|Home | About | Journals | Submit | Contact Us | Français|
Anxiety-like behaviors are integral features of withdrawal from chronic ethanol exposure. In the experiments in the current study, we tested the hypothesis that anxiety can be regulated independently of other withdrawal signs and thus may be responsive to selective pharmacological agents. For 17 days, rats were fed ethanol (8–12 g/kg/day) in a liquid diet. Between 5 and 6 h after cessation of ethanol treatment, rats were tested in either the social interaction or plus-maze test of anxiety-like behavior after treatment with drugs hypothesized to have anxiolytic action. SB242084, flumazenil, and CRA1000—antagonists for 5-hydroxytryptamine (serotonin) (5-HT) 2C (5-HT2C), benzodiazepine, and corticotropin-releasing factor type 1 (CRF1) receptors, respectively—attenuated decreased social interaction without concomitant effects on activity measures. In contrast, ifenprodil, MDL 72222, and zolpidem—antagonists for N-methyl-d-aspartate (NMDA) and 5-HT3 receptors, and agonist for benzodiazepine type 1 receptors, respectively—did not share this effect. Results for SB242084, flumazenil, and ifenprodil in the elevated plus-maze test were comparable to those in the social interaction test. These results support the suggestion that multiple neuronal systems (CRF1, 5-HT2C, and benzodiazepine receptors) contribute to the ethanol withdrawal sign of decreased social interaction. Furthermore, the selective effects of pharmacological agents on social interaction seem to indicate that this behavior can be dissociated from other signs. Because anxiety may be a complicating factor in alcohol withdrawal and relapse, future studies of this type are needed to provide focus for the effort to define selective and novel antianxiety agents for these disorders.
Studies of antianxiety-like actions of pharmacological agents in animals undergoing ethanol withdrawal are challenging, in part because of the co-occurrence of potentially confounding signs, such as tremor, suppressed activity/locomotion, and seizures. Nonetheless, the results of many previous reports show that withdrawal from chronic ethanol treatment results in anxiety-like behavior in both human beings (Koob & Le Moal, 1997; Meyer, 1986; Naranjo & Sellers, 1985) and animal models (Baldwin et al., 1991; Criswell & Breese, 1993; File et al., 1989; Knapp et al., 1998; Moy et al., 1997, 2000). The mechanisms mediating anxiety-like behavior have not been definitively established.
Although benzodiazepines such as diazepam have long been known to be helpful in attenuating anxiety during ethanol withdrawal, it is generally understood that significant tolerance and dependence may result from their use. Novel benzodiazepine agents are actively being pursued as possibly safer treatments [see, for example, June et al. (1998a, 1998b)]. Like benzodiazepines, agents directed at the N-methyl-d-aspartate (NMDA) receptor [e.g., dizocilpine maleate (MK-801)] have also met with difficulties, given evidence for neurotoxicity or lack of efficacy on anxiety-like behavior (Criswell et al., 1994; Gatch et al., 1999; Horváth et al., 1997). However, NMDA receptor-modulatory sites (e.g., ifenprodil binding site or the strychnine-insensitive glycine recognition site) may present safer and more promising targets (Snell et al., 2000; Yang et al., 1996). Novel agents acting on corticotropin-releasing factor type 1 (CRF1), 5-hydroxytryptamine (serotonin) (5-HT) 2C (5-HT2C), or 5-HT 1A (5-HT1A) receptors (Chaki et al., 1999; File et al., 1993b; Griebel et al., 1997; Kennett et al., 1996, 1997; Schaffer & Naranjo, 1998; Wood et al., 2001) also offer promise for safely attenuating the anxiety of ethanol withdrawal. For example, the 5-HT1A receptor agonist buspirone has made the transition from animal studies (File et al., 1993b; Lal et al., 1991) to human beings and is recommended for treatment of chronic anxiety in alcoholism (Schaffer & Naranjo, 1998).
