Ethanol withdrawal and proclivity are inversely related in rats selectively bred for differential saccharin intake

Alcohol 37 (2005) 9–22

Ethanol withdrawal and proclivity are inversely related in rats selectively bred for differential saccharin intake Nancy K. Dess*, Patricia O’Neill, Clinton D. Chapman Occidental College, 1600 Campus Road, Los Angeles, CA 90041, USA Received 22 July 2005; received in revised form 29 September 2005; accepted 29 September 2005

Abstract Withdrawal severity and voluntary alcohol consumption are inversely related in rats and mice. The present study demonstrated this empirical relation and extended it in two ways. First, the rats were selectively bred for low (LoS) and high (HiS) saccharin intake, a phenotype that correlates positively with ethanol intake and inversely with emotional reactivity. Withdrawal has not yet been studied in these rats. Second, proclivity to consume ethanol was measured as conditioned preference for an ethanol-paired flavor. After 2 weeks of forced exposure to ethanol and a period of abstinence, LoS rats showed elevated acoustic startle; HiS rats did not (Exp. 1). When ethanol- and no-ethanol solutions were available freely during conditioning, both LoS and HiS rats preferred a flavor paired with 4% ethanol, but only HiS rats preferred a flavor paired with 10% ethanol (Exp. 2A); when exposure to the two solutions was controlled, all groups except LoS males preferred flavors paired with 4% or 10% ethanol (Exp. 2B). Thus, as predicted, withdrawal was more severe in the line with less ethanol proclivity (LoS). These results implicate basic associative and affective processes in individual differences in patterns of alcohol use. Ó 2005 Elsevier Inc. All rights reserved. Keywords: Saccharin; Withdrawal; Ethanol proclivity; Conditioned flavor preference; Rats

1. Introduction Regular consumption of alcoholic beverages is the culmination of many processes, from those associated with initial ethanol exposure to those that maintain and modulate drinking behavior. A striking feature of human ethanol consumption is the diversity of behavioral trajectories through which these processes reveal themselves. People differ in how early and the extent to which they seek alcohol, how they respond to initial experiences with it, which beverages they prefer, and whether they can moderate consumption after developing a ‘‘taste’’ for alcohol. Clearly, multiple pathways to alcohol use exist. One attempt to identify these pathways is Cloninger’s (1987) distinction between Type 1 and Type 2 alcoholism. In terms of underlying biobehavioral processes, Type 1 alcoholism derives from high harm avoidance and loss of control accompanied by low novelty seeking, whereas Type 2 alcoholism derives from high novelty seeking accompanied by low harm avoidance. The two types can be thought of as

* Corresponding author. Department of Psychology, Occidental College, 1600 Campus Road, Los Angeles, CA 90041, USA. Tel.: 11-323259-2570; fax: 11-323-341-4887. E-mail address: [email protected] (N.K. Dess). 0741-8329/05/$ – see front matter Ó 2005 Elsevier Inc. All rights reserved. doi: 10.1016/j.alcohol.2005.09.006

reflecting ‘‘compulsiveness’’ versus ‘‘impulsiveness,’’ respectively (Cloninger, 1996). A key premise of this approach is that the mechanisms of individual differences in ethanol consumption are not unique to ethanol but rather comprise behavioral systems that modulate responses to diverse stimuli and situations. The particulars of Cloninger’s theory are not examined here, but this key premise underlies our reasoning. Rodent models have been useful in elucidating the mechanisms of individual differences in and phenotypic correlates of ethanol consumption. Stable differences in voluntary ethanol consumption have been observed in many rat and mouse strains (e.g., Adams et al., 1991; Belknap et al., 1993; Eriksson, 1968; Fuller, 1974; Li et al., 1987; McClearn & Rodgers, 1959; Sinclair et al., 1992). Animals with higher voluntary ethanol intake also generally consume more saccharin (Kampov-Polevoy et al., 1995; Sinclair et al., 1992; but see Goodwin et al., 2000). Because saccharin lacks the calories and pharmacological effects of ethanol, this positive correlation likely is mediated by differential responses to taste; specifically, animals that consume more ethanol and saccharin may respond more positively to these solutions’ sweet taste and/or less negatively to their bitterness (Kiefer & Lawrence, 1988; Kiefer & Mahadevan, 1993).

10

N.K. Dess et al. / Alcohol 37 (2005) 9–22

Were saccharin intake the only reliable correlate of voluntary ethanol consumption, the relationship would not be very interesting vis a` vis human alcohol use. Clearly, human drinking is not compelled simply by the availability of tasty libations. However, the association of ethanol and saccharin intake with noningestive phenotypes suggests that taste does more than mediate initial consumption of solutions. Taste also appears to serve as a marker for affective processes that can influence long-term patterns of alcohol use. Though ‘‘emotionality’’ is difficult to characterize and results are not entirely consistent, rat strains with relatively low voluntary alcohol consumption generally score higher on measures of emotionality, such as open field defecation, acoustic startle, than do their higher-consuming counterparts (Badistov et al., 1995; Moeller et al., 1997; Overstreet et al., 1997; but see McKinzie et al., 2000, on alcohol-preferring and alcohol-non-preferring rats). 1.1. Experiment 1: withdrawal severity A primary goal of Experiment 1 was to compare highsaccharin-consuming (HiS) and low-saccharin-consuming (LoS) rats on the severity of ethanol withdrawal. Rats received only water or only 11% ethanol solution for 2 weeks. We previously found no difference in forced-choice ethanol intake between LoS and HiS females (Dess et al., 1998); assuming replication of that finding, this method would allow examination of withdrawal unconfounded by differential exposure to ethanol. Forty-eight hours after ethanol removal, all rats received an acoustic startle test. During withdrawal, startle amplitude increases (van Erp & Miczek, 2001; Macey et al., 1996; Pohorecky et al., 1976). Withdrawal generally is more severe among animals that consume less ethanol voluntarily (Chester et al., 2003; Metten et al., 1998). Because voluntary ethanol consumption is lower among LoS rats (Dess et al., 1998), they were expected to show more evidence of withdrawal than HiS rats. This experiment also served as a preliminary inquiry into three other issues. First, ethanol is a source of calories. Much of our thinking about the pattern of phenotypic correlates in LoS and HiS rats relates to energy regulation (Dess, 1991; VanderWeele et al., 2002). Thus, we were interested in any differential effects of ethanol availability on caloric intake and weight gain in HiS versus LoS rats. Second, dietary attenuation of ethanol consumption and withdrawal severity was of interest. Two candidate substances for reducing intake and withdrawal are sugar and caffeine. Sugar has addictive potential (Avena et al., 2005; Colantuoni et al., 2002), and access to sugar or other sweet solutions reduces self-administration of ethanol (Lankford & Myers, 1994; Kampov-Polevoy et al., 1995). Moreover, both sugar and caffeine are used more during withdrawal (Aubin, 1999; Ayers, 1976; Verinis, 1986), perhaps as an attempt at ‘self-medication’ to alleviate stress or cravings (Gatch & Selvig, 2002; Heinz et al., 2003; Jain

et al., 2004; Miller & Goldsmith, 2001; Minor & Saade, 1994; Minor et al., 2001). During the abstinence period, some rats received a choice between a sugar/caffeine solution and a pharmacologically inert bittersweet solution. The impact of access to those solutions on withdrawal symptoms and ethanol intake was assessed. Finally, in addition to overall startle magnitude, prepulse inhibition was measured. Whereas overall startle magnitude reflects affective state, prepulse inhibition reflects the cognitive process of sensory-gating, that is, how effectively the processing of one stimulus (the prepulse) modifies processing of a subsequent one (the startle stimulus; Dawson et al., 1999). In the present study, acoustic startle testing included trials with and without a prepulse stimulus, so that any differential impact of ethanol withdrawal on prepulse inhibition in LoS and HiS rats could be observed. In one study, prepulse inhibition was suppressed during ethanol withdrawal (Rassnick et al., 1992). However, others have reported that prepulse inhibition also is suppressed by ethanol intoxication in humans and rats (Hutchinson et al., 2003; Jones et al., 2000). Impairment of prepulse inhibition by both ethanol and withdrawal from ethanol, and the general paucity of studies on these effects, makes predicting results for LoS and HiS rats difficult. This aspect of the study, therefore, was entirely exploratory. 2. Materials and methods 2.1. Rats Adult female rats (60–80 days old) from Generations 25 and 26 of the Occidental LoS and HiS lines were used. These rats have been selectively bred bidirectionally for low versus high consumption of saccharin (see Dess & Minor, 1996, for details). Selection is made on the basis of consumption of 0.1% saccharin solution in a 24-h, two-bottle test; the selection phenotype is saccharin consumption relative to a predetermined water intake baseline and body weight (saccharin ml 2 water baseline ml, divided by body weight, 3 100). The line difference in saccharin intake is large and highly reliable; neither baseline water intake nor body weight differs consistently between lines. Group sizes, baseline water intake and body weights, and phenotype scores are shown in Table 1. Rats were housed individually in stainless steel hanging cages on a 12:12 h light/dark schedule, with lights on at 0700. Purina 5001 Rodent Chow was freely available throughout the experiment. In this and the following experiments, humane care and use of the rats were assured through compliance with an Institutional Care and use Committee-approved protocol and an institutional Public Health Service Animal Welfare Assurance. 2.2. Materials and apparatus Ethanol solution was 11% (vol/vol). Two bittersweet solutions were used in the withdrawal phase, one containing

