Abstract
The TASK-3 channel is an acid-sensitive two-pore-domain K+ channel, widely expressed in the brain and probably involved in regulating numerous neuronal populations. Here, we characterized the behavioral and pharmacological phenotypes of TASK-3 knockout (KO) mice. Circadian locomotor activity measurements revealed that the nocturnal activity of the TASK-3 KO mice was increased by 38% (P < 0.01) compared with wild-type littermate controls, light phase activity being similar. Although TASK-3 channels are abundant in cerebellar granule cells, the KO mice performed as well as the wild-type mice in walking on a rotating rod or along a 1.2-cm-diameter beam. However, they fell more frequently from a narrower 0.8-cm beam. The KO mice showed impaired working memory in the spontaneous alternation task, with the alternation percentage being 62 ± 3% for the wild-type mice and 48 ± 4% (P < 0.05) for the KO mice. Likewise, during training for the Morris water-maze spatial memory task, the KO mice were slower to find the hidden platform, and in the probe trial, the female KO mice visited fewer times the platform quadrant than the male KO and wild-type mice. In pharmacological tests, the TASK-3 KO mice showed reduced sensitivity to the inhalation anesthetic halothane and the cannabinoid receptor agonist WIN55212-2 mesylate [(R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanone mesylate] but unaltered responses to the α2 adrenoceptor agonist dexmedetomidine, the i.v. anesthetic propofol, the opioid receptor agonist morphine, and the local anesthetic lidocaine. Overall, our results suggest important contributions of TASK-3 channels in the neuronal circuits regulating circadian rhythms, cognitive functions, and mediating specific pharmacological effects.
The TASK-3 (K2P9.1, KCNK9, KT3.2) channel is an acid-sensitive member of the two-pore-domain background K+ (K2P) channel family (Chapman et al., 2000; Kim et al., 2000; Rajan et al., 2000; Vega-Saenz De Miera et al., 2001). K2P channels contribute to the resting membrane potential by allowing K+ to leak out of the cell. Opening of TASK channels promotes hyperpolarization, whereas their inhibition leads to depolarization (for review, see Goldstein et al., 2001; Bayliss et al., 2003). Native TASK-like currents have been recorded from, e.g., cerebellar granule cells (Millar et al., 2000; Aller et al., 2005; Brickley et al., 2007), hippocampal principal neurons and inhibitory interneurons (Taverna et al., 2005; Torborg et al., 2006), cholinergic cells in the caudateputamen (Berg and Bayliss, 2007), thalamocortical relay neurons (Meuth et al., 2006), hypothalamic orexin/hypocretin neurons (Burdakov et al., 2006), and brainstem noradrenergic, serotonergic, and motor neurons (Sirois et al., 2000; Talley et al., 2000; Washburn et al., 2002).
Diverse neurotransmitters (e.g., acetylcholine, serotonin, noradrenaline, and glutamate) and hormones (e.g., thyrotropin-releasing hormone) act through Gq-coupled receptors to strongly inhibit TASK channels and thus influence neuronal excitability (for review, see Mathie, 2007; Millar et al., 2000; Talley et al., 2000; Chemin et al., 2003; Chen et al., 2006; Meuth et al., 2006). Another interesting feature of TASK channels is that they are activated by clinically relevant concentrations of inhalation anesthetics, such as halothane and isoflurane, suggesting that neuronal hyperpolarization produced by these drugs is partly due to increased K+ flow through TASK channels (for review, see Franks, 2006; Patel et al., 1999; Sirois et al., 2000). In addition, other pharmacological agents modulate TASK function. For example, local anesthetics (lidocaine) and cannabinoid agonists (anandamide, WIN55212-2) inhibit TASK channels in vitro (Kindler et al., 1999; Maingret et al., 2001; Berg et al., 2004). Thus, the working hypothesis is that TASK currents, by providing a mechanism to alter the membrane potential and thus neuronal excitability in many neuronal circuitries, are important mediators of neurotransmitter or drug actions that regulate an animal's physiology and behavior.
Functional TASK channels can be TASK-3 homodimers, heterodimers containing TASK-3 and its closest homolog TASK-1 (K2P3.1, KCNK3), and/or TASK-1 homodimers (Berg et al., 2004; Aller et al., 2005). By in situ hybridization, the TASK-3 gene is widely expressed in the rodent brain (Talley et al., 2001; Vega-Saenz De Miera et al., 2001; Aller et al., 2005; Linden et al., 2006). In the mouse brain, the regional expression patterns of the TASK-1 and TASK-3 genes differ; many forebrain areas have mainly or only TASK-3 expression (Aller et al., 2005; Linden et al., 2006). Other cell types, particularly in the hindbrain areas, e.g., spinal cord and brainstem motor neurons and cerebellar granule cells, strongly coexpress TASK-1 and TASK-3. The channel properties of TASK dimers depend on subunit composition. For example, TASK-3 homodimers are predominantly open at physiological pH 7.4 and, therefore, sensitive to acidification (pK ∼6.8), whereas 50% of TASK-1 homodimers are closed at physiological pH (pK ∼7.4) (Berg et al., 2004; Aller et al., 2005). TASK-3-containing channels and TASK-1 homodimers also differ in sensitivity to endogenous modulators such as Zn2+ (Clarke et al., 2004; Aller et al., 2005).
