Sweetened-ethanol drinking during social isolation: Enhanced intake, resistance to genetic heterogeneity and the emergence of a distinctive drinking pattern in adolescent mice

Abstract

With its ease of availability during adolescence, sweetened ethanol (‘alcopops’) is consumed within many contexts. We asked here whether genetically based differences in social motivation are associated with how the adolescent social environment impacts voluntary ethanol intake. Mice with previously described differences in sociability (BALB/cJ, C57BL/6J, FVB/NJ and MSM/MsJ strains) were weaned into isolation or same-sex pairs (postnatal day 21), and then given continuous access to two fluids on postnatal days 34–45: One containing water and the other containing a ascending series of saccharin-sweetened ethanol (3-to-6-to-10%). Prior to the introduction of ethanol (postnatal days 30–33), increased water and food intake was detected in some of the isolation-reared groups, and controls indicated that isolated mice also consumed more ‘saccharin-only’ solution. Voluntary drinking of ‘ethanol-only’ was also higher in a subset of the isolated groups on postnatal days 46–49. However, sweetened ethanol intake was increased in all isolated strain-by-sex combinations irrespective of genotype. Surprisingly, blood ethanol concentration was not different between these isolate and socially housed groups 4 hours into the dark phase. Using lickometer-based measures of intake in FVB mice, we identified that a predominance of increased drinking during isolation transpired outside of the typical circadian consumption peak, occurring ≈8.5 hours into the dark phase, with an associated difference in blood ethanol concentration. These findings collectively indicate that isolate housing leads to increased consumption of rewarding substances in adolescent mice independently of their genotype, and that for ethanol this may be due to when individuals drink during the circadian cycle.

INTRODUCTION

Although research into the social neurobiology of drug reward is a relatively new development, it is quickly becoming its own field (for reviews see Bardo et al., 2013; Trezza et al., 2014; El Rawas & Saria, 2015). Thus, in fundamental ways there is conceptual agreement that social behavior and addiction are inter-related (Panksepp et al., 1980; Insel, 2003; Burkett & Young, 2012). Depending on the social context, peers can either intensify or moderate drug consumption. Moreover, social isolation/exclusion can also promote drug intake. Adolescence—a developmental stage characterized by dramatic changes of the body, nervous system, and behavior (Sisk & Foster, 2004)—is a period when risk-taking as it pertains to the likelihood of abusing alcohol and other drugs is particularly salient (Spear, 2000). Relative to other drugs, social influences on alcohol intake are especially problematic during this period due to the ubiquity of alcohol (Hopson, 2013) and addition of sweetener (i.e., ‘alcopops’; Mart, 2011).

Animal models for elucidating interactions between drug abuse and social processes have been rigorously developed. For instance, drug responsiveness of laboratory mice has been shown to increase after observing morphine-intoxicated peers (Hodgson et al., 2010). Laboratory rodents also prefer to self-administer psycho-stimulants with a partner (Smith & Pitts, 2014) and express social preferences for conspecfics that have a similar history of psycho-stimulant exposure (Smith et al., 2015; Watanabe, 2015). Social interactions additionally enhance the rewarding effects of cocaine (Thiel et al., 2008) and morphine (Cole et al., 2013) in laboratory rodents. There is also robust evidence that chronic social isolation of laboratory rodents increases responsiveness to, intake of, and preference for drugs of abuse (e.g., Zimmerberg & Brett; 1992; Phillips et al., 1994; Raz and Berger, 2010; Cain et al., 2012; Westenbroek et al., 2013; Whitaker et al., 2013; Meyer & Bardo, 2015; Yorgason et al., 2016).

The well-known relationship of alcohol with the social environment (de Castro, 1990; Kirkpatrick & de Wit, 2013) has also been studied with animal models. In this respect, the first empirical report indicated that young-adult male mice— grouped socially for 24h—subsequently reduced voluntary ethanol intake relative to individuals that remained in isolation (Thiessen & Rodgers, 1965). Even though the authors attributed this finding to the stress of social conflict upon grouping (e.g., Norman et al., 2015), the study nevertheless led to a systematic development into how social versus isolate housing can alter voluntary ethanol drinking. Studies in laboratory rodents have led to a general consensus that isolate housing leads to increased ethanol drinking (Deatherage, 1972; Parker & Radow, 1974; Schenk et al., 1990; Wolffgramm, 1990; Hall et al., 1998; Núñez et al., 2002; Juárez & Vázquez-Cortés, 2003; Doremus et al., 2005; Advani et al., 2007; Deehan et al., 2007; Ehlers et al., 2007; McCool and Chappell, 2009; Lopez et al., 2011; Chappell et al., 2013; Talani et al., 2013; Butler et al., 2014; Lopez & Laber, 2015). However, there are exceptions to these observations; for example, socially paired laboratory rats (McCusker and Bell, 1988; Adams & Oldham, 1996; Tomie et al., 2005; Varlinskaya et al., 2015) and mice (Logue et al., 2014) exhibit increased voluntary ethanol consumption depending on the precise experimental conditions and measurements. Moreover, both inhibitory and facilitating effects of social housing with same-sex siblings on alcohol drinking have been observed in rodent species other than the prototypical laboratory mouse and rat, such as the prairie vole (for a review see Ryabinin & Hostetler, 2016). Interestingly, within this species the direction of the influence can differ between individual animals (Anacker & Ryabinin, 2013). These studies collectively indicate that the social environment heavily affects voluntary ethanol drinking, but the direction of this influence can depend upon additional variables, such as life history (e.g., adolescence versus adulthood) and genetic (individual/species/strain) differences.

With the current experiments, we asked whether differences in sociability expressed during adolescence are associated with how the social environment influences ethanol consumption. In the first study, daily volumes of sweetened ethanol intake (relative to water and food intake) were measured using a continuous, 2-bottle choice procedure in isolated and same-sex, pair housed mice. Using laboratory mice allowed us to utilize well-documented strain differences. We assessed adolescent BALB/cJ (BALB) and C57BL/6J (B6) strains that have been repeatedly shown to express differences in social interest, investigation and reward (Brodkin et al., 2004; Sankoorikal et al., 2006; Moy et al., 2007; Panksepp et al., 2007; 2008). We also evaluated FVB/NJ (FVB) mice, which have been described as highly social (Brodkin et al., 2004; Bolivar et al., 2007; Moy et al., 2007; Panksepp et al., 2013). An additional comparison group included MSM/mJ (MSM), which is a wild-derived inbred strain of a domestic mouse subspecies that harbors substantial genetic variation relative to the 3 other strains tested (Takada et al., 2013), and has been described as highly social compared to B6 mice (Takahashi et al., 2010).

We hypothesized that mice with high levels of social motivation (e.g., B6 and FVB) would be more inclined to voluntarily drink sweetened ethanol in a social context than in isolation whereas less social strains, such as BALB, might exhibit an opposite pattern of intake. We examined MSM mice without a specific hypothesis regarding what type of socially related drinking pattern to expect, but included this strain due to its genetic variation relative to classical inbred strains, with such differences arising in the wild prior to subsequent selection by humans.

In a second study, we employed lickometers to assess the circadian structure of sweetened ethanol drinking in socially versus isolate-housed mice from the highly social FVB strain. Taken together, our results suggest that enhanced voluntary ethanol drinking during adolescence is more strongly associated with the social environment (i.e., isolation) than genetic differences, and that this may be attributable to the emergence of a distinctive drinking pattern during social isolation.

