Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Potassium channels are a diverse family of ion channels with a wide variety of functions in the central nervous system. In different excitable cells, these channels set resting potential and modulate their ability to regulate action potential (Wann, 1993), and some potassium channels have been implicated as targets of anesthetic action (Urban, 1993; Yost, 1999). Inwardly rectifying potassium channels are activated by neurotransmitters and hormones as well as by other messengers such as kinases and G-proteins (Dascal, 1997). G-protein-coupled inwardly rectifying potassium channels (GIRKs) are activated by m₂ muscarinic, ₂ adrenergic, D₂ dopaminergic, histamine, serotonin 5HT_1A, A₁ adenosine, -aminobutyric acid_B, µ-, -, and -opioid, and somatostatin receptors (Mark and Herlitze, 2000). Enhancement of GIRK2 function by ethanol concentrations as low as 10 mM has been shown in Xenopus laevis oocytes expressing homomeric and heteromeric channels (Kobayashi et al., 1999; Lewohl et al., 1999). In contrast, other inwardly rectifying potassium channels (IRK1, ROMK1, ROMK2, and ROMK3) showed no significant effect of ethanol at concentrations up to 200 mM. Enhancement of GIRK2 channel function by ethanol contrasts with effects of alcohols on several other potassium channels, which are either unaffected by ethanol or other alcohols (Anantharam et al., 1992) or are inhibited by high concentrations of ethanol (Covarrubias et al., 1995; Leonoudakis et al., 1998). Thus, GIRKs represent a new class of ethanol-sensitive membrane channels that are potential in vivo targets of ethanol.

A previous study used weaver mice, which have a mutation in GIRK2, and found that these mice lack ethanol analgesia; ethanol-induced loss of righting reflex, hypothermia, bradycardia, and locomotor activity were not different from wild-type mice (Kobayashi et al., 1999). However, the weaver mutant is not ideal to assess the role of GIRK2 because the mutation alters the ion selectivity of GIRK2 and produces neuronal degeneration rather than a simple loss of GIRK2 function (Patil et al., 1995). To explore the role of GIRK channels in mechanisms of ethanol actions in vivo, we used GIRK2- knockout mice, which were generated recently (Signorini et al., 1997). GIRK2 is one of the most prominent GIRKs in brain and is more sensitive to ethanol enhancement than other GIRKs (Kobayashi et al., 1999; Lewohl et al., 1999); thus, mice lacking GIRK2 provide a logical approach to understanding the importance of GIRK signaling in alcohol action. Specifically, if ethanol enhancement of GIRK2 function produces behavioral changes, then mice lacking GIRK2 should show diminution of some actions of ethanol. A key question is which behaviors might be dependent upon GIRK2. Other than increased sensitivity to pentylenetetrazole convulsions, there is little information about the behavioral phenotype of GIRK2 null mutant mice (Signorini et al., 1997), and it is impossible to predict a priori which actions of ethanol are likely to be mediated by GIRKs. Thus, we chose a battery of the most commonly used tests to study alcohol-induced changes in behavior and possible effects of the mutation. In some cases, we also identified effects of the mutation on behaviors observed in the absence of ethanol.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals. The GIRK2 null mutant mice were generated and their genotypes identified by polymerase chain reaction analysis of tail DNA as described by Signorini et al. (1997). The genetic background of both GIRK2 mutant and wild-type control mice is 129Sv X C57BL/6. Heterozygous mating was maintained to generate GIRK2 (/) and GIRK2 (+/+) littermates. To minimize the possible effect of genetic background, only littermates were used in all experiments. Mice were group-housed four to five to a cage based on sex and litter. Food and water were available ad libitum. The vivarium was maintained on a 12:12-h light/dark cycle with light on at 7:00 AM. The temperature and humidity of the room were maintained at 20°C and 50%, respectively. All experiments were performed during the light phase of the light/dark cycle, except where noted. Behavioral testing began when the mice were at least 2 months old. Mice more than 6 months old were not used in these studies. Altogether, 427 (/) and 417 (+/+) mice with approximately equal number of males and females for each genotype were used in this study.

