Pre-training naltrexone increases conditioned fear learning independent of adolescent alcohol consumption history

Abstract

Our previous research has shown a relationship between low voluntary alcohol consumption and high conditioned fear in male Long Evans rats. Here, we examined whether differences in the endogenous opioid systems might be responsible for these differences. Rats received 6 weeks of chronic intermittent to 20% alcohol (v/v) or water-only from PND 26–66. Based on their consumption during the last 2 weeks of alcohol access, the alcohol-access rats were divided into high drinking (>2.5 g/kg/24-h) or low drinking (<2 g/kg/24-h). Rats were then given injections of the preferential mu opioid receptor antagonist naltrexone (1 mg/kg, s.c.) or the selective kappa opioid receptor antagonist LY2456302 (10 mg/kg, s.c.) prior to fear conditioning and were then tested for conditioned fear 2 days later. Pre-training naltrexone increased conditioned suppression of lever-pressing during training and testing, with no differences between high versus low alcohol drinkers or between water-only versus alcohol access groups (averaged across drinking levels). There was no effect of LY2456302 on conditioned fear in any comparison. We also found no differences between high and low alcohol drinkers and no reliable effect of prior alcohol access (averaged across drinking levels) on conditioned fear. Our experiment replicates and extends previous demonstrations that a preferential MOR antagonist can increase fear learning using conditioned suppression of lever-pressing as a fear measure. However, additional research is needed to determine the cause of the differences in conditioned fear that we previously observed (as they were not observed in the current experiments).

1. Introduction

Previous research has shown that post-traumatic stress disorder (PTSD) and alcohol abuse frequently co-occur [1-4]. However, the cause of this co-morbidity is unclear. One possibility is that individuals with trauma or anxiety disorders consume/abuse alcohol as a form of self-medication. For example, prior PTSD diagnosis increases an individual’s likelihood of diagnosis with a new alcohol use disorder [5] and retrospective self-report suggests that traumatic events can increase the likelihood of subsequent development of problematic alcohol use [6]. Alternatively, alcohol abuse is associated with subsequent development of anxiety disorders [7], possibly through alcohol-induced neurobiological alterations. For example, there is neuroimaging evidence for altered developmental trajectories in brain development in adolescents with regular/heavy alcohol use (from baselines assessed prior to regular/heavy alcohol use) [8]. An additional possibility is that a third pre-existing factor may affect both alcohol consumption and anxiety. For example, there is neuroimaging evidence for pre-existing neurobiological differences in adolescents who subsequently increase their level of alcohol consumption, including in brain areas associated with fear and anxiety (as reviewed in [8]). In addition, individuals with a family history of alcoholism exhibit structural and functional differences in brain areas associated with fear-learning and stress responses, whether or not they themselves abused alcohol [9, 10]. The evidence in the human PTSD literature has shown a degree of support for each of these possibilities but is unable to establish causality. The use of animal models provides a unique opportunity to address the issue of causality in the relationship between PTSD and alcohol abuse.

Rodent research examining the relationship between fear and alcohol use/exposure has found evidence for increased [11], decreased [12, 13] or unaltered [14-18] conditioned fear following alcohol exposure. These conflicting results may occur because of differences in the methods, such as route and level of alcohol exposure, the fear conditioning procedure used, and the length of time between alcohol exposure and behavioral training/testing. Our lab has examined the effects of voluntary alcohol consumption using the chronic intermittent access (CIA) model during adolescence/early adulthood from post-natal day PND 26-66, with behavioral training beginning 6-10 days after the last day of alcohol access and the fear conditioning session occurring exactly 15 days after the last day of alcohol access. We use a two-bottle choice CIA procedure in which rats are given 3 24-h periods of access to 20% ethanol in tap water (v/v) interspersed with periods of access to water-only [19, 20]. Our adolescent alcohol access procedure leads to blood ethanol concentrations that are low (approximately 5-15 mg/dl based on prior pilot data from our lab [14]), but drinking under this procedure can lead to long-term alterations in learning in at least one learning assessment- the rate of omission contingency learning [21]. However, most of our experimental results suggest a correlational relationship between alcohol consumption and other behavioral traits (including conditioned fear) without evidence for alcohol exposure altering behavior. To paraphrase, we typically found that the level of alcohol rats will voluntarily consume is associated with specific behavioral traits, although the alcohol access group as a whole does not differ from rats only given access to water. In particular, in two experiments, animals that consumed high levels of alcohol (>1.5-2 g/kg/24-h) exhibited low fear expression, while animals that consumed lower levels of alcohol exhibited high fear expression [14, 15]. In further experiments, we used the same CIA procedure and correlated alcohol consumption with behavior in a go/no-go reversal learning task and a devaluation/operant extinction task. Rats that consumed high levels of alcohol exhibited fast instrumental extinction and low discontinuation/maintenance errors in reversal learning compared with rats that consumed lower amounts [22]. Separately, fast instrumental extinction was associated with low fear expression in alcohol-naive rats [14], suggesting that these behavioral traits are part of a more general phenotype that exists independently of alcohol consumption, rather than resulting from alcohol consumption.

