blockers. (D) Tolbutamide (200 μM) reversed the decrease in firing rate because of hypoxia to an increase in a wild-type neuron. (E) Changes in firing rate of wild-type neurons (solid circles, n = 6 from six mice) due to hypoxia in the presence of 200 μM tolbutamide. (F) As in (E), but with blockers to neurotransmissions added (solid circles, n = 6 from six mice). The data points show means ± SE in (C), (E), and (F).

The cellular mechanisms that produce spike facilitation and membrane depolarization in KO SNr neurons are not clear. Inactivation of Na+-K+ adenosine triphosphatase or participation of other types of ATP-dependent and/or O2-sensitive pathways may be possible (28). However, activation of the KATP channels reversed this facilitatory effect in wild-type mice, showing the crucial role of the KATP channels during hypoxia.

We propose that inactivation of SNr neurons after opening of KATP channels protects against seizure propagation during metabolic stress. Extracellular field potentials in cerebral cortex slices evoked by electrical stimulation of the underlying white matter were not altered during hypoxic conditions (90 s) either in KO or wild-type mice (16). A similar result was reported in hippocampus (29). This suggests that SNr neurons, being extremely sensitive to hypoxia, may act as a sensor for hypoxic conditions, although the possibility that other brain regions also are involved cannot be excluded (30). The SNr has been proposed as a key site of the activity of anticonvulsant drugs that enhance GABA-mediated inhibition of seizures (7, 31), and blockade of excitatory neurotransmission in the nucleus raises the threshold for seizures (8). The present study, therefore, suggests the therapeutic potential of selective agonists to this specific KATP channel in the treatment of brain disorders associated with ATP insufficiency such as stroke and metabolic encephalopathies.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed: E-mail: inagaki{at}


View Abstract

Navigate This Article