Alcohol Drinking. Testing procedures have been described earlier (Belknap et al., 1993). Mice were allowed to acclimate for 1 week to individual housing. Experiments were carried out in standard 7.5- × 12.5-inch cages in sliding racks. Bottles were placed vertically 3.5 inches from the back wall through two holes in the cage top. The distance between two bottles was about 2.0 inches. A feeder was placed on the front wall. Two drinking tubes were continuously available to each mouse, and tubes were weighed daily. Food was available ad libitum, and mice were weighed every 4 days. After 4 days of water consumption (both tubes), mice were offered 3% ethanol (v/v) versus water for 8 days. Tube positions were changed every 2 days to control for position preferences. Quantity of ethanol consumed (gram per kilogram of body weight/24 h) was calculated for each mouse, and these values were averaged for every concentration of ethanol (Phillips et al., 1994). Immediately following 3% ethanol, a choice between 6% ethanol and water was offered for 8 days and finally 10% ethanol versus water for 8 days. Because the mice demonstrated a strong side preference in the first experiment (see Results), a second experiment was performed in which the favored position (highest level of consumption as a criteria determined during first 8 days of presentation of water versus water with alternation of bottles every 2 days) was used for presentation of 4 and 8% alcohol solutions. However, most wild-type mice would not consume the 8% ethanol solution even when presented in the "favored" position, and some mice developed aversion to consuming liquids from that position. Thus, when the solutions were changed from 8% ethanol to 8% ethanol + saccharin, the favored position was redetermined for all mice. This was accomplished by testing consumption of water (both bottles, position with higher level of consumption as the criteria) before presentation of ethanol + saccharin and again before presentation of saccharin and quinine solutions. In another experiment, side preference was determined, water was placed in the favored position, and a 4% ethanol solution was placed in the nonfavored position. Consumption of both solutions was measured daily.

Preference for Nonalcohol Tastants. Wild-type or knockout mice were also tested for saccharin and quinine consumption. Mice were serially offered saccharin (0.033 and 0.066%) and quinine hemisulfate (0.015, 0.03, and 0.1 mM), and intakes were calculated. Each concentration was offered for 6 days, with bottle position changed every 2 days. Within each tastant, the low concentration was always presented first, followed by the higher concentrations. Between tastants mice had two bottles with water for 2 weeks.

Chronic Alcohol Diet. Mice were individually housed and given an ethanol-containing liquid diet (BioServ, Frenchtown, NJ) (0% ethanol for 2 days, 2.3% ethanol for 2 days, 4.5% ethanol for 2 days, and 6.0% ethanol for 5 days) (Snell et al., 1996). Pair-fed control mice were given a volume of a control diet (with maltodextrin equicalorically replacing the ethanol) equal to the average volume that the ethanol-fed mice had consumed on the previous day. Because the mutant mice consumed more of the ethanol diet than the wild-type mice, a "matched" variant of the diet was used in a subsequent experiment. Mutant mice were given the average volume of the ethanol-liquid diet that the ethanol-fed wild-type mice had consumed 1 day before. Handling-induced convulsions (HIC) were scored from 0 to 7 as previously described (Crabbe et al., 1991). Briefly, beginning at 7:00 AM on the 10^th day (9-day ethanol diet), ethanol-containing diet was replaced with control diet. Animals were scored for HIC each 2 h for the first 4 h, then hourly for another 8 h after withdrawal and then at 23 and 24 h after withdrawal. To avoid possible induction of convulsions and loss of animals, we weighed mice only at the first and last day of ethanol exposure. The volume of diet consumed was recorded daily, but calculated per kilogram of body weight only for the last 3 days of exposure.

Alcohol Injections. All alcohol solutions were made in saline and injected i.p. with a volume of 0.2 ml/10 g of body weight. Control mice received the same volume of saline.

Acute Functional Tolerance. Animals were trained to remain on a rotarod for 1 min (Economex, Columbus Instruments, Columbus, OH; speed of rod: 5 rpm), given a dose of ethanol (1.75 g/kg i.p.), and tested for recovery of balance every 5 min. When they could remain on the rotarod for 1 min, a blood sample (BEC1) was taken from the retro-orbital sinus and the animals were immediately injected with a dose of 2 g/kg i.p. ethanol. When the animals could remain on the rotarod again for 1 min, a final blood sample (BEC2) was taken. BEC values, expressed as milligrams of ethanol per milliliter of blood, were determined spectrophotometrically by an enzyme assay (Lundquist, 1959). The difference between BEC1 and BEC2 was taken as an index of the acquisition of acute tolerance (Erwin and Deitrich, 1996).

