VOLATILE anesthetics induce neuron hyperpolarization and consequent depression of the central nervous system. 1–10In addition to the well-known potentiation of γ-aminobutyric acid type A and glycine chloride channels, 2,3evidence demonstrates that in both invertebrates and vertebrates, volatile anesthetics open background K+channels and thus increase the resting membrane potential. 8,9,11,12For instance, in the mollusk Lymnaea , halothane opens a class of baseline K+channels (IKAn) that hyperpolarize and silence pacemaker neurons (figs. 1A and B). 9,11,13In Aplysia californica , halothane-mediated neuronal hyperpolarization is caused by the opening of the background S-type (serotonin-sensitive) K+channel. 12Similarly, opening of baseline acid-sensitive K+channels by volatile anesthetics produces rat hypoglossal and locus coeruleus neuron hyperpolarization. 8,14
Recent reports demonstrate that the volatile anesthetic–sensitive background K+channels belong to the family of mammalian K+channel subunits with four transmembrane segments and two P regions (fig. 2and table 1). 8,15–17Interestingly, the yeast 2P domain K+channel, which is characterized by eight instead of four transmembrane segments, is also activated by volatile anesthetics, demonstrating the conservation of this pharmacology. 18,19The mammalian 2P domain K+channel subunits are characterized by diverse patterns of expression and functional properties. 8,15,20–24Volatile anesthetics selectively open human TREK-1, TREK-2, TASK-1, TASK-2, TASK-3, and TALK-2 channels. 15–17,25,26On the contrary, local anesthetics reversibly inhibit the 2P domain K+channels. 26–29In the current report, the expression and properties of these anesthetic-sensitive K+channels are reviewed, and their possible functional role in the mechanisms of anesthesia and analgesia is discussed.
Mammalian 2P Domain K+Channels
Mammalian K+channels belong to three main structural classes made of two, four, or six transmembrane segments (TMSs). 30,31The common feature of all K+channels is the presence of a conserved motif called the P domain (the K+channel signature sequence or putative pore-forming region), which is part of the K+conduction pathway. 32The two TMS and six TMS classes are characterized by the presence of a single P domain, whereas the most recently discovered class of four TMS subunits is characterized by the presence of a tandem of P domains (fig. 2A). 30,31Functional K+channels are tetramers of pore-forming subunits for the two and six TMS classes and possibly dimers in the case of the four TMS class. 33
The 2TMS–1P K+channels encode the inward rectifiers. These K+channels close with depolarization because of channel block by intracellular Mg2+and polyamines (for review, see report by Ruppersberg 34). The conductance increases on hyperpolarization, and, consequently, the inward K+currents recorded at potentials below the equilibrium potential EK+(approximately −90 mV in a physiological K+gradient) are much larger than the outward K+currents recorded at depolarized potentials. Although the amplitude of the outward currents flowing through the inward rectifiers is limited, they will have a major influence on the resting membrane potential. 35The voltage at which channel gating by intracellular Mg2+and polyamines occurs will set the range in which the K+channel will influence the cell membrane potential. Because they are blocked at depolarized potentials, these channels will have a small but limited role in the repolarization of the action potential.
By contrast, the outward rectifiers encoded by the 6TMS–1P subunits open on depolarization (Kv channels) and after intracellular Ca2+increase (BK and SK channels). Depolarization is sensed by the positively charged fourth TMS of Kv channels, which is coupled to activation gates. Opening of the voltage-gated K+channels is time-dependent (delayed rectifiers) and contributes to repolarize and terminate the action potential. 35Several voltage-gated K+channels are also characterized by a fast (N-type) inactivation process. The inactivation gate is the positively charged amino terminus of these specific subunits (ball-and-chain mechanism).
