, Volume 185, Issue 2, pp 188–200

Previous experience of ethanol withdrawal increases withdrawal-induced c-fos expression in limbic areas, but not withdrawal-induced anxiety and prevents withdrawal-induced elevations in plasma corticosterone


  • Gilyana G. Borlikova
    • Department of Psychology, School of Life SciencesUniversity of Sussex
  • Julie Le Merrer
    • Department of Psychology, School of Life SciencesUniversity of Sussex
    • Department of Psychology, School of Life SciencesUniversity of Sussex
Original Investigation

DOI: 10.1007/s00213-005-0301-3

Cite this article as:
Borlikova, G.G., Le Merrer, J. & Stephens, D.N. Psychopharmacology (2006) 185: 188. doi:10.1007/s00213-005-0301-3



Increased anxiety is a characteristic of the acute ethanol withdrawal syndrome. Repeated exposure of rats to withdrawal from chronic ethanol increases sensitivity to seizures.


We investigated whether repeated withdrawal experience increases withdrawal-induced anxiety and stress, and if it changes withdrawal-induced activation of related brain areas.


Rats were chronically treated with an ethanol-containing liquid diet either for 24 days continuously (single withdrawal, SWD) or interspersed with 2×3-day withdrawal periods (repeated withdrawal, RWD), or with a control diet. Eight hours after ethanol withdrawal, anxiety-like behaviour was tested in the elevated plus-maze, blood corticosterone levels were measured, and expression level of markers of neuronal activity and plasticity, c-fos and zif268, was assessed.


Eight hours after ethanol withdrawal, SWD rats showed increased anxiety on the elevated plus-maze relative to control rats. Rats given previous withdrawal experiences did not show further increases in measures of anxiety. Corticosterone levels were elevated during withdrawal in SWD rats but not in RWD rats. RWD resulted in marked increases in c-fos expression in amygdala, hippocampus, nucleus accumbens and dorsolateral periaqueductal grey. In contrast, zif268 expression was not increased after RWD, and in central amygdala the marked increase in zif268 seen after SWD was absent after RWD.


The data suggest increased ability of withdrawal to activate neuronal circuits but reduced plasticity after RWD. We suggest parallels between the consequences of repeated ethanol withdrawal and repeated exposure to stress, and discuss implications of withdrawal for brain plasticity.


AmygdalaDetoxificationDependenceRepeated withdrawalAnxietyStressc-foszif268


Repeated experience of detoxification from ethanol dependence leads to increased severity of withdrawal symptoms (Baker and Cannon 1979; Brown et al. 1988; Lechtenberg and Worner 1991). Although most studies have concentrated on the effects of repeated ethanol withdrawal on seizure activity during withdrawal (Brown et al. 1988; Lechtenberg and Worner 1991), a phenomenon readily modelled in rodents (Becker and Hale 1993; Ghozland et al. 2005; Mhatre and Gonzalez 1999; Stephens et al. 2001; Veatch and Becker 2002), attention has been given more recently to the effects of repeated withdrawal on other behavioural and physiological symptoms of withdrawal (Holter et al. 1998; Overstreet et al. 2002) and to long-term consequences of repeated withdrawal experience on emotional and cognitive function (Duka et al. 2004; Ripley et al. 2003a; Stephens et al. 2001; Townshend and Duka 2003). A link has also been drawn between repeated withdrawal and binge drinking (Duka et al. 2004; Stephens et al. 2005; Townshend and Duka 2005). We have previously reported that a long-term consequence of repeated withdrawal from ethanol is the impairment of fear conditioning in both rodents (Ripley et al. 2003b; Stephens et al. 2001) and humans (Stephens et al. 2005), though aspects of appetitive conditioning may also be influenced (Ripley et al. 2004).

Increased anxiety is a characteristic of the acute ethanol withdrawal syndrome (DSM-IV 1994). Nevertheless, there is mixed evidence in the literature that the number of previous detoxification experiences influences anxiety levels in alcoholics undergoing acute detoxification. In one study, several days into detoxification of alcoholic inpatients, although we found increased levels of anxiety in patients undergoing detoxification, anxiety levels were not exacerbated by prior withdrawal experience (Duka et al. 2002). In contrast, (Malcolm et al. 2002) found some evidence for a mild increase in anxiety in this population very early in withdrawal. Nevertheless, in a rodent model, Breese and colleagues (Breese et al. 2005; Overstreet et al. 2002, 2003, 2004a,b) consistently found a decrease in social interaction, which has been characterised as an animal model of anxiety (File 1980; File and Hyde 1978), in rats subjected to repeated ethanol withdrawal, and concluded that multiple withdrawals from chronic ethanol led to a sensitisation of withdrawal-induced anxiety. Using a quite different model of lengthy (11-month) chronic ethanol intake, Holter et al. 1998 also reported that withdrawal anxiety was higher after many (perhaps nine) withdrawal episodes than on the first.