METHODS

Subjects and animal husbandry

Mice were received from Jackson Laboratories (Bar Harbor, ME) and bred within our own colony at OHSU. Strains included C57BL/6J (‘B6’), BALB/cJ (‘BALB’) and FVB/NJ (‘FVB’), as well as MSM/Ms (‘MSM’), a wild-derived line originating from Mus musculus molossinus (Takada et al., 2013). All breeding and experiments took place in a standard vivarium (temperature, 21±1° C; humidity, 40–50%) that was maintained under a ‘reversed’ 12:12, light/dark cycle (‘lights on’ at 2100, ‘lights off’ at 0900). Routine colony work was conducted by a senior technician before 0900 and all procedures involving experimental animals were conducted by the first author under dim (≈20 lux) red light (light filters had an emission spectrum <1% overlap with the adsorption spectrum of mouse photoreceptors, LEE Filters Inc., Burbank, CA). To avoid potential influences of genetic drift or mutation, new breeders were routinely introduced to the colony and brother-sister matings were not conducted. Females were isolated ≈15 days post-coitus and allowed to give birth to their litters. Weaning took place on postnatal day (PD) 20–22 and entailed housing 2 same-sex siblings together (herein referred to as ‘Social housing’, Sh) or placing a single juvenile in social isolation (‘Isolate housing’, Ih). Weight measurements of paired Sh mice were always within 5% of each other. All experimental animals were housed in standard polycarbonate boxes (290×180×130 mm; Allentown Inc., Allentown, PA) lined with ≈10 mm of pelleted paper bedding (ECOfresh, Absorption Corp., Ferndale, WA) and containing a nestlet. Mice always had ad libitum access to food (Lab Rodent Diet 5001, Purina Mills) and H₂O. All experimental procedures were approved by the Institutional Animal Care and Use Committee at OHSU and followed the recommendations provided by the NIH Guide for the Care and Use of Laboratory Animals (8^th edition, ISBN 978-0-309-15401-7).

Experimental procedures

Experiment 1

256 mice from the 4 strains listed above were used for this experiment. On PD 29–31, two 25-ml tubes (fitted with rubber stoppers and metallic open-ended sippers bearing 3-mm apertures), each containing H₂O, were placed on the right and left side of each cage, respectively. Fluid volumes (0.2-ml increments), as well as the weight of each mouse and their food, were measured daily at 1700. This time of measurement (8 h into the circadian dark phase) was used to avoid disturbing mice during sleep and the 1^st half of the circadian dark phase, a preferred time for ethanol drinking by mice. Beginning on PD 33–35, one of the bottles was replaced with either a 3% ethanol (E) + 0.2% saccharin (S) solution (herein referred to as ‘E+S’ groups) or 0.2S-only (‘S’ groups). This constituted the beginning of a voluntary choice, continuous access procedure (Fig. 1), where E concentrations were increased (from 3-to-6-to-10% + 0.2S) every 4 days or where bottles in the S groups remained constant at 0.2S. All mice were given access to two E+S (or S-only) sippers to accommodate the high level of drinking that was observed (see Results section) and to eliminate potential competition between co-housed individuals. Thus, mice had access to 2 sippers containing E+S (or S-only) that were situated on the same side of the cage within 1 cm of each other (i.e., a total of three 25-ml tubes were in each cage mice). Following this phase of the experiment, bottles in all cages were transitioned to 10E-only (with no S) vs. H₂O for an additional 4 days. After this phase of the experiment, bottles in each cage were switched back to 10E+S vs. H₂O and drinking was monitored for an additional 1–4 days (2±1 days, mean±std. dev.) prior to blood collection at 1300 (i.e., 4 h into the circadian dark phase). All mice were initially introduced to 3E+S and ended the test phase drinking 10E+S because it was not clear if mice other than those from the B6 strain would drink metabolically relevant amounts of E without the addition of sweetener (e.g., Yoneyama et al., 2008). At the time of blood collection, mice from the E+S groups thus had a total of 18±1 days of voluntary exposure to E whereas mice from the S groups had a total of 6±1 days of voluntary E exposure. The side placement of bottles was rotated every other day throughout the entire experiment. Cage changes occurred 2 days prior to the introduction of H₂O sippers, 8 days later at the transition from 3E+S (or S-only) to 6E+S (or S-only), and another 8 days later at the transition from 10E+S (or S-only) to 10E-only. The entire experiment lasted ≈5 months and consisted of multiple, overlapping passes as experimental mice became available from the colony. In terms of the content of each pass and the position of each cage on the rack, the housing condition and genotype of cages was essentially random. Overall, there were 32 comparison groups (4 genotypes, 2 sexes, 2 housing conditions, 2 bottle conditions (E+S or S)) for which there were N=5–6 cages per group.

Figure 1. Schematic of experimental time line during the 2-bottle choice procedure.

Following weaning, mice remained undisturbed in Sh or Ih for ≈9 days. Bottles containing the solutions indicated above (vs. H₂O) were then introduced to the cages. Above the solid horizontal line is the time line of solutions (vs. H₂O) that were offered to mice in the ‘E+S’ groups and below the line are the solutions (vs. H₂O) that were offered to age-matched mice in the ‘S’ groups. Note that mice in S groups had access to E (10E or 10E+S) for a total of ≈6 days. Mice in the E+S groups had access to E for a total of ≈18 days. Additional details are provided in the Methods section.

Experiment 2

30 FVB mice were used for this experiment. On PD 42, mice that were housed continuously in Ih or Sh since weaning (see above) were injected with 2.5 g/kg E (i.p., 300-μl volume) at 1000 (1 h into the circadian dark phase) to evaluate E elimination (Grisel et al., 2002; Anacker et al., 2011a). Blood collection occurred at 1030 (30 min post-injection) or at 1300 (180 min post-injection). Each post-injection group contained 3–4 mice per housing condition/sex.

Experiment 3

54 FVB mice were used for this experiment and it took place in 3 passes (12 cages/pass, with equal numbers of Ih and Sh mice in each pass). Lickometer systems (Med Associates Inc., St. Albans, VT) were utilized within the home cage. Sippers were fitted to the Allentown cages such that they protruded ≈5 mm into the living space. A stainless steel grid (100 × 180 mm, 2-mm dowel diameter, spaced at intervals 9 mm on-center) was located on the floor near the sippers, allowing completion of the electrical circuit when mice licked. Each sipper was located equidistant (60 mm) from the center of a short side of each cage (20 mm above the steel grid) and licks were registered at a 10-ms resolution. Mice were introduced to cages equipped with the lickometers >5 days prior to the beginning of the experiment. Cages were changed 2 days before baseline H₂O measurements began on PD 41 and remained unchanged for the remainder of the experiment. Both bottles contained H₂O from PD 41–44 (days 1–4) and one bottle was switched from H₂O to 10E+S on PD 45 (10E+S vs. H₂O for days 5–12). As with Experiment 1, bottle orientation was rotated every other day. Volume measurements were taken at the beginning and end of day 9 and day 12. Licking measurements were taken for a final hour (1700–1800) lasting into day 13 (i.e., 8–9 h into the circadian dark phase), and blood collection took place beginning half way through this period. Each experimental group contained 9 cages per housing condition/sex.