N.K. Dess et al. / Alcohol 37 (2005) 9–22 Table 1 Group sizes, mean baseline body weight, water intake, calorie intake, and phenotype score (6standard deviation) for female LoS and HiS rats in Experiment 1 LoS

HiS

n 5 41

Body weight (g) Water intake (% bwt) Caloric intake (cal/day) Saccharin phenotype (D%)

n 5 40

Mean

SD

Mean

SD

290.2 15.3 77.7 7.8

39.3 4.4 8.6 8.5

277.0 14.2 71.2 49.5

29.5 2.2 12.6 12.8

11

and the startle chamber was swabbed with a 5% solution of ammonium hydroxide between rats. The fifth and final phase was an ethanol reinstatement test. Immediately upon return to the home cage after the startle test, each rat was given a preweighed bottle of 11% ethanol and a preweighed bottle of water. The bottles were removed and weighed approximately 24 h later. At least 4 days after the end of the experiment, all rats were given a 24-h, two-bottle test with 0.1% saccharin versus tap water to confirm the selection phenotype.

LoS, low-saccharin-consuming; HiS, high-saccharin-consuming.

2.4. Statistical analyses sucrose (15 g/l) and caffeine (500 mg/l) and one containing sucralose (2 g/l; Splenda, McNeil-PPC, Ft. Washington PA) and quinine hydrochloride (10 mg/l). In a pilot study, HiS rats consumed roughly equal amounts of these solutions in a two-bottle test. All solutions were made with tap water. Solutions were presented in glass bottles with stainless steel spouts and rubber stoppers. Startle testing was conducted in a startle chamber with a piezoelectric sensor and digital display of platform force in arbitrary units (San Diego Instruments, San Diego CA). The startle stimulus was a 40-ms, 95-dB burst of white noise on a 60-dB masking noise background. The prepulse stimulus was a 40-ms, 65-dB burst of white noise that preceded the startle stimulus by 100-ms.

2.3. Procedure The experiment had five phases. Phase 1 was a 2-day baseline assessment phase during which chow intake, water intake, and body weight were measured. In Phase 2, half of the HiS rats and half of the LoS rats were given only ethanol solution to drink for 14 days; the remaining rats remained on tap water. Phase 3 was the ethanol withdrawal phase. During this phase, half of the rats in each group were given a choice between sucrose/caffeine (Suc/Caff) and sucralose/quinine (Sx/Q) to drink, whereas the others were given water. These procedures comprised a 2 3 2 3 2 betweengroups design, with line (HiS vs. LoS), forced ethanol exposure (EtOH vs. No EtOH), and withdrawal solutions (Choice vs. No Choice) as variables. Thus, eight experimental groups were formed, each with 10 rats except for the LoS/EtOH/Choice group, which had 11 rats. Forty-eight hours after ethanol was withdrawn, all rats were given an acoustic startle test (Phase 4). A rat was placed in the startle chamber for 3 min before the first startle trial. Thirty startle trials were run with an intertrial interval of 30 s. Fifteen trials were regular startle trials, and 15 trials were prepulse trials. Regular and prepulse trials were presented in one of two quasirandom orders. Rats from the eight experimental groups were run in quasirandom order,

Between-group and mixed-design analyses of variance (ANOVAs) with Greenhouse–Geisser correction (for repeated measures with df O 1) were followed up with Bonferroni contrasts for post hoc comparisons and, to test the prediction of greater withdrawal in LoS than HiS rats, an a priori contrast using the error term from the ANOVA. All effects and contrasts significant at a 5 .05 are reported in the text.

3. Results 3.1. Phase 1: baseline A line 3 ethanol exposure 3 choice/no-choice ANOVA on each baseline measure confirmed the comparability of experimental groups (EtOH/No EtOH, Choice/No Choice). Water intake and body weight did not differ between HiS and LoS rats. However, baseline calorie intake was lower among HiS than LoS rats [71.2 vs. 77.7 cal/day; F(1, 73) 5 7.13].

3.2. Phase 2: ethanol exposure Data were averaged into seven 2-day blocks for analysis, and total caloric intake and body weight were assessed as changes from baseline values using a line 3 ethanol 3 block mixed-design ANOVA. Caloric intake did not differ from baseline levels or between groups. LoS rats gained more weight than did HiS rats [19.4 vs. 16.0 g, F(1, 77) 5 3.82]. Rats given water to drink gained more weight [19.4 vs. 16.1 g, F(1, 77) 5 9.05], at a faster rate [solution 3 block interaction, F(6, 462) 5 4.20] than did those given ethanol. LoS and HiS rats were not differentially affected by access to ethanol. Ethanol dose consumed daily was examined in the EtOH groups using a line 3 block mixed-design ANOVA. Dose averaged 7.8 g/kg/day overall. It was highest in the first 2-day block (8.5 g/kg/day), then declined to a relatively stable level of 7.7 g/kg/day [block effect, F(6, 228) 5 2.76]. HiS rats and LoS rats consumed the same dose of ethanol.

12

N.K. Dess et al. / Alcohol 37 (2005) 9–22

3.3. Phase 3: bittersweet solutions during withdrawal Consumption of Suc/Caff and Sx/Q solutions was analyzed in three line 3 ethanol 3 day mixed-design ANOVAs. (Groups given only water during this phase are excluded.) The first ANOVA was conducted on preference for Suc/Caff over Sx/Q (Suc/Caff consumed/total consumed). Consistent with the pilot study, HiS rats had no preference or aversion for Suc/Caff; LoS rats, on the other hand, were averse to it [0.55 vs. 0.27, F(1, 37) 5 11.92]. In addition, Suc/Caff preference decreased from the first to the second day [0.46 vs. 0.34, F(1, 37) 5 10.16]. The other two ANOVAs were conducted on caffeine dose (mg/kg/day) and grams of sugar (g/day) consumed each day. Relative to LoS rats, HiS rats consumed larger doses of caffeine [33.8 vs. 21.4 mg/kg/day; F(1, 37) 5 7.38] and more sugar (0.3 vs. 0.2 g; F(1, 37) 5 5.41]. Caffeine dose and sugar intake did not differ significantly between the 2 days.

