Interstrain differences in adolescent fear conditioning after acute alcohol exposure

Abstract

Adolescent sensitivity to alcohol is an important predictor of continued alcohol use and misuse later in life. Thus, it is important to understand the many factors that can impact alcohol sensitivity. Data from our laboratory suggested that susceptibility to alcohol-associated contextual fear learning deficits varied among adolescent and adult mice from two mouse strains. To investigate the extent of genetic background’s influences on adolescent learning after alcohol exposure, we examined how 9 inbred mouse strains differed in vulnerability to alcohol-induced contextual and cued fear conditioning deficits. We demonstrated significant strain- and sex-dependent effects of acute alcohol exposure on adolescent fear learning, with alcohol having most pronounced effects on contextual fear learning. Female adolescents were more susceptible than males to alcohol-induced impairments in contextual, but not cued, fear learning, independent of genetic background. Heritability for contextual and cued fear learning after alcohol exposure was estimated to be 31% and 18%, respectively. Learning data were compared to Blood Ethanol Concentrations (BEC) to assess whether strain differences in alcohol metabolism contributed to strain differences in learning after alcohol exposure. There were no clear relationships between BEC and learning outcomes, suggesting that strains differed in learning outcomes for reasons other than strain differences in alcohol metabolism. Genetic analyses revealed polymorphisms across strains in notable genes, such as Chrna7, a promising genetic candidate for susceptibility to alcohol-induced fear conditioning deficits. These results are the first to demonstrate the impact of genetic background on alcohol-associated fear learning deficits during adolescence and suggest that the mechanisms underlying this sensitivity are distinct from alcohol metabolism.

1. INTRODUCTION

Over 15 million people in the United States have been diagnosed with an Alcohol Use Disorder (AUD; SAMHSA, 2012). Most people with AUD experience some degree of cognitive impairment and related brain damage (Parsons, 1998). Even drinking patterns characterized as light-to-moderate have been shown to produce cognitive impairments over time, suggesting that alcohol-related damage to learning-related brain regions is not limited to those with AUD (Parsons & Nixon, 1998). These consequences are not limited to those currently using alcohol. Adults who were repeatedly exposed to alcohol during adolescence can exhibit long-lasting learning impairments (Mahedy et al., 2018). Poor cognitive functioning can contribute to poor job performance and increased risk of relapse during alcohol withdrawal (Evren et al., 2012; McIntyre et al., 2015). Because of their prevalence and role in the maintaining addiction, learning impairments after alcohol exposure are a major national health problem.

Adolescence is a period of rapid development for numerous brain regions, such as the hippocampus, which may make adolescents more susceptible to cognitive consequences of alcohol use (Spear, 2000). This is especially concerning as one in four teenagers age 16 to 17 report alcohol use. Additionally, those who use alcohol before the age of 17 are over twice as likely to become alcohol dependent than those who only use alcohol after turning 21 (SAMHSA, 2012). Adolescent alcohol use has been linked to impaired learning and executive function during adolescence (Hanson et al., 2011; Thoma et al., 2011) and adulthood (Mahedy et al., 2018). While it is known that genetic factors can influence alcohol sensitivity and consumption during adolescence, the genetic factors mediating susceptibility to alcohol-induced learning deficits during adolescence have not yet been examined.

AUD are heritable (Verhulst et al., 2015), with genetic factors influencing various aspects of AUD vulnerability during adolescence, including age of initiation and frequency of alcohol use (Thompson et al., 2020; Ystrom et al., 2014). These findings suggest that age and genetic factors may interact to determine sensitivity to alcohol, which can influence vulnerability to AUD. While a growing body of research is investigating the genetics of alcohol sensitivity, little is known about the genetic factors influencing alcohol’s cognitive effects during adolescence. These can be studied in rodent models, where cognitive effects of other drugs, such as nicotine, have been shown to be determined by genetic background (Portugal et al., 2012). Understanding the genetic factors underlying susceptibility to alcohol’s effects during adolescence is critical for understanding adolescent alcohol use vulnerability and outcomes.

The role of cognition in the development and maintenance of AUD is complex. Inbred mouse strain comparisons are a useful tool for investigating the role of genes and drug exposures in behavioral outcomes. These studies provide the opportunity to examine specific contributions of exposure and genetic background on phenotypes of interest in a controlled laboratory setting. Prior research has used mouse strain panels to demonstrate the impact of genetic background on alcohol conditioned taste aversion (CTA) during adolescence (Moore et al., 2013). Strain comparisons in adults have also successfully identified genetic factors involved in alcohol-induced hypothermia (Chesler et al., 2012; Crabbe et al., 1994), loss of righting reflex (Chesler et al., 2012), ataxia (Chesler et al., 2012), anxiety-related behavior (Putman et al., 2016), locomotor activity (Phillips et al., 1995), conditioned taste aversion (Risinger & Brown, 1996), conditioned place preference (Cunningham, 1995), and withdrawal-associated convulsions (Crabbe, 1998). Our laboratory has used strain comparisons to examine genetic contributions to fear learning after nicotine exposure (Portugal et al., 2012). Such studies demonstrate the utility of strain comparisons in addiction genetics. Recently, we completed a study comparing alcohol-induced contextual fear learning deficits in adolescent C57BL/6J and DBA/2J mice. After an acute alcohol treatment, C57BL/6J mice experienced a significantly greater contextual learning deficit than DBA/2J mice (Seemiller & Gould, 2021). Additionally, adult C57BL/6J mice experienced larger contextual learning deficits than adolescent C57BL/6J mice, while adolescent and adult DBA/2J mice experienced similar contextual learning deficits. These findings suggest that age and genetic background influence susceptibility to alcohol-induced fear conditioning deficits. This is important as alcohol-induced cognitive impairments have been shown to be positively associated with alcohol consumption in humans (Weafer & Fillmore, 2008).