The current experiments were designed to test whether there is a common mechanism behind these co-occurring individual differences in alcohol consumption and conditioned fear. Although one possible strategy to examine these differences would be to engage in a selective breeding program to create separate rats lines with divergent phenotypes (as is often done in research on rats screened for alcohol consumption or preference [23-26]), we chose to examine these differences in outbred samples. While selectively bred rats would likely exhibit more extreme differences in neurobiology and behavior, there are several examples of strong differences in behavioral traits and their neurobiological substrates that can be observed within outbred samples, with one prominent example being the sign-tracking/goal-tracking phenotypes [27]. Our multiple previous findings of strong and reliable individual difference effects [14, 15], suggested that there may be significant neurobiological differences that could be observed in these two phenotypes in outbred rats.

One candidate for a common mechanism would be opioid receptor activity. Endogenous opioid activity is involved in the reinforcing properties of alcohol consumption [28, 29] and affects learning and/or expression of conditioned fear [30-33]. For example, administration of the opioid receptor antagonist naltrexone (an opioid receptor antagonist that is preferentially effective on mu opioid receptors [MORs]) decreases alcohol consumption and self-administration in a dose-dependent manner [19, 34-37] and facilitates fear acquisition [33]. Previous research has also shown that human individuals with a family history of alcoholism exhibit greater release of cortisol after administration of the opioid receptor antagonists naltrexone or naloxone than individuals without a family history of alcoholism [46, 47] (but see [48]). Conversely, antagonists of another type of opioid receptor, the kappa opioid receptors (KORs), decrease fear learning and/or expression [31, 38] and either decrease or have no effect on alcohol consumption or alcohol self-administration [39-45]. Therefore, it is possible that the high alcohol drinking rats have different endogenous opioid activity or sensitivity compared with their low alcohol drinking counterparts.

Here, we antagonized opioid receptors during conditioned fear training in high alcohol drinkers, low alcohol drinkers, and water-only exposed rats and then tested these rats (without pre-test injections) to isolate possible effects of opioid receptor activity during fear learning. We tested effects of MOR and KOR antagonists, as prior research suggests that these receptors may play opposing roles in fear regulation. The literature examining the role of MOR in fear learning suggests that pre-training MOR antagonism facilitates fear acquisition, as measured with conditioned freezing and conditioned suppression of lever-pressing in rats [30, 31, 33, 49, 50] and with reaction times in humans [51]. The literature on the role of KORs is less clear. Much of the literature on KOR antagonist effects on fear conditioning utilizes norbinaltorphimine (norBNI) or JDTic (which are both antagonists with effects lasting for several weeks) [52] and pre-training injections could affect fear learning and/or expression on subsequent days. Although it is difficult to determine the time-point when effects occur, pre-training injections of KOR antagonists decrease fear learning and/or expression, as assessed with conditioned freezing or fear-potentiated startle [31, 38]. However, in experiments that have examined effects of post-training/pre-test KOR antagonist administration, the effects on conditioned fear are often only observed under a narrow set of parameters (ex: renewal of fear in the original training context, but not fear in the extinction context or a novel context) [53], while other experiments found no significant effect [32]. In the current studies, we gave injections of the MOR antagonist naltrexone or the KOR antagonist LY2456302 (with a shorter half-life than norBNI and JDTic- the half-life is 1.8 h after a 1 mg/kg intravenous injection in rats [42]) before fear training, but not testing, in order to determine potential roles of these receptors in regulating acquisition of conditioned fear in high alcohol-drinking, low alcohol-drinking, or alcohol naïve rats.