In the current series of experiments, we tested potential antianxiety agents, acting at a range of neurotransmitter receptors (CRF1, 5-HT2C, and 5-HT3 receptors; the benzodiazepine type 1 receptor; and the polyamine-site on the NMDA receptor), in a modification of the social interaction test. The social interaction test is a simple and elegant animal model that has been used in studies of a host of anxiogenic or anxiolytic drugs of various classes [see, for example, Costall et al. (1988), Dunn et al. (1989), File et al. (1996, 1999), and Overstreet et al. (2000)] and ethanol or benzodiazepine withdrawal [see, for example, Andrews et al. (1997), File (1997), File et al. (1989, 1993a, 1993b), and Kampov-Polevoy et al. (2000)]. In the current studies, we attempted (1) to assess anxiety-like behavior with the social interaction test after withdrawal from shorter-term ethanol drinking (17 days), in which low-to-moderate levels of ethanol withdrawal severity are manifest; (2) to confirm the induction of anxiety-like behavior with a second established animal model; and (3) to determine whether this approach could generate anxiety-like behavior that was separable (operationally/pharmacologically) from effects of ethanol withdrawal on locomotor behavior. These objectives were based, in part, on the premise that anxiety exists in various forms that are differentially elicited in different anxiety models and that manipulation of specific receptors in brain can alter the expression of specific anxiety states. Positive results from this type of behavioral study would set the stage for future complementary studies to identify neuroanatomic substrates of the anxiety component of withdrawal and effective pharmacological actions.
For 1 week before ethanol treatment, groups of 40, adult male, Sprague–Dawley rats (each weighing between 160 and 180 g), purchased from Charles River (Raleigh, NC), had ad libitum access to food and water in a temperature-and humidity-controlled environment on a normal 12-h light/12-h dark cycle, with lights on at 0900. For each group of 40 rats, 8 received control diet, and 32 received an ethanol-containing diet. After receiving this diet for 17 days, the rats were withdrawn and treated with vehicle or one of several treatments described below. Animal care and use was approved by the Institutional Animal Care and Use Committee at the University of North Carolina as per the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, Commission on Life Sciences, National Research Council, 1996).
Ethanol (from 95% stock; Aaper, Shelbyville, KY) was administered by means of a nutritionally complete liquid diet. This standard dietary ethanol procedure involves administration of 7% [weight/volume (wt./vol.)] ethanol in a lactalbumin/dextrose-based diet for 17 days, during which time rats generally consume ethanol at 8 to 12 g/kg/day and achieve blood ethanol levels up to 200 mg/dl (Criswell & Breese, 1993; Frye et al., 1981; Knapp et al., 1998). The diet was nutritionally complete (with concentrations of vitamins, minerals, and other nutrients derived from ICN Research Diets) and calorically balanced (with dextrose) across ethanol-treated rats and control rats. Intake matching was achieved by giving the control diet–treated rats a volume of diet equivalent to the average intake of the ethanol diet–treated rats the day before. Both groups readily gained weight during the ethanol exposure period.
In the first series of experiments, specific receptor antagonist or agonist treatments were given after removal of chronic ethanol diet exposure on day 17. Each rat received only one of the drugs and was tested only one time during withdrawal. The maximal withdrawal syndrome resulting from this procedure (including anxiety-like behavioral responses) occurs 5 to 7 h after removal of the ethanol diet from the rats’ cages (Criswell & Breese, 1993; Knapp et al., 1998; Moy et al., 2000). Attempts to block behavioral changes induced by withdrawal during this period were made with intraperitoneal injections of the 5-HT2C receptor antagonist SB242084 (6-chloro-5-methyl-1-[6-(2-methylpyridin-3-yloxy) pyridin-3-yl carbomyl] indoline) (Smith-Kline Beecham, Harlow, UK); the 5-HT3 receptor antagonist MDL 72222 (3-tropanyl-3,5-dichlorobenzoate); the NMDA receptor antagonist