Blood ethanol concentration assay

Samples were taken via trunk blood following CO₂ euthanasia. Cages and bottles remained in place until blood collection, and were then moved to a separate room located two (closed) doors from the testing room. For Experiment 1, blood collection began at 1300. Spacing of blood collection between cages was random across the genotype and social housing condition of each cage, and lasted no longer than 1400 (i.e., 5 h into the dark phase). Blood collection during Experiment 2 took place from 1030–1045 or 1300–1315. Blood collection during Experiment 3 took place from 1730–1800. Blood ethanol concentration (BEC) was determined via the alcohol oxidase reaction using the Analox GL5 system (Analox Instruments Ltd., Lunenburg, MA). An E standard (100 mg/dl) was run every 10 samples and the electrode was recalibrated if a standard was >5% outside of the target value. All samples were run in duplicate, each on a different day. Values in the figures and statistical outcomes were based on the average of these duplicate measurements. Estimates of inter-assay variation (R²) are presented in the respective figure legends.

Data analyses

For Experiments 1 and 3, data used for all graphical presentations and statistical outcomes regarding daily intakes and licks during Sh, respectively, were divided by 2.

For Experiment 1, daily H₂O and food intake from PD 30–33 was evaluated with a 3-factor ANOVA (housing condition × genotype × sex). During the E+S, S-only and E-only phases of the experiment, some intake values were too high to be considered genuine despite a lack of patent fluid leakage. Outliers were identified using Tukey box plots of total fluid consumed (ml) per gram body weight; data points ±1.5 of the inter-quartile range (roughly >2 std. deviations) of any experimental group (genotype-by-sex-by-housing condition) were excluded (i.e., 4.9 and 4.5% of the data from the E+S and S groups, respectively). 3-factor ANOVAs with drinking day as a repeated measure were used to evaluate daily E intake, S intake and preference ratios. Due to the large number of comparisons and potential differences in acquisition of voluntary fluid intake between strains, statistical analyses of E intake, S intake and preference ratios were restricted to the PD 42–45 (10E+S or S-only) and PD 46–49 (10E-only) time points. Differences between Sh and Ih mice during the earlier parts of the experiment nevertheless appeared very similar to the later parts of the experiment (see Figs. 2–4). Post-hoc comparisons were first explored with Tukey’s HSD matrices and then specific comparisons were evaluated with orthogonal contrasts. The final day of 10E+S intake and BECs were assessed with 4-factor ANOVAs (housing condition × genotype × sex × previous drinking history (E+S or S)) and post-hoc comparisons were conducted as described above.

Figure 2. Voluntary E intake as a function of housing condition, genotype, sex and bottle concentration across adolescent development.

Daily E intake for B6 (A, E), BALB (B, F), FVB (C, G) and MSM (D, H) are provided for females (A–D) and males (E–H), respectively. Sh mice from E+S groups (closed blue circles), Ih mice from E+S groups (open red circles) Sh mice from S groups (closed blue triangles) and Ih mice from S groups (open red triangles) are depicted in each panel. Intake values are provided on the ordinate and bottle concentration (age) is provided on the abscissa. All data are presented as the mean ± std. error and each value is constructed from 5–6 cages. *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing Sh vs. Ih in the E+S groups at PD 42–45 and PD 46–49, respectively. ⁺⁺P<0.01, ⁺⁺⁺P<0.001 for orthogonal contrasts comparing Sh vs. Ih in the S groups at PD 46–49.

Figure 4. Voluntary S intake as a function of housing condition, genotype and sex across adolescent development.

Daily S intake for B6 (A, E), BALB (B, F), FVB (C, G) and MSM (D, H) are provided for females (A–D) and males (E–H), respectively. Sh mice from E+S groups (closed blue circles), Ih mice from E+S groups (open red circles) Sh mice from S groups (closed blue triangles) and Ih mice from S groups (open red triangles) are depicted in each panel. Intake values are provided on the ordinate and bottle concentration (age) is provided on the abscissa. All data are presented as the mean ± std. error and each value is constructed from 5–6 cages. **P<0.01, ***P<0.001 for orthogonal contrasts comparing Sh vs. Ih in the E+S groups at PD 42–45. Note that these comparisons are numerically equivalent to the matched groups depicted in Figure 2. ⁺P<0.05, ⁺⁺P<0.01, ⁺⁺⁺P<0.001 for orthogonal contrasts comparing Sh vs. Ih in the S groups at PD 42–45.

For Experiment 2, a 3-factor ANOVA (housing condition × sex × time post-injection) was used to evaluate BECs and post-hoc comparisons were conducted via orthogonal contrast.

For Experiment 3, 2-factor ANOVAs (housing condition × sex) with drinking day as a repeated measure were used to evaluate licking and preference for H₂O and 10E+S, respectively. The final hour of licking and BECs were evaluated with 2-factor ANOVAs (housing condition × sex). Licking was also binned into 288 five-min epochs across the circadian cycle (see Results section) and differences between Ih and Sh for each epoch were evaluated via unpaired two-tailed t-test. Using this approach Type I error was corrected with a Bonferonni step-down procedure, beginning with a H_ø rejection criterion of α′=0.0001736. 3-factor ANOVAs (housing condition × sex × hour) were used to evaluate licking during specific hours of the circadian cycle. All post-hoc testing was conducted as described above.

For all experiments, linear regression analysis and Pearson’s correlations were conducted as needed.

RESULTS

Experiment 1: Voluntary drinking as a function of social vs. isolate housing

Water and food intake prior to introduction of sweetened solutions

Following weaning on PD 21, mice remained in Ih or Sh for ≈2 weeks prior to beginning the continuous access procedure on PD 34. Food and H₂O consumption was monitored on PD 30–33. At this age, all mice in Ih drank more H₂O than sex- and genotype-matched Sh mice (Table 1; effect of housing, F(1, 644)=240.3, P<0.0001), with the intake difference enhanced in females relative to males (housing × sex interaction, F(1, 644)=12.9, P=0.0004).

Table 1. Water intake as a function of genotype, sex and housing condition at PD 30–33.

All measurements were taken prior to the introduction E+S or S. Intake values are presented in fluid volume consumed per gram body weight. Values in the last column utilized the average from Sh groups for conversion into a percent-change. All data are presented as the mean ± std. error and each value is constructed from 10–12 cages.

Genotype	Sex	Housing	H2O intake (ml/g)	% change (from social)
B6	Female	Social	0.34±0.010
		Isolate	0.41±0.010	+20.6***
	Male	Social	0.34±0.011
		Isolate	0.39±0.011	+14.7***
BALB	Female	Social	0.37±0.011
		Isolate	0.49±0.010	+32.4***
	Male	Social	0.38±0.011
		Isolate	0.42±0.011	+10.5**
FVB	Female	Social	0.35±0.010
		Isolate	0.45±0.009	+28.6***
	Male	Social	0.33±0.010
		Isolate	0.39±0.009	+18.2***
MSM	Female	Social	0.44±0.011
		Isolate	0.54±0.011	+22.7***
	Male	Social	0.47±0.010
		Isolate	0.56±0.011	+19.1***

^**P<0.01,

^***P<0.001 for orthogonal contrasts comparing Sh vs. Ih.

Food intake was also increased in Ih mice at PD 30–33 (effect of housing, F(1, 662)=30.9, P<0.0001), but this effect was more sensitive to the particular genotype and sex of mice (Table 2).