We found in previous experiments that repeated withdrawal from another class of sedative/hypnotic drugs, the benzodiazepines, whilst like ethanol withdrawal, sensitising mice to convulsants, reduced their anxiety-like behaviour in the elevated plus-maze and four-plate tests (Ward and Stephens 1998). If ethanol withdrawal resembles benzodiazepine withdrawal, repeated withdrawal experience from ethanol may even result in habituation to the withdrawal-induced stress and perhaps lower responsiveness to stress-related anxiety. In this context, it is interesting that repeated exposure to restraint stress also protects against future stressors (Belda et al. 2004; Carter et al. 2004; Chen and Herbert 1995; Stamp and Herbert 1999).

In the present study, we therefore tested whether repeated ethanol withdrawal led to the sensitisation of the anxiety state, as suggested by Overstreet, Breese and colleagues (Overstreet et al. 2002; Breese et al. 2005), or, as we have reported for repeated benzodiazepine withdrawal, if it lowered anxiety level in rodents. For this purpose, we tested animals in the elevated plus-maze, a task frequently used to assess anxiety (Pellow et al. 1985; Rodgers et al. 1999), including anxiety induced by ethanol withdrawal (Baldwin et al. 1991). Additionally, we measured blood levels of the stress-related hormone, corticosterone (Herman and Cullinan 1997). To understand more precisely the brain processes influenced by repeated withdrawal, the ability of the final withdrawal to affect expression levels of the immediate early genes that are markers of neuronal activity, c-fos and zif268, was screened in different brain areas related to stress, fear and anxiety. The amygdala is strongly implicated in fear- and anxiety-related processes, as well as showing activation during stress processing (Dayas et al. 2001; LeDoux 2000; Singewald et al. 2003), and the hippocampus also plays a role in these processes, especially in case of conditioned cues and cognitive stressors (Herman et al. 1998; Linden et al. 2004). The hypothalamus is important for regulation of the hormonal response to stress and to fear-related stimuli (Dayas et al. 2001; Herman and Cullinan 1997; Linden et al. 2004; Singewald et al. 2003), and the periaqueductal grey is implicated in the response to stressors and anxiety-related behaviour (Dayas et al. 2001; Linden et al. 2004; Salome et al. 2004). Changes in the expression of c-fos and zif268 are widely used as indicators of brain activation as a result of different manipulations (Herdegen and Leah 1998; Knapska and Kaczmarek 2004; Mutschler et al. 2000). Stress-related changes in the level of c-fos and zif268 expression are well documented (Day et al. 2005; Herdegen and Leah 1998; Keay et al. 2001; Knapska and Kaczmarek 2004; Melia et al. 1994; Senba and Ueyama 1997).

Materials and methods


The subjects of experiment 1 were 36 male Lister hooded rats (130–160 g at the beginning of the experiment, Harlan, Bicester, UK). For experiments 2 and 3, 54 male Lister hooded rats (350–400 g at the beginning of the experiment, University of Sussex, Brighton, UK) were used. All animals were pair-housed and maintained on a 12-h light/dark cycle (lights off at 19:00; temperature 21±2°C; humidity 50±10%) with ad libitum access to water. All experiments were carried out under the authority of the UK Animal (Experimental Procedures) Act, 1986.

Chronic ethanol treatment

For each of the experiments described below, separate groups of rats were randomly assigned to one of three treatment conditions: single withdrawal (SWD), repeated withdrawal (RWD) and control (CON) (according to Ripley et al. 2004). Briefly, the animals in the first two groups received a nutritionally complete liquid diet (Dyets, Bethlehem, PA, USA), containing 7% ethanol as their only food source. Control animals were pair-fed with a calorifically matched control diet. During 30 days of treatment, the RWD group had two intermediate withdrawal episodes (days 11–13 and 21–23), when animals received the control diet instead of the ethanol diet. The SWD group remained on ethanol-containing diet for 24 consecutive days.

On the final day, animals were withdrawn from the liquid diet at 06:00–08:00 h (counterbalanced between groups) and remained in their home cages with ad libitum access to water but no food for 8 h until they were tested in the elevated plus-maze or killed for blood collection or immunohistochemistry. In our experience, food restriction during this period heightens the ethanol withdrawal syndrome, so that signs of withdrawal could be observed throughout the 8-h period (unpublished observations from Ripley et al. 2004).