We have recently generated both TASK-1 KO (Aller et al., 2005; Linden et al., 2006; Meuth et al., 2006) and TASK-3 KO mouse lines (Brickley et al., 2007). The TASK-1 KO mice were grossly normal, but they were more reactive to thermal nociceptive and novel environmental stimuli (Linden et al., 2006). In pharmacological tests, the TASK-1 KO mice showed reduced sensitivity to inhalation anesthetics, the α2-adrenergic receptor agonist dexmedetomidine, and the cannabinoid receptor agonist WIN55212-2 (Linden et al., 2006). Motor coordination was modestly disturbed in the TASK-1 KO mice in line with unaltered resting membrane potentials of their cerebellar granule neurons but enhanced sensitivity to modulatory actions of Zn2+ (Aller et al., 2005). In contrast, resting membrane potentials of cerebellar granule neurons of the TASK-3 KO mice were on average 10 mV more depolarized, causing the granule cells to fire action potentials more easily (Brickley et al., 2007). However, the ability of the cerebellar granule neurons to fire continuously action potentials at high frequency without accommodation is abolished in the TASK-3 KO, whereas it is normal in the TASK-1 KO and wild-type mice (Brickley et al., 2007). Thus, in certain situations, TASK-3 activity can, because of its biophysical properties, promote rather than inhibit cell excitability (Brickley et al., 2007). Therefore, several findings prompted the characterization of behavioral and pharmacological phenotype of adult TASK-3 KO mice.
Materials and Methods
TASK-3 KO Mice. TASK-3 KO mice were generated as described previously with the genetic background being C57BL/6J × 129S1/SvJ (Brickley et al., 2007). Homozygous KO and wild-type littermates used in the present study were from the 5th and 6th generations of heterozygous breeding. Mice were bred and genotyped in Heidelberg (Germany). For behavioral studies, they were transferred to Helsinki (Finland), where they were allowed to equilibrate in their new environment for at least 2 weeks. Adult male and female TASK-3 KO and littermate wild-type mice were maintained in the standard animal facilities in groups of one to five in polypropylene macrolon cages with food pellets and tap water available ad libitum. Lights were on from 6:00 AM to 6:00 PM. All other experiments except the circadian activity were performed during the light phase. All animal tests were approved by the Laboratory Animal Committee of the University of Helsinki and the Southern Finland Provincial Government. Basal behavioral characterization was performed before pharmacological studies. When the same mice were used for multiple tests, a washout period of at least 1 week was kept between experiments, and the order of tests was from less stressful to more stressful.
Basal Behavior. Behavioral and physiological characterization of the TASK-3 KO phenotype was performed similarly as for TASK-1 KO mice (Linden et al., 2006) using a modified version of the primary observational screen described in the SHIRPA protocol (Rogers et al., 1997). The person who observed and recorded the behavior was not aware of the genotype of the tested animals. To measure the body temperature and stress-induced hyperthermia, rectal temperature was recorded, and a mouse was placed immediately in an empty 500-ml glass jar for 10 min, after which the rectal temperature was measured again.
Animal behaviors were recorded using a video tracking system with a charge-coupled device video camera. The animal was detected, and its movements were analyzed automatically using EthoVision Color-Pro 3.0 software (Noldus Information Technology, Wageningen, The Netherlands). Locomotor activity and exploratory behavior in a staircase for 3 min was analyzed as described earlier (Linden et al., 2006). To examine anxiety-related behaviors of TASK-3 KO mice, knockout mice and their wild-type littermate controls were tested using elevated plus-maze and light-dark choice tests. The elevated plus maze test was performed as described earlier (Linden et al., 2006). The mice were placed individually on the central platform facing an open arm and allowed free exploration of the maze for 5 min. The light-dark choice box consisted of a light compartment (30 × 27 cm; walls, 27 cm; painted white), a dark compartment (15 × 27 cm; walls, 27 cm; painted black) covered with an opaque top and an opening between compartments (7.5 × 7.5 cm). The mice were placed in the light compartment, and their behavior was recorded for 5 min.
Circadian Activity. The mice were placed individually in transparent polypropylene cages (38 × 22 × 15 cm) with standard beddings on the floor. Food pellets and water were available ad libitum. Locomotor activity of the mice was measured using photobeam frames (PhotoBeam Activity System; San Diego Instruments Inc., San Diego, CA). A computer control unit registered and converted the breaks of photobeams to the total counts, including horizontal, vertical, and fine movements of the animal. The habituation to novel environment was measured for 4 h at 15-min intervals during the light phase, and the circadian rhythm of locomotor activity was measured after at least 8-h acclimation for 24 h at 60-min intervals.
Motor Performance. To investigate motor coordination and learning, the mice were trained for 7 days (six trials a day) to walk on a rotating rod (Rotamex 4/8; Columbus Instruments, Columbus, OH) for 180 s while the rotation speed accelerated from 5 to 40 rpm (dowel, 4 cm in diameter). The latency to fall from the rod was recorded, and a daily average of six trials was calculated for each animal. The mice were also trained to walk along 1-m wooden beams (1.2 and 0.8 cm in diameter, 84 cm above the floor) back to their home cages. Beam training was performed once a day for 7 days.