Acute Withdrawal Severity. Mice were scored for HIC severity 20 min before and immediately before i.p. ethanol administration. The two predrug baseline scores were averaged. A dose of 4.0 g/kg of ethanol in saline was injected i.p., and the HIC score was tested every hour until the HIC level reached the baseline. Acute withdrawal was quantified as area under the curve but above predrug baseline level (Crabbe et al., 1991).

Ethanol Sleep Time. Animals were injected with ethanol (3.8 g/kg i.p.), and the length of ethanol-induced loss of righting response (sleep time) was measured. Upon loss of the righting reflex, mice were placed supine in a sleep trough (~90° angle) and the time to regain the righting reflex measured. Loss of righting reflex was defined as inability of a mouse to right itself within 30 s. Return of the righting response was defined as the ability of a mouse to right itself twice in 1 min. Sleep time was defined as the time between loss of righting reflex and return of the righting response.

Open Field Motor Activity. The assessment of locomotor activity was conducted in a circular arena measuring 1 M in diameter with a surrounding wall 50 cm in height; the arena was well illuminated by overhead lights from three bulbs (100 W each). The inside metal surfaces of the arena were painted white. A video tracking system, which included an overhead camera, image analyser (Videomax I), and IBM PC computer running software was used to monitor activity. The arena was divided into two concentric zones, and behavior was estimated for each of these zones separately. Vertical activity has been taken from visual observation. Each mouse was tested in the open field in six 2.5-min consecutive sessions. At the start of the first session, the mouse was placed in the same location near the wall and allowed to move freely. The image analysis system recorded the path traversed by the mouse during each session. Path length, time spent in defined zones of the arena, and the numbers of visits into the each zone were the primary dependent measures for behavioral analysis. After each mouse completed a session, the arena was cleaned with a 90% ethanol solution.

Spontaneous Motor Activity Testing. Locomotor activity was measured in the standard mouse cages in Opto-microvarimex (Columbus Instruments). Each cage had bedding and food and was covered by a heavy plastic lid with holes for ventilation and a bottle of water. Each mouse was prehabituated to the experimental cage for 3 h. Next, the mice were removed from the experimental cages, weighed, and injected with ethanol or saline (i.p.). After drug administration, mice were placed immediately in individually prehabituated cages, and the activity was monitored every 5 min for 30 min.

Elevated-Plus Maze. The elevated-plus maze was a modification of that validated by Lister (1987) and comprised two open (30 × 5 × 0.25 cm) and two enclosed (30 × 5 × 5 cm) arms that extended from a common central platform (5 × 5 cm). The apparatus was constructed from black Plexiglas and was elevated to a height of 60 cm above floor level. All testing was conducted under room light. In accordance with established procedure (Rodgers and Johnson, 1995), mice were individually placed on the central platform of the maze facing an open arm. A normal 5-min test duration was used, with maze thoroughly cleaned between subjects. All test sessions were recorded by a vertically mounted camera linked to a monitor and VCR in an adjacent laboratory. Parameters scored from videotape were the conventional spatiotemporal measures and a range of specific behaviors related to the defensive repertoire of the mouse (Rodgers and Johnson, 1995).

Statistical Analysis. Data are reported as the mean ± S.E.M value. The statistics software program GraphPad (Jandel Scientific, Costa Madre, CA) was used throughout. To evaluate differences between groups, analysis of variance (one-or two-way ANOVAs) and post hoc Dunnett's test were carried out.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Ethanol Preference. We first asked whether loss of GIRK channels changed alcohol consumption. In a two-bottle, free-choice paradigm in which mice could drink either water or an ascending series of ethanol concentrations (0, 3, 6, and 10%), there were no differences in ethanol intake or preference between knockout (/) and wild-type (+/+) mice (Fig. 1). Males consumed less ethanol per kilogram of body weight than did females. Both male and female mutant mice showed an increase in total intake of liquid (ethanol solution and water; Fig. 1, E and F). Two weeks after the ethanol-drinking study, the same mice were tested for saccharin (sweet) and quinine (bitter) intake and preference in an order-balanced experimental design. There were no significant differences in preference (for saccharin) or avoidance (for quinine) between mutant and wild-type females, but male mutant mice showed higher avoidance of quinine than wild-type animals (P < 0.05) (data not shown).