Experiment 1

Behaviour in the elevated plus-maze

The animals (n=12 per treatment group) were tested between 14:00 and 17:00 h in the plus-maze, 8 h after the final withdrawal. The plus-maze consisted of two open (44×12 cm) and two enclosed (44×12×26 cm) arms that were connected by a common central platform (12×12 cm). The arms and central platform had white Plexiglas covering; the walls of the enclosed arms were from grey Plexiglas. The maze was elevated 82 cm above the floor level and illuminated using diffuse overhead fluorescent lighting. Testing was conducted under low light illumination (2×32 W, 2 m above the maze). The animals were placed on the central platform facing an open arm and their behaviour in the maze was recorded for 5 min by a video camera. The maze was wiped with cleaning liquid after each animal was tested. Behavioural parameters, scored by observation of videotapes, (according to Cruz et al. 1994, Rodgers et al. 1999), were the number of open and closed arm entries, time spent in different arms, the number of crossings of the hypothetical line dividing an arm into equal proximal and distal parts, time spent in distal parts of open and closed arms, the number of head dips from open arms and the number of “protected” dips, the number of rears, the number of risk assessments [investigations of the open arms from the central platform or a closed arm with the forepaws and head only (Cruz et al. 1994; Rodgers et al. 1999)] and the number of grooming episodes. Time spent in distal parts of open arms was calculated as percentage of the total time spent in distal parts of both types of arms. The number of rears and head dips was calculated as a ratio of the number of rears (dips) in the open/closed arms to the total time spent in that kind of arm/area.

Experiment 2

Collection of blood samples and determination of plasma corticosterone

The animals (n=10 per group) were killed by decapitation 8 h after ethanol withdrawal. Trunk blood was collected into iced, heparinised tubes and centrifuged at 3,500×g for 15 min at room temperature to remove cells, and the plasma was stored at −18°C. Plasma corticosterone levels were determined using a commercially available enzyme immunoassay kit (Immunodiagnostic Systems, UK). Intra- and inter-assay coefficients of variation were less than 5%.

Experiment 3

c-fos and zif268 immunohistochemistry

c-fos and zif268 expression was assessed in the same animals (n=8 per group). Eight hours after final ethanol withdrawal, animals were deeply anaesthetised with Avertin (40 mg/kg, i.p.) and immediately transcardially perfused with paraformaldehyde in 0.1 M phosphate-buffered saline (PBS). The brains were excised and placed in 4% paraformaldehyde in PBS for 24 h, then cryoprotected in sucrose before freezing in isopentane at −50°C and stored at −80°C for up to 2 months before being coronally sectioned at 30 μm using a freezing microtome. Sections were collected in PBS (alternative sections through the region of interest stained for c-fos or for zif268), incubated with 0.3% H2O2 for 10 min, and washed in PBS. After 1 h incubation with 1.5% normal goat serum (NGS, Vector Laboratories), sections were washed again and incubated overnight in PBS containing 0.5% NGS, 0.02% sodium azide (Sigma), 0.3% Triton X-100 (Sigma) and 1:10,000 c-fos primary antibody (rabbit polyclonal serum, Oncogene, Ab-5) or 1:1,600 zif268 primary antibody (Egr-1 (588), rabbit polyclonal serum, Santa Cruz Biotechnology, sc-110). On the next day, sections were washed and incubated for 1 h in PBS containing 1.5% NGS and 1:300 biotinylated anti-rabbit secondary antibody (Vector laboratories, BA-1000). After washing, sections were incubated for 1 h in PBS containing 1:1,000 avidin-biotinylated–horseradish peroxidase complex (Vectastain ABC kit Elite, Vector Laboratories). The reaction was visualised using a standart glucose oxidase-3,3-diaminobenzidine method. The reaction was terminated by extensive washing.

Fos- and zif-positive nuclei were quantified from images of sections captured using AxioCam HRc digital camera mounted on a Zeiss Akioskop 2 plus microscope (Carl Zeiss, UK) using AxioVision 3.1 software (Imaging Associates, Bicester, UK) with ×100 magnification. The regions studied were basolateral (BLA, 0.15 mm2 region), central (CeA, 0.13 mm2) and lateral (LA, 0.18 mm2) nuclei of the amygdala, core (AcbC, 0.13 mm2) and dorsal shell of nucleus accumbens (AcbSh, 0.13 mm2), CA1 (0.12 mm2) and CA3 (0.08 mm2) fields of hippocampus, medial parvacellular nucleus of paraventricular nucleus of hypothalamus (PVN, 0.05 mm2), dorsal caudate putamen (dCPu, 0.18 mm2) and dorsolateral periaqueductal grey (dlPAG, 0.23 mm2). Counting was performed using Scion Image for Windows (Scion Corporation, MD, USA). The locations of the areas used for quantifying Fos- and zif-positive nuclei were taken from Paxinos and Watson (1998) and are shown in Fig. 3. Counts were taken bilaterally in two sections for each brain region and the mean of these four values was taken as the number of nuclei for that brain region.