Here, we examined alcohol’s effects on contextual and cued fear learning in male and female adolescent subjects from nine inbred mouse strains. Learning data were compared to BEC to assess the impact of strain differences in alcohol metabolism on strain differences in alcohol’s behavioral effects. Because the preliminary study found differences in contextual fear learning after alcohol treatment in C57BL/6J and DBA/2J mice and because prior research has shown that adolescents from the chosen strains varied in other alcohol phenotypes (Moore et al., 2013), we expected that the chosen strains would show variation in contextual fear learning after alcohol treatment. Additionally, because prior research suggested that strain differences in alcohol metabolism at this dose are minor (Linsenbardt et al., 2009), we predicted that learning outcomes would not be associated with BEC differences between groups. This research is the first to establish the genetic basis of alcohol-induced fear learning deficits in adolescent subjects and provides an important first step to better understanding the genetic factors involved in adolescent alcohol sensitivity and AUD.

2. METHODS

2.1. Subjects

Adolescent subjects (PND 38 +/− 3) were male and female mice from 9 inbred strains: C57BL/6J, C57BL/6NJ, DBA2/J, 129S1/SvImJ, A/J, BALB/cByJ, BTBR T+ tf/J, C3H/HeJ, and FVB/NJ (The Jackson Laboratory, Bar Harbor, ME, USA; JAX). This adolescent age was chosen based on our laboratory’s previously published studies examining drug effects in adolescent subjects (Connor & Gould, 2017; Gitik et al., 2018; Holliday & Gould, 2017; Kutlu et al., 2016, 2018; Seemiller & Gould, 2021). Eight of these strains were chosen because prior studies have shown that they differ in sensitivity to alcohol during adolescence (Moore et al., 2013). The 9th strain, C57BL/6NJ, was also examined in this study because of a mutation in Gabra2 that was found in the C57BL/6J strain and not the related C57BL/6NJ strain (Mulligan et al., 2019).

Group sizes for behavior were 9 per sex per strain per treatment group based on power analyses using preliminary data from our laboratory. Group sizes for BEC analyses were variable (7-9 per sex per strain) due to some unsuccessful blood collections. Subjects were housed in groups of three and had continuous access to water and food (LabDiet 5053, Lab Diet, St. Louis, MO, USA). Lights were on a 12-hour light/dark cycle, and procedures took place between 8 AM and 5 PM (during the light period). All animal use was performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and approved by the Pennsylvania State University Institutional Animal Care and Use Committee.

2.2. Experimental timeline

Subjects arrived from JAX on PND 31 +/− 3 and acclimated to the facility for seven days prior to behavioral procedures. On approximately PND 38, subjects received an acute saline or ethanol injection and underwent fear conditioning training fifteen minutes after treatment. On the next day (~PND 39), subjects underwent contextual testing followed by cued testing. Fear conditioning behavior was video-recorded for later analysis. One day after testing (~PND 40), subjects that were previously treated with saline were treated with an acute ethanol injection. Fifteen minutes after treatment, subjects were euthanized and blood was collected for assessment of ethanol metabolism (Figure 1).

Figure 1:

Timeline of behavioral experiments and blood collection in adolescent subjects. Only alcohol-naïve subjects (from the saline control fear conditioning group) were treated with alcohol on ~PND 40 for BEC assessment.

2.3. Alcohol dose and route of administration

Prior to fear conditioning training or BEC assessment, subjects were treated with an acute intraperitoneal injection of 0.9% saline solution or 1.5 g/kg 20% w/v ethanol in 0.9% saline solution. The dose and concentration of alcohol was based on prior publications (Gould, 2003; Gulick & Gould, 2008). Subjects received ethanol 15 minutes prior to fear conditioning on training day only, and not on testing day, based on prior data showing that alcohol preferentially impairs memory acquisition during fear conditioning (Gould, 2003).

2.4. Fear conditioning procedure

Subjects were fear conditioned according to behavioral procedures previously published by our laboratory (Portugal et al., 2012; Gould, 2003; Gould & Lommock, 2003; Gulick & Gould, 2007, 2008). Fear conditioning took place in noise-attenuating chambers (18.8 × 20 × 18.3 cm; MED Associates, St. Albans, VT, USA) located in a private behavioral suite. A 65 dB background noise was produced by background fans. To train subjects, mice were placed in chambers and observed for baseline freezing over 120 seconds. Two conditioned stimulus (CS, 30 seconds, 85-dB white noise)–unconditioned stimulus (US, 2 seconds, 0.45-mA foot shocks) pairings were presented, separated by 120 seconds. To test contextual fear retention, mice were returned to the same conditioning chambers one day later and monitored for 300 seconds. Later that same day, mice were placed in a novel context and observed for 180 seconds without a CS followed by 180 seconds of CS presentation to test cued fear retention. Fear conditioning training and cued testing contexts were distinguished by the use of different floor textures, wall patterns, and scents, as previously described (Kutlu & Gould, 2014). Behavioral responses were video-recorded and Ethovision software (Noldus Information Technology, Leesburg, VA, USA) was used to quantify freezing behavior across stages of behavioral testing. Freezing was defined as the absence of voluntary movement aside from respiration.