Table 2. Food intake as a function of genotype, sex and housing condition at PD 30–33.

All measurements were taken prior to the introduction E+S or S. Intake values are presented in kilocalorie units per gram body weight. Values in the last column utilized the average from Sh groups for conversion into a percent-change. All data are presented as the mean ± std. error and each value is constructed from 10–12 cages.

Genotype	Sex	Housing	Chow intake (kcal/g)	% change (from social)
B6	Female	Social	0.91±0.040
		Isolate	1.05±0.039	+15.4*
	Male	Social	0.86±0.045
		Isolate	0.95±0.044	+10.5
BALB	Female	Social	0.99±0.041
		Isolate	1.02±0.039	+1.0
	Male	Social	0.96±0.044
		Isolate	1.02±0.042	+6.3
FVB	Female	Social	0.87±0.039
		Isolate	0.89±0.040	+1.0
	Male	Social	0.80±0.040
		Isolate	0.98±0.039	+22.5**
MSM	Female	Social	1.20±0.046
		Isolate	1.46±0.043	+21.7***
	Male	Social	1.15±0.042
		Isolate	1.29±0.043	+12.2*

^*<0.05,

^**P<0.01,

^***P<0.001 for orthogonal contrasts comparing Sh vs. Ih.

Body weight on the day prior to the introduction of sweetened solutions (PD 33) was not different between Ih and Sh mice (F(1, 170)=1.4. P=0.23), and there were no interactions of the housing conditions with genotype or sex. Regarding body weight, no difference between Ih and Sh mice was also found on the last day (PD 51±1) of the 2-bottle choice procedure (effect of housing, F(1, 170)=0.15. P=0.70).

Voluntary drinking and preference during access to a sweetened 10% ethanol solution

Regardless of genotype or sex, Ih mice consumed more E than age-matched (PD 42–45) mice in Sh when 10E+S was available (Figs. 2A–H, effect of housing, F[1,323]=290.0, P<0.0001). Strain-dependent variation in 10E+S intake also existed (effect of genotype, F[3, 321]=106.2, P<0.0001), with MSM mice (49±0.8 g/kg) drinking more than all other genotypes (orthogonal contrast, F[1, 309]=294.9, P<0.0001) and BALB mice (30±0.8 g/kg) drinking less than all other genotypes (orthogonal contrast, F[1, 309]=110.3, P<0.0001).

Across genotypes, females consumed more 10E+S than males (40±0.5 vs. 34±0.6 g/kg, respectively; effect of sex, F[1,323]=64.0, P<0.0001). In addition to finding that mice from all genotype-by-sex combinations drank more in Ih, the genotype and sex of mice also interacted with the housing conditions (housing × genotype × sex interaction, F[3, 321]=3.5, P=0.015). For instance, sex differences were not detectable for BALB mice in Sh (orthogonal contrast, F[1, 309]=0.92, P=0.34), but emerged when in Ih, with females drinking more than males (orthogonal contrast, F[1, 309]=29.4, P<0.0001). This same general pattern was true for adolescent FVB mice, but not B6 or MSM, where a sex difference (i.e., females > males) was consistent across the Ih and Sh conditions (statistics not shown). Notably, MSM mice from both sexes appeared to be highly sensitive to Ih, as individuals from the male and female groups consumed more 10E+S relative to all other comparison groups (orthogonal contrast, F[1, 309]=381.0, P<0.0001). A majority of the statistical effects (reported above) were similar when the proportion of daily calories derived from E was utilized as the dependent variable rather than 10E+S intake (Suppl. Fig. 1), indicating that increased 10E+S drinking during Ih was not due to reallocation of calorie intake to E.

Preference for 10E+S over H₂O (Figs. 3A–H) also exhibited strain dependent variation (effect of genotype, F[3, 321]=13.7, P<0.0001). For example, while the 10E+S preferences of B6 and FVB adolescent mice were similar (orthogonal contrast, F[1, 309]=0.53, P=0.47), they were both higher than those of BALB and MSM mice (orthogonal contrast, F[1, 309]=36.8, P<0.0001).

Figure 3. Preference for different concentrations of E+S, S-only or E-only as a function of housing condition, genotype and sex across adolescent development.

Fluid preferences for B6 (A, E), BALB (B, F), FVB (C, G) and MSM (D, H) are provided for females (A–D) and males (E–H), respectively. Sh mice from E+S groups (closed blue circles), Ih mice from E+S groups (open red circles) Sh mice from S groups (closed blue triangles) and Ih mice from S groups (open red triangles) are depicted in each panel. Preferences ratios are provided on the ordinate and bottle concentration (and age) is provided on the abscissa. Note that values at PD 30–33 assess a side preference for H₂O whereas values at PD 46–49 assess preference for 10E vs. H₂O in all groups. The 3 middle time points assess preference for E+S (circles) or S (triangles) vs. H₂O. All data are presented as the mean ± std. error and each value is constructed from 5–6 cages. *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing Sh vs. Ih in the E+S groups at PD 42–45 and PD 46–49, respectively. ⁺P<0.05, ⁺⁺P<0.01 for orthogonal contrasts comparing Sh vs. Ih in the S groups at PD 42–45 and PD 46–49, respectively.

In contrast to 10E+S intake, a main effect of housing on 10E+S preference was not detected (F[1, 323]=0.30, P=0.58), but females did express a higher 10E+S preference than males (90±0.6 vs. 85±0.6%, respectively; effect of sex, F[1, 323]=23.1, P<0.0001). Non-monotonic influences on 10E+S preference were manifest in a genotype × housing interaction (F[3, 321]=3.2, P=0.02) and a near significant 3-way interaction (genotype × housing × sex interaction, F[3, 321]=2.2, P=0.08). For instance, female MSM (Fig. 3D; F[1, 309]=5.3, P=0.02) and male FVB (Fig. 3G; F[1, 309]=4.1 P=0.04) mice had higher 10E+S preferences in Sh relative to Ih. By contrast, BALB females had higher preferences in Ih compared to Sh (Fig. 3B; F[1, 309]=4.1 P=0.04).

Voluntary drinking and preference during access to a non-sweetened 10% ethanol solution

When bottles were transitioned to 10E-only, many of the effects on fluid intake were similar to those in the 10E+S phase of the experiment: Ih mice from 5 of 8 genotype-by-sex combinations expressed higher levels of E drinking compared to age-matched Sh mice. (Figs. 2A–H, effect of housing, F[1,331]=59.8, P<0.0001). Furthermore, a strain dependence continued into the 10E phase of the experiment (effect of genotype, F[3,329]=15.2, P<0.0001), with MSM mice continuing to drink more than B6 and FVB mice (F[1,317]=29.1, P<0.0001), but not individuals from the BALB genetic background (F[1,317]=0.1, P=0.92).

A main effect of sex on 10E intake was not detected (F[1,331]=2.2, P=0.14); however, there was a 3-way interaction (genotype × housing × sex interaction, F[3,329]=3.3, P=0.02), with no differences in intake between Ih and Sh male mice of the B6, BALB and FVB genotypes (see Figs. 2E–G).