Data collected were analysed using SPSS 11.5 for Windows. For each set of data, means were calculated and data that were greater than or less than mean ±2 SD were excluded from the analysis. All data analysed were initially checked for normality of distribution by Kolmogorov–Smirnov’s test of normality. Then data were analysed using one-way or two-way ANOVA and repeated measures ANOVA. Significant interactions were followed by post hoc tests (Student–Newman–Keuls test, t test). All data analysed using a repeated measures ANOVA were initially checked by Mauchly’s test of sphericity. A significant p value indicated that there was sufficient evidence to reject the sphericity assumption and the p value for repeated measures was calculated using the Greenhouse–Geisser Epsilon correction. All tests of significance were performed at α values of 0.05.


Chronic ethanol treatment

In the subgroup of rats later tested on the elevated plus-maze, the mean ethanol consumption over the final week of chronic treatment in both SWD and RWD groups was approximately 18 g ethanol per kilogram body weight per day. An analysis of the overall ethanol consumption with repeated measures ANOVA showed that, whilst consumption changed with treatment (Fig. 1; main effect of day F23, 230=13.16, p<0.05), there was no significant difference in ethanol intake between the two groups (F1,10=4.39, ns), but there was a significant group by day interaction (F23, 230=2.00, p<0.05). Post hoc analysis revealed that SWD and RWD groups consumed different amount of ethanol on days 1, 2, 11, 15 and 17 (F1,10=69.63, F1,10=7.54, F1,10=33.12, F1,10=24.72, p<0.05) of the 24-day treatment. In experiments 2 and 3, over the final week of chronic treatment, SWD and RWD animals consumed approximately 12.5 g kg−1 day−1 of ethanol. Analysis of the 24-day consumption showed changes in the consumption level with treatment (F23,138=6.97, p<0.05), but no difference between SWD and RWD groups and no group by day interaction (F1,6=1.09, F23,138=0.83, ns).
Fig. 1

Ethanol consumption in animals tested on the elevated-plus-maze (gram per kilogram of body weight per day of treatment). Data represent the mean±SEM. Arrows indicate starting points of the withdrawal episodes in repeatedly withdrawn group. There was no overall difference between groups in the consumption level; during the last week of treatment SWD and RWD groups consumed an average 18 g kg−1 day−1 of ethanol (*different from SWD group, p<0.05)

Experiment 1

Behaviour in the elevated plus-maze

The effect of single and repeated withdrawal from chronic ethanol on behaviour in the elevated plus-maze is shown in Fig. 2. A significant effect of arm (open vs closed) (F1,64=27.36, p<0.001) indicated less exploration of the open arms (Fig. 2a). Whilst there was no main effect of treatment (F2,64=0.14, ns), a significant group by arm interaction (F2,64=3.39, p<0.05) suggested that ethanol withdrawal decreased exploration of the open arms. Whilst under the experimental conditions used, controls showed no differences in time spent exploring open and closed arms (T20=1.27, ns), both SWD and RWD rats spent significantly less time exploring the open arms (SWD: T22=4.82, p<0.001; RWD: T22=2.88, p<0.01). A clear group difference emerged in the percentage of time that animals from different groups spent in the distal part of open arms (Fig. 2b, main effect of treatment; F2,32=4.52, p<0.05), post hoc tests revealing that animals from SWD and RWD groups spent significantly less time in the distal parts of open arms than control animals (p<0.05). The analysis of the number of closed and open arm entries (Fig. 2c) showed significant effect of treatment group (F2,64=6.21, p<0.01), but no effect of arm (F1, 64=1.46, ns), and no group by arm interaction (F2,64=1.03, ns). Post hoc tests revealed that both ethanol-treated groups were significantly different from the CON group, indicating reduced activity, but did not differ from each other. The analysis of the entries into the distal parts of arms as a percentage of the total number of arm entries (Fig. 2d) further supported findings from the analysis of the time spent in different arms. It showed that all animals visited distal parts of the open arms on fewer occasions than the distal parts of the closed arms (arm: F1,64=57.33, p<0.001; group: F2,64=2.21, ns; group by arm interaction: F2,64=4.58, p<0.05). Consequent separate analysis for each type of arms revealed that both RWD and SWD animals avoided the most anxiogenic distal parts of the open arms (F2,32=4.15, p<0.05; post hoc p<0.05), whilst in the closed arms, animals from all treatment groups showed the same activity (F2,32=2.39, ns).
Fig. 2

Measures of anxiety-related behaviour in the plus-maze. Data show the mean±SEM, n=11–12 per group. a Time spent in different arms (**p<0.01 compared to closed arms). b Time spent in distal parts of open arms as percentage from the total time in distal parts of arms (*p<0.05 compared to CON). c The number of entries into different arms (SWD and RWD groups less active than CON, p<0.05). d The number of entries into distal parts of different arms as percentage of the total number of arm entries (all animals entered distal parts of open arms on fewer occasions, p<0.001; *p<0.05 compared to CON). e Rates of rearing in open and closed arms (all animals performed more rearings in closed arms, p<0.001; *p<0.05 compared to SWD in open arms). f Rates of dipping from the “protected” (central platform and proximal parts of closed arms) and open areas (*p<0.05 compared to CON in “protected” area; **p<0.01 compared to CON in open arms)