Fear conditioning provides an opportunity to examine dorsal hippocampus-dependent contextual fear learning and dorsal hippocampus-independent cued fear learning (Logue et al., 1997; Maren & Fanselow, 1997; Phillips & LeDoux, 1992). Our laboratory has consistently demonstrated that higher doses of alcohol impair and lower doses of alcohol enhance fear conditioning, preferentially affecting dorsal hippocampus-dependent contextual fear learning (Gould, 2003; Gould & Lommock, 2003; Gulick & Gould, 2007, 2008).

Because interpretation of fear conditioning in this paradigm relies upon using freezing as a proxy for learning (Fanselow, 1980a, 1994), it is important to consider how strain, sex, or alcohol exposure could affect freezing behavior independent of learning. Thus, we examined strain, sex, and alcohol effects on freezing during the baseline (first 120 seconds in training context, prior to shock or cue exposure), immediate (120 seconds in between shock-cue pairings), and pre-cue (first 180 seconds in cued testing chamber, prior to cue exposure and testing) periods of the fear conditioning protocol.

2.5. Ethanol metabolism assessment

Ethanol-naïve subjects were given an acute ethanol injection and euthanized fifteen minutes after the injection. This timepoint was chosen to match the timepoint of fear conditioning training after ethanol injections, which would allow estimation of BEC at the time of fear conditioning training. Blood was collected using cardiac puncture and samples were placed on ice for later processing. To isolate serum, samples were left to clot at room temperature for thirty minutes prior to being centrifuged for 10 minutes at 3000 g and 4°C. Serum was pipetted off the top of samples and stored at −80°C for later analysis. BEC was measured by Analox AM1 Alcohol Analyzer (Analox Instruments, Lunenburg, MA, USA) and recorded as mg/dl.

2.6. Statistical analysis

Fear conditioning data were analyzed two ways. First, strain, sex, and treatment effects on freezing frequency (% time) across all stages of fear conditioning (baseline, immediate, contextual, pre-cue, and cued) were analyzed using a three-way ANOVA (Strain x Sex x Drug). Baseline, immediate, and pre-cue data are included in Supplemental Materials 1. Second, a difference score was calculated for contextual and cued tests to control for learning differences between strains and address each strain’s unique sensitivity to alcohol-induced learning changes. To calculate alcohol-induced changes in contextual freezing, the average freezing score of the saline-treated group was subtracted from each subject’s individual freezing score in the alcohol-treated group, within strain and sex groups. That number was divided by the average freezing score of the saline-treated group and multiplied by 100 to calculate the relative percent reduction in contextual freezing after alcohol treatment [((individual freezing score of alcohol-treated subject – average saline freezing score) / average saline freezing score) x 100], as we have published previously (Seemiller & Gould, 2021). Using these values, a two-way ANOVA (Strain x Sex) was conducted to assess group differences in learning changes.

Any drug and sex interactions were followed with post-hoc Bonferroni-corrected independent samples t tests. Strain means, with combined male and female data, are shown in all figures. However, where sex effects were detected, data are shown separated by sex in Supplemental Materials 1. Strain and sex effects on BEC (mg/dl) were examined using a two-way ANOVA (Strain x Sex). For BEC data, drug was not included as a factor in the statistical model because all subjects were treated with alcohol. Statistical analyses were conducted in SPSS v26 (IBM, Armonk, NY, USA; α = 0.05) and figures were made using GraphPad Prism (La Jolla, CA, USA).

2.7. Heritability calculations and correlations

The heritability of fear conditioning behaviors was estimated based on observed strain differences. This was calculated by dividing the variance between strains (σ2st) by the sum of within-strain (σ2r) and between-strain variance: (h2) = σ2st/ (σ2st+ σ2r) (Owen et al., 1997; Rai et al., 2012). This represents the variance of the data set that can be attributed to genetic differences. Additionally, we tested for Pearson correlations in our fear conditioning data.

2.8. SNP analysis

To identify genetic variation related to phenotypic variation in fear conditioning across strains, a GenomeMUSter search was conducted using Mouse Phenome Database (Bogue et al., 2020; Keane et al., 2011; Yalcin et al., 2011), as we have done previously (Mooney-Leber et al., 2021; Seemiller et al., 2021, 2022). Across inbred strains, genetic variation was surveyed in alcohol- and learning-associated genetic targets identified through literature searches (Browman & Crabbe, 2000; Peirce et al., 1998; Wehner et al., 2004) including Alcp1 and Alcp3 (alcohol preference-associated loci), Dlg2 and Dlg3 (Discs Large MAGUK Scaffold Proteins), Gabra1, Gabra2, and Gabra3 (Gamma-Aminobutyric Acid Type A Receptor Subunits), Adh1 (Alcohol dehydrogenase), ApoE (Apolipoprotein E), ApoB (Apolipoprotein B), and Chrna7 (Cholinergic Receptor Nicotinic Alpha 7 Subunit). For some variants, functional consequences were predicted using Variant Effect Predictor (VEP) by Ensembl (Cunningham et al., 2022). These data are included in Supplemental Materials 2.