Cognitive Functions. A T-maze was used to study working memory (Gerlai, 1998). The T-maze was made of gray plastic and consisted of three arms (50 × 10 cm; walls, 15-cm height). A door separated a 12-cm compartment of the start arm, and two other doors separated 24-cm compartments of the other arms. The mouse was placed in the start compartment with the door closed for 5 s, after which the door was opened, and the latency for the mouse to visit the left arm (a door to the right arm was closed) and to return to the start arm was recorded (test trial). If the test trial took longer than 5 min, the mouse was not accepted for the spontaneous alternation test because of too low exploratory activity. After the test trial, to test spontaneous alternation, the mouse was first confined in the start arm for 5 s, then the door was opened to allow the mouse to visit either the left or right arm. After entering either arm, the opposite arm was closed. Every time the mouse returned to the start arm after entering either arm, it was confined for 5 s in the start arm, after which the doors were again opened for the mouse to visit either arm. The spontaneous alternation was measured for 10 min. If the animal performed less than five trials, it was excluded from the study due to low exploratory activity. A video camera was located above the T-maze, and the experimenter observed the behavior through a video monitor system and remotely operated the doors from an adjacent room.
In a Morris water-maze test (Morris et al., 1982), the mice were trained for 4 days (six trials per day) to find a hidden platform (14 × 14 cm) submerged 1 cm below the surface of water (temperature, 20 ± 2°C). Water was made opaque with milk powder. The diameter of the pool was 120 cm. Visual markers were placed on the walls to facilitate spatial learning. During the training, the platform was constantly in the same position, but the quadrant in which the mouse was placed into water varied. If the mouse did not find the platform within 3 min, it was gently guided on it. The mouse was allowed to stay on the platform for 10 s and then towel-dried and placed on a heating pad (37°C) for several minutes before returning to the home cage. The intertrial interval of six daily trials was prolonged to 10 min to avoid hypothermia. During the training, latencies to escape on the hidden platform were recorded. On the 5th day, a probe trial of 1-min duration was performed in which the platform was removed, and a number of visits to the former platform position, time the mouse spent in each quadrant, and swimming distance were recorded using the EthoVision software. Next day, latencies for reaching a visible platform placed in new positions were tested twice to determine visual capabilities of the mice.
Hot-Plate and Tail-Flick Tests. Thermal analgesia was tested as described earlier in studies of TASK-1 KO mice (Linden et al., 2006). To avoid stress-induced analgesia, the mice were habituated to the hot-plate test apparatus (Hot Plate Analgesic Meter; Harvard Apparatus, Edenbridge, UK) and tail-flick apparatus (Model-DS20; Ugo Basile, Comerio, Italy) for 7 days. During the experiment, the surface of the hot-plate was maintained at 52 ± 0.2°C. Latency to react was scored, and the mouse was removed when the mouse rapidly moved or licked its hind paw or jumped. The cut-off time was 60 s. Before drug or vehicle administration, two basal reaction times, 15 min apart, were determined for each mouse. Then the latencies to react were analyzed 30 min after vehicle (5% Cremophor in saline or saline for WIN55212-2 and morphine studies, respectively), WIN55212-2 mesylate (3 mg/kg s.c.), or morphine hydrochloride (6 mg/kg s.c.) injections. In the tail-flick experiment, the heat intensity was adjusted so that the basal reaction time was 2 to 5 s, and the cut-off time was 9 s. Before drug or vehicle administration, two basal reaction times, 15 min apart, were determined. Then the latencies to the withdrawal reaction were analyzed 30 min after vehicle and WIN55212-2 (3 mg/kg s.c.) or morphine (6 mg/kg s.c.) injections. The analgesic responses were calculated as percentage of maximal possible effect [% analgesia = (test latency – basal latency)/(cut-off time – basal latency) × 100].
Sensitivity to Inhalation Anesthetics. The experiments were performed similarly as in TASK-1 KO studies (Linden et al., 2006). Halothane or isoflurane anesthesia was induced and maintained in a 5-liter chamber that was warmed (33–34°C) by a heat radiator placed above to maintain normal body temperature during anesthesia. Rectal temperatures after the experiments ranged from 35.4 to 38.5°C in the halothane test (mean ± S.E.M. for wild-type, 36.5 ± 0.2 versus KO, 36.9 ± 0.2°C) and from 33.2 to 39.4°C in the isoflurane test (wild-type, 36.6 ± 0.3 versus KO, 37.1 ± 0.4°C). Halothane was added to the air stream (4 l/min) by a vaporizer (Fluotec 3; Ohmeda, BOC Health Care, Westyorkshire, UK). For isoflurane (Isotec 5 vaporizer; Ohmeda, BOC Health Care), an air stream of 7 l/min was used. Carbon dioxide and anesthetic concentrations were monitored by Capnomax-Ultima (Datex, Instrumentarium, Helsinki, Finland). Carbon dioxide concentration was maintained at <0.3% by placing soda lime granules onto the floor of the chamber. Anesthetic concentrations were gradually increased with an equilibrium time (20 min for halothane, 12 min for isoflurane) for each concentration. Halothane concentrations used were 0.5, 0.7, 0.9, 1.1, 1.3, and 1.5%, and isoflurane concentrations were 0.6, 0.8, 1.0, 1.2, 1.4, 1.6, and 1.8%. Tail-clamp withdrawal reflex was tested at the end of equilibrium times. Animals were considered to have lost the tail-clamp withdrawal reflex (LOTW) when they did not clearly respond to the clamp of the proximal part of the tail. Thus, our endpoint was immobility in response to nociception. For dose-response curves, the percentage of animals with LOTW (Y) was fitted to logarithms of anesthetic concentrations (X) using variable slope sigmoidal equation of the form: Y = bottom + (top – bottom)/[1 + 10(LogEC50 – X) × Hill slope] with the bottom and top set to 0 and 100, respectively (GraphPad Prism 3.03; GraphPad Software Inc., San Diego, CA). The EC50 values and their 95% confidence intervals were obtained from the nonlinear regression fit.