All animals performed more rears (Fig. 2e) in the closed than in the open arms (F1,64=203.18, p<0.001). Although no effect of treatment group was found, (F2,32=0.90, ns), there was a significant group by arm interaction (F2, 64=3.59, p<0.05). Subsequent analyses of each arm separately showed no group differences in the number of rears in the closed arms, but a significant difference in the rate of rearing between SWD and RWD groups in the open arms (F2,32=1.60, ns, closed arms and F2,32=3.89, p<0.05; post hoc p<0.05, open arms). The picture for head dips was rather different (Fig. 2f). Unsurprisingly, head dipping was mostly confined to the open arms. One-way ANOVA of dips from the open arms showed an effect of treatment, with both alcohol-treated groups exhibiting fewer head dips (F2,32=6.32, p<0.01; post hoc p<0.05, RWD and SWD vs CON). Additionally, SWD animals performed fewer dips than control animals from the central area of the maze (F2,32=5.36, p<0.05). There was no difference between experimental groups in the number of risk assessments and groomings (Table 1, risk assessments: F2,32=0.67, ns; groomings: F2,32=0.30, ns).
Table 1

Additional exploratory measures from the plus-maze


Treatment group




Risk assessments








Data presented as means±SEM. Risk assessments represented investigation of the open arms from the central platform or a closed arm (Cruz et al. 1994; Rodgers et al. 1999)

Experiment 2

Blood corticosterone level

Differences in the plasma corticosterone level between the three chronic treatment groups (Fig. 3) were significant (F2,26=4.08, p<0.05), post hoc Student–Newman–Keuls indicating that SWD animals differed from both CON and RWD animals in having higher level of corticosterone at 8 hr after withdrawal, whilst there was no difference between RWD and CON groups.
Fig. 3

Blood corticosterone levels, 8-h after withdrawal. SWD animals had higher level of corticosterone than both CON and RWD animals. n=9–10 per group (*p<0.05)

Experiment 3

c-fos and zif268 expression

Areas in which immunostaining for Fos and zif 268 were examined are shown in Fig. 4. Mean numbers of Fos-positive nuclei for different brain areas studied are shown in Fig. 5. The patterns of c-fos expression in different amygdala nuclei are shown in Fig. 5a and typical photomicrographs of c-fos expression in CeA in Fig. 6. Two-way ANOVA revealed significant main effect of amygdala region (F2,34=15.88, p<0.01), region by group interaction (F4,34=5.47, p<0.01) and a significant difference between treatment groups (F2,17=16.88, p<0.01). Post hoc test showed that the RWD group differed from both CON and SWD animals. Subsequent separate analysis of c-fos data from each nucleus showed that, whilst there was no difference among groups in the level of this gene expression in LA, in CeA and BLA nuclei of RWD animals, level of c-fos expression was much higher than in CON and SWD animals (one-way ANOVA, LA: F2,20=0.64, ns; BLA: F2,18=8.36, p<0.05, post hoc RWD differed from CON and SWD; CeA: F2,18=16.82, p<0.01, post hoc RWD differed from CON and SWD; see Fig. 5 for typical photomicrographs of Fos immunostaining in CeA).
Fig. 4

Schematic diagrams, adapted from the atlas of Paxinos and Watson (1998), showing the areas in which c-fos and zif268 expression was quantified. Bregma 1.60: AcbC Nucleus accumbens core, AcbSh nucleus accumbens dorsal shell, dCPu dorsal caudate putamen. Bregma −1.80: PVN Medial parvacellular part of paraventricular nucleus of hypothalamus. Bregma −3.14: LA lateral amygdala nucleus, BLA basolateral amygdala nucleus, CeA central amygdala nucleus, CA1 CA1 field of hippocampus, CA3 CA3 field of hippocampus. Bregma −6.8: dlPAG dorsolateral periaqueductal grey
Fig. 5

Effect of single and repeated withdrawal from ethanol on the number of c-fos positive cells in amygdala (a), hippocampus (b), nucleus accumbens (c), paraventricular nucleus of hypothalamus (d), periaqueductal grey (e) and caudate putamen (f). Data represent the mean±SEM of the number of c-fos positive cells in an indicated brain region, n=7–8 per group (*different from CON and SWD groups, **different from CON group, p<0.05)
Fig. 6

Representative photomicrographs (magnification ×100) of the central nucleus of the amygdala illustrating immunostaining of c-fos (left column) and zif268 (right column) in the three treatment groups. Controls (a and d), SWD (b and e) and RWD (c and f)

In the hippocampus (Fig. 5b), analysis showed a difference between c-fos expression in CA1 and CA3 field (F1,17=12.8, p<0.01), but no field by group interaction (F2,17=1.62, ns), and no effect of treatment group (F2,17=2.17, ns). Subsequent one-way ANOVA found no significant difference between groups in the CA1 field (F2,19=1.41, ns), whereas in the CA3 field pattern was rather different—the RWD group had a significantly higher level of c-fos expression than the CON group, with SWD group intermediate (F2,18=4.63, p<0.05, post hoc RWD differed from CON).