3. RESULTS

3.1. Contextual learning

Behavior during all stages of fear conditioning was analyzed for group differences, but because contextual fear learning was the primary phenotype of interest, results from this test will be discussed first. First, a three-way ANOVA (Strain x Sex x Drug) was used to analyze percent freezing during contextual testing among adolescent subjects. Main effects of strain (F_8,288=58.519, p<0.001) and drug (F_1,288=141.914, p<0.001), and strain x drug (F_8,288=5.993, p<0.001) and sex x drug (F_1,288=4.525, p=0.034) interactions were detected. Thus, strain, alcohol treatment, and sex significantly impacted contextual learning outcomes. No three-way interactions were detected. The main effect of drug represented alcohol-treated subjects exhibiting less freezing behavior than saline-treated subjects. To assess the strain x drug interaction, saline- and alcohol-treated subjects were compared within strain to determine which strains were significantly impacted by alcohol exposure. After Bonferroni correction, this analysis indicated that all strains except FVB/NJ and C57BL/6NJ experienced significant impairments in contextual learning after alcohol treatment (Table 1). To better understand the sex by drug interaction, a two-way ANOVA (Strain x Drug) was performed within sex groups. In both males and females, significant main effects of strain (males: F_8,144=31.690, p<0.001; females: F_8,144=27.649, p<0.001) and drug (males: F_1,144=47.537, p<0.001; females: F_1,144=99.274, p<0.001) and strain x drug interactions (males: F_8,144=2.263, p=0.026; females: F_8,144=4.855, p<0.001) were found. Across both sexes, freezing means of alcohol-treated subjects were lower than those of saline-treated subjects (male saline = 37.74%, male EtOH = 24.28%, female saline = 41.62%, female EtOH = 22.31%). Thus, alcohol impaired adolescent male and female contextual learning in a strain-dependent manner (sexes shown together in Figure 2A; sexes shown separately in Supplemental Materials 1).

Table 1:

Results from post-hoc Bonferroni-corrected independent samples t tests comparing saline and alcohol means within strain groups.

Strain	t	p
BALB/cByJ	t18=6.902	p<0.001
BTBR T+ tf/J	t34=3.922	p<0.001
129S1/SvImJ	t34=6.091	p<0.001
C57BL/6J	t34=3.787	p<0.001
FVB/NJ	NS	NS
C3H/HeJ	t28=3.625	p<0.001
DBA/2J	t34=3.624	p<0.001
A/J	t34=3.593	p<0.001
C57BL/6NJ	NS	NS

Figure 2:

A) Adolescent contextual freezing after pre-training saline or alcohol treatment. Main effects of strain and drug (EtOH<saline), and interactions of strain and drug and sex and drug were detected. B) Adolescent contextual difference scores, or percent reduction in contextual freezing due to pre-training alcohol treatment. Main effects of strain and sex (females having larger deficits than males) were detected. Strains are listed in order of proportionally greatest to smallest alcohol-related contextual learning deficit and remain in this order for later figures. n=9 per sex per strain per treatment. Double asterisks indicate p<0.001. Data are shown as mean +/− SEM.

Second, a difference score was calculated to assess alcohol-induced learning deficits within strains. Difference scores (representing percent reductions in learning due to alcohol) were analyzed using a two-way ANOVA (Strain x Sex). Main effects of strain (F_8,144=8.848, p<0.001) and sex (F_1,144=4.051, p=0.046) were detected, with females experiencing greater learning deficits than males. This analysis supported the results of the previous analysis, again demonstrating that learning effects of alcohol exposure are strain- and sex-dependent in the surveyed mouse strains (Figure 2B; Supplemental Materials 1).

Difference scores were used to calculate the heritability of alcohol-related learning deficits during adolescence. Heritability of adolescent contextual learning deficits after alcohol exposure was estimated to be 31%, indicating that genetic background was responsible for approximately 31% of the variance seen in adolescent contextual learning outcomes. Cumulatively, these finding confirm a significant impact of genetics and sex on susceptibility to alcohol’s effects on contextual fear learning among adolescents.

3.2. Cued learning

Cued fear learning was analyzed in the same manner as contextual fear learning. First, percent time freezing was analyzed using a three-way ANOVA (Strain x Sex x Drug). Main effects of strain (F_8,288=50.969, p<0.001) and drug (F_1,288=46.867, p<0.001) were found, with alcohol-treated groups showing less freezing behavior than saline-treated groups. No interactions were detected. This demonstrated that strain and alcohol treatment significantly influenced adolescent cued learning outcomes (Figure 3A).

Figure 3:

A) Adolescent cued freezing after pre-training saline or alcohol treatment. Main effects of strain and drug (EtOH<saline) were detected. B) Adolescent cued difference scores, or percent reduction in cued freezing due to pre-training alcohol treatment. A main effect of strain was detected. n=9 per sex per strain per treatment. Double asterisks indicate p<0.001. Data are shown as mean +/− SEM.

Then, difference scores were also calculated for cued testing to assess reductions in adolescent cued learning due to alcohol. A two-way ANOVA (Strain x Sex) found a main effect of strain (F_8,144=4.312, p<0.001) and no effects of sex on difference scores, suggesting that strain significantly impacted alcohol’s effects on cued learning (Figure 3B). This supports the initial analysis and suggests that both strain and alcohol treatment can influence cued learning outcomes. However, these results suggest that strain and alcohol may interact to alter cued learning, while the initial analysis of freezing did not detect an interaction of strain and drug treatment.

Difference scores were used to calculate the heritability of cued learning deficits due to alcohol treatment. Heritability of cued deficits was estimated to be 18%, meaning that strain effects could explain about 18% of the variability in the data set. Thus, while genetic background significantly impacted cued fear learning responses after alcohol, reduced heritability suggests a smaller genetic component for cued learning deficits after alcohol when compared to contextual learning deficits.