Sensitivity to Dexmedetomidine, Propofol, and Lidocaine. The sedative effect of dexmedetomidine was determined by recording locomotor activity in a novel environment. The mouse was placed individually in an open arena 30 min after vehicle or dexmedetomidine hydrochloride (0.03 mg/kg s.c.) administration, its behavior was recorded for 5 min, then the mouse was returned to its home cage, and the rectal temperature was measured 30 min later.
The sensitivities to propofol and lidocaine were determined by measuring the latency to and duration of loss of righting reflex (LORR). Every 4 min, after an injection of propofol (150 mg/kg i.p.) or lidocaine hydrochloride (125 mg/kg s.c.), righting reflex was tested by placing the mouse on its back on a plastic V-shaped trough. When the animal could not right itself with in 5 s in three successive trials, the righting reflex was considered lost. Thereafter, the test was repeated every 4 min until regaining the righting reflex. The body temperature was maintained using a heat radiator placed above the mice or using a heating pad. The rectal temperatures after the propofol-induced LORR ranged from 37.2 to 38.9°C, and there was no difference between the wild-type and KO mice (37.8 ± 0.2 versus 38.0 ± 0.1°C, respectively). The rectal temperatures after the lidocaine-induced LORR ranged from 35.2 to 38.4°C, and there was no difference between the wild-type (males, 37.0 ± 0.4; females, 36.8 ± 0.2°C) and KO (males, 36.6 ± 0.2; females, 37.3 ± 0.2°C) mice.
Drugs. WIN55212-2 mesylate (Tocris, Bristol, UK) was dissolved first in 100% Cremophor EL (polyoxyethyleneglycerol triricinoleate 35; Sigma, St. Louis, MO) and then diluted with saline to the final concentration of 0.3 to 0.6 mg/ml (Cremophor, 5–10%). Morphine hydrochloride (20 mg/ml injection solution; Leiras, Turku, Finland) was diluted with saline to the final concentration of 0.6 mg/ml. Halothane was obtained from Rhodia (Bristol, UK), and isoflurane was from Abbott Laboratories (Queensborough, Kent, UK). Dexmedetomidine hydrochloride (0.1 mg/ml injection solution; Orion Pharma, Espoo, Finland) was diluted with water to the final concentration of 0.006 mg/ml. Lidocaine hydrochloride (20 mg/ml injection solution; Orion) was diluted with saline to the final concentration of 12.5 mg/ml. Propofol (20 mg/ml injection solution; Fresenius Kabi, Uppsala, Sweden) was injected undiluted at the volume of 7.5 ml/kg. Other drugs were injected at the volume of 10 ml/kg.
Statistical Testing. Three- and two-way analyses of variance (ANOVA) were used to test main sex, genotype, and drug effects (SPSS 10.0.7 for Windows; SPSS Inc., Chicago, IL). If no significant sex effect was found, the data from males and females were combined and tested further using two- and one-way ANOVA followed by Newman-Keuls or Dunnett's post hoc tests or using Student's t test or Mann-Whitney analyses (GraphPad Prism 3.03). Data from motor training and antinociception tests were tested using repeated measures ANOVA. A limit for significance was P < 0.05. All data are given as means ± S.E.M.
Results
TASK-3 KO Mice Display a Largely Normal Behavioral Phenotype. The SHIRPA primary screening protocol revealed no obvious physiological abnormalities or defects in the adult TASK-3 KO mice. When the body weight was determined at the age of 19 to 20 weeks (n = 13–20), two-way ANOVA revealed significant effects of sex and genotype (Table 1), the KO males being on the average 9.6% and KO females 7.1% heavier than the wild-type littermates. The weight differences, however, were not significant when males and females were separately tested (Student's t tests). The KO mice showed slightly increased novelty-induced locomotor activity in the SHIRPA screening because they crossed more squares than the wild-type controls in an open arena during a 30-s observation period (Table 1). However, in the staircase test of exploration, in which the behavior was measured for 3 min, the KOs did not significantly differ from the wild-type mice (Table 1), and no significant differences between the genotypes were found in vehicle-treated animals in a 5-min open arena test (Fig. 6A). No significant differences were observed between the KO and wild-type mice in elevated plus-maze or light-dark choice anxiety tests (Table 1). In addition, the KO mice had similar responses to thermal nociception in both hot-plate and tail-flick tests as the wild-type controls (Table 1).
Locomotor Activity and Circadian Rhythm. We wanted to analyze the circadian rhythm of the locomotor activity of the TASK-3 KO mice because TASK-3 channels have been suggested to control the activity of the hypothalamic orexin neurons (Burdakov et al., 2006). The test began by analyzing the novelty-induced locomotion and its decline during a 4-h habituation period at the beginning of the light phase (starting between 9:00 AM and 10:00 AM). No difference in locomotor activity was observed between the TASK-3 KO mice and wild-type controls in this part of the experiment (Fig. 1A). The analysis of the circadian rhythm started when lights were turned off at 6:00 PM. The TASK-3 KO mice moved significantly more than the wild-type controls during the dark phase (repeated measures one-way ANOVA, F1,466 = 5.45, P= 0.025), the difference being largest 6 h after the lights were turned off (Fig. 1B). Both mouse lines showed a normal circadian rhythm by moving less during the light phase and more during the dark phase (Fig. 1, B and C). Interestingly, the effect of TASK-3 channel inactivation observed during the dark phase was totally absent during the light period.