View larger version (33K): [in this window] [in a new window]   Fig. 1.   Enhancement of total intake of liquid but no difference in ethanol preference or consumption in GIRK2 null mutant mice. A and B, wild-type females show greater ethanol consumption than wild-type males (P = 0.0015), and mutant females show greater consumption than mutant males (P = 0.007). E and F, mutant females (P = 0.025) and mutant males (P < 0.0001) have a higher total fluid intake than wild-type. For this experiment, bottle position was alternated and the data are an average of consumption from all days and positions (see Materials and Methods for details). A, C, and E, data for females, n = 15 (+/+) and n = 22 (/); B, D, and F, data for males, n = 15 (+/+) and n = 15 (/).

View larger version (44K): [in this window] [in a new window]   Fig. 2.   When presented in the favored position, GIRK mutant mice consumed more ethanol than wild-type, but consumption of saccharin or quinine was not different. The bottle containing the ethanol solution was placed in the favored position and the water bottle in the nonfavored position for each mouse (see Fig. 2 and Materials and Methods for details). A and B, ethanol consumption and preference, respectively, were higher for mutant mice versus wild-type: P = 0.013 and P = 0.007 for 4% ethanol; P < 0.0001 and P < 0.0001 for 8% ethanol. C, no differences in total intake of liquid during all periods of ethanol presentation. D, no differences in saccharin consumption. E, mutant mice tended to drink less of the highest concentration of quinine, but an overall ANOVA indicated no significant difference, P = 0.064. All mice were females, n = 20 (+/+) and n = 25 (/) at the start of experiment; #, death of mutant mice.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   When presented not in the favored position, wild-type and GIRK mutant mice did not consume ethanol. The bottle containing the ethanol solution was placed in the nonfavored position and the water bottle in the favored position for each mouse. A and B, ethanol consumption and preference, respectively, were not different for mutant and wild-type mice during 2 weeks of presentation of 4% ethanol. After change of ethanol position on the 15th day, GIRK2 null mutant females consumed more ethanol (P = 0.01) and demonstrated higher preference (P = 0.007) than wild-type mice. D, slightly increased total intake in wild-type mice for the first 14 days (P = 0.001). After change of ethanol position on the 15th day, total intake in wild-type and mutant mice was identical. All mice were females, n = 10 (+/+) and n = 10 (/).

Ethanol Acute Withdrawal. Because GIRK2 null mutant mice are more sensitive than wild-type to pentylenetetrazole convulsions (Signorini et al., 1997), we studied HIC before and after acute and chronic alcohol treatments.

View larger version (31K): [in this window] [in a new window]   Fig. 4.   Withdrawal severity (HIC) after acute administration of ethanol is reduced in GIRK2 mutant mice, and basal HIC level is increased. HIC were measured for 15 h after injection of 4.0 g/kg ethanol i.p. Acute withdrawal was quantified as area under the curve (AUC) but above predrug HIC level corresponding to each genotype. C, the area under the curve values calculated from A and B. The AUC was lower for mutant than for wild-type (P < 0.0001) and lower for females than for males (P = 0.015) (C). D, the basal HIC score (before ethanol injection) was higher for mutants than for wild-type (P < 0.0001). Mutant mice are different from wild-type mice of correspondent sex (Student's t test). For females, n = 20 (+/+) and n = 20 (/); males, n = 20 (+/+) and n = 20 (/). **P < 0.01, ***P < 0.001.