3.3. Baseline, immediate, and pre-cue freezing

To account for broader strain differences in freezing responses that could affect interpretation of learning, freezing responses to all stages of fear conditioning (baseline, immediate, and pre-cue) were analyzed using three-way (Strain x Sex x Drug) ANOVA. Additional analyses were completed to test the possible influence of generalized freezing on observed alcohol-induced learning deficits. Details are provided in Supplemental Materials 1, but in short, strain, sex, and drug effects were seen throughout baseline, immediate, and pre-cue stages of fear conditioning. Correlations of these behaviors with learning measures are analyzed later in Section 3.6 (Behavior Correlations). Strain and drug effects on learning were still observed after subtraction of generalized (pre-cue) freezing from contextual and cued freezing scores (Supplemental Materials 1).

3.4. Heritability of fear learning in saline-treated subjects

Fear conditioning behavior independent of drug exposure has been shown to be heritable in prior research (Mooney-Leber et al., 2021; Owen et al., 1997; Wehner et al., 1997). Here, we calculated heritability estimates of fear conditioning in saline-treated subjects to assess consistency between our study and previous studies. Heritability of contextual freezing in the saline-treated group was estimated to be 60% and heritability of cued freezing in saline-treated subjects was estimated to be 50%, consistent with prior reports (Mooney-Leber et al., 2021).

3.5. Blood ethanol concentration

Alcohol-related behaviors can be affected by the rates of alcohol absorption and metabolism. Thus, we examined BEC of alcohol-naïve subjects (from the saline-treated fear conditioning control group) treated with the same dose of alcohol, 1.5 g/kg, that was used for behavioral studies, at a timepoint (15 minutes post-injection) representative of what BEC were at the start of fear conditioning training in alcohol-treated subjects. A two-way ANOVA (Strain x Sex) of adolescent BEC found no significant impact of strain or sex (Figure 4). This suggests that adolescent BEC were similar across strain and sex groups at the start of fear conditioning, suggesting that other biological mechanisms are responsible for observed strain and sex differences in learning after alcohol.

Figure 4:

Adolescent BEC 15 minutes after alcohol treatment. No effects of strain or sex were detected. n=7-9 per sex per strain. Data are shown as mean +/− SEM.

While initial analyses did not detect strain or sex differences in BEC, follow-up comparisons with the adolescent data set more directly addressed the question of whether alcohol metabolism differences among adolescents could be responsible for group differences in learning after alcohol. Adolescent strain means of contextual and cued difference scores were compared with BEC. Neither contextual (R=0.303, p=0.427) nor cued (R=0.620, p=0.075) difference scores significantly correlated with BEC, indicating that strain differences in alcohol metabolism were unlikely to be related to learning outcomes after alcohol exposure among adolescents (Supplemental Materials 1).

3.6. Behavior correlations

To better understand how genetic susceptibility to alcohol-induced learning deficits relate across learning tests and to other alcohol phenotypes, adolescent learning difference scores were compared. First, adolescent contextual and cued difference scores were compared to one another. There was not a significant correlation between adolescent contextual and cued difference score strain means (R=0.665, p=0.051), although there was a trend supporting a positive relationship between these outcomes (Supplemental Materials 1). This supports prior literature suggesting that contextual and cued learning tasks have some overlapping but still unique neural mechanisms (Hunsaker & Kesner, 2008; Maren & Fanselow, 1997; Phillips & LeDoux, 1992).

Additionally, to assess whether group differences in freezing behavior influenced interpretation of group differences in learning outcomes, strain means for freezing within saline- or alcohol-treated groups across all stages of fear conditioning were compared (Supplemental Materials 1). Among saline-treated subjects, contextual freezing strain means positively correlated with immediate (R=0.743, p=0.022) and pre-cue (R=0.855, p=0.003) freezing means, while cued freezing did not correlate with freezing during any other fear conditioning stages. Among alcohol-treated subjects, for contextual freezing strain means, significant positive correlations were observed with baseline (R=0.796, p=0.010), immediate (R=0.842, p=0.004), and pre-cue (R=0.901, p<0.001) freezing. For cued freezing strain means of alcohol-treated subjects, significant positive correlations were observed with immediate (R=0.724, p=0.027) and pre-cue (R=0.681, p=0.043) freezing.

3.7. SNP analysis

To assess possible genetic substrates involved in strain differences in alcohol’s effects on fear conditioning, polymorphisms in relevant genes were found using publicly available data for all surveyed strains. This analysis identified polymorphisms across inbred strains in all 13 surveyed alcohol- and learning-related genetic targets, including genes broadly related to GABAergic signaling and cholinergic signaling such as Gabra1, Gabra3, Dlg2, Dlg3, and Chrna7 (Supplemental Materials). A follow-up analysis in a subset of these strains, C57BL/6J and C57BL/6NJ, utilized VEP to predict functional consequences of the sequence variation in select genes, including Chrna7. This analysis revealed potential genetic regulatory consequences of strain-specific polymorphisms in ApoB and Chrna7, which could be related to observed behavioral differences across strains (Supplemental Materials 2).

4. DISCUSSION

4.1. Overview

This study characterized the influence of genetic background on learning deficits after adolescent alcohol exposure. Adolescent mice from multiple genetic backgrounds were treated with alcohol and underwent fear conditioning. To understand the specificity of alcohol’s effects on contextual fear learning, strain differences in related measures were also assessed, including cued learning, freezing during non-testing stages of fear conditioning procedures, and alcohol metabolism. Genetic background consistently impacted fear conditioning responses after alcohol exposure, with heritability estimates for contextual and cued fear conditioning after alcohol exposure being 31% and 18%, respectively. Fear conditioning performance was not clearly associated with strain differences in alcohol metabolism as measured by BEC. Together, these experiments demonstrate that genetic variability contributes to adolescent learning deficits after alcohol exposure in a manner that is independent of alcohol metabolism.