3.4. Phase 4: acoustic startle testing Startle amplitude habituated over trials [F(14, 1008) 5 9.02], less so for prepulse trials [trial type 3 block interaction, F(14, 1008) 5 2.28]. Because these effects involving trial block were mundane and did not interact with variables of primary interest (line and ethanol exposure), means shown in Fig. 1 are averaged across all trials. Startle amplitude was greater on trials without a prepulse stimulus [i.e., prepulse inhibition; F(1, 72) 5 41.14] and was greater among LoS than HiS rats, F(1, 72) 5 7.00. Prior ethanol exposure/withdrawal differentially affected LoS and HiS rats in two ways. First and most importantly, among LoS rats, those withdrawn from ethanol startled more than did noethanol rats. In striking contrast, HiS rats withdrawn from ethanol startled slightly less than did their no-ethanol HiS counterparts. The line 3 ethanol exposure interaction was significant, F(1, 72) 5 4.31; an a priori contrast comparing the withdrawal effect in LoS rats (LoS EtOH 2 No EtOH) to the effect in HiS rats (HiS EtOH 2 No EtOH) was significant, t(72) 5 2.94. The second differential effect of ethanol exposure concerned the magnitude of prepulse inhibition. LoS rats withdrawn from ethanol showed greater prepulse inhibition than did water-only controls. Among HiS rats, prepulse inhibition did not differ between ethanol-exposed rats and water-only controls. The line 3 ethanol exposure 3 trial type interaction was significant, F(1, 72) 5 4.59. Contrasts on prepulse inhibition scores (regular 2 prepulse) showed that EtOH and No-EtOH LoS groups differed from each other, whereas the EtOH and No-EtOH HiS groups did not. As noted above, LoS rats gained more weight during the 2-week water or ethanol exposure period. Body weight differences, however, cannot account for the differential startle amplitude of LoS and HiS rats as a function of ethanol exposure and trial type for two reasons. First, the line

Fig. 1. Mean acoustic startle amplitude (arbitrary units, 6standard error) among low-saccharin-consuming rats (upper panel) and high-saccharinconsuming rats (lower panel) after weeks of exposure to water (No Ethanol) or 11% ethanol (Ethanol Exposed) in Experiment 1.

difference in body weight did not differ as a function of ethanol exposure [line 3 ethanol exposure interaction not significant, F(1, 77) 5 0.14]. Second, the line 3 ethanol exposure interaction was significant even when prestartle body weight was statistically controlled by using it as a covariate in the ANOVA [F(1, 71) 5 5.04 vs. F(1, 72) 5 4.31 without the covariate]. It should also be noted that in another study, LoS female rats hyperstartled despite being slightly lighter than HiS female rats, and the much heavier males did not startle more than females (Dess et al., 2000). Availability of the bittersweet solutions had no significant effect on startle amplitude (data not shown). 3.5. Phase 5: ethanol reinstatement test Preference for ethanol in the final 24-h two-bottle test is shown in Fig. 2. Overall, ethanol preference was lower among LoS rats than HiS [line effect, F(1, 72) 5 5.30] and among rats given access to bittersweet solutions [choice/no-choice effect, F(1, 72) 5 5.13]. However, both these differences were eliminated by ethanol exposure: The line 3 ethanol exposure and choice 3 ethanol exposure interactions were significant, Fs(1, 72) 5 5.53 and 4.80, respectively; contrasts showed that LoS rats drank less than HiS rats, and that Choice rats drank less than No Choice rats, only when they were ethanol naı¨ve.

N.K. Dess et al. / Alcohol 37 (2005) 9–22

13

Fig. 2. Mean preference score (6standard error) for 11% ethanol over water among low-saccharin-consuming and high-saccharin-consuming rats previously maintained on water (left) or on 11% ethanol (right). In the 48 h preceding this test, half of each group had access to a sucrose/caffeine and a sucralose/ quinine solution (Caff/Suc) and half had only water.

Analysis of ethanol dose yielded a similar pattern of results. Overall, a smaller dose was consumed by LoS rats (2.5 g/kg vs. 4.3 g/kg for HiS) and by rats given the bittersweet solutions (2.3 g/kg vs. 4.5 g/kg for No Choice rats) [line and choice/no choice effects, Fs(1, 72) 5 7.45 and 11.07, respectively]. Prior ethanol exposure/withdrawal eliminated the difference between Choice and No Choice rats [choice 3 ethanol exposure interaction, F(1, 72) 5 6.07; comparison of choice/no choice groups significant only in the ethanol-naı¨ve condition]. Ethanol exposure/ withdrawal also tended to reduce the line difference, but the line 3 ethanol exposure interaction was only marginally significant, F(1, 72) 5 3.52, P 5 .06. Finally, access to the bittersweet solutions eliminated the line difference [line 3 choice interaction, F(1, 72) 5 4.94; comparison of lines significant only in the no choice condition]. 4. Experiment 1 discussion At 48 h after removal of ethanol, only LoS rats displayed withdrawal, despite having consumed a dose of ethanol comparable to that consumed by HiS rats (as previously reported for a one-bottle, forced exposure test; Dess et al., 1998). This finding demonstrates in a new population of rats the inverse relationship between voluntary alcohol consumption and withdrawal severity reported in the literature. One possible mediator of heightened withdrawal among LoS rats is corticosterone, as it increases withdrawal severity (Roberts et al., 1994), and LoS are hypercorticosteronemic (VanderWeele et al., 2002). The absence of withdrawal symptoms among HiS rats does not necessarily mean that they experienced no withdrawal: A relatively long abstinence period (Macey et al., 1996; Metten et al., 1998; Pohorecky et al., 1976) preceded the test, and HiS rats might have shown withdrawal symptoms if tested sooner. Indeed, peak withdrawal could be as intense among HiS rats as it is among LoS rats but diminish

faster (van Erp & Miczek, 2001). Another design is needed to map the time course of withdrawal, such that the recruitment, diminution, and peak intensity of symptoms can be examined. The present study does show that withdrawal persists among LoS rats for at least 48 h, at which time it is not present among HiS rats. The ethanol-exposure method employed here – access to only ethanol solution – has advantages and disadvantages relative to other methods (Cicero, 1979). For instance, compared to ethanol vapor exposure, allowing rats to drink ethanol better preserves the behavioral and metabolic sequelae of human alcohol consumption. A potential drawback this method shares with liquid-diet ethanol administration is the development of nutritional differences from controls that can complicate interpretation of group differences. It seems unlikely that nutritional differences (or hydration) account for the line difference in withdrawal reported here. First, LoS and HiS rats did not drink different amounts of ethanol, and ethanol exposure did not differentially affect weight gain in LoS and HiS rats. Second, the ethanol exposure phase was shorter than the ‘‘few weeks to 5 to 6 months,’’ referenced by Cicero (1979) in connection with nutritional deficiencies. Finally, all rats had free access to food and water for approximately 48 h prior to withdrawal testing, a period sufficient for rehydration and probably to restore any modest nutritional differences that may have begun to emerge. Nonetheless, replicating the line difference in withdrawal with another ethanol-exposure method would be informative. Overall startle amplitude was higher among LoS rats than HiS rats. Whereas the main effect of line appears to replicate an earlier finding (Dess et al., 2000), inspection of Fig. 1 indicates virtually no line difference among ethanol-naı¨ve rats. This is not, however, a direct failure to replicate the earlier study. In that study, rats received more startle trials, all with no prepulse trials. With 20–30 identical trials, habituation proceeded quickly among HiS rats,

14

N.K. Dess et al. / Alcohol 37 (2005) 9–22

and the consistently higher startle scores of LoS on later trials contributed most heavily to the overall difference between lines. Thus, in the present study, using only 15 regular startle trials intermixed with prepulse trials probably precluded observation of a line difference in the ethanol-naı¨ve groups. Prior ethanol exposure augmented prepulse inhibition among LoS rats and slightly suppressed it among HiS rats. One explanation for this difference appeals to the time course of withdrawal, that is, LoS rats are still in withdrawal and thus show augmentation, whereas HiS rats are more completely recovered and thus show prepulse inhibition that is normal or even suppressed (‘‘rebound’’). Alternatively, this finding could reflect qualitative differences in the impact of ethanol exposure. This interpretation is more consistent with the observation in another study that rats tested near peak withdrawal times tend to show reduced prepulse inhibition, as do HiS rats (Rassnick et al., 1992). Grillon et al. (2000, 1994) have reported that nonalcoholic people with a family history of alcoholism show generally lower acoustic startle and reduced prepulse inhibition relative to people with a family history of anxiety disorders; this pattern was observed in both the presence and absence of alcohol. Possible mechanisms for enhanced prepulse inhibition among ethanol-exposed LoS rats and humans whom they model include reduced gamma-aminobutyric acid (Fendt, 1999) or dopamine (Swerdlow et al., 2005; Zhang et al., 2000) activity. Additional work is necessary to understand the nature of this intriguing line difference in connection with alcohol. Access to ethanol after the startle test constituted the first exposure for No Ethanol groups and a reinstatement procedure for Ethanol groups. Among the No Ethanol groups, HiS rats preferred ethanol more strongly and drank a larger dose of it than did LoS rats, replicating prior results from two-bottle tests (Dess et al., 1998). After 2 weeks of ethanol consumption, however, LoS rats behaved like HiS rats. Determining whether this convergence of lines results directly from ethanol exposure or is secondary to differential withdrawal requires a design including ethanol-exposed rats tested prior to any period of abstinence. Preliminary exploration of effects of access to bittersweet solutions after ethanol exposure yielded some interesting results. Whereas it had no impact on startle, it did affect alcohol preference and dose in the two-bottle reinstatement test. Specifically, it reduced alcohol consumption among animals (a) naı¨ve to alcohol or (b) prone to high voluntary alcohol consumption in the first place, that is, HiS rats. HiS rats drank more sucrose and caffeine, which could account for the greater impact on their alcohol consumption and preference. However, the self-administration choice procedure used here does not permit conclusions as to whether these results are attributable to pre- or postabsorptive effects of sucrose and/or caffeine per se or more generally to the opportunity to drink any kind of (bitter)sweet solutions. The potential utility of this kind of dietary