The class of mammalian 4 TMS K+channel subunits has expanded to include 14 members (fig. 2Band table 1). These subunits share the same structural motif with 4TMS–2P, an extended M1P1 extracellular loop, and both amino and carboxy termini facing the cytosol. However, low sequence identity is found outside the pore domain. 30The mammalian 2P domain K+channels are classified into five main structural subgroups: (1) TWIK-1, TWIK-2, and KCNK7 36–40(KCNK7 is not functional); (2) TASK-1, TASK-3, and KT 3-3 (Acc No.: NM 022358; expression has not been reported) 21,23,28,41–43; (3) TREK-1, TREK-2, and TRAAK 17,20,44,45; (4) TASK-2, TALK-1, and TALK-2 (also called TASK-4) 16,25,46,47; and (5) THIK-1 and THIK-248(THIK-2 is not functional) (fig. 2B). 2P domain K+channels have also been identified in Drosophila and Caenorhabditis elegans . 49–51In the nematode, approximately 50 of 80 K+channel subunits belong to the 2P domain family. 49,52In the Drosophila , 11 of 30 K+channel subunits belong to the 2P domain family. 53Very little conservation (< 35%) is found between nematode, Drosophila , and human sequences.
TWIK-1, the first mammalian 2P domain K+channel to be identified, self-associates to form disulfide-bridged homodimers. 33The extracellular M1P1 interdomain, predicted to form an amphipathic helix, promotes self-dimerization. A cysteine located in this domain appears to be important for the dimerization of some, but not all 2P domain K+channels, for instance, TASK-1. 33A dimer contains 4P domains, which are essential in the formation of K+-selective pores. No evidence for heteromultimerization has yet been reported for the 2P domain K+channels.
Leak channels, which are opened pores in the membrane, have no voltage or time dependency by definition. 35Activation is instantaneous on depolarization as the channels are always opened at rest. Because of an asymetrical physiological K+gradient (approximately 150 mm intracellular and 5 mm extracellular), the current-to-voltage relation of a K+leak channel is predicted to be outwardly rectifying (Goldman, Hodgkin, and Katz constant-field theory, also called the open rectification) (fig. 2C). 35In a symmetrical K+gradient, the current-to-voltage relation of the leak channel is expected to be linear, as observed for TASK-1, TRAAK, and TALK-1 (fig. 2C). These K+channels thus behave as open rectifiers. Because of this leak characteristic, the background 4TMS–2P K+channels are expected to influence both the resting membrane potential (along with the inward rectifiers) and the repolarization phase of the action potential (along with the voltage-gated and Ca2+-activated outward rectifiers). Several 4TMS–2P background K+channels are, however, more than a simple open rectifier leak channel (figs. 2D and E). For instance, TREK-1 is a K+channel that displays a strong outward rectification in a symmetrical K+gradient (fig. 2D). 20,29This rectification is at least partly caused by a voltage-dependent gating. 29Similar outward rectification has also been observed for TALK-2, TASK-2, TREK-2, and TASK-3. 17,25,28,46Moreover, the activation of TASK-2 and TREK-2 is time-dependent. 16,17,46On the contrary, TWIK-2 shows a mild inward rectification when recorded in a symmetrical K+gradient (fig. 2E). 37Mild inward rectification is also typical to TWIK-1 and THIK-1. 36,48TWIK-2 is additionally characterized by a time-dependent inactivation. 37The various 4TMS–2P channels will differentially tune the resting potential or the action potential duration because of these particular rectification-, time-, and voltage-dependent properties.
Mechano-gated TREK-1 and TREK-2 K+Channels are Opened by Inhalational Anesthetics
Human TREK-1 is highly expressed in brain and ovary and to a lesser extent in kidney and small intestine. 15,17,54,55In human brain, TREK-1 shows the greatest expression in the caudate nucleus, the putamen, the spinal cord, and the dorsal root ganglia (fig. 3). 17,55At the protein level, TREK-1 is present at both synaptic and nonsynaptic sites in mouse brain. 22,56,57Significant expression is also detected in both small and medium sensory neurons of mouse dorsal root ganglia. 22Human TREK-2 (78% of homology with TREK-1) is abundantly expressed in kidney and pancreas and more moderately in testis, brain, colon, and small intestine. 17In human brain, TREK-2 shows the strongest expression in the caudate nucleus, the cerebellum, the corpus callosum, and the putamen (fig. 3). 17,55Some tissues only express TREK-1 (ovary) or TREK-2 (pancreas, colon). 17Other tissues do not express these channels or only to very modest levels (heart, skeletal muscle, lung, blood leukocytes, and spleen). Finally, some tissues present overlapping expression (brain, kidney, small intestine).