Motor Coordination and Motor Learning. The genetic inactivation of TASK-3 affects the firing properties of cerebellar granule neurons so that the resting membrane potential recorded from the TASK-3 KO cerebellar granule neurons is on average about 10 mV more depolarized; moreover, these cells have lost the ability to keep firing action potentials at high frequencies during sustained depolarization (Brickley et al., 2007). On the basis of these findings, we analyzed how the TASK-3 KO mice perform in the Rotorod and horizontal beam walking tests. Unexpectedly, the TASK-3 KO mice showed no defects in learning to adjust their stepping patterns to match with accelerated speed of the rotating rod (Fig. 2, A and B).
Both wild-type and TASK-3 KO mice learned to walk along the thicker 1.2-cm beam back to their home cage in a few seconds after the 1st training day (Fig. 2, C and D) without significant differences between the genotypes (repeated measures ANOVA). However, if the time needed to traverse the beam on the 1st day was analyzed separately, the genotype effect emerged with the KO males being faster than wild-type males to reach the home cage (P < 0.05, Student's t test; Fig. 2C). No similar difference was found in the females (Fig. 2D). Although traversing the beam faster, the TASK-3 KO males also fell more often from the beam on the 1st training day (two-way ANOVA, genotype effect F1,34 = 8.53, P = 0.006). The mean number of falls from the 1.2-cm beam was: wild-type males, 0.1 ± 0.1 versus KO males, 0.8 ± 0.3 (P < 0.05, Student's t test); and wild-type females, 0.0 ± 0.0 versus KO females, 0.3 ± 0.2 (P > 0.05) (Fig. 2, E and F).
Walking on the narrower beam (0.8 cm in diameter) was more difficult, and especially the TASK-3 KO females required more training before they learned to walk on it without falling several times (repeated measures one-way ANOVA, genotype × training interaction F1,124 = 2.61; P = 0.022; Fig. 2H). There was also a significant interaction between genotype and training in males (F1,117 = 6.74, P < 0.001; Fig. 2G), the KO males falling more frequently on the 1st day. The time to cross the narrower beam did not significantly differ between the wild-type and KO mice (data not shown).
Impaired Working Memory of TASK-3 KO Mice. The T-maze spontaneous alternation test indicated that the disruption of TASK-3 channel may affect working memory (Fig. 3A). Both male and female TASK-3 KO mice visited more often than the wild-type mice the same arm as on the previous trial, leading to a significantly lower percentage of spontaneous alternation (genotype effect F1,52 = 6.59, P = 0.013; Student's t test, P < 0.01, males and females together; Fig. 3A). Thus, the TASK-3 KO mice choose either arm at the 50% chance level, whereas the wild-type mice appeared to remember better which arm they had previously visited. Figure 3, B and C, show that the exploratory activity of those mice that were accepted to the alternation test did not differ between the genotypes. The parameters measuring locomotor and exploratory activity of all animals before the exclusion of those that moved too little were not significantly different between the wild-type and TASK-3 KO mice, although the TASK-3 KO male mice tended to be slightly more active (Table 1).
Impaired Learning of TASK-3 KO Mice in the Morris Water Maze. Already after 1 day of training, both wild-type and TASK-3 KO mice improved their performance in finding the hidden platform (Fig. 4A). However, on each training day, the TASK-3 KO mice were slightly slower to find the platform producing a significant genotype effect in a repeated measures two-way ANOVA (F2,145 = 5.01, P = 0.032). The difference between genotypes did not reach significance on any single day in subsequent Newman-Keuls post hoc tests. In a probe trial, the number of visits in a position where the platform used to be was significantly lower in the TASK-3 KO females compared with the wild-type females (genotype × sex interaction F1,35 = 6.16, P = 0.018, Student's t test, P < 0.01) (Fig. 4B). No difference was observed between the wild-type and KO males. In addition, no significant genotype or sex effect was observed in percentage of time spent in any of the quadrants (Fig. 4C). In the female TASK-3 KO mice, the swimming velocity was significantly reduced (genotype F1,35 = 7.66, P = 0.009) (Fig. 4D). The next day, after the probe trial, we tested whether the mice could find the visible platform to exclude vision-related factors affecting the results. All other mice, except one TASK-3 KO male and one KO female, reached the visible platform within a 3-min trial with the latencies being 35 ± 10 s for the wild-type males, 27 ± 10 s for the wild-type females, 24 ± 6 s for the TASK-3 KO males, and 30 ± 4 s for the TASK-3 KO females. The two TASK-3 KO mice not escaping on the visible platform were excluded from all the data. The same two mice showed strong thigmotaxis and returned immediately to swim when placed on the platform during 4-day training.
Reduced Sensitivity to Halothane and Isoflurane in TASK-3 KO Mice. Halothane and isoflurane sensitivity was reduced in the TASK-1 KO mice (Linden et al., 2006). These earlier findings led us to test the sensitivity of TASK-3 KO mice to halothane and isoflurane by determining the anesthetic concentrations that was required to suppress the withdrawal reaction to tail-clamp in 50% of mice [EC50 = minimal alveolar concentration (MAC)]. The genotype, but not the sex, significantly affected the halothane concentration at which the mice lost their tail-withdrawal reflex (F1,22 = 8.46, P = 0.009). The TASK-3 KO mice needed a higher halothane concentration for LOTW than the wild-types (Fig. 5A, P < 0.01). In the dose-response curve, the percentage of animals with LOTW was fitted to halothane concentrations (Fig. 5B). A shift to the right in the dose-response curve indicates the reduced sensitivity in the TASK-3 KO mice (MAC = 1.48; 95% confidence interval, 1.43–1.53) compared with the wild-type mice (MAC = 1.25; 95% confidence interval, 1.22–1.28). The mean concentration of isoflurane at which the mice had LOTW was not significantly different between the wild-type and TASK-3 KO mice (Fig. 5C). However, a modest shift to the right in the LOTW dose-response curve was observed, suggesting that the KO mice had a slightly reduced sensitivity to isoflurane (MAC = 1.50; 95% confidence interval, 1.49–1.50) compared with the wild-type mice (MAC = 1.38; 95% confidence interval, 1.35–1.42) (Fig. 5D).