intervention in prophylactically reducing alcohol consumption by individuals with a high proclivity to drink may warrant further exploration (Colby-Morley, 1982). 5. Experiment 2: conditioned flavor preference Experiment 2 concerned the role of initial experience with ethanol on proclivity to prefer alcoholic beverages. Rodents’ proclivity to consume ethanol is most commonly assessed by giving them ethanol in water and measuring how much they drink, either as volume, dose, or preference relative to water. Although a valuable tool, such a test does not distinguish between unconditioned responses to ethanol and learning from experience with ethanol. Learning, such as habituation of neophobia and conditioned preferences or aversions for ethanol-associated flavors, likely plays a role in the proclivity toward sustained ethanol consumption. Counterconditioning to ethanol’s aversive taste properties may be particularly important because, unlike flavors unique to specific beverages, conditioning to ethanol’s flavor would generalize across alcoholic beverages. We developed a simple protocol for flavor preference conditioning based on the work of Cunningham and Neihus (1997) and Mehiel and Bolles (1984). In each of four experiments, male and female HiS and LoS rats were exposed overnight to two distinctively flavored solutions, five times each, in strict alternation. One solution contained ethanol and one did not. Later, a two-bottle flavor preference test was conducted. In Experiment 2A, ethanol concentration was manipulated (4% and 10%) in two successive direct replications. Low concentrations of ethanol are acceptable or palatable, and concentrations above 6% are rejected and unpalatable (Dess et al., 1998). Manipulating concentration provided a means of exploring the relationship between amount of solution and dose of ethanol consumed during conditioning and flavor preference conditioning. In Experiment 2A, rats generally did consume different amounts of ethanol and no-ethanol conditioning solutions, raising the possibility that the relative novelty of the two flavors influenced preferences expressed in the test. Given the evidence of LoS rats’ greater reactivity to environmental change of many sorts, they might be especially likely to avoid a flavor to which they had limited their exposure during conditioning. In Experiment 2B, consumption of the two conditioning solutions was equated for each rat, allowing evaluation of the role of familiarity in later flavor choices by HiS and LoS rats. 6. Method 6.1. Rats Female and male HiS and LoS rats (60–80 days of age) from Generations 27 and 29 were used. Sample sizes, baseline water intake and body weights, and phenotype scores for Experiments 2A and 2B are shown in Table 2.

N.K. Dess et al. / Alcohol 37 (2005) 9–22

15

Table 2 Group sizes, mean baseline body weight and water intake, and phenotype score (6standard deviation) in Experiments 2A and 2B Body weight (g)

Experiment 2A LoS females LoS males HiS females HiS males Experiment 2B LoS females LoS males HiS females HiS males

Water (% bwt)

Phenotype (D%)

n

Mean

SD

Mean

SD

Mean

SD

47 47 47 48

295.2 442.9 277.2 443.4

31.0 43.4 33.9 62.7

15.0 13.2 15.8 11.5

2.5 2.1 4.5 2.6

7.9 3.8 54.0 32.2

11.4 6.0 13.4 10.7

22 19 22 19

307.4 413.7 301.9 392.3

31.2 43.3 44.1 54.7

14.2 13.2 14.1 12.9

2.1 3.0 1.9 1.4

12.5 7.4 49.9 34.6

11.2 13.5 16.0 11.6

LoS, low-saccharin-consuming; HiS, high-saccharin-consuming.

Rats were housed individually in stainless steel hanging cages on a 12:12 h light/dark schedule, with lights on at 0700. Purina 5001 Rodent Chow was freely available. 6.2. Materials Six KoolAidÒ solutions (2.5 g/l) were used during training: cherry and grape with 4% ethanol (vol/vol), cherry and grape with 10% ethanol (vol/vol), and cherry and grape with no ethanol, each with 1 g/1 sucralose to increase acceptability. Test solutions were plain cherry and grape KoolAidÒ (2.5 g/l). All solutions were made with tap water. At the end of each experiment, saccharin testing was done with sodium saccharin (Sigma Chemical Co., St. Louis, MO; 1 g/l). 6.3. Procedure Average water intake and body weight were measured in a 2-day baseline period. All rats then received five exposures to a cherry solution and five exposures to a grape solution in strict alternation, for a total of 10 conditioning trials. One solution contained ethanol (4% or 10%, vol/ vol) and the other did not; which flavor was paired with ethanol and which was presented first were balanced. Thus, half of the rats received cherry–ethanol solution and grape–no-ethanol solution, and the other rats received grape–ethanol solution and cherry–no-ethanol solution. In each of those groups, half of the rats received grape on the first trial, and the others received cherry on the first trial. For each of the 10 conditioning trials, each rat’s water bottle was removed at approximately 1600 and replaced with a preweighed bottle of flavored solution. The bottles were removed and weighed, and water bottles were returned to the cages the next morning at approximately 0800. After the last conditioning trial, water was available for 2 days. A two-bottle flavor preference test was then conducted. At approximately 1600, each rat received a bottle containing cherry solution and a bottle containing grape

solution. The bottles were removed and weighed, and water bottles were returned to the cages the next morning at approximately 0800. At least 3 days after the completion of preference testing, all rats received a 24-h two-bottle test with 0.1% saccharin solution and water to confirm the selection phenotype. In Experiment 2A, rats had unlimited access to ethanol solution and no-ethanol solution on all conditioning trials. This experiment was conducted as two direct replications. Replication 1 consisted of an experiment with 4% ethanol solution (ns 5 10) followed by an experiment with 10% ethanol solution (ns 5 17–18). Replication 2 consisted of an experiment in which rats were randomly assigned to either a 4% or a 10% ethanol solution condition (ns 5 9–11). Experiment 2B was the same as Experiment 2A with one difference: During conditioning, consumption of ethanol and no-ethanol solutions was matched for each rat on a trial-by-trial basis. Matching was accomplished by alternating free access to the less palatable of the two solutions with yoked rations of the other solution. A ration equaled the volume of solution the rat had consumed freely the previous night. Because the rationed solution was the more palatable of the two, rations were depleted. Thus, for rats trained with 4% ethanol (more palatable than water), free access to no-ethanol solution alternated with rations of 4% ethanol solution. For rats trained with 10% alcohol solution (less palatable than water), free access to 10% alcohol solution alternated with rations of no-alcohol solution.

7. Statistical analyses ANOVAs were conducted with all applicable between- and with-groups variables, as indicated in Table 3. Trial effects were evaluated for significance with Greenhouse–Geisser correction. Significant effects involving more than two levels of a repeated measure were interpreted with multiple comparisons using Bonferroni adjustment;

16

N.K. Dess et al. / Alcohol 37 (2005) 9–22

Table 3 Analyses of variance conducted on data from Experiments 2A and 2B

Experiment 2A Experiment 2B

Conditioning solution intake

Preference score

Replic 3 Line 3 Sex 3 Conc 3 Soln Line 3 Sex 3 Conc

Replic 3 Line 3 Sex 3 Conc Line 3 Sex 3 Conc

because trial effects during conditioning were numerous but not of particular interest, only the highest order interaction involving trial was subjected to analysis with multiple comparisons. A few data were lost for mechanical reasons; thus, degrees of freedom are not identical for all analyses. Effects and contrasts significant at a 5 .05 are reported. For each experiment, data from the flavor preference test were transformed to preference scores (ethanol-paired flavor consumption/total consumption), by which 0.5 indicates indifference. To simplify interpretation of these complex designs, preference scores were averaged for variables other than line that yielded no main effect and interacted either not at all or only ordinally (direction and significance of all differences preserved) with other variables. Preliminary analyses showed no significant differences in consumption of cherry versus grape solutions. Thus, no distinction is made between cherry and grape in the analyses below.