2. Methods

2.1. Subjects

Male naïve Long Evans rats (n= 196; 72 for Exp 1 and 124 for Exp 2) from an in-house breeding program were used for all experiments. These rats were selected from a larger pool of rats screened for alcohol consumption or water-only exposure based on their drinking phenotype and whether there were corresponding rats in the other groups in the same cohort simultaneously screened for consumption (to control for seasonality effects). The female breeders were from Charles River Laboratories (Kingston, NY) and the male breeders were from Charles River Laboratories (Raleigh, NC) to ensure that inbreeding would be avoided. Following weaning, animals were individually housed and maintained on a 12-hour reverse light-dark cycle with lights off at 07:00 am in a temperature and humidity controlled room. Rats were weaned on PND 21 and received CIA starting on PND 26. Water and food were available ad libitum during this 6-week period. Three days after the final alcohol access period, body weights were recorded and rats were food-restricted and subsequently allowed to grow 1.5 g/day from this initial weight. These conditions were initially chosen because maintaining growth between 1.5 and 2 g/day maintains lever-pressing at a relatively constant rate across a fear incubation/retention interval [54]. While we do not compare lever-press rates or conditioned fear responses across incubation/retention intervals in the current experiments, these were the food-restriction conditions used in our previous two published papers on alcohol consumption and conditioned fear [14, 15], and keeping the same food-restriction conditions simplifies comparisons with our previous research. Water was available ad libitum throughout the period of food restriction. All procedures and animal care were in accordance with the Kansas State University Institutional Animal Care and Use Committee guidelines, the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, and United States federal law.

2.2. Apparatus

Experiments were conducted in 12 operant chambers (Med Associates, St Albans, VT). Each chamber had two retractable levers 9 cm above the floor, but only the right lever was extended into the chamber during behavioral sessions. Responding on the active lever activated the pellet dispenser, delivering 45-mg precision pellets (#1811155, 5% fat, 66% carbohydrate, 20.3% protein; TestDiet, Richmond, IN). A red house-light was located in the center at the top of the back wall of the chambers. A tone generator that delivered a 2900 Hz tone (20 dB above background) was located directly to the right of the house-light. The chambers had grid floors connected to electric shock generators that were capable of delivering a 0.5-mA scrambled foot-shock.

2.3. Alcohol access

Rats received 6 weeks of alcohol (using CIA) or water-only access beginning in adolescence and extending into early adulthood (PND 26-66). A two-bottle choice procedure was used in which all rats had access to water in at least one bottle at all times. Rats (n = 48 in Exp. 1 and 80 in Exp. 2) (the Alcohol group) received 24-h access to 20% (v/v) ethanol mixed with tap water 3X per week in one bottle, while the other bottle contained tap water. Animals in the Alcohol group received 24-h alcohol access three times per week with 24- or 48-h periods of water-only in-between. every other day. The Water group (n=24 in Exp. 1 and 44 in Exp. 2) received two water bottles each day during the six weeks. Bottles were weighed and changed (for alcohol groups) starting between 1-1:30 pm, and placement each day was counterbalanced to control for any side preference. Following the end of CIA, we selected rats to be fear conditioned that were either low drinkers (<2 g/kg/24-h) or high drinkers (>2.5 g/kg/24-h) based on the average of the last two weeks of alcohol consumption. In Experiment 1, the rats were selected from a population in which 34% of rats that received alcohol access were classified as low drinkers. In Experiment 2, the rats were selected from a population in which 33% of rats were classified as low drinkers.