ifenprodil (2-(4-benzyl-piperidino)-1-(4-hydroxyphenyl)-1-propanol) (Research Biochemicals International, Natick, MA); the nonpeptide CRF antagonist CRA1000•H2SO4 (2-[N-(2-methylthio-4-isopropylphenyl)-N-ethylamino]-4-[4-(3-fluorophenyl)-1,2,3,6-tetrahydropyridin-1-yl]-6-methylpyrimidine) (Taisho Pharmaceutical Co., Ltd., Omiya, Japan); the benzodiazepine receptor antagonist flumazenil (RO15-1788, Hoffman-La Roche Inc., Nutley, NJ); and the benzodiazepine type 1 receptor agonist zolpidem (Synthelabo Recherche, Cedex, France). SB242084 (1 mg/kg), flumazenil (5 mg/kg), CRA1000 (1 mg/kg), and MDL 72222 (1 mg/kg) were prepared as fine suspensions after sonication in a 0.5% solution of carboxymethylcellulose (CMC), whereas ifenprodil (5 mg/kg) was prepared in distilled water. Zolpidem (3 mg/kg) was dissolved in saline. All drugs were administered as a 2-ml/kg dose between 5 and 6 h into withdrawal and 7 min (flumazenil), 15 min (zolpidem), or 30 min (CRA1000, SB242084, ifenprodil, and MDL 72222) before behavioral testing. All doses of drugs were carefully chosen on the basis of evidence of anxiolytic activity in other models or on the basis of information documenting the highest doses that one could expect would limit sedative or other generally disruptive effects on rat behavior (Criswell & Breese, 1993; Ebert et al., 1997; Harro et al., 2001; Mazzola-Pomietto et al., 1995; Millan et al., 2001; Moy et al., 1997, 1998). To confirm drug effects and dose responsivity, these doses were tested in a replication experiment with additional doses as follows: ifenprodil, 2.5, 5.0, and 10.0 mg/kg; zolpidem, 1.5 and 3.0 mg/kg; MDL 72222, 0.3, 1.0, and 3.0 mg/ kg; flumazenil, 1.25, 2.50, and 5.00 mg/kg; CRA1000, 0.25, 0.50, and 1.00 mg/kg; SB242084, 0.25, 0.50, and 1.00 mg/ kg. To determine the potential effects of these drugs in non–ethanol-treated rats, select doses of drugs were examined as follows in a final experiment: SB242084, 1 mg/kg; CRA1000, 1 mg/kg; flumazenil, 5 mg/kg; and ifenprodil, 5 mg/kg.
The primary behavioral test of anxiety used was the social interaction test (Duxon et al., 1997; File et al., 1996, 1999; File & Hyde, 1977; Gonzalez et al., 1996). Between 5 and 6 h into withdrawal, two identically treated pairs of rats, naive to the test and approximately equal in body weight, were placed into the center of a test box (a 60 × 60-cm square open field with 16 15 × 15-cm squares marked on the floor) simultaneously for a 5-min period. The time each rat engaged in social interaction (conspecific grooming, sniffing, following, crawling over/under) with its partner was recorded by an observer blind to the treatment conditions. Increased anxiety was inferred from reduced social interaction times. Locomotor activity, recorded as the number of squares entered during the 5-min session, was also recorded. It is important to emphasize that, unlike the plus-maze test behavior, all behavioral subcomponents of social interaction except “following” are independent of forward locomotion. Thus, when social interaction is present the animals tend to stay within the same square or two. Then, when a bout of interaction ends animals may or may not ambulate from square to square. Furthermore, results of other recent work in this laboratory have established that social interaction behavior within rat pairs was not correlated (Overstreet et al., 2002), a finding that reflects the fact that one animal may be actively engaged in a particular behavior such as grooming while the other is not. On the basis of these observations, as well as analyses with the use of individual animal data or data representing the combined data from the pair give comparable results (Overstreet et al., 2003), scores from individual rats were used in all analyses.
A second behavioral test of anxiety, the elevated plus-maze, was used in a subgroup of rats immediately before the social interaction test (Moy et al., 1997; Pellow et al., 1985). Rats were placed in the maze for 5 min and scored for the number of entries into and time spent in the open arms of the maze. Locomotor activity in the maze was determined from the number of closed-arm entries (File et al., 1999; Pellow et al., 1985), whereas the measure of anxiety-like behavior was inferred from diminished time in the open arms (Gonzalez et al., 1998).