Chronic Ethanol Diet. In our first study, knockout mice fed an ethanol-containing diet demonstrated a more severe withdrawal syndrome (HIC) than wild-type animals. Ethanol exposure significantly increased in the HIC score (Fig. 5, A and B) as well as area under the HIC withdrawal curve (Fig. 5, C and D) with a clear effect of genotype and a tendency for dependence on gender (P = 0.06). Pair-fed mice showed neither genotype- nor gender-dependent differences in area under the HIC withdrawal curve (Fig. 5, C and D). Six female knockout mice (but none from the other groups) died from seizures during last 5 days of ethanol consumption. Knockout mice consumed more of the ethanol-containing liquid diet (Fig. 6, C and D), and females consumed more ethanol per kilogram of body weight than did wild-type mice (Fig. 6A). There was a trend toward increased consumption in males for the mutant mice (Fig. 6B). The level of ethanol consumption was gender-dependent for knockout mice. Males consumed less ethanol per kilogram of body weight than did females (P < 0.01).

View larger version (35K): [in this window] [in a new window]   Fig. 5.   Withdrawal severity is greater in GIRK2 mutant mice if diet consumption is not matched. HIC were measured after chronic consumption of an ethanol-containing liquid diet. A and B, mutant male or female mice showed higher withdrawal scores than wild-type (P < 0.0001). C and D, the area under the curve (AUC) calculated from A is greater for null mutant female (P = 0.003) and male (P < 0.0001) mice. *P < 0.05, **P < 0.01, mutant mice are different from wild-type mice of correspondent sex (Student's t test). A and C, for females, n = 10 (+/+ and /) for the ethanol diet; n = 11 (+/+) and n = 12 (/) for the pair-fed group. B and D, for males, n = 10 (+/+) and n = 12(/) for the ethanol diet, n = 12 (+/+ and /) for the pair-fed group.

View larger version (34K): [in this window] [in a new window]   Fig. 6.   GIRK2 mutant mice consume more of an ethanol-containing liquid diet than wild-type. Ethanol and diet consumption from the mice shown in Fig. 5 are presented here. A and B, ethanol intake during the last 3 days of liquid diet administration was greater for mutant females versus wild-type (P = 0.04) and tended to be higher for males (P = 0.067). C and D, for the entire period of ethanol administration, the amount of liquid diet consumed was greater for mutant females versus wild-type (P = 0.006) and also for males (P = 0.0005).

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Withdrawal severity for mutant and wild-type mice with matched consumption of the ethanol-containing liquid diet. A, the observed HIC score after withdrawal in mutant and wild-type mice consuming the same amount of the liquid diet (see Materials and Methods for details). B, the area under the curve calculated from A. Females, n = 12 (+/+) and n = 20 (/), were used.

Open Field Locomotion. We used several tests that are sensitive to motor activity and anxiety-related behaviors to determine whether the mutation affected basal behavior or ethanol effects on these domains. Evaluation in the open field for 15 min showed that mutant and control mice traveled a similar distance in the peripheral ring (Fig. 8A), but distance traveled in the central area and number of rearings was about twice as high in knockout animals compared with wild-type (Fig. 8, B and C). Ethanol had a strong stimulatory effect on peripheral activity only in knockout mice. Ethanol enhanced peripheral activity in knockout mice in a wide range of doses (from 0.5-2.25 g/kg) without any activating effect in wild-type animals. Only the highest dose of ethanol (3 g/kg) produced a decrease of activity in wild-type mice. Two-way ANOVA showed that the effect of ethanol on the distance in peripheral ring was dose- and genotype-dependent. Similar genotype- and dose-dependent differences were demonstrated in the effect of ethanol on the distance traveled in the central area as well as number of rearings (Fig. 8, B and C). We found an inhibitory effect of ethanol on distance in the central area and on vertical activity for wild-type and knockout mice, but knockout mice required a higher dose of ethanol for a significant decrease in motor activity (Fig. 8, B and C).

View larger version (27K): [in this window] [in a new window]   Fig. 8.   Ethanol-induced motor activation is revealed in GIRK2 mutant mice in the open field test. A, B, and C, activation of mutant but not wild-type mice by ethanol (significant effects of dose: P < 0.0001; and genotype: P < 0.0001). Significant effects of individual doses of ethanol are indicated as follows: *P < 0.05, **P < 0.01, mutant mice are different from control (mutant mice with injection of saline); &P < 0.05, &&P < 0.01, wild-type mice are different from control (wild-type mice with injection of saline) (post hoc Dunnett's test). Each point represents an average from independent group of animals. Data from females and males were combined as there were no sex differences; n = 15 to 25 for each group.