Sensitivity of TASK-3 KO Mice to Dexmedetomidine and Propofol. We also analyzed whether TASK-3 disruption attenuates the effects of other sedative or anesthetic agents. The α2 adrenoceptor agonist dexmedetomidine is a clinically used potent sedative agent that produces strong sedation and hypothermia in mice (Sallinen et al., 1997). The wild-type and TASK-3 KO mice were injected with vehicle or dexmedetomidine (0.03 mg/kg s.c.), after which their locomotor activity and rectal temperature were analyzed 30 and 60 min later, respectively. Three-way ANOVA indicated a significant effect of dexmedetomidine treatment, but no sex or genotype effect, on the percent duration of moving (Fig. 6A). Other parameters measured (total movements in centimeters and number of rearings) were not different between genotypes (data not shown). As expected, the body temperature was affected by dexmedetomidine treatment (F2,66 = 261.2, P < 0.001) but also by the genotype (F2,66 = 7.87, P = 0.007), and there was a significant interaction between the drug treatment and genotype (F2,66 = 8.75, P = 0.004). The hypothermic effect of dexmedetomidine was significantly weaker in the TASK-3 KO mice, although clearly induced also in them (Fig. 6B).
Next, we tested whether the disruption of the TASK-3 channel affects the sleep time induced by the injectable anesthetic agent propofol that acts mainly by enhancing GABAA receptor function (Jurd et al., 2003). The effect of propofol (150 mg/kg i.p.) was unaltered by TASK-3 disruption, the latency to and duration of LORR being similar in the wild-type and TASK-3 KO mice (Fig. 6C). Two wild-type mice (a male and a female) did not loose the righting reflex, perhaps due to misplaced injection, and were excluded from the study.
Sensitivity of TASK-3 KO Mice to Lidocaine. Local anesthetics have been reported to inhibit TASK channels in vitro (Kindler et al., 1999). We, therefore, tested here whether the inactivation of TASK-3 has any effect on in vivo actions of lidocaine (125 mg/kg s.c.). The latency to lidocaine-induced LORR did not differ between the study groups (data not shown). The duration of LORR was unaltered by TASK-3 disruption, but significantly affected by the sex (F1,30 = 10.49, P = 0.003), being shorter in females than in males (Fig. 6D). In post hoc tests, differences between males and females within the genotypes did not reach significance. Two wild-type females and one KO male and one KO female did not loose righting reflex and were excluded from the study. One KO male mouse died during LORR and was excluded from the study.
Reduced Analgesic Effect of the Cannabinoid Receptor CB1 Agonist WIN55212-2 in TASK-3 KO Mice. The analgesic and hypothermic effects of WIN55212-2 were reduced in the TASK-1 KO mice (Linden et al., 2006). It is unclear whether only TASK-1- or also TASK-3-containing channels are inhibited by cannabinoid agonists (Maingret et al., 2001; Berg et al., 2004; Aller et al., 2005; Meuth et al., 2006). WIN55212-2 (3 mg/kg s.c.) produced a significant antinociception (repeated measures two-way ANOVA, F2,69 = 30.83, P < 0.001) in the hot-plate test (Fig. 7). A significant genotype effect was also observed (F2,69 = 5.37, P = 0.027). Because no sex effect emerged, the results of males and females were combined. The drug and genotype effects were observed also in the pooled results (F1,70 = 30.44, P < 0.001; F1,70 = 5.61, P = 0.024, respectively), and the Newman-Keuls post hoc test confirmed that the administration of WIN55212-2 produced weaker antinociception in the TASK-3 KO mice (Fig. 7A). In the other thermal nociception test, the tail-flick test, WIN55212-2 (3 mg/kg s.c.) administration produced significant antinociception (F2,67 = 8.10, P = 0.008), but no sex or genotype effects were observed. The pooled data of males and females are shown in Fig. 7.
The effect of WIN55212-2 (6 mg/kg s.c.) was also tested on locomotor activity and on body temperature to investigate whether its other actions differ in the TASK-3 KOs in addition to the supraspinal antinociception. Three-way ANOVA did not reveal any sex effect letting us test pooled data that showed that the hypomotility and hypothermic actions of WIN55212-2 were not significantly affected by the genetic disruption of TASK-3 (Fig. 8). WIN55212-2 produced significant behavioral effects compared with the vehicle (10% Cremophor in saline)-treated mice in both genotypes.