Social interaction time in seconds and the number of squares entered are normally distributed scores. Therefore, data were analyzed by analysis of variance (ANOVA). A similar strategy was used for plus-maze test data, on which analyses were conducted of the closed-arm entries and the time in open arms. When significant, Tukey tests were used to compare pairs of groups (n = 8–10).
Rats experiencing ethanol withdrawal consistently reduced their time in social interaction across all experiments in the current study. Normally, rats engaged in 30 to 40 s of social interaction during the 5-min test period (Fig. 1), whereas ethanol-withdrawn rats typically displayed only 5 to 10 s of this behavior [t(18) = 6.19, P < .001]. Ethanol withdrawal also reduced locomotor activity (Fig. 1), as reflected in lower numbers of squares entered during the social interaction test [t(18) = 3.22, P < .01]. However, the reduction in social interaction was not simply a consequence of the decreased locomotor activity, because the groups were still different when the scores for social interaction and locomotor activity were compared as a ratio. Rats that received control diet had a ratio of 0.32, whereas those exposed to the ethanol diet had ratios of only 0.18 (indicating that the withdrawal preferentially affected the social interaction measure). A different profile of drug action emerged for effects on locomotor activity.
The initial series of pharmacological manipulations comparing results with the social interaction test revealed specific receptor-dependent and behavior-dependent effects during ethanol withdrawal. Among the group differences in the social interaction test [F(4,36) = 10.29, P < .0001], the benzodiazepine antagonist flumazenil and the 5-HT2C receptor antagonist SB242084 attenuated ethanol withdrawal–induced deficits in social interaction time (P < .05 vs. vehicle; Fig. 2). This effect was also observed when ratios of social interaction to locomotor activity were compared (data not shown). In contrast, the NMDA receptor antagonist ifenprodil exerted a limited effect in the social interaction test (Fig. 2; P > .05 vs. vehicle). In the social interaction test, a significant group effect on the locomotor activity measure was found [Fig. 2; F(4,36) = 5.05, P < .005]. Drug effects on social interaction were not accompanied by reduction of activity deficits. The pattern of results in the plus-maze test experiment mirrored the social interaction results (Fig. 3). For example, significant differences were found in open-arm time [F(4,31) = 5.31, P < .005] and closed-arm entries [F(4,31) = 7.49, P < .001] among treatment groups. Ethanol withdrawal reduced the number of closed-arm entries and the amount of time that the rats spent in the open arms by 57% and 76%, respectively (Fig. 3). Flumazenil and SB242084 were active in the test, but ifenprodil was not.
Given the parallel pharmacological profiles of behavioral activity in the social interaction test and the plus-maze test, other agents were examined only in the social interaction test. Among the group differences seen during withdrawal [F(4,33) = 18.93, P < .0001], the CRF1 receptor antagonist CRA1000 attenuated social interaction deficits (P < .05 vs. vehicle; Fig. 4). The 5-HT3 receptor antagonist MDL 72222 did not attenuate social interaction deficits (P > .05 vs. vehicle; Fig. 4). In the social interaction test, a significant group effect on the locomotor activity measure was found [Fig. 4; F(4,33) = 10.99, P < .0001]. Drug effects on social interaction were not accompanied by reduction of activity deficits.
The ameliorative effects of flumazenil on social interaction deficits were again revealed (P < .05 vs. vehicle) among the group effects [F(4,34) = 11.85, P < .0001] in Experiment 4 (Fig. 5). The benzodiazepine type 1 receptor agonist zolpidem did not attenuate social interaction deficits in this experiment (P > .05 vs. vehicle; Fig. 5). In the social interaction test, a significant group effect on the locomotor activity measure was found [Fig. 5; F(4,34) = 5.43, P < .005]. Drug effects on social interaction were not accompanied by reduction of activity deficits.
In control rats that had not been exposed to ethanol, no significant effect of the drugs on social interaction behavior was observed, with the exception of ifenprodil, which reduced social interaction behavior (Table 1). Ifenprodil and flumazenil significantly attenuated locomotor behavior in control rats, but CRA1000 and SB242084 did not (Table 1).