8. Results 8.1. Experiment 2A: free access to ethanol and no-ethanol solutions At baseline, females weighed less than males, F(1, 185) 5 583.70; body weight did not differ between lines. Baseline water intake was higher among females than males, F(1, 185) 5 46.35. It was also higher among HiS females than among LoS females but did not differ between HiS and LoS males. The line 3 sex interaction was significant, F(1, 185) 5 8.02; contrasts showed that the line difference was significant only for females. Dose of ethanol consumed on conditioning trials (g/kg) is shown in Fig. 3. No effects involving replication were significant, and data are shown collapsed across replications. Overall, larger doses of ethanol were consumed by HiS rats than by LoS rats, by females than males, and in the 10% condition than the 4% condition [Fs(1, 173) 5 8.89, 25.50, and 57.86, respectively, for these main effects]. The line 3 concentration interaction was also significant, F(1, 173) 5 6.39; contrasts showed that the line difference was significant only in the 10% condition. Dose consumed generally increased over trials, F(4, 692) 5 7.61, and contrasts showed that average dose consumed was smaller on Trials 1–3 than on Trials 4–5. Interactions involving trial included trial 3 sex, trial 3

Fig. 3. Mean dose of ethanol consumed (g/kg) on conditioning trials by female and male low-saccharin-consuming and high-saccharin-consuming rats in Experiment 2A trained with 4% ethanol (‘‘A,’’ upper panel) or 10% ethanol (‘‘B,’’ lower panel).

sex 3 concentration, and trial 3 sex 3 concentration 3 line, Fs(4, 692) 5 2.86, 4.14, and 8.01, respectively. The four-way interaction was interpreted by conducting separate trial 3 concentration 3 line ANOVAs for females and males and examining interactions involving line, trial, or concentration. The females’ ANOVA yielded a concentration 3 trial interaction, F(4, 360) 5 3.53; contrasts showed that the increase in dose consumed after Trial 1 was not significant until Trial 5 in the 4% ethanol condition but was significant at Trial 2 in the 10% condition. The males’ ANOVA yielded a concentration 3 line interaction, F(1, 91) 5 6.20, and contrasts showed that the lines differed only in the 10% ethanol condition. Thus, the four-way interaction apparently derives from the interaction of ethanol concentration with a different variable in females (trial) and males (line). Volume of flavored solutions consumed during conditioning is shown in Fig. 4 as change from water baseline. Data are shown collapsed across replications because only two trivial effects involving replication were significant: a main effect [4.0 ml greater suppression from baseline overall in Replication 1, F(1, 173) 5 11.04] and an ordinal solution 3 replication interaction [F(1, 173) 5 9.39; the difference between 4% and 10% solutions was somewhat larger in Replication 2, but contrasts showed that the difference was significant in each replication].

N.K. Dess et al. / Alcohol 37 (2005) 9–22

17

Fig. 4. Mean volume consumed (ml, as change from water baseline 6 standard error) of ethanol (EtOH) and no-ethanol (NE) solutions on conditioning trials by female (left) and male (right) low-saccharin-consuming and high-saccharin-consuming rats in Experiment 2A trained with 4% ethanol (‘‘A,’’ upper panel) or 10% ethanol (‘‘B,’’ lower panel).

Rats generally consumed less of both conditioning solutions than they had of water in the baseline period. Intake was suppressed less among females than males, in the 4% ethanol (vs. the 10% ethanol) condition, and on no-ethanol (vs. ethanol) trials, Fs(1, 173) 5 19.87, 16.04, and 63.44, respectively. Line interacted with sex and with concentration, Fs(1, 173) 5 7.62 and 4.43, respectively. Contrasts showed that the greater suppression among LoS rats was obtained only for males and only at 10% ethanol. Type of solution also interacted with line and concentration [solution 3 concentration, solution 3 line, and solution 3 line 3 concentration, Fs(1, 173) 5 128.65, 50.06, and 18.99, respectively]. Contrasts showed that (a) drinking was suppressed less on ethanol than no-ethanol trials in the 4% condition but more on ethanol than no-ethanol trials in the 10% condition, (b) drinking was suppressed more on ethanol than no-ethanol trials among LoS rats but not HiS rats, and (c) HiS and LoS rats differed significantly only on ethanol trials in the 10% ethanol condition. Suppression of drinking lifted over trials [main effect of trial, F(4, 692) 5 21.45] but the degree to which it lifted varied as a function of all the other variables (significant three-, four-, and five-way interactions). The highest order interaction (trial 3 line 3 sex 3 solution 3 concentration) was interpreted by conducting separate trial 3 line 3 solution 3 concentration ANOVAs for females and males, both

of which yielded significant four-way interactions, F(4, 360) 5 2.79 and F(4, 364) 5 3.31, respectively. Contrasts among trials for each solution showed an idiosyncratic pattern of changes for each group, all involving less suppression on earlier trials than later trials. Suppression lifted over trials for females in all conditions but two: 4%-trained LoS females increased intake only on no-ethanol trials, and 10%-trained HiS females did not change intake of either solution significantly. Suppression lifted in fewer conditions for males, changing significantly only on ethanol trials among 10%-trained HiS rats and among 4%-trained LoS rats. Preference for the ethanol-paired flavor during the twobottle flavor test (ethanol flavor divided by total intake) is shown in the upper panel of Fig. 5. The ANOVA yielded an ordinal replication 3 concentration interaction, F(1, 172) 5 5.20; contrasts showed that preference was significantly greater after training with 4% ethanol than 10% ethanol in both replications, but the difference was more robust in Replication 2. Because no other effect involving replication was significant, and no effect involving sex was significant (all ps O .10), data are shown collapsed across replication and sex. Overall, rats preferred the ethanolpaired flavor to the unpaired flavor [grand mean 0.65 6 0.02 vs. 0.50, t(187) 5 8.07]. The preference was stronger among HiS rats than LoS rats and after training

18

N.K. Dess et al. / Alcohol 37 (2005) 9–22

differed whereas HiS and LoS females did not. No effects involving concentration were significant. 9. Experiment 2 discussion

Fig. 5. Mean preference (6standard error) of low-saccharin-consuming and high-saccharin-consuming rats for a flavor previously paired with 4% ethanol or 10% ethanol in Experiment 2A.

with 4% ethanol (vs. 10% ethanol), Fs(1, 172) 5 7.03 and 24.72, respectively. In addition, the line effect varied with concentration [line 3 concentration interaction, F(1, 172) 5 8.57]; contrasts showed that the ethanol-paired flavor preference was stronger among HiS than LoS rats only after training with 10% ethanol. 8.2. Experiment 2B: yoked conditioning intake At baseline, females weighed less than males, F(1, 78) 5 102.46, and had higher water intake than males, F(1, 78) 5 5.30. No line differences were significant. Suppression of intake of unrestricted solutions and dose of ethanol consumed during conditioning are shown (Fig. 6). Suppression was greater on early trials, among males, and among rats trained with 10% ethanol. A line 3 sex 3 concentration ANOVA yielded significant main effects of trial, F(4, 292) 5 5.84, sex, F(1, 73) 511.37, and concentration, F(1, 73) 5 11.74; contrasts showed that suppression was greater on Trials 1 and 2 than on Trials 4 and 5. Dose consumed (lower panel) was larger on later trials, among females, among animals trained with 10% ethanol, and among HiS rats [main effects of trial, F(4, 296) 5 9.69; sex, F(1, 74) 5 10.596; concentration, F(1, 74) 5 30.32; and line, F(1, 74) 5 4.32]. In addition, the trial 3 concentration interaction was significant, F(4, 296) 5 3.16. Contrasts showed that dose consumed only changed significantly across trials among animals trained with 10% ethanol: It was lower on Trial 1 than on later trials. Preference for the ethanol-paired flavor is shown in Fig. 7. As in Experiment 2A, HiS females and males preferred the ethanol-paired flavor, regardless of whether they were trained with 4% or 10% ethanol. The same was true of LoS females, who previously only preferred a flavor paired with 4% ethanol. LoS males, who also previously preferred a flavor paired with 4% ethanol, did not prefer a flavorpaired with either 4% or 10% ethanol. A line 3 sex 3 concentration ANOVA yielded a line 3 sex interaction, F(1, 74) 5 4.94; contrasts showed that HiS and LoS males