4.2. Contextual fear learning

All learning measures were analyzed two different ways: as raw freezing percentages (percent time spent freezing during the test) and as difference scores (percent reduction in freezing in the alcohol groups when compared to the saline control groups). Using raw freezing data in the saline-treated animals, we estimated heritability of contextual freezing to be 60%. This estimate is higher than previous reports in populations with less genetic diversity (Wehner et al., 1997) while similar to prior reports in similarly heterogenous populations (Mooney-Leber et al., 2021).These analyses also confirmed significant variation in adolescent contextual learning due to alcohol across genetic backgrounds. Adolescent contextual learning deficits after alcohol were pronounced across nine strains, with reductions in learning (difference scores) ranging from an average of −25% in the C57BL/6NJ strain to −81% in the BALB/cByJ strain. Notably, genetically similar C57BL/6J and C57BL/6NJ exhibited different contextual learning deficits after alcohol, with difference scores of −45% and −25%, respectively, which could be related to known substrain differences in GABAergic function and alcohol responses (Chavez et al., 2021; Mulligan et al., 2019). Strain, or genetic background, could account for 31% of the variance in the adolescent contextual learning dataset, suggesting that variability in the effects of alcohol on adolescent contextual learning are 31% heritable. Prior studies in mice have found that alcohol responses and contextual fear learning are heritable (Linsenbardt & Boehm, 2013; Owen et al., 1997), and our laboratory previously found that nicotine withdrawal-induced changes in contextual learning are heritable (Goldberg et al., 2021). However, the heritability of acute alcohol-induced changes in adolescent contextual learning was not previously known, making this study a valuable first step in characterizing genetic influence on this phenotype.

Sex effects were also observed among adolescents, with adolescent females generally experiencing greater reductions in contextual learning due to alcohol than adolescent males. While noteworthy, these sex differences were observed at a time of rapid sexual development and may require more careful study at other adolescent ages. However, our findings provide an example of protection from alcohol’s effects that is male-specific, which may contribute to explanations of women showing faster progression to treatment for AUD and higher risk for negative consequences of alcohol including cognitive impairments (Agabio et al., 2017; Hernandez-Avila et al., 2004; Miller et al., 2009).

4.3. Cued fear learning

Prior research reported that cued fear learning is less susceptible to disruption by alcohol than contextual fear learning in adult and adolescent C57BL/6J and DBA/2J mice (Gould, 2003; Seemiller & Gould, 2021). Across the surveyed adolescent population, this finding remained consistent. Adolescent cued fear learning deficits (difference scores) were smaller than adolescent contextual fear learning deficits after alcohol, ranging from an average of −7% in the 129S1/SvImJ strain to −38% in the BTBR T+ tf/J strain. A comparison of strain means for contextual and cued deficits after alcohol exposure (difference scores) revealed that the two measures did not correlate, suggesting that a significant common genetic basis does not underlie contextual and cued learning outcomes after alcohol exposure.

The heritability of cued learning deficits after adolescent alcohol exposure was estimated to be 18%. This is much lower than the heritability estimate for cued freezing among saline-treated subjects (50%), suggesting that genetics contribute more dramatically to cued freezing generally than susceptibility to alcohol-indued cued learning deficits. The heritability estimate for cued learning deficits due to alcohol (18%) was also much lower than the heritability estimate for contextual learning deficits (31%) due to alcohol, supporting previous literature suggesting that alcohol preferentially affects contextual fear learning (Gould, 2003). It also helps explain the lack of consistency in strain interactions with drug treatment that were detected in the analysis of cued learning via raw freezing vs difference scores, where an effect of strain is seen in cued difference scores (with difference scores representing drug responses, suggesting that strain determines drug responses) but there is no interaction of strain and drug in raw cued freezing values (suggesting that strain does not determine drug response). While these findings provide some support for an interaction effect of strain and alcohol on cued learning, their synergistic effects on this learning outcome appear to be more subtle than for contextual fear conditioning. This may suggest that the dorsal hippocampus, which is critically involved in contextual fear conditioning but not cued fear conditioning (Hunsaker & Kesner, 2008; Maren & Fanselow, 1997; Phillips & LeDoux, 1992), may be highly vulnerable to the effects of alcohol compared to neural areas involved in both contextual and cued fear conditioning. Consistent with this idea, other research supports hippocampal vulnerability to alcohol, showing that acute alcohol exposure suppresses hippocampal place cell activity (White & Best, 2000) and inhibits hippocampal long-term potentiation (Blitzer et al., 1990). Further, adolescent hippocampal tissue appears to be more susceptible to inhibition of long-term potentiation by alcohol than adult hippocampal tissue, suggesting that hippocampal sensitivity to acute alcohol may be even more pronounced in adolescents (Pyapali et al., 1999).

While sex differences in cued fear conditioning have been reported in other studies (Goldberg et al., 2021; Seemiller & Gould, 2021), none were detected in this study. This is likely due to the focus on adolescent subjects and the number of strains surveyed. In comparison to adolescent males, adolescent females were more susceptible to alcohol’s effects on contextual learning but equally susceptible to alcohol’s effects on cued learning. Thus, we identified adolescent sex differences that were specific to contextual, and not cued, fear conditioning.