In nucleus accumbens (Fig. 5c), there was a significant difference in c-fos expression between core and shell regions (F1,19=62.14, p<0.01), no region by group interaction (F2,19=2.86, ns) and a significant main effect of treatment group (F2,19=5.44, p<0.05). Post hoc tests showed that RWD group differed from CON. One-way ANOVA clarified that in AcbC, RWD animals had higher c-fos expression than CON, whereas SWD animals were again intermediate (F2,20=3.63, p<0.05, post hoc RWD differed from CON); there was no significant difference between treatment groups in AcbSh (F2,19=2.32, ns).

One-way ANOVA did not reveal any significant difference between c-fos expression in PVN (Fig. 5d) of ethanol- and control-diet treated animals (F2,18=0.35, ns). Nevertheless, RWD animals had higher c-fos level in dorsolateral PAG (Fig. 5e, F2,18=4.61, p<0.05, post hoc RWD differed from CON). In the dorsal part of caudate putamen (Fig. 5f), despite some tendency, ANOVA analysis did not show any significant difference between groups (F2,20=1.71, ns).

The profile of zif268 expression 8 h after the final withdrawal from chronic ethanol in different brain structures is shown in Fig. 7. The patterns of zif268 expression in different amygdala nuclei are shown in Fig. 7a, and typical photomicrographs of zif268 expression in CeA in Fig. 6. Analysis of the data with two-way ANOVA showed that level of zif268 expression varied between different amygdala nuclei (F2,38=106.43, p<0.01) and there was significant nucleus by group interaction (F4,38=3.43, p<0.05), but no main effect of group (F4,19=2.15, ns). Subsequent analysis showed that in the CeA, single withdrawal from ethanol caused significant increase in zif268 expression; however, experience of repeated withdrawal did not affect expression of this gene (F2,20=5.47, p<0.05, post hoc SWD differed from CON and RWD). There was no difference in the level of zif268 expression between treatment groups in the other amygdala nuclei (one-way ANOVA, BLA: F2,20=0.48, ns; LA: F2,21=0.13, ns).
Fig. 7

Effect of single and repeated withdrawal from ethanol on the number of zif268 positive cells in amygdala (a), hippocampus (b), nucleus accumbens (c), paraventricular nucleus of hypothalamus (d), periaqueductal grey (e) and caudate putamen (f). Data represent the mean±SEM of the number of zif268 positive cells in an indicated brain region, n=7–8 per group (*different from CON group, p<0.05)

In hippocampus (Fig. 7b), expression of zif268 was significantly higher in the CA1 field than in the CA3 field (F1, 21=594.58, p<0.01, nearly all nuclei were stained), but there were no differences in expression between groups (F2,21=1.06, ns) and no region by group interaction (F2,21=0.75, ns). One-way ANOVA analysis confirmed, that, in contrast to the c-fos pattern, zif268 expression in the brains of RWD and SWD rats was not changed in either field of hippocampus examined (CA1: F2,21=1.05, ns; CA3: F2,21=0.79, ns).

In nucleus accumbens (Fig. 7c), the pattern was similar: There was main effect of region (F1,21=54.3, p<0.01) but no differences between groups (F2, 21=0.11, ns) and no region by group interaction (F2,21=0.08, ns). Further analysis once again confirmed absence of differences between treatment groups (AcbC: F2,21=0.09, ns; AcbSh: F2,21=0.12, ns).

In PVN of hypothalamus (Fig. 7d), there was a numerically higher level of zif268 expression in SWD rats, but it was not significant (F2,18=0.79, ns). Level of zif268 expression in dlPAG (Fig. 7e) and dCPu (Fig. 7f) was similar in all treatment groups (dlPAG: F2,19=1.24, ns; dCPu: F2,21=0.2, ns).


In keeping with previous studies demonstrating increased anxiety-like behaviour during alcohol withdrawal, rats, on their first experience of ethanol withdrawal, showed increased anxiety-like behaviour (spent less time in the open arms of the elevated plus-maze) than controls. This became especially apparent when time spent in and entries into the distal parts of the open arms were considered. Although entries to the distal parts of the arms did not form part of the classical analyses of plus-maze behaviour (Handley and Mithani 1984; Pellow et al. 1985), a number of researchers have subsequently introduced this measure (Dere et al. 2002; File et al. 1999), which factor analysis reveals to be heavily loaded under the factor anxiety (Cruz et al. 1994). Rats that had previously experienced withdrawal (RWD group) showed similar anxious-like behaviour in these measures to SWD rats. Data on the number of entries into different arms showed that both ethanol-treated groups were generally less active than controls. Similar reductions in activity in the plus-maze after withdrawal from chronic ethanol treatment have been reported previously (File et al. 1993; Gatch et al. 1999, 2000; Watson et al. 1997; Wilson et al. 1998). The fact that, in our experiment, ethanol-treated groups tended to show reduced locomotor activity in some measures puts some constraints on the interpretation of the elevated plus-maze data. But as a greater locomotor suppression in RWD group should bias the entry towards a higher anxiety-like profile (Dawson and Tricklebank 1995; Kliethermes 2005; Kliethermes et al. 2005), our data are at least clear in showing no evidence for sensitisation of the anxiety-like behaviour with repeated withdrawal.