HiS and LoS rats differ in proclivity for consuming flavors paired with ethanol. HiS males and females prefer flavors paired with either a low- or a high-ethanol concentration. This finding is inconsistent with two possible mechanisms for the preference. The first is the conditioning of a positive taste–taste association (Fanselow & Birk, 1982). Rats generally prefer 4% ethanol solution to water and prefer water to 10% ethanol; HiS and LoS rats both show this pattern (Dess et al., 1998). In Experiment 2A, consumption of 4% ethanol solution was suppressed relative to baseline, presumably due to citric acid in the KoolAid (Dess, 2000). Consistent with the relatively high acceptability and palatability of 4% ethanol, however, consumption of 4% solution was less suppressed from baseline than was 10% ethanol solution. It is possible, then, that a preference for a 4%-paired flavor would have developed based on its association with a palatable concentration of ethanol. However, HiS rats also preferred a flavor paired with 10% ethanol, which these and earlier data indicate is unpalatable to them. Thus, no parsimonious explanation of HiS rats’ preference for ethanol-paired flavors at high or low concentrations can turn on pairing with ethanol’s positive taste qualities. The second unsatisfactory explanation is relative novelty. During conditioning, HiS rats drank more 4% solution than no-ethanol solution, and differential exposure therefore could potentially account for their subsequent flavor preference: They could simply reject the more novel, no-ethanol solution flavor in favor of the less novel, 4%solution flavor. However, HiS rats drank less 10% solution than no-ethanol solution, yet still preferred the relatively novel ethanol-paired flavor to the no-ethanol associated flavor. Whereas HiS rats’ preference for a flavor paired with 10% ethanol could have been underestimated due to its relative novelty, the fact that it was preferred despite being less familiar points to some other basis of the preference. LoS rats tell a different story. When allowed to freely drink during conditioning, they behaved much as did HiS rats: They drank at least as much of the 4% solution as the no-ethanol solution, with LoS males eventually consuming more of it than no-ethanol solution, and they drank less of the 10% solution than the no-ethanol solution. LoS rats’ subsequent preference for a familiar flavor paired with an acceptable concentration of ethanol (4%) and lack of preference for a relatively novel 10% paired flavor in Experiment 2A are consistent with a role for taste–taste associations or relative novelty in their flavor choices. Thus, the enhanced sensitivity to aversive taste mixtures and novelty of LoS rats observed in other behavioral domains (e.g., novel open-field defecation; Dess, 2000; Dess & Minor, 1996) manifest in this paradigm as well.

N.K. Dess et al. / Alcohol 37 (2005) 9–22

19

Fig. 6. Mean ethanol consumption (6standard error) during the 10-day conditioning procedure by low-saccharin-consuming and high-saccharin-consuming rats in Experiment 2B. Shown as volume consumed (freely available solution in ml, as change from water baseline; upper panel) and dose (g/kg; lower panel).

Experiment 2B, however, implicates different mechanisms in the flavor preferences of LoS males and LoS females. LoS males’ preference for a 4%-paired flavor in Experiment 2A was controlled primarily by its being less novel than the no-ethanol paired flavor, because the preference was eliminated when the two flavors were equally novel. Their failure to prefer a 10%-paired flavor, on the other hand, was not due to its novelty, because they failed to prefer such a flavor even when it was no more or less novel than the alternative flavor. Though LoS males trained with 10% ethanol did consume a smaller ethanol dose than did the other three groups, dose cannot easily account for their failure to acquire a preference: The average dose they consumed (3.4 g/kg) was comparable to or exceeded doses that supported a conditioned preference in 4%-trained LoS females (3.5 g/kg) and HiS rats (3.9 g/kg females, 2.2 g/kg males) in Experiment 2B. However, a larger volume of 4% solution yields such a dose. Perhaps, then, the amount of 10% ethanol solution consumed by LoS males was below some minimum necessary to develop and/or express a preference for it. Like LoS males, LoS females preferred a 4%-paired flavor to an unpaired flavor and did not prefer a 10%-paired flavor to less novel alternative when all solutions were freely available during training (Exp. 2A). Unlike LoS

males, they preferred both a 4%-paired and a 10%-paired flavor to an unpaired flavor when the two flavors were equally novel (Exp. 2B). That controlling for novelty, and not ethanol dose, was key to the expression of a 10%-paired flavor preference in Experiment 2B is clear from the fact that average dose consumed by LoS female groups trained with 10% ethanol was no higher in that study (4.9 g/kg in Exp. 2B vs. 4.8 g/kg in Exp. 2A). Among LoS females, then, relative novelty was irrelevant to their preference for a 4%-paired flavor (Exp. 2A and 2B) but effectively prevented expression of a preference for a 10%-paired flavor in Experiment 2A. A ‘‘preference’’ score can reflect conditioned preference, conditioned aversion, or a mix of both. No group in Experiment 2 showed a frank aversion to the ethanol-paired flavor (mean preference score ! 0.50). However, ad hoc inspection of the data showed that many 10%-trained LoS rats in Exp. 2A (51%) had preference scores below 0.50, compared to their 4%-trained LoS (15%) or 10%-trained HiS (18%) counterparts. Those low scores do not necessarily indicate aversion; random variation around the mean of an indifferent group would ensure that about half the scores fell below 0.50. On the other hand, Hood and Buck (2000) reported that mice more prone to withdrawal developed stronger conditioned taste aversions. Thus, the lower

20

N.K. Dess et al. / Alcohol 37 (2005) 9–22

Fig. 7. Mean preference score (6standard error) of low-saccharin-consuming and high-saccharin-consuming rats for a flavor previously paired with 4% or 10% ethanol in Experiment 2B.

‘‘preference’’ scores of LoS rats, that are more withdrawalprone, could to some extent reflect development of an aversion. Conditioned aversion cannot account for LoS females expressing a preference when flavor novelty is controlled (Exp 2B), but it may contribute to the LoS males’ failure to prefer a flavor paired with 10% ethanol regardless of its relative novelty. In summary, HiS rats show robust preferences for ethanol-paired flavors. LoS rats can also develop and express preferences for ethanol-paired flavors. Whether they do, however, is (a) mediated by behavior toward, and thus exposure to, the solutions during initial experiences with them and (b) moderated by sex. LoS males prefer a freely consumed flavor paired only with a low-ethanol concentration, and that preference is a byproduct of greater prior exposure to it versus the alternative. In contrast, LoS females prefer a flavor paired with either low- or high-ethanol concentrations as long as they are not allowed to differentially avoid the latter during earlier experiences with it. With novelty controlled, LoS females look like HiS rats. An important question concerns the extent to which these group differences in ‘‘preference’’ scores reflect hedonic evaluation (‘liking’) versus incentive value (‘wanting’), which have distinct neurochemical mechanisms (Berridge & Robinson, 2003). This question pertains to the nature and behavioral implications of the preferences and, possibly, aversions expressed in this study. Research using methods appropriate to dissociating liking and wanting is needed to answer that question.

10. General discussion Withdrawal was more severe and preference for ethanolpaired flavors was less robust among LoS rats than HiS rats. These findings replicate and extend the widely reported inverse relation between withdrawal and voluntary ethanol consumption to animals selectively bred on a phenotype related to, but different from, ethanol intake or its effects.

As such, they encourage the further elaboration of models that incorporate noningestive markers for and mediators of proneness to the development of alcohol behavior (Cloninger, 1987; Kampov-Polevoy et al, 1999). Experiment 2 suggests that flavor conditioning could contribute to the shift from highly flavored, sweeter, lowerconcentration alcoholic beverages favored by inexperienced drinkers, such as cider, beer, and fortified wine, to less highly flavored, drier, higher-concentration beverages favored by experienced drinkers, such as wine and distilled spirits (Lolli et al., 1958; Lolli et al., 1960; Measham, 1996; Parreiras et al., 1956; Terry et al., 1957). Though the mechanisms appear to vary, male and female rats of both lines preferred an ethanol-paired flavor when the concentration of alcohol was low (see Dudley, 2001). Once a preference for alcohol-associated flavors has been established – even if only by virtue of its familiarity – a higher concentration of alcohol in cocktails may be tolerated. To the extent that counterconditioning to ethanol’s initial aversiveness also is occurring (a premise not examined directly here), the conditioned flavor preference would generalize, supporting the development of a preference for other, stronger drinks. Consumption of larger doses of ethanol then would promote further modification of drinking behavior based on ethanol’s pharmacological properties, including those that promote the development of tolerance and withdrawal and those that foster incentive salience and cravings. The argument here, then, is not that people drink simply because of their flavor preferences or that these preferences derive from ethanol’s pharmacological effects and therefore directly promote drug craving. Indeed, the pattern of conditioned preferences reported here does not covary reliably with dose of ethanol consumed, so pharmacological effects are implicated weakly if at all. Rather, the key idea is that the establishment of conditioned flavor preferences can influence amount consumed (Dawson, 1993; Kidorf et al., 1990), thus laying the foundation for further changes in drinking patterns. Additional research is needed to test the usefulness of this ‘‘scaffolding’’ model of alcohol use. A final note concerns individual differences in propensity to develop problem drinking. The most obvious rodent candidates for modeling alcoholism are those with relatively high voluntary ethanol intake – such as, HiS rats. However, animals more prone to withdrawal during abstinence – such as, LoS rats – also are candidates, because a good antidote to withdrawal is drinking alcohol (e.g., Gibbins et al., 1971). The present study and related work reinforce the view that multiple pathways to alcohol use, and thus multiple vulnerable phenotypes, exist. Expression of those vulnerabilities occurs in different experimental procedures, which model different kinds of life experiences. A particularly interesting aspect of the vulnerability landscape in the present study is the interaction of line and sex. HiS rats developed and expressed preferences for ethanol-paired flavors across a range of parameters, and