4.4. Baseline, immediate, and pre-cue freezing

In contextual and cued fear learning tests, freezing behavior is used as an index of learning, where higher freezing represents stronger fear learning. However, expression of freezing behavior unrelated to conditioned context or cue exposure can vary between inbred mouse strains (Seemiller et al., 2021). This can complicate the direct interpretation of strain differences in freezing as strain differences in learning. Freezing behavior throughout all stages of fear conditioning was analyzed to examine the influence of these potential group differences.

Various effects of strain, sex, and alcohol were seen on baseline, immediate, and pre-cue freezing in adolescent and adult subjects. Freezing frequency during all stages of fear conditioning (baseline, immediate, contextual, pre-cue, cued) was found to be strain-dependent (at least a main effect of strain), demonstrating a robust influence of genetic background on conditioned and non-conditioned freezing behavior throughout fear conditioning as we have described previously (Seemiller et al., 2021). Sex effects on baseline, immediate, and pre-cue freezing were variable, with main effects on immediate (males<females) and pre-cue (females<males) freezing. Alcohol consistently decreased baseline freezing, likely due to alcohol administration being shortly before the baseline period and because of alcohol’s motor-impairing effects (Crabbe et al., 2005). However, alcohol’s effects on immediate freezing were strain-dependent. Alcohol administered prior to training (but not testing) consistently decreased pre-cue freezing, suggesting a possible reduction in a generalized fear response after fear conditioning. Note, prior work demonstrated that alcohol disrupts fear conditioning whether given before training only or both training and testing (Gould, 2003).

Comparisons of freezing across stages of fear conditioning were conducted to determine whether alcohol’s effects on freezing behavior generally could explain some of alcohol’s effects on learning-associated freezing behaviors (during contextual and cued tasks). Generally, contextual and cued freezing correlated with freezing in other stages more frequently among alcohol-treated subjects than among saline-treated subjects. Comparisons among saline-treated subjects revealed positive correlations between contextual freezing and immediate and pre-cue freezing, while no correlations were found with cued freezing. Among alcohol-treated subjects, contextual freezing significantly correlated with baseline, immediate, and pre-cue freezing, and cued freezing correlated with immediate and pre-cue freezing. This could suggest that alcohol affects freezing behavior separately from its effects on learning. However, additional analyses examining alcohol’s effects on contextual and cued freezing with pre-cue freezing subtracted from those analyses revealed that alcohol’s effects on learning are robust to this adjustment.

Considering these findings, it is possible that group differences could have altered propensity for freezing behavior in a way that changed interpretation of learning responses. However, it would be difficult to parse this apart further because while these measures were not used as learning metrics in the current study, some may also represent forms of learning. Specifically, immediate (post-shock) freezing represents a subject’s first response to shock and cue exposure, and other studies have used freezing during this period to measure the strength of that association (Fanselow, 1980b). Pre-cue freezing, while not a direct response to a context or cue, could also be interpreted as a generalized fear response to diffuse fear conditioning stimuli. It is important to note that alcohol was only given on training day (which includes baseline and immediate stages) and not on testing day (which includes contextual, pre-cue, and cued stages). If alcohol intoxication exerted locomotor effects on training day that affected baseline or immediate freezing, this was unlikely to be the case on testing day, as alcohol was not given on that day. A possible explanation for the stronger correlations of freezing behaviors observed in the alcohol-treated group relative to the saline-treated group could be that alcohol produced anxiolytic and/or locomotor effects (Durcan & Lister, 1988) that reduced freezing during baseline and immediate periods. However, alcohol-induced contextual learning deficits observed during the contextual test were greater than changes in freezing during other periods. Thus, decreased freezing, possibly related to increased locomotion, cannot solely explain the deficits in contextual fear conditioning. It may be that if the alcohol-treated mice have not learned the context well on training day, they may view the context as novel on testing day and have more novelty-induced exploratory behavior (Gould et al., 2009).

To account for the possible issue of strain- or sex-specific tendencies to freeze more or less, contextual and cued fear learning were also examined as difference scores, or freezing frequencies after alcohol exposure represented in proportion to freezing of the control groups. This allowed us to examine alcohol-related changes in freezing while controlling, to some degree, for strain- and sex-specific freezing patterns. Results from the analysis of difference scores mostly reinforced those of the analysis of raw freezing values, showing strain- and sex-specific effects of alcohol on learning outcomes, even while controlling for group freezing differences. This supported the initial conclusions that alcohol impacts learning to variable degrees across strains and sexes.

4.5. Alcohol metabolism

It is well-understood that inbred mouse strains have variable rates of alcohol metabolism that can influence observed strain differences in alcohol-related behaviors (Bagley et al., 2021; Crabbe et al., 2003; Linsenbardt et al., 2009). Therefore, it was crucial to assess alcohol metabolism across all strains and sexes included in this study. BEC measurement was taken after the same alcohol dose and time period as subjects that underwent fear conditioning in the alcohol group to approximate the BEC and level of intoxication of subjects during fear conditioning training. No significant strain or sex effects were detected at this dose and time point. Comparisons of strain means for BEC and contextual or cued difference scores found that BEC did not correlate with learning variables. Additionally, prior literature assessing BEC after comparable alcohol doses have reported similar lack of variability across inbred strains (Crabbe et al., 2003; Linsenbardt et al., 2009). Thus, while future studies could examine this issue more thoroughly by assessing BEC at additional time points, current evidence suggests that strain differences in fear learning after alcohol exposure are largely unrelated to strain differences in alcohol metabolism in adolescent subjects. This could imply that factors other than metabolism, such as alcohol effects on the hippocampus, are mainly responsible for adolescent strain differences in learning after alcohol.