All main effects and interactions evaluating preference for 10E vs. H₂O were significant, including a 3-way interaction (genotype × housing × sex, F[3,329]=18.6, P<0.0001). Similar to the 10E+S phase of the experiment, preference for 10E in B6 females (Fig. 3A, orthogonal contrast, F[1, 309]=2.34, P=0.13) and B6 males (Fig. 3E, orthogonal contrast, F[1, 309]=2.33, P=0.13) was not sensitive to differential social housing. Ih females from the BALB (Fig. 3B) and FVB (Fig. 3C) genetic backgrounds, as well as Ih males of the MSM genotype (Fig. 3H), expressed higher preferences for 10E than sex-matched Sh mice. FVB males exhibited the opposite pattern, with higher preferences for 10E in Sh relative to Ih (Fig. 3G).

Comparisons to groups with access to a sweetened, ethanol-free solution

To determine the specificity of differential social housing on voluntary fluid intake, 10E+S drinking was compared with S drinking in age-matched groups. Ih mice (PD 42–45) drank more S than mice in Sh (Figs. 4A–H; effect of housing, F(1, 319)=58.8, P<0.0001). Increased S drinking in Ih was also dependent on the genotype and sex of mice (housing × genotype × sex interaction, F(3, 317)=3.6, P=0.013), with no differences between Ih vs. Sh males from the BALB (Fig. 4F) or MSM (Fig. 4H) genotypes. Additionally, females drank more S than males (effect of sex, F(1, 319)=6.9, P=0.009) and there were strain dependent differences between all genotypes (MSM>FVB>B6>BALB, statistics not shown). Preference for S over H₂O was slightly increased in Sh mice (92±0.06%) relative to Ih mice (90±0.06%) at PD42–45 (Figs. 3A–H; effect of housing, F(1, 319)=4.7, P=0.03).

Differences in voluntary fluid intake between Ih mice drinking 10E+S vs. S are illustrated in Figure 5 as a percentage-change in S consumption relative to age-matched individuals in Sh. Six of 8 genotype-by-sex combinations in Ih exhibited higher levels of S intake during 10E+S access relative to S-only access (Fig. 5).

Figure 5. Voluntary S intake as a function of genotype, sex and bottle content at PD 42–45.

The average daily level of S intake by age-, genotype- and sex-matched Sh mice (see Fig. 4) was used as a baseline value for converting daily amounts of S intake by individual Ih mice to a percent-change. Graphical organization follows the same structure of Figures 2–4, with genotypes aligned in columns and the sex of mice in rows. Percent-changes in isolation-induced S intake are provided on the ordinate and bottle conditions (10E+S; open red bars, or S; closed red bars) are presented on the abscissa. All data are presented as the mean ± std. error and each value is constructed from 5–6 cages of Ih mice. *P<0.05, **P<0.01, ***P<0.001 for one-tailed orthogonal contrasts comparing 10E+S vs. S-only.

On PD 46–49, mice in the S drinking groups were also transitioned to 10E-only access. Within this drinking context, Ih mice from all genotype-by-sex combinations consumed more 10E (Fig. 2; F(1, 302)=150.8, P<0.0001) and 4 out of of 8 genotype-by-sex groups in Ih exhibited a higher preference for 10E (Fig. 3; F(1, 302)=28.5, P<0.0001) than age-matched Sh mice.

Blood ethanol concentration

Following the E-only phase of the experiment, all mice were introduced to voluntary 10E+S access for another 2±1 days and blood samples were collected 4 h into the dark phase of the circadian cycle. Consistent with earlier parts of the experiment, 10E+S intake on the final day was strain dependent (effect of genotype, F(3,160)=14.0, P<0.0001), and Ih mice from all genotypes drank more 10E+S than Sh mice (Fig. 6; effect of housing, F(1,162)=66.6, P<0.0001). A sex difference in 10E+S intake neared significance on the final day of drinking (F(1,162)=3.8, P=0.05), with females (32±1.2 g/kg) drinking more than males (28±1.2 g/kg). Moreover, another near-significant effect (F(1,162)=3.6, P=0.06) was detected between mice from the E+S groups (32±1.2 g/kg, individuals with 18±1 prior days of E exposure; see Methods section) and the S groups (28±1.2 g/kg, 6±1 prior days of E exposure). There were no interactions between the independent variables (statistics not shown).

Figure 6. 10E+S intake during the final day of voluntary drinking.

E intake on the final day is provided on the ordinate, and former bottle condition (E+S or S) and genotype are presented on the abscissa. Note that intake values correspond to the final 20 h of drinking as daily volume measurements were taken at 1700, while the last volume measurement presented here was taken at 1300 (i.e., 4 h into the dark phase of the circadian cycle). All data are presented as the mean ± std. error, and each value is collapsed across sex and constructed from 10–12 cages. *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing Ih (red bars) vs. Sh (blue bars).

BEC (Fig. 7) was also strain dependent (effect of genotype, F(3, 251)=8.5, P<0.0001) and genotype interacted with whether mice had been previously exposed to E+S or S during the 2-bottle choice procedure (F(3, 251)=16.0, P<0.0001). Despite no differences in 10E+S intake (see above), B6 mice formerly of the S group (6±1 days of prior E exposure) had a higher BEC than B6 mice from the E+S group (18±1 days of prior E exposure; orthogonal contrast, F(1, 223)=12.9, P<0.001). By comparison, BALB (orthogonal contrast; F(1, 223)=5.5, P=0.02), FVB (orthogonal contrast; F(1, 223)=36.7, P<0.0001) and MSM (orthogonal contrast; MSM F(1, 223)=9.2, P=0.003) mice formerly of the E+S group had higher a BEC than their respective counterparts from the S group.

Figure 7. BEC 4 h into the dark phase on the final day of voluntary drinking.

BECs are provided on the ordinate, and former bottle condition (E+S or S) and genotype are presented on the abscissa. All data are presented as the mean ± std. error. Each value is collapsed across sex, and constructed from 10–12 mice for Ih (red bars) and 20–24 mice for Sh (blue bars). Values and statistical outcomes are derived from an average between duplicate BEC measurements (R²=0.97). *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing former E+S groups vs. former S groups.

In contrast to the final day of 10E+S intake (Fig. 6), BEC 4 h into the dark phase did not differ between Ih and Sh mice (Fig. 7; effect of housing, F(1, 253)=1.2, P=0.28). Overall, females had higher BECs than males (effect of sex, F(1, 253)=5.6, P=0.02). BECs of mice housed together were positively correlated (Pearson’s correlation, r=⁺0.62, d.f.=83, P<0.0001) whereas matched siblings that were housed in Ih did not exhibit significantly correlated BECs (r=⁺0.22, d.f.=30, P=0.23).

Experiment 2: Ethanol elimination as a function of social vs. isolate housing

To determine whether differential housing influenced E elimination, adolescent FVB mice were maintained in Sh or Ih from weaning (PD 21) to PD 42, and then injected with 2.5 g/kg E (i.p.). BEC was not different between Sh and Ih mice (F(1,28)=0.4, P=0.83) at 30 min (Ih – 267±12.2 mg/dl; Sh - 262±11.6 mg/dl) or 180 min (Ih – 47±11.9 mg/dl; Sh - 51±9.5 mg/dl) post-injection, respectively (housing × time interaction F(1,28)=0.1, P=0.76), indicating similar E elimination rates in Sh and Ih mice (effect of time, F(1,28)=345.4, P<0.0001). There was no effect of sex and no interactions (statistics not shown).