TREK-1 and TREK-2 channels are mechano-gated K+channels opened by membrane stretch. 17,22,29,44At the whole cell level, TREK-1 is modulated by cellular volume. 22,29Mechanical force is believed to be transmitted directly to the channel via the lipid bilayer. 29,58Lowering intracellular pH shifts the pressure–activation relation of TREK-1 and TREK-2 toward positive values and ultimately leads to channel opening at atmospheric pressure. 17,59Acidosis essentially converts a TREK mechano-gated channel into a constitutively active background K+channel. 17,59TREK-1 is also gradually and reversibly opened by heat. An increase in temperature of 10°C enhances TREK-1 current amplitude by approximately sevenfold. 22Finally, TREK-1, TREK-2, and TRAAK are reversibly opened by polyunsaturated fatty acids and lysophospholipids, including arachidonic acid and lysophosphatidylcholine (fig. 4). 17,29,44,45,60Deletional analysis demonstrates that the carboxy terminus, but not the amino terminus and the extracellular M1P1 loop, is critical for activation of TREK-1 by stretch, arachidonic acid, lysophospholipids, intracellular acidosis, and temperature. 22,29,59TREK-1 and TREK-2 activation are reversed by protein kinase A stimulation. 17,20,22,29,60Protein kinase A–mediated phosphorylation of Ser333 in the carboxy terminus mediates TREK-1 closing. 29TREK-1 and TREK-2 are insensitive to most of the classical K+channel blockers, including tetraethylammonium (TEA+; 10 mm), 4-aminopyridine (4-AP; 3 mm), Ba2+(1 mm), glibenclamide (10 μm), charybdotoxin (1 μm), and apamin (10 μm). TREK-1 is reversibly blocked by Gd3+(30 μm), amiloride (2 mm), and chlorpromazine (10 μm). 29,45,60
TREK-1 and TREK-2 are opened by chloroform, diethyl ether, halothane, and isoflurane in transfected mammalian cells (figs. 4 and 5A). 15,17,29Opening of these channels by anesthetics induces cell hyperpolarization (fig. 5B). Interestingly, the other structurally and functionally related 2P domain K+channel, TRAAK, is insensitive to volatile anesthetics. 15Human TREK-1 is most sensitive to chloroform (2.3-fold increase in current amplitude at 1 mm), whereas halothane is the strongest opener of human TREK-2 (2.3-fold increase at 1 mm). 15,17In excised outside-out patches, the 48 pS TREK-1 channel is reversibly opened in a dose-dependent manner by chloroform and halothane (fig. 5C). 15No channel activity is observed in the absence of anesthetic, suggesting that it converts inactive channels to active ones. Deletion of the amino terminus does not affect anesthetic-induced TREK-1 opening. 15In contrast, deletion of the carboxy terminus at Thr 322 completely suppresses responses to both chloroform and halothane. 15These results demonstrate that the carboxy terminus, but not the amino terminus, of TREK-1 is critical for anesthetic activation.
TREK-1 and TREK-2 share all the functional properties of the anesthetic-sensitive background K+channels in Lymnaea pacemaker neurons and the S channel in Aplysia sensory neurons (fig. 1B). 9,11,12Both endogenous and cloned background K+channels are opened by volatile anesthetics in excised patches, suggesting a direct mechanism of action (fig. 5C). 12,15Moreover, the lack of effect of volatile anesthetics on TRAAK, another mechano-gated polyunsaturated fatty acid–sensitive 2P domain K+subunit, suggests that an indirect membrane effect is unlikely. 15Pure optical isomers of the volatile anesthetic isoflurane show clear stereoselectivity in activating background K+currents in pacemaker neurons of the Lymnaea . 13The + isomer is more potent than the − isomer in hyperpolarizing neurons, suggesting that volatile anesthetics act by direct binding to the protein rather than a nonspecific perturbation of lipids. 13Although the stimulation by inhalational anesthetics seems to be direct, one cannot fully rule out possible indirect effects. 11,12,15Indeed, it has been shown in Lymnaea neurons that, although the activation of IKAn by volatile anesthetics is independent of the lipoxygenase and cyclooxygenase pathways, it might involve the cytochrome P450 pathway. 11Moreover, TREK-1 is not sensitive to volatile anesthetics when expressed in Xenopus oocytes. 16These negative results may indicate that either a specific membrane environment or critical cofactors, which are absent in Xenopus oocytes but present in mammalian cells, may be required.