N.K. Dess et al. / Alcohol 37 (2005) 9–22

LoS females were their equal as long as novelty was controlled. By contrast, LoS males developed and expressed preferences for ethanol-paired flavors in even more limited circumstances. Hence, expressed phenotypic diversity was greater among males than females. This pattern is reminiscent of findings from surveys of drinking behavior in the United States and Wales (e.g., American Medical Association, 2004; Dawson, 1993; Roberts et al., 1999). Teenage girls report drinking more than teenage boys, and their preferred drinks are sweet, fruity cocktails, so-called alcopops or girlie drinks. An overall higher level of consumption among girls would occur if current social contexts – including beverage fads and conformity pressures – favored expression of phenotypic diversity among boys more than among girls. In short, only some ‘‘types’’ of boys would develop alcohol-based flavor preferences whereas more ‘‘types’’ of girls would do so. This proposal is highly speculative. Human alcohol consumption is enormously complex. Yet, in light of the devastating impact of excess alcohol use – especially and increasingly among women – modeling early stages of alcohol use with heterogeneous rats and basic behavioral paradigms as in this study seems a promising and worthwhile endeavor.

Acknowledgments Support for this project from the Howard Hughes Medical Institute and the Undergraduate Research Center at Occidental College is gratefully acknowledged, as is the able assistance of Cameryn Garrett, Vincent Chen, Stephanie Freidburg-Strauss, Rita Molestina, Cheryl Prigodich, Candace Ryan, Larissa Gibson, and Mitzi Gonzales.

References Adams, N., Shihabi, Z. K., & Blizard, D. A. (1991). Ethanol preference in the Harrington derivation of the Maudsley Reactive and Non-Reactive strains. Alcohol Clin Exp Res 15, 170–174. American Medical Association (2004). Teenage girls targeted for sweetflavored alcoholic beverages. Retrieved July 12, 2005, from. http:// www.ama-assn.org/ama/pub/category/print/14425.html. Avena, N. M., Long, K. A., & Hoebel, B. G. (2005). Sugar-dependent rats show enhanced responding for sugar after abstinence: evidence of a sugar deprivation effect. Physiol Behav 84(3), 359–362. Badistov, B. A., Overstreet, D. H., Kashevskaya, O. P., Viglinskaya, I. V., Kampov-Polevoy, A. B., Seredenin, S. B., & Halikas, J. A. (1995). To drink or not to drink: open field behavior in alcohol-preferring and -nonpreferring rat strains. Physiol Behav 57, 585–589. Belknap, J. K., Crabbe, J. C., & Young, E. R. (1993). Voluntary consumption of ethanol in 15 inbred mouse strains. Psychopharmacology 112, 503–510. Berridge, K. C., & Robinson, T. E. (2003). Parsing reward. Trends Neurosci 26(9), 507–513. Chester, J. A., Blose, A. M., & Froehlich, J. C. (2003). Further evidence of an inverse genetic relationship between innate differences in alcohol preference and alcohol withdrawal magnitude in multiple selectively bred rat lines. Alcohol Clin Exp Res 27(3), 377–387.

21

Cicero, T. J. (1979). A critique of animal analogues of alcoholism. In E. Majchrowicz, & E. P. Noble (Eds.), Biochemistry and Pharmacology of Ethanol Vol. 2 (pp. 533–560). New York: Plenum Press. Cloninger, C. R. (1987). Neurogenetic adaptive mechanisms in alcoholism. Science 236(4800), 410–416. Cloninger, C. R. (1996). Assessment of the impulsive–compulsive spectrum of behavior by the seven-factor model of temperament and character. In J. M. Oldham, E. Hollander, & A. E. Skodol (Eds.), Impulsivity and Compulsivity (pp. 59–95). Washington, DC: American Psychiatric Association. Colantuoni, C., Rada, P., McCarthy, J., Patten, C., Avena, N. M., Chadeayne, A., & Hoebel, B. G. (2002). Evidence that intermittent, excessive sugar intake causes endogenous opioid dependence. Obes Res 10, 478–488. Colby-Morley, E. (1982). The reflection of hypoglycemia and alcoholism on personality: nutrition as a mode of treatment. J Orthomol Psychiatry 11(2), 132–139. Cunningham, C. L., & Neihus, J. S. (1997). Flavor preference conditioning by oral self-administration of ethanol. Psychopharmacology 134(3), 293–302. Dawson, D. A. (1993). Patterns of alcohol consumption: beverage effects on gender differences. Addiction 88(1), 133–138. Dawson, M.E., Schell, A.M., & B€ohmelt, A.H. (1999). Startle Modification: implications for Neuroscience, Cognitive Science, and Clinical Science (383 14th ed.). Dess, N. K. (1991). Ingestion and emotional health. Human Nat 2, 235–269. Dess, N. K. (2000). Responses to basic taste qualities in rats selectively bred for high versus low saccharin intake. Physiol Behav 69, 247–257. Dess, N. K., Arnal, J., Chapman, C. D., Siebel, S. I., VanderWeele, D. A., & Green, K. (2000). Exploring adaptations to famine: rats selectively bred for differential saccharin intake differ on deprivation-induced hyperactivity and emotionality. Int J Comp Psychol 13, 34–52. Dess, N. K., Badia-Elder, N., Thiele, T. E., Kiefer, S. W., & Blizard, D. (1998). Ethanol consumption in rats selectively bred for differential saccharin intake. Alcohol 16, 275–278. Dess, N. K., & Minor, T. R. (1996). Taste and emotionality in rats selectively bred for high versus low saccharin intake. Anim Learn Behav 24, 105–115. Dudley, R. (2001). Fermenting fruit and the historical ecology of ethanol ingestion: is alcoholism in modern humans an evolutionary hangover? Addiction 97, 381–388. Eriksson, K. (1968). Genetic selection for voluntary alcohol consumption in the albino rat. Science 159, 739–741. van Erp, A. M. M., & Miczek, K. A. (2001). Persistent suppression of ethanol self-administration by brief social stress in rats and increased startle response as index of withdrawal. Physiol Behav 73(3), 301–311. Fanselow, M. S., & Birk, J. (1982). Flavor–flavor associations induce hedonic shifts in taste preference. Anim Learn Behav 10, 223–228. Fendt, M. (1999). Enhancement of prepulse inhibition after blockade of GABA activity within the superior colliculus. Brain Res 833(1), 81–85. Fuller, J. L. (1974). Single locus control of saccharin preference in mice. J Heredity 65, 33–36. Gatch, M. B., & Selvig, M. (2002). Theophylline blocks ethanol withdrawal-induced hyperalgesia. Alcohol Alcohol 37(4), 313–317. Gibbins, R. J., Kalant, H., LeBlanc, A. E., & Clark, J. W. (1971). The effects of chronic administration of ethanol on startle thresholds in rats. Psychopharmacologia 19(2), 95–104. Goodwin, F. L. W., Bergeron, N., & Amit, Z. (2000). Differences in the consumption of ethanol and flavored solutions in three strains of rats. Pharmacol Biochem Behav 65(3), 357–362. Grillon, C., Sinha, R., Ameli, R., & O’Malley, S. S. (2000). Effects of alcohol on baseline startle and prepulse inhibition in young men at risk for alcoholism and/or anxiety disorders. J Stud Alcohol 61(1), 46–54. Grillon, C., Sinha, R., & O’Malley, S. S. (1994). Effects of ethanol on the acoustic startle reflex in humans. Psychopharmacology 114(1), 167–171.