2.4. Drugs

Naltrexone (Sigma-Aldrich, St. Louis, MO) was diluted in 0.9% sterile saline and given at a dose of 1 mg/kg (s.c., in 1 ml/kg of sterile saline). LY2456302 (NIDA Drug Supply Program) was diluted in 5% dimethyl sulfoxide (Fisher Scientific, Waltham, MA) and 5% cremophor EL (EMD Millipore, Billerica, MA) in sterile saline, and given at a dose of 10 mg/kg (s.c., in 1 ml/kg of vehicle). Dosing, route of administration, and timing of administration were selected based on methods from Bossert and colleagues [55]. Although our 2-h pretreatment time for LY2456302 is longer than the 1.8-h half-life of LY2456302 when given through the intravenous route [42], the half-life of drugs tends to be longer if given through the subcutaneous route. In addition, we used the same pre-treatment time as Bossert and colleagues [55] (with a shorter behavioral session than was used in that paper) and their doses and pretreatment times were based on information from AstraZeneca.

2.5. Behavioral training in fear experiments

First, rats were given a 40-min food-cup training session, with pellet deliveries every 125 sec. The following day, the rats received 2 sessions on a fixed-interval-1 (FI-1) reinforcement schedule for lever-presses on the right lever (lever-presses could earn a pellet each sec) 2–4 h apart. These sessions ended when rats received 50 pellets (with a limit of 1 h). On the subsequent days (one session per day), the rats were given one 51-min session in which pellets were earned on a variable-interval-30 (VI-30) reinforcement schedule (pellet availability for lever-presses ranging from 1 to 59 sec), and 2 daily 51-min sessions on a VI-60 schedule (pellet availability ranging from 1 to 119 sec). Rats were maintained on the VI-60 schedule for the remainder of the behavioral training.

During the fear conditioning session, rats earned pellets on a VI-60 schedule. Sessions began with the extension of the active lever and illumination of a red house-light. On this day, they received 8 20-second tones (2900 Hz, 20 dB above background) ranging from 2 to 7 min apart, co-terminating with 0.5-mA, 0.5-sec foot-shocks (adjusted for inter-chamber variability) pseudo-randomly throughout a 51-min session.

Rats then received a lever reacquisition session (one day following fear conditioning) and a cued fear test in extinction (two days following fear conditioning). The lever reacquisition session (51 min) consisted of VI-60 lever-press training during which no tones or shocks were presented. The extinction test lasted 25 min, during which rats lever-pressed on a VI-60 schedule and 4 presentations of the tone fear cue were presented without any shock deliveries

2.6. Individual Experiments

2.6.1. Experiment 1 – Pre-training injections of naltrexone, a MOR antagonist

Rats (n=72) received CIA during adolescence/early adulthood. The Alcohol group (n=48) received alcohol access as described above and the Water group (n=24) received water-only access. Of the 6 cohorts used in experiment 1, 5 of the cohorts contained at least 1 high drinking-, at least 1 low drinking-, and at least 1 water-only-rat, while only one cohort did not contain a representative from each of the 3 groups (this cohort contained low drinkers only). Three days after the last alcohol access period, all rats were food-restricted and behavioral training procedures began 8-10 days after the last alcohol access period. The rats received the behavioral procedures described above. Ten minutes prior to the beginning of fear conditioning, in a 2 X 3 between-subjects design, the rats (6 groups: 2 naltrexone doses [0 or 1 mg/kg] X 3 consumption groups [High Drinkers, Low Drinkers, or Water-Only]; 12/group) received a s.c. injection (1 ml/kg) of either a 1 mg/kg dose of naltrexone or sterile saline. Injections were not administered prior to the lever reacquisition session or the cued fear test.

2.6.2. Experiment 2 – Pre-training injections of LY2456302, a KOR antagonist

Rats (n=124) received CIA during adolescence/early adulthood. The Alcohol group (n=80) received alcohol access as described above and the Water group (n=44) received water-only access. Of the 8 cohorts used in experiment 2, 7 of the cohorts contained at least 1 high drinking-, at least 1 low drinking-, and at least 1 water-only-rat, while only one cohort did not contain a representative from each of the 3 groups (this cohort contained water-only rats). Three days after the last alcohol access period, all rats were food-restricted and behavioral training procedures began 6-10 days after the last alcohol access period. The rats received the behavioral procedures described above. Two hours prior to the beginning of fear conditioning, in a 2 X 3 between-subjects design, the rats (6 groups: 2 LY2456302 doses [0 or 10 mg/kg] X 3 consumption groups [High Drinkers, Low Drinkers, or Water-Only]; 17-20/group) received a s.c. injection (1 ml/kg) of either a 10 mg/kg dose of LY2456302 or vehicle. Injections were not administered prior to the lever reacquisition session or the cued fear test.