Considering the so-called ethological measures of behaviour in the plus-maze, animals from all three treatment groups emitted similar numbers of rears, indicating little effect of ethanol withdrawal in this type of exploratory activity (Rodgers et al. 1999); however, a small increase in numbers of rears was seen in RWD, compared to SWD animals, in open arms. The number of head dips from open arms was lower in the both RWD and SWD groups than controls. Surprisingly, there was no significant difference between groups in the number of risk assessments; number of grooming episodes was also similar in all groups.

In summary then, whilst withdrawal from chronic treatment with ethanol gave rise to anxious-like behaviour in the plus-maze, there was no evidence that rats that had previously experienced withdrawal show heightened levels of anxiety, compared to SWD rats, even though this regimen of repeated withdrawal from alcohol sensitises rats to seizures (Stephens et al. 2001). This conclusion also fits with our clinical observations that withdrawal-associated anxiety is not increased following multiple withdrawals (Duka et al. 2002).

Nevertheless, the results obtained here using a 7% ethanol diet differ from those of Overstreet et al. (2002, 2004a,b) when lower ethanol concentrations were given. In this latter case, an increase in anxiety-like behaviour after repeated withdrawal was found. There are several possible accounts of these discrepancies. Firstly, differences may be a result of different treatment protocols, as levels of intake induced by Overstreet and colleagues (Overstreet et al. 2002) were lower than those that we achieve. Indeed, when a liquid diet regimen using 7% ethanol concentrations was used (Overstreet et al. 2002), whilst withdrawal increased anxiety relative to controls, no differences were seen between SWD and RWD groups. Secondly, there is a possibility that the elevated plus-maze and the social interaction test model different aspects of anxiety (File 1992), and repeated withdrawal from ethanol affects one of these aspects more than the other. Thirdly, in the experiments of Overstreet, Knapp and Breese, RWD animals not only exhibited suppressed social interaction compared with controls and SWD animals, which was considered as an indication of increased anxiety, but simultaneously ethanol-withdrawn animals showed a decrease in locomotion (Overstreet et al. 2002, 2003, 2004a,b), a common observation during ethanol withdrawal (File et al. 1993; Gatch et al. 1999; 2000; Watson et al. 1997; Wilson et al. 1998). Reductions in activity may give rise to a false positive result in the social interaction test (File and Seth 2003; Kliethermes 2005; Koss et al. 2004), though by including the reduced locomotion as a factor in an analysis of covariance, Breese, Overstreet and colleagues attempted to control for this aspect. Although other groups have reported increased anxiety-like behaviour after repeated withdrawal from ethanol (Ghozland et al. 2005; Holter et al. 1998). Holter et al. (1998) used a much longer period of ethanol exposure, while Ghozland et al. (2005) did not control for length of exposure to ethanol, so that it cannot be interpreted whether the increased anxiety-like profile is the result of multiple withdrawals or of a more prolonged ethanol exposure.

In addition to behavioural indices of anxiety, we also investigated the effect of withdrawal on blood levels of the stress-related hormone corticosterone (Herman and Cullinan 1997). In keeping with the aversive and stressful nature of withdrawal and as has been reported by others (Janis et al. 1998; Tabakoff et al. 1978), corticosterone levels were increased in SWD animals. However, there was no evidence for increased circulating levels of corticosterone in RWD rats. The failure of withdrawal to increase corticosterone levels in RWD rats suggests that the withdrawal experience was less stressful in this group. These results fit with more general observations of habituation to repeated stress (Belda et al. 2004; Carter et al. 2004; Chen and Herbert 1995; Martinez et al. 1998; Melia et al. 1994; Senba and Ueyama 1997; Stamp and Herbert 1999; Umemoto et al. 1997).

Withdrawal from ethanol in RWD rats (but not in SWD animals) led to a significant increase in the number of neurons expressing Fos in the central and the basolateral nuclei of amygdala, in the CA3 field of hippocampus, in the core of nucleus accumbens and in the dorsolateral PAG. Surprisingly, no evidence was seen of withdrawal-induced activation of the PVN. Fos expression in dCPu was also not changed. Zif268 expression was increased only in the central nucleus of amygdala of the SWD group. In all other structures studied, level of zif268 expression after both RWD and SWD was similar to that in the control group.