The analgesic effect of morphine (6 mg/kg s.c.) was compared between the wild-type and TASK-3 KO mice in the hot-plate and tail-flick tests. In the hot-plate test, repeated measures two-way ANOVA revealed a significant treatment effect (F2,65 = 54.30, P < 0.001) and a treatment × sex interaction (F2,65 = 4.77, P = 0.037), probably due to higher percentage of analgesia in females (wild-type, 23 ± 11%; knockout, 11 ± 4%) than in males (wild-type, –14 ± 8%; knockout, 3 ± 3%) after vehicle administration. Only the treatment effect, but no genotype effect, was found after pooling the results of males and females in the morphine sensitivity test (Fig. 7C). In the tail-flick test, repeated measures two-way ANOVA revealed a drug effect (F2,61 = 97.21, P < 0.001) and a significant treatment × genotype interaction (F2,61 = 4.80, P = 0.037). This interaction remained significant also after pooling the results of males and females (F1,62 = 5.06, P = 0.032) and seemed to be due to slightly greater antinociceptive effect after vehicle injections and slightly lower antinociception after morphine injections in the TASK-3 KO mice (Fig. 7D). However, these differences did not reach significance in post hoc tests.
Discussion
Given the regionally wider expression profile of the TASK-3 gene in the brain compared with TASK-1 (Talley et al., 2001; Vega-Saenz De Miera et al., 2001; Aller et al., 2005; Linden et al., 2006), it was unexpected that the TASK-3 KO mice were behaviorally almost as normal as the TASK-1 KO mice we reported on earlier (Linden et al., 2006). Nevertheless, several interesting phenotypic features were observed both in naive and drug-treated TASK-3 KO mice, attesting to the value of the TASK channel-deficient mouse models in understanding the significance of K2P channel function.
The TASK-3 KO mice showed increased locomotor activity during the dark phase, whereas their locomotor activity during the light period was unaltered. In addition, we found that the TASK-3 KO mice tended to be slightly heavier than their wild-type littermates. These findings are especially interesting because TASK channels, particularly TASK-3-containing channels, have been suggested to be effectors in a glucose sensing cascade that convey the inhibition of hypothalamic orexin/hypocretin neurons when extracellular glucose levels are increased (Burdakov et al., 2006). It is thus possible that increasing blood glucose levels during the dark phase, when the mice are normally highly active and feed themselves, fail to inhibit orexin/hypocretin neurons properly, leading to ex-aggerated nocturnal activity as seen here in the TASK-3 KO mice. Based on the physiological role of the orexin/hypocretin neurons, their overactivity could lead to increased feeding, enhanced arousal and activity, as well as disturbed sleep-wake cycle (see e.g., Sakurai, 2007). It is important to note that the TASK-1 KO mice had a normal electrocorticogram pattern, and they showed normal sleep-wake behaviors (Meuth et al., 2006), suggesting that the enhanced nocturnal activity in the TASK-3 KO mice is indeed due to a specific loss of TASK-3-mediated currents.
Further behavioral abnormalities in the TASK-3 KO mice were found in relation to memory functions. The TASK-3 KO mice performed poorly in the spontaneous alternation test, suggesting impaired working memory, and they had difficulties during the training phase in the Morris water-maze test. TASK-3 mRNA is expressed at high levels in the hippocampal CA1 pyramidal neurons, stratum oriens interneurons, and dentate gyrus granule cells (Talley et al., 2001); correspondingly, TASK-like currents have been detected in the CA1 pyramidal cells and inhibitory interneurons (Taverna et al., 2005; Torborg et al., 2006). It is possible that the firing properties of hippocampal neurons, either pyramidal or interneurons, are altered in a way that disturbs short-term memory formation and/or retrieval. Another way how lack of TASK-3 channels could affect memory formation is from disruption of acetylcholine signaling. Increased levels of acetylcholine in the hippocampus improve, whereas the antagonists of muscarine receptors (e.g., scopolamine) disturb, the performance of mice in the spontaneous alternation test (Ragozzino et al., 1998). The effects of muscarine applications on TASK currents in cerebellar granule neurons and thalamocortical relay neurons in vitro have shown that TASK channel inhibition via a Gq-coupled mechanism is a part of acetylcholine signaling (Millar et al., 2000; Meuth et al., 2006). It remains to be tested whether a similar mechanism operates in brain regions involved in short-term memory and whether it would be deficient in the TASK-3 KO mice.
When motor coordination and balance were tested, the TASK-3 KO mice had more difficulties in walking without falling along the narrow 0.8-cm wooden beam, but other motor tests showed no clear genotype effect. Almost normal motor performance was unexpected because a remarkable inability of TASK-3 KO cerebellar granule neurons to sustain high frequency firing of action potentials during long suprathreshold depolarization in vitro (Brickley et al., 2007). Our results indicate that this nonaccommodating firing of the cerebellar granule neurons observed in vivo in anesthetized animals and in vitro cerebellar slices (Chadderton et al., 2004; Brickley et al., 2007) is not required in vivo for adjusting stepping patterns on the basis of mostly proprioceptive cues to match with the accelerated speed of the rotating rod or for learning to improve this task. However, our data from the narrower 0.8-cm beam suggest that the TASK-3 deficiency has a subtle effect on the balance or motor coordination. The minor problems of motor performance in the TASK-3 KO mice might also be due to TASK channel inactivation in e.g., the striatal cholinergic interneurons, which have TASK channels with properties of homomeric TASK-3 channels (Berg and Bayliss, 2007) or in motor neurons, which have TASK-1/TASK-3 heterodimers (Berg et al., 2004). Further motor functional and challenge tests should be carried out with the TASK-3 KO mice; for example, using Parkinson's disease models with a lesioned nigrostriatal dopaminergic pathway. If TASK-3 deficiency leads to any dysregulation of striatal cholinergic activity, it could also be one mechanism producing defects in cognitive performances in the T- and water-maze tests because the release of acetylcholine from striatal interneurons is implicated especially in procedural learning (Chang and Gold, 2003; Pych et al., 2005).