The number of subjects was higher in Exp. 2 than in Exp. 1. After we had tested a few cohorts for Exp. 2, we observed some promising trends in the behavioral data. As a result, we decided to collect brain tissue for immunohistochemical processing from that point on. Because we had already collected behavioral data for a number of subjects (without collecting brains for immunohistochemistry), we needed to conduct testing on additional animals (with collection of brains) in order to achieve the desired 11-12/group for the neurobiological data. However, upon completion of behavioral data collection, we no longer observed effects of LY2456302 on conditioned fear and we decided against pursuing immunohistochemical processing.

2.7. Behavioral measures

For alcohol consumption, the alcohol consumed was calculated as the bottle weight difference between the start and end of the access periods, minus 2 g for spillage/evaporation, and then multiplied by 0.162 for the weight of alcohol in 1 g of a 20% v/v alcohol solution. The 2 g subtracted for spillage was determined based on pilot research conducted in our laboratory during which bottles were placed on empty cages and weighed daily to determine general spillage/evaporation.

We used conditioned suppression of lever-pressing to assess fear response during conditioning and testing [56-58]. Lever-presses were recorded during the 40 sec prior to tone presentation (Precue) and during the 20-sec tone presentation (Cue), and the rate of lever-pressing/min were used to calculate a suppression ratio that normalizes lever-press suppression during the tone based on baseline responding: Suppression ratio = ((Precue−Cue)/(Precue+Cue)) [59]. A value of 1 indicates total conditioned suppression of lever-pressing during tone presentation (high fear), while a value of 0 reflects no lever-press suppression during tone presentation (low fear). The suppression ratio formula utilized produces statistical results that are identical to those using the traditional Annau-Kamin ratio [60], but is easier to conceptualize as higher fear is represented by higher ratios, and no change in lever-press rate is represented by a ratio of zero in our ratio.

2.8. Statistical analyses

Data were analyzed by Statistica 13.3 software (Palo Alto, CA). The factors used in the statistical analyses are described in the Results section for each ANOVA and significant effects (p<0.05) in the different ANOVAs were followed by post-hoc Tukey’s HSD tests.

3. Results

3.1. Experiment 1

3.1.1. Alcohol Consumption

For the alcohol consumption data, a mixed-effects ANOVA was conducted using the within-subjects factor of Week (the 6 weeks of alcohol consumption) and the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle; this represents their group assignments after alcohol consumption was complete). There was a significant main effect of Consumption, F(1, 44) = 18.48, p < 0.01, such that High drinkers drank more than Low drinkers (Fig. 1A). There were no other significant effects or interactions (p’s = 0.07 – 0.81). Therefore, the two subgroups within our alcohol access rats differed in their level of alcohol consumption. Importantly, the main effect of Injection and the interaction effects including it were not significant, suggesting there were no pre-existing differences between animals that later received different treatments during fear conditioning.

Figure 1.

Data from Experiment 1. A. Weekly alcohol consumption (mean±SEM) across the 6 weeks of pre-training access during adolescence and early adulthood. * indicates p < 0.05 significant difference between the high and low alcohol drinking groups (a main effect across the 6 weeks). Gray circles indicate low alcohol drinkers (< 2 g/kg/24-h) and black circles/bars indicate high alcohol drinkers (> 2.5 g/kg/24-h). B. Average lever-presses/min (mean±SEM) during the precue period during training (left) and testing (right) for alcohol versus water-only groups (collapsing across high and low drinkers in the alcohol group). C. Average suppression ratio (mean±SEM) during the cue during training (left) and testing (right) for alcohol versus water-only groups (collapsing across high and low drinkers in the alcohol group). * indicates p < 0.05 significant difference between the naltrexone and vehicle groups. # indicates p < 0.05 significant difference between the alcohol and water groups. D. Average lever-presses/min (mean±SEM) during the precue period during training (left) and testing (right) for high vs. low alcohol drinkers. E. Average suppression ratio (mean±SEM) during the cue during training (left) and testing (right) for high- vs. low-drinkers. * indicates p < 0.05 significant difference between the naltrexone and vehicle groups. For B-E, black bars indicate rats that received pre-training naltrexone injections and white bars indicate rats that received pre-training vehicle injections.