Experiment 3: Voluntary drinking of a sweetened 10% ethanol solution as a function of differential social housing and the circadian cycle

To characterize the enhanced drinking phenotype of Ih mice in relation to the circadian cycle, differentially housed adolescent FVB mice were monitored with lickometers during a continuous access, 2-bottle choice (10E+S vs. H₂O) procedure. Volume measurements on select days confirmed that licking and fluid intake were highly correlated (Suppl. Fig. 2) Consistent with Experiment 1, mice in Ih expressed higher levels of licking for 10E+S than Sh mice (Fig. 8A; effect of housing, F(1, 430) = 69.7, P<0.0001) and 10E+S licking increased across multiple days of exposure (housing × day interaction, F(11, 420) = 4.1, P<0.0001). For instance, Ih mice increased 10E+S licking from the first to last day of access (orthogonal contrast, F(1, 384) = 8.7, P=0.004) whereas daily 10E+S licking by Sh mice remained stable (F(1, 384) = 1.1, P=0.29). Fluid volume and body weight measurements after the final day of exposure confirmed higher 10E+S intake and preference in Ih mice compared to Sh mice (Suppl. Fig. 3). A main effect of sex on 10E+S licking was not detected (F(1, 430) = 2.8, P=0.10) and there were no interactions with other variables (statistics not shown).

Figure 8. Daily lick patterns and preference for 10E+S in differentially housed FVB mice.

Licking for 10E+S (Ih; open red circles, or Sh; closed blue circles) and H₂O (Ih; open red squares, or Sh; closed blue squares) are presented on the ordinate in (A). Preference ratios of Ih and Sh mice are presented on the ordinate in (B). The age of mice is presented on the abscissa. All data are presented as the mean ± std. error. Each value is collapsed across sex and constructed from 18 cages. *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing Sh vs. Ih for 10E+S licking and preference ratios. ⁺P<0.05, ⁺⁺P<0.01 for orthogonal contrasts comparing Sh vs. Ih for H₂O licking.

When one bottle was switched from H₂O to 10E+S, licking at the other H₂O bottle decreased (Fig. 8A) substantially for mice in Ih (orthogonal contrast for day 5 vs. 12, F(1, 384) = 9.5, P=0.002), but non-significantly for Sh mice (orthogonal contrast for day 5 vs. day 12, F(1, 384) = 0.1, P=0.82). Overall, these patterns of licking for 10E+S and H₂O resulted in a higher 10E+S preference for Ih mice relative to mice in Sh (Fig. 8B; effect of housing, F(1, 430) = 31.2, P<0.0001), and this difference increased from the first to final day of 10E+S access (orthogonal contrast for day 5 vs. day 12, F(1, 384) = 5.2, P=0.02).

To evaluate the circadian structure of 10E+S drinking, licks were tallied into 5-min epochs (i.e., bins) and plotted across the circadian cycle (Fig. 9A). Licking for 10E+S exhibited a biphasic circadian pattern, with the highest levels of licking occurring within 2 broad peaks that were roughly centered at ‘lights off’ and ‘lights on’. Seventy-two of 288 (25%) bins were higher for Ih mice than Sh mice at a <0.05 level-of-probability. However, utilizing a more conservative statistical approach (see Methods section), 7 bins were identified as being significantly higher in Ih vs. Sh. A majority of this increased licking in Ih transpired during a period located 8–9 h into the dark phase of the circadian cycle (i.e., zeitgeber time (ZT) 20–21; see Fig. 9A). This finding was confirmed during the last hour of measurement, with the licking rate of Ih mice ≈3-fold higher than Sh mice at ZT 20–21 (Fig. 9B; effect of housing, F(1,34)=14.4, P<0.001). At this circadian time BEC measurements for Ih mice were higher than for Sh mice (Fig. 9C; effect of housing, F(1,52)=10.5, P=0.002), with the BEC of females higher than males (effect of sex, F(1,52)=4.9, P=0.03), but there was no housing-by-sex interaction (F(1,52)=2.5, P=0.12). Notably, at ZT 20–21 more Ih mice (28%, 5 of 18 individuals) than Sh mice (0%, 0 of 36 individuals) reached a BEC ≥80mg/dl (Fisher’s exact test, P=0.003). Licking for 10E+S during the final hour was correlated with BEC during Ih (Pearson’s correlation, r=⁺0.59, d.f.=17, P=0.01) and Sh (r=⁺0.43, d.f.=35, P=0.01).

Figure 9. Circadian structure of 10E+S licking in differentially housed FVB mice.

Licking for 10E+S (Ih; red trace, or Sh; blue trace) is presented on the ordinate and circadian time in zeitgeber units (5-min intervals) is presented on the abscissa of (A). 10E+S licking during the last hour is presented on the ordinate of (B). BEC during the last hour is presented on the ordinate of (C). All data are presented as the mean ± std. error. Each value is collapsed across sex and constructed from 18 cages for (A–B), and from 18 mice (Ih; red bar) and 36 mice (Sh; blue bar) for (C). The first dashed horizontal line in (A) represents when BEC measurements were taken in Experiment 1 whereas the second line represents when BEC measurements were taken in (C). Values and the statistical outcome in (C) were derived from an average between duplicate BEC measurements (R²=0.94). *P<0.05, **P<0.01 for main effects comparing Sh vs. Ih on 10E+S licking and BEC. ⁺P<0.0001786 for Bonferroni corrected two-tailed t-tests comparing Sh vs. Ih on 10E+S licking.

The differential licking phenotype of Ih mice vs. Sh mice at ZT 20–21 appeared to develop over the 8 days of 10E+S access. For example, at ZT 16–17 both Ih mice (65±20.2% increase) and Sh mice (41±17.0% increase) exhibited higher levels of licking during the last 4 days of 10E+S access (PD 49–52) compared to the first 4 days (PD 45–48; effect of day, F(1,286)=10.0, P=0.002). While a similar effect of day was detected at ZT 20–21 (F(1,286)=19.8, P<0.0001), there also was a day-by-housing interaction (F(1,286)=11.7, P=0.0001) that was not found at ZT 16–17 (F(1,286)=1.0, P=0.31), indicating that increased licking over multiple days of 10E+S access was greater during a specific part of the circadian cycle. Overall this finding demonstrates that 10E+S drinking (at ZT 21–22) is enhanced in a time-dependent manner during Ih (225±49.6% increase at PD 49–52) compared to Sh (103±40.2% increase).

DISCUSSION

Our data collectively indicate that adolescent social isolation promotes consumption of sweetened ethanol across four genetically distinct mouse strains. We anticipated finding genotype- and sex-dependent effects, as these factors have been extensively described in previous studies (Belknap et al., 1993; Lancaster et al., 1996; Almeida et al., 1998; Cailhol and Mormède, 2001; Rhodes et al., 2007; Yoneyama et al, 2008; Butler et al., 2014; Varlinskaya et al., 2015). Despite finding genetic and sex differences that are consistent with previous work (e.g., B6 mice drinking more 10E+S than BALB (Belknap et al., 2003) levels comparable to FVB (Yoneyama et al., 2008), females generally drinking more than males), the most robust and consistent influence on adolescent drinking was the social environment: When 10E+S was available, Ih mice from all genotype-by-sex groups exhibited increased intake relative to mice in Sh. This finding was unexpected because strain selection for the study was based upon reports of genetically based differences in adolescent mouse sociability (Brodkin et al., 2004; Sankoorikal et al., 2006; Bolivar et al., 2007; Moy et al., 2007; Panksepp et al., 2007, 2008; Takahashi et al., 2010). Using a validated behavioral model (Panksepp et al., 2008), we previously found substantial differences in sociability between these strains (FVB > B6 = MSM > BALB) with the FVB vs. BALB comparison representing a ≈3.5-fold difference and no overlap between the respective distributions (Panksepp et al., 2013). Thus, counter to our primary hypothesis, different levels of sociability appear not to moderate the impact of the adolescent social environment on voluntary ethanol consumption.