Fos expression increases as a result of neuronal activation, and Fos is widely used for monitoring the areas of the brain activated as a result of different manipulations (Chen and Herbert 1995; Dunworth et al. 2000; Herdegen and Leah 1998). Surprisingly, a single withdrawal did not give rise to reliable increases in Fos expression in any area, though trends were observed. We used an immunohistochemical method to visualise Fos protein, and the time course of expression therefore reflects the time required for the translation and transcription processes. We have chosen 8 h after withdrawal of ethanol as a time point for all our measures on the basis of data showing that blood alcohol levels are similar immediately after withdrawal from SWD and RWD treatment (Stephens et al. 2001), the rate of reduction of blood alcohol levels after withdrawal of 7% liquid diet treatment does not differ between SWD and RWD rats, and elimination is largely complete at 4 h (Overstreet et al. 2002), and correspondingly, this is a time at which withdrawal of rats from liquid diet gives rise to several withdrawal signs that are both intense and stable over a few hours (Macey et al. 1996; Rassnick et al. 1992). Perhaps a later time point may have allowed Fos accumulation to have occurred even in the SWD group. Nevertheless, the numbers of neurons expressing Fos was reliably increased in several areas of the RWD group, reflecting the expected increases in excitability following sensitisation, or “kindling” of withdrawal that results in increased seizure sensitivity (Ballenger and Post 1978; Becker and Hale 1993; Brown et al. 1988; Duka et al. 2004; Lechtenberg and Worner 1991; Mhatre and Gonzalez 1999; Pinel et al. 1975; Veatch and Becker 2002). Areas showing clear effects of repeated withdrawal included CeA and BLA nuclei of amygdala, n. accumbens core, CA3 field of hippocampus, and the periaqueductal grey of the midbrain, though other areas showed similar, if non-significant trends. Thus, although we found no evidence that anxiety-inducing effects of withdrawal increased with RWD, there was clear evidence for increased withdrawal-induced neuronal activation in several brain areas. Such increases may thus reflect the kindling of seizure activity induced by repeated withdrawal from ethanol.

In contrast, expression of another immediate early gene, zif268, was generally insensitive to withdrawal. Nevertheless, zif268 showed a reliable increase in the central amygdala after single withdrawal but failed to increase after repeated withdrawal. This pattern of zif268 expression in CeA parallels the corticosterone effects. Zif268 has been suggested to be a marker of synaptic plasticity rather than neuronal activity per se (Cole et al. 1989; Jones et al. 2001; Knapska and Kaczmarek 2004; Lee et al. 2004; Thomas et al. 2003). We have recently reported that withdrawal from ethanol results in reduced capacity of rat hippocampus and lateral amygdala to display long-term potentiation, and that, at least in hippocampus, the extent of this deficit increases with increased numbers of withdrawals (Stephens et al. 2005). The pattern of zif268 expression in CeA, showing an increase following a single, but not repeated withdrawal, may thus reflect withdrawal-induced plasticity after a single withdrawal (withdrawal non-specifically strengthens weak synapses) but reduced capacity for plasticity following multiple withdrawals. That is, following several withdrawals as a result of previous strengthening, the pool of synapses available for strengthening with future learning may decrease, and the room for further plasticity narrows, thus reducing the capacity for further LTP, as seen in the previous study. These neuronal processes in the amygdala may constitute the basis for impaired fear conditioning associated with multiple withdrawals observed in rats (Ripley et al. 2003b; Stephens et al. 2001) and human binge drinkers (Duka et al. 2004; Stephens et al. 2005).

This suggestion that the absence of an increase in zif268 expression after RWD reflects impaired plasticity is weakened by our surprising failure to find reliable effects of withdrawal on zif268 expression in other brain areas. However, in certain areas such as hippocampus CA1 and lateral amygdala, zif268 expression was already high in the control animals, so that further increases induced by withdrawal may have been obscured. As with c-fos, we also do not know whether immunohistochemistry for zif268 was performed at the optimal withdrawal time.

In conclusion, the present report demonstrates that whilst withdrawal from chronic ethanol treatment gives rise to anxiety-like behaviour in the elevated plus-maze, increases corticosterone blood level, and zif268 expression in central amygdala, previous experience of withdrawal from chronic ethanol treatment does not exacerbate withdrawal-induced anxiety, protects against increases in corticosterone level, possibly as a result of habituation to ethanol withdrawal-induced stress, and simultaneously gives rise to increased neuronal activity as indicated by c-fos expression in amygdala, hippocampus, nucleus accumbens and dorsal PAG. This complex pattern indicates there is no simple relationship between behavioural measures thought to reflect anxiety (such as the plus-maze), hormonal stress responses and activation of brain areas involved in mediating responses to fearful stimuli.


Supported by MRC Programme Grant G9806260

Copyright information

© Springer-Verlag 2006