3.1.2. Behavioral Results

3.1.2.1. Group Differences Analyses

When we compared water-only and alcohol access groups (collapsing across high and low drinkers) during fear training, alcohol consumption decreased suppression ratios and naltrexone exposure increased suppression ratios. However, neither alcohol consumption nor naltrexone affected precue lever-pressing.

None of the groups differed in precue lever-pressing on the first training trial or across the conditioning session. Precue lever-pressing on the first trial of fear conditioning did not differ between groups (lever-presses/min): Alcohol-Naltrexone = 11.44 ± 1.42, Water-Naltrexone = 17.00 ± 2.28, Alcohol-Vehicle = 14.06 ± 1.79, and Water-Vehicle = 12.13 ± 2.25. A 2x2 ANOVA of the trial 1 precue lever-press rates with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle) found no significant effects or interaction (p’s = 0.19 - 0.98). Next, we ran a 2x2 ANOVA on the average precue lever-press measure across all trials of fear training with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle). There were no significant effects or interactions (p’s = 0.41 - 0.70) (Fig. 1B, left).

There were, however, differences between groups in conditioned suppression of lever-pressing. The suppression ratio on the first trial of fear conditioning was lower for alcohol-exposed rats than for water-only rats: Alcohol-Naltrexone = 0.39 ± 0.10, Water-Naltrexone = 0.64 ± 0.12, Alcohol-Vehicle = 0.31 ± 0.10, and Water-Vehicle = 0.54 ± 0.11. A 2x2 ANOVA of the trial 1 suppression ratios with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle) found a significant main effect of Consumption F(1, 68) = 4.04, p < 0.05, such that alcohol drinkers showed lower levels of fear on the first trial of training than water drinkers. The effect of Injection and the Consumption*Injection interaction were not significant (p’s = 0.31 - 0.69). Then we ran a 2x2 ANOVA on the average suppression ratio measure across all trials of fear training with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle). We found significant main effects of Consumption, F(1, 68) = 7.06, p < 0.01, and Injection, F(1, 68) = 4.08, p < 0.05, indicating that rats exposed to alcohol showed lower levels of fear than those exposed to water-only, and naltrexone-exposed rats exhibited higher levels of fear than vehicle-exposed rats (Fig. 1C, left). The Consumption*Injection interaction was not significant (p = 0.55).

When we examined effects of naltrexone on conditioned fear in alcohol access groups (collapsing across high and low drinkers) in the fear test, naltrexone increased, but alcohol had no effect on, suppression ratios. Neither alcohol consumption nor naltrexone administration affected precue lever-pressing. We ran a 2x2 ANOVA on the precue lever-press measure during testing with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle). There were no significant effects or interactions (p’s = 0.19 - 0.97) (Fig. 1B, right). Next, we ran a 2x2 ANOVA on the suppression ratio measure during testing with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle). We found a significant effect of Injection, F(1, 68) = 9.52, p < 0.01, indicating that naltrexone-exposed rats showed higher levels of fear than vehicle-exposed rats (Fig. 1C, right). The main effect of Consumption and the Consumption*Injection interaction were not significant (p’s = 0.43 – 0.97).

3.1.2.2. Individual Differences Analyses

When we compared the high and low alcohol consumption groups during fear training, naltrexone administration increased suppression ratios (but not precue lever-pressing) in High and Low drinking rats, but there was no relationship between alcohol drinking group and suppression ratios. Neither alcohol drinking group nor naltrexone administration affected precue lever-pressing.