It is worth noting that the social behavior of adolescent mice in their home cage (undisturbed by an experimenter for long periods of time) may differ from that measured during a relatively short laboratory test (where anxiety related to cage movement and experimenter handling may occur; but see Panksepp et al., 2008; Fairless et al., 2013). Although this possibility is deserving of additional consideration, including direct comparisons between ‘short-term’ and ‘long-term’ voluntary drinking procedures, our finding of isolation-induced consumption-increases across four genetically distinct strains underscores the significance of social contact during adolescence on voluntary drinking. Either the genetic substrates underlying social influences on adolescent drinking are largely conserved (even for the highly divergent MSM strain/sub-species) and/or the presence of peers during adolescence has a stronger effect compared to variation at relevant genetic loci.

There is a clear genetic contribution to alcohol intake in adults (Walters, 2002; Crabbe, 2014). Work in humans suggests that genetic contributions to alcohol abuse may be considerably stronger during adulthood compared to adolescence (van Beek et al., 2012, Edwards et al., 2015; also see Chorlian et al., 2013; Guo et al., 2015). Indeed, adolescent B6 mice voluntarily drink more ethanol than adults, an ontogenetic difference not observed in mice from the DBA/2J strain (Moore et al., 2010). Ethanol-related phenotypes, such as ataxia (Linsenbardt et al., 2009) and taste aversion (Moore et al., 2013), are also influenced by a gene-by-development interaction in mice. In the future, it will be important to systematically assess whether differences in sociability in both adolescents and adults affect voluntary drinking. Such a study would be essential to draw conclusions regarding the lack of influence of genetically based differences in sociability on ethanol drinking across the life span. Of note is that genetic variation appears to have a stronger effect on sociability during early adolescence compared to mice nearing sexual maturity (Panksepp et al., 2007)

It is well known that exposure to ethanol during adolescence can predict patterns of consumption in adulthood (e.g., Broadwater et al., 2013). Similarly, the most standard experimental approach for evaluating adolescent social experiences on ethanol consumption in laboratory rodents involves restricting social access during adolescence and subsequently evaluating drinking outcomes in adulthood (for a review see Butler et al., 2016). Despite the relevance to modeling peer influences on drug experimentation, there are relatively few animal studies that have concurrently examined the relationships between social variables and voluntary ethanol drinking during adolescence. Varlinskaya et al. (2015) recently found that both adolescent and adult male rats voluntarily drink more sweetened ethanol when tested in groups of 4–5 individuals relative to alone, whereas the opposite was true for adolescent females. Similarly, triads of adolescent male mice, but not females or adults, appear to consume more ethanol than matched groups tested alone (Logue et al., 2014). By contrast, but in agreement with previous work employing two rat strains (Hall et al., 1998), the present data in adolescent mice demonstrate that prolonged social isolation promotes ethanol drinking, with a much weaker contribution of genetic background. In addition to obvious procedural differences, comparison of the precise timing and duration of the isolation periods may help reconcile the apparent differences among these studies. The Logue and Varlinskaya studies included rodents that were socially reared, and then intermittently exposed as groups or alone to a single concentration of ethanol for a short period. The procedure used in the Hall study and our own entailed rearing individuals into dyads or social isolation, and then following a period of 8 weeks (Hall et al., 1998) or ≈2 weeks (the present study) monitoring daily intake during continuous access to an ethanol solution (ascending concentrations) and water in the home cage. A logical interpretation arising from these considerations is that voluntary ethanol drinking during adolescence can be modulated by the social environment in a dynamic manner, with the direction of changes dependent upon an interaction between the timing of social manipulation, it’s duration, and the timing and fashion in which ethanol is made available. This notion is consistent with studies that have identified variation in drug responsiveness across and within the classically defined stages of development (Hall, 1998; Terranova and Laviola, 2001; Laviola et al., 2003; Spear, 2015).

The amount of voluntary ethanol intake measured here was remarkably high. When bottles were switched to 10E+S adolescent BALB, B6 and FVB mice consumed sweetened ethanol upwards of ≈30–40 g/kg/day. Measurements of voluntary drinking in adolescent MSM mice were even higher, reaching >50 g/kg/day, but caution is suggested in interpreting this particular finding because these mice are <50% the weight of individuals from the other 3 strains. Thus, any spillage from bottle-sippers was over-represented in the consumption values for this genotype relative to the other genetic backgrounds. Although we did not measure spillage, it was nevertheless small, as evidenced by the very high preference (>90%) for all strains drinking from bottles that contained saccharin (i.e., levels of intake were low) For instance, if all H₂O intake (across the independent variables) during the 10E+S phase of the experiment is used as a liberal estimate of spillage, E+S intake would be reduced by 12.2%. Four hours into the dark phase, BEC reached upwards of 150 mg/dl in some groups, with continued high-levels of voluntary drinking after this time point (see Fig. 9). Higher BECs, along with relatively lower daily levels of voluntary ethanol intake, have been reported for mice selectively bred for high ethanol preference (Matson & Grahame, 2013). Interestingly, in an older study, adolescent mice were provided an all-ethanol diet and isolated mice from 2 inbred strains drank significantly more then socially grouped mice (≈43–47 g/kg/day in social isolation; Yanai & Ginsburg, 1976).

Our interpretation of these very high levels of voluntary ethanol intake and BEC derives from four considerations. First, well documented is that the addition of sweetener to ethanol promotes intake in numerous mouse strains (e.g., Yoneyama et al., 2008). This may be particularly true when ethanol access begins at low concentrations (e.g., 3%) since rodents perceive such stimuli as possessing a ‘sweet’ component (Kiefer and Lawrence, 1988). Second, chronically isolated rodents routinely consume substantially more ethanol than socially reared individuals (Deatherage, 1972; Parker & Radow, 1974; Schenk et al., 1990; Wolffgramm, 1990; Hall et al., 1998; Núñez et al., 2002; Juárez and Vázquez-Cortés, 2003; Doremus et al., 2005; Advani et al., 2007; Deehan et al., 2007; Ehlers et al., 2007; McCool & Chappell, 2009; Lopez et al., 2011; Chappell et al., 2013; Talani et al., 2013; Butler et al., 2014; Lopez & Laber, 2015). Third, adolescents are particularly prone to drink alcohol to intoxication (Hopson, 2013; see Broadwater et al., 2013), perhaps due to multiple differences in alcohol sensitivity compared to adults (Spear, 2014). Finally, the mere presence of multiple ethanol sippers in the home cage relative to water sippers can increase ethanol intake up to ≈20 g/kg/day in adult B6 mice (Tordoff and Bachmanov, 2003). Taken together, we hypothesize that the confluence of [i] sweetened ethanol (beginning at 3%) availability and [ii] social isolation [iii] during adolescence, along with [iv] two ethanol sippers, which constitutes four independent susceptibility factors, accounts for the high levels of voluntary consumption reported here. Interestingly, isolate-housed FVB mice in Experiment 3 drank much lower amounts of 10E+S than they did in Experiment 1 (≈17 g/kg/day; see Suppl. Fig. 3). These mice started drinking sweetened 10% ethanol (relative to 3% in Experiment 1) at PD45 (relative to PD34 in Experiment 1) with the presence of one ethanol sipper (instead of two as in Experiment 1).