Acid-sensitive Background K+Channels TASK-1 and TASK-3 Are Opened by Volatile Anesthetics
Human TASK-1 is particularly abundant in the pancreas and the placenta. 21Lower levels of expression are found in brain, lung, prostate, heart, kidney, uterus, small intestine, and colon. High concentrations of TASK-1 are found in cerebellar and olfactory granule cells, olfactory tubercles, scattered neurons through all layers of cerebral cortex, intralaminar thalamic nuclei, pontine nuclei, and the locus coeruleus of the rat. 8,21Brainstem and spinal cord motorneurons display the strongest expression of TASK-1. 8Motor nuclei with high concentrations of TASK-1 include facial, hypoglossal, ambigual, and motor trigeminal, as well as the vagal motor nucleus. TASK-1 is also particularly abundant in rat carotid body cells. 61In human brain, the strongest expression is found in the cerebellum, thalamus, and pituitary gland (fig. 3). 55TASK-1 is also particularly abundant in human dorsal root ganglia. 55Human TASK-3, which is 62% identical to TASK-1, is largely and selectively expressed in the cerebellum. 42,55
TASK-1 and TASK-3 K+currents are instantaneous (or at least very rapidly activating 28,62) and non-inactivating. 8,21,23,26,41,43TASK-1 and TASK-3 are background K+currents that are very sensitive to variations in extracellular pH (fig. 2C). 8,21,23,26,28,41Half of TASK-1 and TASK-3 channels are closed at pH values of 7.3 and 6.5, respectively. 8,21,23,26,41,43TASK-1 is inhibited by Zn2+, the local anesthetic bupivacaine, the anticonvulsant phenytoin, quinidine, and Ba2+. 8,21,23,43However, it is resistant to TEA+and 4-AP. TASK-1 is inhibited by receptors coupled to Gq proteins, including the M3 muscarinic receptor. 63,64The second messenger involved in channel inhibition is still unknown. The endocannabinoid anandamide has recently been shown to be a direct and selective blocker of TASK-1. 65It has been suggested that anandamide might be involved in receptor-induced down-modulation of TASK-1. 65TASK-3 is blocked by lidocaine, bupivacaine, barium, and quinidine but resistant to tetraethylammonium. 26,28,41
TASK-1 is opened by halothane and isoflurane but is insensitive to chloroform and partially inhibited by diethylether (figs. 6A and B). 8,15Halothane opens TASK-1 in the excised patch configuration despite a strong rundown of channel activity on excision (fig. 6C). 15Deletion of the last 147 amino acids in the carboxy terminus of TASK does not alter halothane sensitivity, whereas further deletion kills channel activity. Fusing the carboxy terminus of TREK-1 restores basal but not anesthetic-stimulated activity, demonstrating that the region located between residues 242 and 248 (VLRFMT) is critical for anesthetic sensitivity. 15
A background K+current sets the resting membrane potential in type I carotid body chemoreceptor cells. 61,66Reversible inhibition of this background K+current by hypoxia or acidosis induces membrane depolarization. Depolarization of type I cells leads to opening of voltage-gated Ca2+channels, an increase in intracellular Ca2+, and release of neurotransmitters, including dopamine. 67The released neurotransmitters stimulate sinus nerve endings and trigger the reflex increase in respiration. The endogenous background K+channel in type I cells shares the biophysical and pharmacologic properties of TASK-1. 61It is reversibly inhibited by mild external acidosis, time- and voltage-independent, resistant to TEA+and 4-AP, but blocked by Ba2+, Zn2+, bupivacaine, and quinidine. Moreover, the type I cell background K+current is enhanced by halothane but is insensitive to chloroform. 61Opening of TASK-1–like background K+channels in type I carotid body cells may be partially responsible for the suppression of hypoxic ventilatory drive during general anesthesia.