In corroboration of the single-dose study results, MDL 72222 and zolpidem were also not active at any dose when evaluated in multiple-dose experiments (Table 2) and were not pursued further. The lack of effect of ifenprodil in the social interaction test was replicated at three different doses (Table 2). Flumazenil, CRA1000, and SB242084 reversed withdrawal-induced deficits in social interaction time, as in the single-dose experiments (Table 2). The most robust effects were observed at the highest doses of these drugs.
The replicated results from experiments in the current study demonstrate that a model incorporating relatively short-term (17 days) dietary ethanol intake and the social interaction test can be used to study anxiety-like behavior during ethanol withdrawal in rats. These results also demonstrate that pharmacological blockade of 5-HT2C, CRF1, or benzodiazepine receptors, but not blockade of 5-HT3 or stimulation of specific benzodiazepine type 1 receptors, during withdrawal from ethanol can attenuate anxiety-like responding in the relative absence of drug effects on withdrawal-induced suppression of general locomotor activity. The independence of locomotor activity from anxiety-like responding observed in the experiments in the current study supports previous findings in this model [Breese et al. (online publication 10 July 2003 at http://www.acnp.org/citations/Npp07100303159/default.pdf); Overstreet et al. (2002, 2003)]. Thus, anxiety-like behavior can be manipulated in a sign-specific manner in the social interaction test. These results emphasize the importance of breaking down the withdrawal syndrome into its respective signs for study and intervention. The findings should promote efforts to develop pharmacotherapies for specific ethanol withdrawal signs that otherwise may independently promote further cycles of intake and withdrawal.
The role of benzodiazepine receptors in the expression of anxiety-like behavior during ethanol withdrawal was consistently shown in the results of previous experiments [see, for example, Criswell & Breese (1993), File et al. (1992), Jung et al. (2000), and Moy et al. (1997, 2000)]. However, a general agonist action at benzodiazepine receptors may be insufficient to encompass the relevant mechanisms operating in the current experiments. For example, gamma-aminobutyric acid B (GABAB) receptors may also be engaged at this time because baclofen antagonized ethanol withdrawal–induced deficits in behavior in social interaction and plus-maze tests (File et al., 1991, 1992). With regard to GABAA receptors, zolpidem preferentially binds to α1 GABA subunit–containing receptors (benzodiazepine type 1 receptors) (Doble, 1999; Mereu et al., 1990), an effect that may bias zolpidem’s behavioral actions toward sedation rather than anxiolysis (McKernan et al., 2000). This action of zolpidem differs from that of the classic benzodiazepines, which are less specific. However, previous study findings from our laboratory showed that zolpidem possesses anxiolytic activity in the elevated plus-maze during withdrawal, and other study findings showed that zolpidem possesses this behavioral action in normal (control) animals as well [see, for example, Griebel et al. (1996) and Pellow and File (1986)]. Because the ethanol withdrawal–induced social interaction deficits were somewhat less severe in the first zolpidem experiment (i.e., 47% reduction), it is possible that the potential anxiolytic action of zolpidem might have been more apparent against a typical background of more severe deficits (70%–90% reductions). However, in the replication experiment, the more typical withdrawal deficit was not blocked by either of two doses tested. Thus, that zolpidem is apparently not clinically useful in human beings for anxiety, is sedative at higher doses, and is not active against social interaction deficits during ethanol withdrawal underscore the potential importance of the actions of flumazenil and benzodiazepine type 2 agents in the anxiety associated with alcoholism.