The general pattern of increased intake during social isolation was not specific to sweetened ethanol access. Isolated mice from some groups consumed more food and water prior to introducing the test-solutions, and controls indicated that groups of isolated mice also drank more saccharin without ethanol and vice versa. Importantly, these isolation-related increases in consumption were more variable across the genotype-by-sex comparison groups and the associated effect sizes were smaller compared to sweetened ethanol. Mice had ad libitum access to food and water from birth, which suggests these stimuli imparted little salience during testing, and accordingly isolated mice only expressed modest increases in consumption of these ‘rewards’. By contrast isolation-related increases in saccharin-only and ethanol-only intake were larger, but not to the degree of sweetened ethanol drinking. Ethanol and sweeteners are highly novel to laboratory mice, and can be addictive (Avena et al., 2008; Koob, 2009). Moreover, the addition of sweetener to ethanol promotes intake in a majority of mouse strains (Yoneyama et al., 2008) and rodents prefer sweetened ethanol to an iso-caloric sucrose solution (Heyman, 1997), which indicates that an ethanol-saccharin ‘cocktail’ is likely more rewarding than either substance by itself. Thus, adolescent social isolation appears to result in a ‘hyper-consumptive’ state, with the magnitude of increased intake modulated by the desirability of the available reward. Changes in weight were not observed throughout the study, which is indicative of a high rate of adolescent metabolism.

In addition to the principal effect of differential social housing on intake, several additional influences on voluntary ethanol drinking and preference were detected. For instance, all isolated mice that drank saccharin-only for 12 days continued to consume more (relative to socially housed mice) when bottles were switched to ethanol-only. This ubiquitous effect may be related to ‘cross-sensitization’, which has been observed previously between calorically dense food-rewards and drugs of abuse (see Avena et al., 2008 for a review). Moreover, compared to all other socially housed groups, which exhibited reduced preference when bottles were switched from sweetened ethanol to ethanol-only, socially housed FVB males maintained very high levels of preference for ethanol-only. Thus, FVB males could be used in the future to study how the presence of social peers supports a strong preference for ethanol. Finally, during social housing, a correlation between the BEC of cage mates 4-h into the dark phase was found. This suggests that although socially housed adolescent mice drink less than their chronically isolated counterparts, they nevertheless drink in a temporally coordinated fashion. Although more detailed behavioral analyses are required to characterize this finding, it is particularly interesting because human adolescents are highly capable of influencing the drinking levels of peers (Astudillo et al., 2013; Lau-Barraco & Linden, 2014; Robinson et al., 2015; see Anacker et al., 2011b for an example in rodents), which could also affect when individuals drink.

In this respect, using lickometers in a replication experiment we found that isolated FVB mice consume more sweetened ethanol in part because they drink at times when socially housed individuals did not. Compared to socially housed FVBs, isolated individuals drank further into the dark (‘active’) phase of the circadian cycle, but they also exhibited a clear episode of voluntary drinking that occurred ≈8.5 h into the dark, resulting in heightened BEC. This increased drinking was distinct from the crepuscular rhythm found in both isolated and socially housed mice, which explains why no differences in BEC were found at 4 h into the dark phase.

Why do chronically isolated, adolescent mice approach and drink from a sipper that provides sweetened ethanol at a circadian time when socially housed mice do not? One possibility is that the presence of a social peer can serve as a protective factor with respect to ethanol withdrawal. Not exclusive from the latter possibility is that socially housed adolescent mice may be engaged in social interaction two-thirds of the way through the dark phase of the circadian cycle, perhaps deriving reward from such experiences (Panksepp & Lahvis, 2007). There is a rich empirical literature regarding interactions between natural (including social) and drug rewards (see Introduction) along with an underlying theoretical framework (Kelley and Berridge, 2002). The present findings suggest that growing up in a social group may buffer individuals from excessive reward seeking.

Our results also raise several additional issues: Are there other strains or mouse lines bearing mutations or transgenes that exhibit altered sensitivity to differences in the social environment on alcohol drinking? Does social isolation during adolescence result in similar circadian changes on voluntary drinking of ethanol-only or saccharin-only? Are there other changes in the behavioral repertoire of socially isolated adolescent mice, and might they relate to increased drinking (Butler et al., 2016)? Overall, the present findings support the long known fact that the presence of social companions and time into-the-day are some of the post influential factors on voluntary ethanol intake in humans (de Castro, 1990). Changes in the circadian structure of drinking, including drinking into one’s typical sleep-time and ‘morning drinking’, can typify alcohol use disorders (e.g., DSM-5 and CAGE questionnaire). Our findings suggest that the psychosocial factors contributing to the emergence of a distinctive alcohol-drinking pattern can be modeled in laboratory mice, and that the aggregate group of independent variables used here can be employed to promote very high levels of voluntary ethanol drinking in general.

Supplementary Material

Supp Fig S1. Supplemental Figure 1. Proportion of daily caloric intake from ethanol as a function of housing condition, genotype, sex and bottle concentration across adolescent development.

Caloric intake from E for B6 (A, E), BALB (B, F), FVB (C, G) and MSM (D, H) are provided for females (A–D) and males (E–H), respectively. Sh mice (closed blue circles) and Ih mice (open red circles) from E+S groups are depicted in each panel. Percentage of daily calories taken from E is provided on the ordinate and bottle concentration (age) is provided on the abscissa. All data are presented as the mean ± std. error and each value is constructed from 5–6 cages. At PD 42–45, there was a main effect of housing (F(1,319)=47.4, P<0.0001) and genotype (F(3,317)=13.1, P<0.0001). At PD 46–49, there was a main effect of housing (F(1,325)=38.9, P<0.0001), genotype (F(3,317)=13.1, P<0.0001) and a housing-by-genotype interaction (F(3,323)=8.4, P<0.0001). *P<0.05, **P<0.01, ***P<0.001 for orthogonal contrasts comparing Sh vs. Ih at PD 42–45 and PD 46–49, respectively.

Supp Fig S2. Supplemental Figure 2. Relationship between daily licking and consumed volumes.

Total volumes of 10E+S (Ih; open red circles, or Sh; closed blue circles) and H₂O (Ih; open red triangles, or Sh; closed blue triangles) consumed per cage during days 9 and 12 of the experiment are presented on the ordinate and total contacts detected by the lickometers are presented on the abscissa. Pearson’s correlation, R=0.81, d.f.=143, P<0.0001.

Supp Fig S3. Supplemental Figure 3. Voluntary 10E+S intake and preference as a function of housing condition.

E intake (A) and preference ratios (B) on the last day of the lickometer experiment are provided on the ordinate and housing condition is provided on the abscissa. All data are presented as the mean ± std. error. Each value is collapsed across sex and constructed from 18 cages. There were main effects of housing condition on E intake (F(1,34)=18.2, P=0.0002) and preference (F(1,34)=14.5, P=0.0006). ***P<0.001 for the main effects comparing Sh vs. Ih.