Spontaneous Motor Activity. To determine the role of the novel environment in the differences in ethanol-stimulated motor activity seen in the open field test, we studied the effect of ethanol on motor activity in the home cage after habituation. Knockout mice demonstrated higher baseline motor activity, but no activation by ethanol. The highest dose of ethanol (2.0 g/kg) decreased motor activity to a level lower than baseline for knockout, but not wild-type, mice (Fig. 9A). Wild-type mice were weakly, but significantly, activated by ethanol at a dose of 1.25 g/kg. This activating effect of ethanol was observed after 20 (Fig. 9B) and 30 min (Fig. 9C) of monitoring.

View larger version (24K): [in this window] [in a new window]   Fig. 9.   GIRK2 mutant mice do not show ethanol-induced motor activation after habituation to the test cage. A, there was a significant effect of genotype (P = 0.0002) and dose (P = 0.04). One dose of ethanol inhibited activity in the mutant mice (*P < 0.05) and one dose enhanced activity in the wild-type mice (&&P < 0.01) (post hoc Dunnett's tests). B and C, there was a significant dependence only on genotype (P < 0.0001 and P = 0.0001, respectively). One dose enhanced activity in the wild-type mice (&&P < 0.01) (post hoc Dunnett's tests). Each point represents an average from independent group of animals. Data are the average from 20 to 35 mice for each point; there were no sex differences, and data for females and males were combined.

Elevated-Plus Maze. Results from open field and home cage activity measurements suggested that the mutation increases activity and decreases anxiety. To explore this further, we used a standard test of anxiety-related behavior, the elevated-plus maze. GIRK2 knockout mice (injected with saline) demonstrated higher locomotor activity (Fig. 10D), spent more time in the open arms (Fig. 10A), and showed a higher percentage of entries into the open arms (Fig. 10C). Ethanol increased the total number of entries and the entries into the open arms for both mutant and wild-type mice. Ethanol increased time spent in the open arms only for wild-type mice. Ethanol markedly increased the percentage of entries into the open arms in wild-type mice but produced only a small increase in mutant mice. Thus, mutant mice display much less anxiety than wild-type and do not show the anxiolytic action of ethanol.

View larger version (27K): [in this window] [in a new window]   Fig. 10.   Ethanol reduced anxiety in wild-type but not mutant mice in the elevated-plus maze. A, mutant mice spent more time in open arms (P < 0.0001, two-way ANOVA), but there was no effect of ethanol treatment. B, C, and D, mutant mice entered the open arms more often, preferred the open arms, and had a greater number of total entries (P < 0.0001, two-way ANOVA). Ethanol enhanced these measures (B, C, and D) (P < 0.0001, except panel C where p = 0.025, two-way ANOVA). **P < 0.01, ***P < 0.001, significant differences from wild-type control (saline injection) (Student's t test); &&&P < 0.001, significant differences from mutant control (saline injection) (Student's t test). Data from females and males were combined, as there were no sex differences. Control, n = 10 for (+/+) and for (/); ethanol, n = 11 for (+/+) and n = 9 for (/).

Other Effects of Ethanol. Duration of loss of righting response (sleep time) and development of acute functional tolerance on the rotarod test were also studied. Neither genotype nor gender differences were found in the anesthetic effects of ethanol or blood ethanol concentration at time of regain of righting reflex (Table 1). Acute functional tolerance to ethanol as measured by the rotarod test was similar in knockout and wild-type mice as well as blood ethanol concentration at time of recovery after injection of the first dose.

View this table: [in this window] [in a new window]   TABLE 1 Ethanol sleep time and acute functional tolerance (rotarod test) do not differ between control and GIRK2 null mutant (/) mice Values are means ± S.E.M.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

These results show that the GIRK2 channel deletion changes several behavioral effects of ethanol. These include increased consumption of ethanol, decreased ethanol dependence (withdrawal severity), stimulation of motor activity by ethanol, and reduction of the antianxiety action of ethanol. These are discussed below.