22

N.K. Dess et al. / Alcohol 37 (2005) 9–22

Heinz, A., L€ ober, S., Georgi, A., Wrase, J., Hermann, D., Rey, E. R., Wellek, S., & Mann, K. (2003). Reward craving and withdrawal relief craving: assessment of different motivational pathways to alcohol intake. Alcohol Alcohol 38(1), 35–39. Hood, H. M., & Buck, K. J. (2000). Allelic variation in the GABA-sub(A) receptor 32 subunit is associated with genetic susceptibility to ethanolinduced motor incoordination and hypothermia, conditioned taste aversion, and withdrawal in BXD/Ty recombinant inbred mice. Alcohol Clin Exp Res 24(9), 1327–1334. Hutchinson, K. E., McGeary, J., Wooden, A., Blumenthal, T., & Ito, T. (2003). Startle magnitude and prepulse inhibition: effects of alcohol and attention. Psychopharmacology 167(3), 235–241. Jain, R., Mukherjee, K., & Singh, R. (2004). Influence of sweet tasting solutions on opioid withdrawal. Brain Res Bull 64(4), 319–322. Jones, A. E., McBride, W. J., Murphy, J. M., Lumeng, L., Li, T.-K., Shahkar, A., & McKinzie, D. L. (2000). Effects of ethanol on startle responding in alcohol-preferring and -non-preferring rats. Pharmacol Biochem Behav 67(2), 313–318. Kampov-Polevoy, A. B., Garbutt, J. C., Janowsky, D. S. (1999). Association between preference for sweets and excessive alcohol intake: a review of animal and human studies. Alcohol Alcohol 34(3) (May–June), 386–395. Kampov-Polevoy, A. B., Overstreet, D. H., Rezvani, A. H., & Janowsky, D. S. (1995). Suppression of ethanol intake in alcoholpreferring rats by prior voluntary saccharin consumption. Pharmacol Biochem Behav 52, 59–64. Kidorf, M., Lang, A. R., & Pelham, W. E. (1990). Beverage preference, beverage type and subject gender as determinants of alcohol consumption in the laboratory. J Stud Alcohol 54(4), 331–335. Kiefer, S. W., & Lawrence, G. J. (1988). The sweet-bitter taste of alcohol: aversion generalization to various sweet-quinine mixtures in the rat. Chem Senses 13, 633–641. Kiefer, S. W., & Mahadevan, R. S. (1993). The taste of alcohol for rats as revealed by aversion generalization tests. Chem Senses 18, 509–522. Lankford, M. F., & Myers, R. D. (1994). Genetics of alcoholism: simultaneous presentation of a chocolate drink diminishes alcohol preference in high drinking HAD rats. Pharmacol Biochem Behav 49(2), 417–425. Li, T.-K., Lumeng, L., McBride, W. J., & Murphy, J. M. (1987). Rodent lines selected for factors affecting alcohol consumption. Alcohol (Suppl 1), 91–96. Lolli, G., Golder, G. M., Serianni, E., Bonfiglia, G., & Balboni, C. (1958). Choice of alcoholic beverage among 178 alcoholics in Italy. Q J Stud Alcohol 19, 303–308. Lolli, G., Schesler, E., & Golder, G. M. (1960). Choice of alcoholic beverage among 105 alcoholics in New York. Q J Stud Alcohol 21, 475–482. Macey, D. J., Schulteis, G., Heinrichs, S. C., & Koob, G. F. (1996). Timedependent quantifiable withdrawal from ethanol in the rat: effect of method of dependence induction. Alcohol 13, 163–170. McClearn, G. E., & Rodgers, D. A. (1959). Differences in alcohol preference among inbred strains of mice. Q J Stud Alcohol 20, 691–695. McKinzie, D. L., Sajdyk, T. J., McBride, W. J., Murphy, J. M., Lumeng, L., Li, T. K., & Shahkar, A. (2000). Acoustic startle and fear-potentiated startle in alcohol-preferring (P) and -nonpreferring (NP) lines of rats. Pharmacol Biochem Behav 65(4), 691–696.

Measham, F. (1996). The ‘‘big bang’’ approach to sessional drinking: changing patterns of alcohol consumption among young people in North West England. Addiction Res 4(3), 283–299. Mehiel, R., & Bolles, R. C. (1984). Learned flavor preferences based on caloric outcome. Anim Learn Behav 12(4), 421–427. Metten, P., Phillips, T. J., Crabbe, J. C., Tarantino, L. M., McClearn, G. E., Plomin, R., Erwin, V. G., & Belknap, J. K. (1998). High genetic susceptibility to ethanol withdrawal predicts low ethanol consumption. Mamm Genome 9, 983–990. Miller, N. S., & Goldsmith, R. J. (2001). Craving for alcohol and drugs in animals and humans: biology and behavior. J Addict Dis 20(3), 87–104. Minor, T. R., Rowe, M. K., Job, R. F. S., & Ferguson, E. C. (2001). Escape deficits induced by inescapable shock and metabolic stress are reversed by adenosine receptor antagonists. Behav Brain Res 120(2), 203–212. Minor, T. R., & Saade, S. (1994). Poststress glucose mitigates behavioral impairment in rats in the ‘‘learned helplessness’’ model of psychopathology. Biol Psychiatry 42(5), 324–334. Moeller, C., Wiklund, L., Thorsell, A., Hyytiae, P., & Heilig, M. (1997). Decreased measures of experimental anxiety in rats bred for high alcohol preference. Alcohol Clin Exp Res 21(4), 656–660. Overstreet, D. H., Halikas, J. A., Serednin, S. B., Kampov-Polevoy, A. B., Viglinskaya, I. V., Kashevskaya, O., Badishtov, B. A., Knapp, D. K., Mormede, P., Kiianmaa, K., Li, T.-K., & Rezvani, A. H. (1997). Behavioral similarities and differences among alcohol-preferring and -nonpreferring rats: confirmation by factor analysis and extension to additional groups. Alcohol Clin Exp Res 21, 840–848. Parreiras, D., Lolli, G., & Golder, G. M. (1956). Choice of alcoholic beverage among 500 alcoholics in Brazil. Q J Stud Alcohol 17, 629–632. Pohorecky, L. A., Cagan, M., Brick, J., & Jaffe, L. S. (1976). The startle response in rats: effect of ethanol. Pharmacol Biochem Behav 4(3), 311–316. Rassnick, S., Koob, G. F., & Geyer, M. A. (1992). Responding to acoustic startle during chronic ethanol intoxication and withdrawal. Psychopharmacology 106, 351–358. Roberts, A. J., Crabbe, J. C., & Keith, L. D. (1994). Corticosterone increases severity of acute withdrawal from ethanol, pentobarbital, and diazepam in mice. Psychopharmacology 115(1–2), 278–284. Roberts, C., Blakey, V., & Tudor-Smith, C. (1999). The impact of ‘alcopops’ on regular drinking by young people in Wales. Drugs Educ Prevent Policy 6(1), 7–15. Sinclair, J. D., Kampov-Polevoy, A., Stewart, R., & Li, T.-K. (1992). Taste preferences in rat lines selected for low and high alcohol consumption. Alcohol 9(2), 155–160. Swerdlow, N. R., Kuczenski, R., Goins, J. C., Crain, S. K., Ma, L. T., Bongiovanni, M. J., & Shoemaker, J. M. (2005). Neurochemical analysis of rat strain differences in the startle gating-disruptive effects of dopamine agonists. Pharmacol Biochem Behav 80(2), 203–211. Terry, J., Lolli, G., & Golder, G. M. (1957). Choice of alcoholic beverage among 531 alcoholics in California. Q J Stud Alcohol 18, 417–428. VanderWeele, D. A., Dess, N. K., & Castonguay, T. W. (2002). Ingestional responses to metabolic challenges in rats selectively bred for high- and low-saccharin intake. Physiol Behav 75, 97–104. Zhang, J., Forkstam, C., Engel, J. A., & Svensson, L. (2000). Role of dopamine in prepulse inhibition of acoustic startle. Psychopharmacology 149, 181–188.