Because the anxiolytic action of benzodiazepines in both control and withdrawn animals is most likely related to their ability to enhance GABA function in key brain regions, one might predict that the benzodiazepine antagonist flumazenil would have no effect on, or perhaps would exacerbate, anxiety-like behavior during withdrawal. However, in the current study and other studies [see, for example, Criswell & Breese (1993), File et al. (1989, 1992), and Moy et al. (1997, 2000)], flumazenil blocked anxiety-like behavior in different tests involving different withdrawal paradigms. These findings argue for a second mechanism of action, perhaps related to the blockade of endogenous benzodiazepine inverse agonists that may be present at the benzodiazepine receptor at this time (Buck et al., 1991; File et al., 1989; Moy et al., 1997, 2000). This effect is interesting in light of the fact that flumazenil has a half-life of 16 min (Lister et al., 1984) and yet can reduce anxiety-like behavior of ethanol withdrawal when administered many hours or days before a single withdrawal (Buck et al., 1991; File et al., 1989) or during multiple withdrawals before a final withdrawal (Knapp et al., 2001). This persistent effect of flumazenil in the absence of drug underscores the possibility that an important and persistent ethanol-dependent phenomenon is at work that may relate to specific interactions with the rich structural and functional diversity of GABA and benzodiazepine receptors. The inactivity of zolpidem in the current experiments also supports the suggestion of involvement of non-type 1 benzodiazepine receptors in this flumazenil effect, which was most marked at higher doses. It is known that various binding sites have differential affinities for and responses to flumazenil (Barnard et al., 1998), and potential structural adaptations of GABA receptors in select cells or brain regions after chronic ethanol treatment [see, for example, Grobin et al. (2000), Papadeas et al. (2001), and Petrie et al. (2001)] may mediate these differences. Furthermore, results of the dose-ranging and drug-treated control experiments demonstrate that the antianxiety actions during withdrawal can occur without correcting locomotor deficits in the plus-maze or social interaction test or without having actions on anxiety in control animals. One can propose, then, that flumazenil and the endogenous inverse agonists that it may block [e.g., octadecaneuropeptide (ODN), diazepam binding inhibitor (DBI), or triakontatetraneuropeptide (TTN)] compete for unique receptor sites differentially in chronic ethanol–treated versus control states. This proposition may be particularly important if it can be shown that these endogenous benzodiazepines have preferential actions on benzodiazepine type 2 receptors.
The potential role of 5-HT in ethanol withdrawal–associated anxiety was supported in previous reports (Gatch et al., 2000; Lal et al., 1993; Overstreet et al., 2000; Prather et al., 1991; To et al., 1999). Results of experiments in the current study extend previous research findings by showing an impressive ability of the selective 5-HT2C receptor antagonist SB242084 to ameliorate deficits in social interaction and plus-maze open-arm time induced by ethanol withdrawal. Moreover, the drug was without significant effect on either locomotor activity or social interaction in control rats. In the plus-maze test, reduced locomotor activity (reduced closed-arm entries) arguably complicates the selectivity of the test for anxiety-like behavior during withdrawal. However, that both SB242084 and flumazenil were shown to increase the time in open arms despite the ethanol-induced reduction in closed-arm entries supports the suggestion that the anxiety-like behavior is to a significant extent independent of the locomotor activity. In other studies, lower doses of the 5-HT2C receptor agonist m-chlorophenylpiperazine (mCPP) were required to induce a maximum anxiogeniclike response in ethanol-withdrawn rats relative to control rats (Rezazadeh et al., 1993). Together, these results seem to indicate that 5-HT2C receptors, perhaps in the amygdala [see Pompeiano et al. (1994)], may play an important role in the expression of the anxiety-like signs of ethanol withdrawal. These results are also consistent with the proposed role for 5-HT2A/2C receptor–mediated protein kinase C (PKC)/phosphoinositide (PI) signaling in ethanol dependence (Pandey, 1998; Pandey & Pandey, 1996).
The failure of the 5-HT3 receptor antagonist MDL 72222 to block social interaction deficits supports the suggestion that these receptors are not involved significantly in the expression of anxiety-like behavior during ethanol withdrawal. However, because this 5-HT3 receptor antagonist blocked anxiety-like behavior in other contexts (Bilkei-Gorzó et al., 1998; Higgins et al., 1991) and chronic treatment with another 5-HT3 antagonist, ondansetron, attenuated social interaction deficits in ethanol-withdrawn rats (Costall et al., 1990), there seem to be specific complex actions of 5-HT at this receptor that as yet need to be resolved. One could argue that the anxiolytic dose range of this drug may not include the dose used in the current study. However, evidence seems to indicate that higher doses can be sedative (Mazzola-Pomietto et al., 1995). Therefore, it is unlikely that MDL 72222 could reverse social interaction deficits in this model. Overall, these observations of SB242084 and MDL 72222 seem to indicate that the action of 5-HT during ethanol withdrawal on anxiety-like behavior may depend on the 5-HT receptor subtype and the subtype of anxiety-like behavior assessed in different tests.