Pharmacological tests revealed that the sensitivity to the inhalation anesthetics halothane and isoflurane was reduced in the TASK-3 KO mice, whereas their sensitivity to the GABAA receptor agonist propofol and the local anesthetic lidocaine were unaltered. The TASK-3 KO mice showed similarly reduced sensitivity to inhalation anesthetics as we observed earlier for the TASK-1 KO mice (Linden et al., 2006). This is consistent with results from recordings of in vitro expression systems and slice preparations from different brain regions (motor neurons, cerebellar granule cells, locus coeruleus, raphe neurons, hippocampus, thalamus) showing increased TASK-like currents after application of halothane or isoflurane (Patel et al., 1999; Sirois et al., 2000; Berg et al., 2004). Our results, together with the reports of reduced sensitivity to inhalation anesthetics in TREK-1 KO mice (Heurteaux et al., 2004), suggest that activation of several K2P channels together (TASK-1, TASK-3, TREK-1) contribute to the anesthetic actions of these agents (for review, see Franks, 2006). Which brain regions or cell types inhibited by inhalation anesthetics do actually mediate different components of anesthesia such as sedation, unconsciousness, antinociception, and immobility remain unclear. Our endpoint was the loss of tail-withdrawal reflex, which can be regarded as corresponding to surgical anesthesia and requires strong spinal cord inhibition (Sonner et al., 2003). Therefore, it is likely that TASK channels on the spinal motor neurons are involved in this anesthetic action (Berg et al., 2004).
In pharmacological tests, the TASK-3 KO mice showed reduced sensitivity to the antinociceptive effects of the cannabinoid agonist WIN55212-2 in the hot-plate test, but not in the tail-flick test, suggesting supraspinal mechanisms. TASK channels are inhibited by the endogenous cannabinoids anandamide and methanandamide and by the synthetic WIN55212-2 ligand independently of cannabinoid receptor activation (Maingret et al., 2001). Although Maingret et al. (2001) implicated this effect as TASK-1-specific based on results from human recombinant TASK-1 and -3 preparations, both rodent TASK-1 and -3 channels have been shown to be inhibited by these cannabinoids at similar concentration ranges (Berg et al., 2004; Aller et al., 2005). Our results of the reduced antinociception of WIN55212-2 in the TASK-3 KO mice as well as in the TASK-1 KO mice (Linden et al., 2006) suggest that WIN55212-2 inhibits both channels at behaviorally meaningful levels in vivo. However, inhibitory actions of WIN55212-2 on locomotion or body temperature of TASK-3 KO mice did not differ from that of the wild-type mice, whereas the TASK-1 KO mice were slightly less sensitive also to the hypomotility and hypothermic effects of WIN55212-2 (Linden et al., 2006). These findings are consistent with a TASK-1-preferring effect (Maingret et al., 2001) and also in line with a recent report of Meuth et al. (2006) showing that the inhibition of TASK-mediated currents by anandamide recorded from wild-type geniculate thalamocortical neurons were absent in the TASK-1 KO mice, suggesting that, at least in these neurons, TASK-3 homodimers are not sensitive to endocannabinoids. Thus, it seems that the behavioral actions of cannabinoids mediated by TASK-channels can be divided into those where TASK-1-containing channels are selectively involved and into supraspinal antinociception where deletion of either subunit causes a reduced response.
In conclusion, the TASK-3 KO mice were less sensitive to anesthetic actions of inhalation anesthetics resembling two other K2P channel KO mice, TASK-1 and TREK-1 mice (Heurteaux et al., 2004; Linden et al., 2006). The basic behavioral phenotype of TASK-3 KO mice was surprisingly normal, suggesting that TASK-1 homodimers, other K2P channels, or other mechanisms can largely compensate for the loss of TASK-3-containing channels in most neurons and also in peripheral tissues. It is important to note that even dramatic changes observed in the electrophysiological properties in the cerebellar motor circuitry in the TASK-3 KO mice (Brickley et al., 2007) did not produce a clear motor behavioral phenotype. Nevertheless, the TASK-3 KO mice showed impaired working memory and exaggerated nocturnal activity, both of which may be due to impaired function of selected neuronal populations in the hippocampus and hypothalamus, respectively. Thus, the TASK-3 KO mouse model provides us with novel tools that may be critical in understanding detailed brain mechanisms associated with several drug effects and physiological functions.
Footnotes
-
This work was supported in part by the German Research Council (Grant DFG WI 1951/1-2 to W.W.), by the Volkswagen Stiftung (Grant I/78 554 to W.W.), by the Academy of Finland (to E.R.K.), and by the Sigrid Juselius Foundation (to E.R.K.).
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.107.129544.
-
ABBREVIATIONS: K2P, two-pore-domain background K+ channel; WIN55212-2 mesylate, (R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanone mesylate; KO, knockout; LOTW, loss of tail-withdrawal reflex; LORR, lossof righting reflex; ANOVA, analysis of variance; MAC, minimal alveolar concentration.
-
↵1 Current affiliation: Institut des Neurosciences Cellulaires et Intégratives, Centre de Neurochimie, Strasbourg, France.
-
↵2 Current affiliation: Instituto de Neurociencas de Alicante, Consejo Superior de Investigaciones Cientificas-Universidad Miguel Hernándes, Campus de San Juan, Sant Joan d'Alacant, Spain.
-
↵3 Current affiliation: Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen, United Kingdom.
- Received August 1, 2007.
- Accepted September 14, 2007.
- The American Society for Pharmacology and Experimental Therapeutics