4.6. Behavior correlations

To assess whether genetic vulnerability to alcohol-induced learning deficits generalized to other alcohol phenotypes, adolescent strain means were compared within our behavioral data set. No significant correlation was detected between strain means for contextual and cued difference scores (representing learning deficits due to alcohol), suggesting that the strains with the most severe contextual learning deficits did not necessarily experience the most severe cued learning deficits. This supports prior literature suggesting unique neurocircuitry underlying contextual and cued fear learning (Hunsaker & Kesner, 2008; Maren & Fanselow, 1997; Phillips & LeDoux, 1992).

4.7. SNP analysis

To investigate possible genetic substrates underlying the strain differences in behavior observed in these studies, we identified polymorphisms in alcohol- and learning-related genes. Substantial variability was detected across strains, suggesting many possible mechanistic explanations for strain differences in behavior. The observed variation is likely to contribute to the observed strain differences in alcohol’s effects on fear conditioning as variation in Gabra genes have been associated with alcohol sensitivity and dependence (Covault et al., 2004; Yang et al., 2017) and variation in Dlg genes has been associated with alcohol consumption and cognition (Hitzemann et al., 2020; Nithianantharajah et al., 2013). Notably, variation was detected in Chrna7 which has been associated with alcohol-induced fear conditioning deficits previously. Specifically, prior literature showed that mice lacking functional Chrna7 were less susceptible to alcohol-induced contextual fear learning deficits (Wehner et al., 2004). Levels of nicotinic acetylcholine receptors, including those containing α7 subunits encoded by Chrna7, vary substantially across inbred mouse strains and in the hippocampus (Marks et al., 1989; Stevens et al., 1996). Subsequent VEP analysis revealed potential regulatory consequences of the variants observed in Chrna7 between C57BL/6J and C57BL/6NJ strains. Chrna7 is highly expressed in the hippocampus (Whiteaker et al., 1999), making it likely that the detected polymorphisms and predicted variant effects contribute to the large range of effects of alcohol observed in the contextual fear conditioning task. Thus, the role of Chrna7 in alcohol-associated learning deficits is an intriguing avenue for further study.

4.8. Limitations and future directions

A prior study from our laboratory (Seemiller & Gould, 2021) found that strain effects on contextual fear learning after alcohol exposure vary by age. Specifically, we found that adult C57BL/6J mice experience larger contextual learning deficits than adolescent C57BL/6J mice, while DBA/2J mice experienced similar contextual learning deficits across adolescence and adulthood. This demonstrated that genetic background can have unique effects on fear learning outcomes during adolescence, which demonstrated the importance of studying strain differences during adolescence, as was done in the current study. Future studies can use these studies to identify the mechanisms underlying observed strain differences. However, strain differences identified in this adolescent study may not generalize to adults, which may also be examined separately in later studies.

Males and females were included in all described experiments. Thus, it is possible that sex hormones may have influenced these data. Studying effects of estrous cycle or strain differences in puberty onset was beyond the scope this experiment. We also assessed learning deficits after 1.5 g/kg alcohol treatment, a dose sufficient to demonstrate strain differences in our preliminary work (Seemiller & Gould, 2021). While it may have been ideal to use multiple doses to examine strain differences in alcohol responses, this dose was used to establish heritability. Intraperitoneal injections were chosen to allow standardized dosing between groups. A voluntary access administration paradigm, while translationally relevant, would have been vulnerable to known strain and sex effects on voluntary alcohol intake (Caruso et al., 2018; Chester et al., 2008; He et al., 1997; Strong et al., 2010; Yoneyama et al., 2008). Additionally, this study examined how one acute alcohol exposure could affect one learning event. While it would be valuable to examine how repeated alcohol exposures affect learning at protracted timepoints, this study established a foundation for genetic influences on acute alcohol effects. This is a valuable first step in understanding the early stages of alcohol use and initial sensitivity to alcohol.

While present data suggest that this is not the case, it is possible that differences in alcohol metabolism may have contributed to differences in alcohol responses. Our analysis indicated that strains did not significantly differ in BEC, but we assessed one time point to represented BEC at the time of fear conditioning training. Future studies could test BEC at additional time points or use additional doses to control for different BEC after alcohol treatment.

4.9. Conclusions

Alcohol sensitivity is impacted by genetic background, and previous work from our laboratory suggested that genetic impacts on learning after alcohol exposure may be unique in adolescent subjects (Seemiller & Gould, 2021). The described studies evaluated alcohol’s impacts on fear conditioning and BEC in adolescent mice from different genetic backgrounds. A survey of contextual and cued fear learning among adolescent mice from nine inbred strains revealed significant variation in contextual and cued fear learning outcomes due to genetic background and a significant difference in the effects of alcohol across strains, indicating genetic contribution to phenotypic differences. Polymorphisms in relevant genes such as Chrna7 were identified and may underlie observed phenotypic differences across strains. Strain differences in contextual fear learning outcomes were not clearly related to measured BEC or cued fear conditioning. We estimated heritability of adolescent alcohol-induced learning deficits to be 31% for contextual and 18% for cued learning tasks. These studies are the first to demonstrate robust effects of alcohol exposure on adolescent fear conditioning across a genetically diverse population with greater effects in contextual fear conditioning than cued fear conditioning. More broadly, this work contributes to our understanding of how learning and memory processes work across genetically heterogenous populations, and it demonstrates how drugs such as alcohol can interact with these complex processes.