Overactivity of the CRF system in anxiety occurring during ethanol withdrawal, as well as in non–ethanol-treated animals, has been implicated in a number of studies. Alcohol or a peptide CRF antagonist attenuates CRF-induced or withdrawal-induced anxiety-like behavior (Baldwin et al., 1991; Thatcher-Britton & Koob, 1986). Results of similar studies support the suggestion that amygdala CRF may play an important role in this effect (Menzaghi et al., 1994; Merlo Pich et al., 1995; Rassnick et al., 1993). Results of the current studies are consistent with those findings, in that the novel selective nonpeptide CRF1 receptor antagonist CRA1000 (Chaki et al., 1999; Okuyama et al., 1999) dose-dependently blocked the expression of social interaction deficits in withdrawing animals. This effect seemed to be specific for anxiety-like behavior in withdrawing animals, because there were no significant effects on locomotor deficits in withdrawing rats or normal locomotor or social interaction behaviors in control rats.
Because the NMDA receptor antagonist ifenprodil was not active in ethanol-withdrawn rats in the current study at any dose, it would appear that NMDA receptors are not exerting significant effects on anxiety-like behavior during ethanol withdrawal. Ifenprodil was not active in either the plus-maze or social interaction test, despite being used at or above a dose that is behaviorally active against the diazepam withdrawal syndrome and hypoxia (Eraković et al., 1997; Tsuda et al., 1998). This observation, combined with the anxiogenic-like response seen in control rats, supports the suggestion that the drug is not likely to be a good anxiolytic prospect in alcoholism. The drug also inhibited locomotor activity in control rats, but no differential effect was seen in withdrawing rats treated with ifenprodil or vehicle. These results are important in determining receptor targets relevant to expression of withdrawal signs because evidence consistently shows that NMDA receptors play a role in action of glutamate receptor–mediated hyperexcitability during withdrawal (Kumari & Ticku, 2000). In other studies, the competitive NMDA receptor antagonists 2-amino-7-phosphonoheptanoic acid (AP-7) and d,l-(E)-amino-4-methyl-5-phosphono-3-pentenoic acid (CGP 37849), but not the noncompetitive NMDA receptor channel blocker MK-801, attenuated the plus-maze deficits during ethanol withdrawal in the elevated plus-maze (Criswell et al., 1994; Gatch et al., 1999). These results support the suggestion that glutamate action on this behavior during ethanol withdrawal depends on the receptor subtype and the mechanism of receptor blockade.
In summary, results of experiments in the current study revealed anxiety-like behavior during withdrawal from short-term ethanol intake that was attenuated by antagonists of benzodiazepine, CRF, and 5-HT2C receptors, but not by antagonists of NMDA-type glutamate or 5-HT3 receptors or an agonist of benzodiazepine type 1 receptors. These results support the suggestion that multiple neuronal systems may contribute to the specific ethanol withdrawal sign of anxiety and that anxiety can be manipulated in a pharmacologically selective manner. These results are consistent with the general theoretical principle that negative reinforcement (e.g., tension reduction, anxiety reduction, and alleviation of other signs and symptoms of acute or chronic ethanol withdrawal) is a potentially important motivator of pathologic levels of alcohol consumption (Cappell & LeBlanc, 1981; Hershon, 1977; Koob & Le Moal, 1997).
We thank Lee Gause for assistance with preparing and administering the ethanol diets. This work was supported by National Institute on Alcohol Abuse and Alcoholism grants AA00214 (D.J.K.) and AA11605/AA10025/AA00253/ AA12655 (G.R.B.).