In rat somatic motoneurons, locus coeruleus neurons, and cerebellar granule neurons, inhalational anesthetics similarly activate a background TASK-1–like conductance, causing membrane hyperpolarization and suppressing action potential discharge (figs. 7A and B). 8,65These effects occur at clinically relevant anesthetic concentrations with the steep dose dependence expected for anesthetic effects of these compounds. 8External acidosis to pH 6.5 completely blocks the current activated by anesthetics (fig. 7B, inset). 8,65In motoneurons and cerebellar granule neurons, opening of TASK-1 channels may contribute to anesthetic-induced immobilization, whereas in the locus coeruleus, it may support analgesic and hypnotic actions. 8
Application of 1 mm halothane reversibly potentiates human TASK-3 current amplitude by 66%. 26The onset of halothane stimulation in Xenopus oocyte is rapid (τ: 62 s), whereas full recovery takes as long as 5 min. By contrast, the neurosteroidal anesthetic alphaxolone inhibits TASK-3. 26Pentobarbital and ketamine do not significantly affect TASK-3 at a concentration of 100 μm. 26
Spinal Cord and Dorsal Root Ganglion Background K+Channel TASK-2 Is Opened by Inhalational Anesthetics
Human TASK-2 is found in kidney, pancreas, lung, and placenta. 46Although TASK-2 expression is weak in whole brain, it can be detected by polymerase chain reaction in rat spinal cord. 16TASK-2 is found throughout the spinal cord in both ventral and dorsal sections. 16Moreover, TASK-2 is expressed in high concentrations in human spinal cord as well as in dorsal root ganglia (fig. 3). 55TASK-2 produces non-inactivating, outwardly rectifying K+currents with activation potential thresholds that closely follow the K+equilibrium potential. 16,46TASK-2 activation is time-dependent with a fast time constant of approximately 60 ms at 0 mV. TASK-2 currents are blocked by quinine and quinidine but not by the classical K+channel blockers TEA+, 4-AP, Cs+, and Ba2+. 46TASK-2 is inhibited by external acidosis with a half-inhibition at pH 7.8. 16,46Application of volatile anesthetics causes a concentration-dependent increase in TASK-2 currents in a range overlapping minimum alveolar concentrations. 16TASK-2 is more sensitive to halothane than isoflurane. 16Unlike TASK-1, TASK-2 is also stimulated by chloroform (fig. 4). Site-directed mutagenesis has been used to delete the carboxy terminus of TASK-2. 16The truncated TASK-2 channel does not express a spontaneous or volatile anesthetic–evoked activity, further demonstrating the critical role of the carboxy terminus in the function of 2P domain K+channels. 15,16,29
Volatile Anesthetic–inhibited 2P Domain K+Channels
TWIK-2 is a background K+channel that is highly expressed in both visceral and vascular smooth muscle. 37Human TWIK-2 is absent in the brain and in the cerebellum (fig. 3). 37–39,55However, a moderate to strong expression is found in the human spinal cord and in the dorsal root ganglia. 55Chloroform (300 μm) and halothane (750 μm) reversibly inhibit TWIK-2 by 32 and 27%, respectively. 37
THIK-1 expression is ubiquitous, with a substantial expression in some restricted areas of the rat brain (fig. 3). 48The strongest concentrations are found in the olfactory bulb granule cell layer, the lateral septal nucleus dorsal, the ventromedial hypothalamic nucleus, the thalamus reticular and reunions nuclei, and, finally, the parabrachial nuclei. 48THIK-1 is a weak inward rectifier that is stimulated by arachidonic acid but inhibited by halothane, with an IC50value of 2.83 mm (fig. 4). 48Interestingly, 1 mm chloroform fails to affect THIK-1. 48
Human TALK-1 is exclusively expressed in the pancreas. 25Human TALK-2 is similarly found in high concentrations in the pancreas but is also present in liver, placenta, heart, and lung. 25TALK-1 and TALK-2 are background K+currents that are activated by alkaline pH but are insensitive to arachidonic acid. Both channels are inhibited by 800 μm chloroform (−21 and −44%, respectively) and 800 μm halothane (−27 and −56%, respectively) (fig. 4). 25,47Interestingly, TALK-1 is not sensitive to 800 μm isoflurane, whereas TALK-2 is stimulated (+58%). 25
Inhibition of Background K+Channels by Local Anesthetics
Bupivacaine, tetracaine, and lidocaine inhibit several mammalian 2P domain K+channels, including TASK-1, TASK-3, TREK-1, and TWIK-1. 26–29The presence of the uncharged form of the local anesthetic is important for channel inhibition. TASK-1 inhibition is greater at alkaline pH values (28% inhibition with 10 μm bupivacaine at pH 8.4). 27The potency of inhibition is directly correlated with the octanol:buffer distribution coefficient of the local anesthetic. The IC50values for TASK-1 are 709 μm mepivacaine, 222 μm lidocaine, 51 μm R (+)-ropivacaine, 53 μm S (−)-ropivacaine, 668 μm tetracaine, 41 μm bupivacaine, and 39 μm etidocaine. 27The Hill coefficient of the dose–effect curves is close to 1, suggesting that a single local anesthetic molecule binds to each 2P domain K+channel. The stereoisomers of ropivacaine are equipotent, demonstrating that the inhibition is not stereoselective. 27Local anesthetics significantly depolarize cells expressing TASK-1. The membrane depolarization may enhance the binding of local anesthetics to the open and inactivated states of the voltage-dependent Na+channels. 27Therefore, the inhibition of background 2P domain K+channels by local anesthetics should contribute to enhance the conduction block of peripheral nerves.