It is clear that this null mutation increases both forced consumption of an ethanol-containing liquid diet as well as free-choice consumption of ethanol solutions; however, the results from the two-bottle choice paradigm are dependent upon the experimental design. The most important condition for revealing increased ethanol consumption by the mutant mice was placement of the ethanol solution in the favored position for drinking. In contrast, placement of water in the favored position resulted in complete avoidance of ethanol. Using a standard protocol with alternation of bottle positions (Belknap et al., 1993), we did not see differences in drinking behavior because of the very strong side preference demonstrated by most of the animals.

For wild-type mice, alcohol consumption was low, and the average level did not exceed 2 to 3 g/kg/day regardless of whether the bottle position was changed or whether ethanol was available only in the favored position. Even with the addition of saccharin, their consumption did not exceed 3 g/kg/day. It is known that alcohol consumption depends on genotype (McClearn, 1968). The mice used in our experiments had a mixed genetic background from 129/SvJ and C57BL/6J strains, which differ in voluntary consumption of ethanol (Belknap et al., 1993), and the consumption of control mice reflects this mixed background. However, the mutant mice consumed 3 to 5 g/kg/day when ethanol was available in the favored position and about 7 g/kg/day when saccharin was added to the ethanol. This level of consumption may have been sufficient to produce physical dependence, as about 30% of the mutant mice (and none of the wild-type mice) died during free-choice ethanol consumption. Males of both genotypes consumed less ethanol than females, consistent with published data (Juarez and Barrios de Tomasi, 1999). Mutant mice did not consume more sweet (saccharin) or bitter (quinine) tastants than did wild-type mice regardless of whether the bottle positions were alternated or tastants placed in the favored positions. Furthermore, knockout mice showed slightly higher avoidance for the bitter solutions of quinine than wild-type mice but nevertheless consumed more ethanol. Thus, the increased consumption of ethanol by GIRK2 knockout mice did not generalize to other tastants.

After chronic consumption of an ethanol diet, GIRK2 knockout mice demonstrated more severe withdrawal signs than wild-type mice. However, these differences were the result of the higher amount of diet consumed by mutant mice. When the amount of diet consumed by mutant mice was matched with the amount of diet consumed by wild-type animals, there were no differences in severity of withdrawal between two genotypes. After acute administration of ethanol, GIRK2 knockout mice did not demonstrate any signs of withdrawal in comparison with wild type. This suggests that the GIRK2 channel is an important mediator of HIC. It is possible that ethanol withdrawal seizures in mutant mice are masked by the higher basal HIC, but this basal HIC is modest and it seems unlikely that it would prevent a higher score during ethanol withdrawal. It is interesting that the GIRK2 null mutants are relatively resistant to withdrawal HIC despite their higher sensitivity to pentylenetetrazole convulsions (Signorini et al., 1997). In addition, it is puzzling that the mutant mice died, apparently of seizures, during chronic alcohol consumption, yet did not display high levels of HIC.

GIRK2 null mutant mice have less anxiety-related behavior, as indicated by the open field test and the elevated-plus maze. In the open field test, mutant mice demonstrated higher activity in the center zone than did control mice, and more center entries indicate less anxiety (Grossen and Kelley, 1972; Treit and Fundytus, 1989). This is supported by pharmacological studies showing that anxiolytic and anxiogenic drugs increase or reduce, respectively, the number of center entries (Treit and Fundytus, 1989; Kvist and Selander, 1992; Simon et al., 1994). However, this interpretation is dependent on whether general activity was also affected. This possibility cannot be completely ruled out, but there was no difference in activity in the peripheral area between knockout and control mice. Data from the elevated-plus maze also indicate a reduction in anxiety in the mutant mice because of the greater percentage of entries into open arms and longer time spent in the open arms.