Conclusions and Perspectives
These recent findings provide strong evidence that 2P domain K+channels are sensitive molecular targets for volatile anesthetics. Together with the known modulation by neurotransmitter receptors, 2,3opening of these channels will contribute to the hyperpolarizing action of general anesthetics. 15
Opening of the 2P domain K+channels is agent-specific. For instance, TREK-1, TREK-2, and TASK-2 are opened by chloroform; TASK-1, THIK-1, and TRAAK are unaffected; and TWIK-2, TALK-1, and TALK-2 are inhibited. 15–17,25,37,48Interestingly, TALK-2 is inhibited by chloroform and halothane while it is stimulated by isoflurane. 25The volatile anesthetic–inhibited 2P domain K+channels, including TWIK-2, THIK-1, TALK-1, and TALK-2, are mostly nonneuronal (THIK-1 is restricted to some brain areas) and strongly expressed in peripheral organs. 25,37,48TREK-1, TREK-2, and TASK-1 are more sensitive to halothane than isoflurane. 15,17This difference in potency between the two volatile anesthetics indicates that opening of 2P domain K+channels will probably be prevalent during halothane-induced anesthesia. Clearly, further work is necessary to map the putative anesthetic binding site and understand the role of the carboxy terminus in the mechanism of channel activation. 15
Opening of 2P domain K+channels will have profound hyperpolarizing effects at both presynaptic and postsynaptic levels. Because of the leak behavior of the 2P domain K+channels, even moderate stimulation, as observed for clinical doses of volatile anesthetics, will have a major effect on the membrane potential. 30,31The discovery of this class of anesthetic-sensitive K+channels with distinct patterns of expression may provide a basis for how inhalational anesthetics depress the central nervous system. 15-17,55In the human brain, opening of TREK-1 and TREK-2 in the caudate nucleus and the putamen, opening of TREK-2 and TASK-3 in the cerebellum, and opening of TREK-2 in the corpus callosum may be functionally important during general anesthesia by volatile anesthetics (fig. 3). 15,17,55Opening of TREK-1, TASK-1, and TASK-2 in the sensory neurons of human dorsal root ganglia may block sensory inputs and contribute to loss of consciousness. 15–17,55Finally, opening of TREK-1 in motoneurons by inhalational anesthetics may depress mobility (fig. 3). 15,55
Local anesthetics inhibit background K+channels and induce cell depolarization. 27These results suggest that closing of 2P domain K+channels by local anesthetics could contribute to peripheral analgesia by augmenting conduction blockade, whereas opening by volatile anesthetics may contribute to general anesthesia. 15,27These recent results contribute to the understanding of the molecular and cellular mechanisms of action of anesthetics.
The authors thank Michel Lazdunski, Ph.D. (Director, Institut de Pharmacologie Moléculaire et Cellulaire–Centre National de la Recherche Scientifique, Valbonne, France), for continual support; François Maingret, Ph.D. (Student, Institut de Pharmacologie Moléculaire et Cellulaire–Centre National de la Recherche Scientifique, Valbonne, France), for his contribution to the manuscript; and Martine Jodar, Valérie Lopez, and Franck Aguila (Technical Staff, Institut de Pharmacologie Moléculaire et Cellulaire–Centre National de la Recherche Scientifique, Valbonne, France) for excellent technical assistance.