In both the open field and elevated-plus maze, ethanol appeared to have no antianxiety action in the mutant mice. In the open field, ethanol did not change central or vertical activity, but it increased the distance traveled in the peripheral area by mutant mice. Therefore, ethanol produced a stimulatory (activating) effect but did not decrease the anxiety in knockout mice. In wild-type mice, ethanol had no stimulatory effect but gradually and dose dependently decreased the central activity and number of rearings, indicative of sedation. Consistent with these results, in wild-type animals ethanol exerted a clear anxiolytic effect in the elevated-plus maze as indicated by an increased percentage of open-arm entries and of time spent in the open arms. Ethanol also produced an activating effect (increased total number of entries) in wild-type mice, but in mutant mice ethanol did not change the time in open arms or percentage of open-arm entries. Thus, the GIRK2 null mutation prevents the anxiolytic effect of ethanol, but this may be due to a marked reduction in anxiety in the mutant mice in the absence of ethanol.

It is of interest to consider possible signaling systems that might account for the increased alcohol consumption and motor activity seen in the mutant mice. GIRKs may be activated by wide variety of receptors, including m₂ muscarinic, a₂ adrenergic, D₂ dopaminergic, histamine, 5HT_1A, A₁ adenosine, -aminobutyric acid_B, µ-, -, and -opioid, and somatostatin receptors (Mark and Herlitze, 2000). Theoretically, signaling by any of these receptors might be impaired in GIRK2 knockout mice, and dopaminergic, serotonergic, and opioid systems have received considerable attention for their role in alcohol actions, particularly consumption. However, most of these studies suggest that decreased signaling by these receptors (as would be expected in mice lacking GIRK2) would decrease ethanol consumption and motor activity. For example, µ-opioid receptor knockout mice did not show any evidence of ethanol self-administration (Roberts et al., 2000), and the 5-HT_1A receptor antagonist WAY 100635 reduced 24-h ethanol drinking in alcohol-preferring (P) rats (Zhou et al., 1998). Phillips et al. (1998) reported a marked aversion to ethanol in mice lacking dopamine D₂ receptors and D₂ dopamine receptor antagonists reduced ethanol-induced hyperactivity (Liljequist et al., 1981; Shen et al., 1995). However, changes in the function of D₁ and D₃ receptors in GIRK2 mutant mice may explain some of the phenotypes. Our pharmacological studies with dopamine receptor drugs indicate that GIRK2 knockout mice are more sensitive to D₁ receptor agonist but that D₃ receptor function is impaired (Y. A. Blednov, in preparation). Both activation of D₁ receptors and block of D₃ receptors produce motor activation (Geter-Douglass et al., 1997; Manzanedo et al., 1999), and Steiner et al. (1997) showed hyperactivity in D₃ receptor knockout mice. Furthermore, voluntary ethanol consumption and preference were markedly reduced in D₁ null mutant mice, and dopamine D₁ receptor blockade reduced alcohol intake in D₁ +/+ and D₁ +/ mice to the level seen in untreated D₁ / mice (El-Ghundi et al., 1998). Blockade of D₃ receptors enhanced ethanol reward (Boyce and Risinger, 2000); thus, at least some of the behavioral phenotypes observed in the present study may be accounted for by increased D₁ and decreased D₃ receptor function in the GIRK2 mutant mice.

Another possible explanation of increased ethanol consumption in GIRK2 knockout mice is a functional inhibition of kappa opioid and/or nociceptin receptors. Kappa opiate (Ulens et al., 1999) and nociceptin/orphanin FQ (Ikeda et al., 1997) receptors are coupled with GIRK, and activation of these receptors decreases ethanol consumption and preference (Nestby et al., 1999; Ciccocioppo et al., 2000). It has been postulated that µ- and -opioid receptors mediate the rewarding effects of ethanol, whereas kappa receptors are thought to mediate the aversive effects of opioids (see Van Ree et al., 1999 for review). Interestingly, increased consumption of ethanol by GIRK2 knockout mice appears to be the result of a decrease in the aversive properties of ethanol.

Of course, an unanswered question is whether compensatory changes in expression of other genes occur as a result of deletion of GIRK2. This issue has yet to be explored with these mice except to show that GIRK1 is also decreased (Signorini et al., 1997).

Taken together, our data suggest that the GIRK2 channel is important for anxiety, motor activity, and seizure sensitivity and may mediate effects of ethanol on these behaviors.