None of the groups differed in precue lever-pressing on the first training trial or across the conditioning session. Precue lever-pressing on the first trial of fear conditioning did not differ between groups (lever-presses/min): High-Naltrexone = 11.00 ± 1.70, Low-Naltrexone = 11.88 ± 2.34, High-Vehicle = 14.00 ± 3.09, and Low-Vehicle = 14.13 ± 1.98. A 2x2 ANOVA of the trial 1 precue lever-press rates with the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle) found no significant effects or interaction (p’s = 0.27 - 0.87). Next, we ran a 2x2 ANOVA on the average precue lever-press measure across all trials of fear training with the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle). There were no significant effects or interactions (p’s = 0.23 - 0.95) (Fig. 1D, left).

There were, however, differences between groups in conditioned suppression of lever-pressing. The suppression ratio on the first trial of fear conditioning did not differ between groups: High-Naltrexone = 0.40 ± 0.12, Low-Naltrexone = 0.37 ± 0.16, High-Vehicle = 0.15 ± 0.17, and Low-Vehicle = 0.47 ± 0.11. A 2x2 ANOVA of the trial 1 suppression ratios with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle) found no significant effects or interaction (p’s = 0.21 - 0.61). Next, we ran a 2x2 ANOVA on the average suppression ratio measure across all trials training with the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle). We found a significant main effect of Injection, F(1, 44) = 4.72, p < 0.05, such that naltrexone administration was associated with higher fear in both High and Low drinkers (Fig. 1E, left). The effect of Consumption and the Consumption*Injection interaction were not significant (p’s = 0.15 – 0.41).

Similar to our findings during training, rats that received naltrexone before training had increased suppression ratios during testing compared to their vehicle counterparts, with no effect of drinking group. Neither alcohol drinking group nor naltrexone administration affected precue lever-pressing. We ran a 2x2 ANOVA on the precue lever-press measure during testing with the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle). There were no significant effects or interactions (p’s = 0.66 – 1.00) (Fig. 1D, right). Next we ran a 2x2 ANOVA on the suppression ratio measure during testing with the between-subjects factors of Consumption (High vs. Low) and Injection (Naltrexone vs. Vehicle). We found a significant main effect of Injection, F(1, 44) = 7.11, p < 0.01, such that naltrexone increased conditioned fear in both High and Low drinkers (Fig. 1E, right). The effect of Consumption and the Consumption*Injection interaction were not significant (p’s = 0.12 – 0.81).

3.1.2.3. Correlations

We then examined whether there was a relationship between the amount of alcohol a rat consumed with conditioned fear during training and testing. We correlated the average alcohol consumed (g/kg/24-h) across the 6 weeks of access or during weeks 5 and 6 of access with suppression ratios during training or testing. We performed separate correlations for rats that received naltrexone or vehicle. There were no significant correlations between the amount of alcohol consumed and conditioned fear in any of our analyses (r = −0.15 – 0.34, p’s = 0.10 – 0.94, Tables 1 and 2).

Table 1.

Experiment 1 correlations during training.

		6 week		2 week
Consumption	Injection	r	p	r	p
Alcohol	Nal	0.02	0.94	0.05	0.81
Alcohol	Veh	0.08	0.72	0.34	0.10

Table 2.

Experiment 1 correlations during testing.

		6 week		2 week
Consumption	Injection	r	p	r	p
Alcohol	Nal	−0.08	0.70	−0.04	0.86
Alcohol	Veh	−0.15	0.49	−0.07	0.74

3.2. Experiment 2

We excluded 12 rats due to experimenter error, resulting in 37 low drinkers (20 LY2456302, 17 vehicle), 37 high drinkers (19 LY2456302, 18 vehicle), and 38 water-only rats (19 LY2456302, 19 vehicle) receiving conditioned fear training.

3.2.1. Alcohol Consumption

For the alcohol consumption data, a mixed-effects ANOVA was conducted using the within-subjects factor of Week (the 6 weeks of alcohol consumption) and the between-subjects factor of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle; this represents their group assignments after alcohol consumption was complete). There were significant main effects of Consumption, F(1, 70) = 70.56, p < 0.01, and Week, F(5, 350) = 16.86, p < 0.01, and the interaction between the two was also significant, F(5, 350) = 35.62, p < 0.01. Post hoc testing found that Low and High drinkers had similar levels of alcohol consumption during the first week of access, but starting on Week 2, the High drinkers consumed significantly more alcohol each week for the rest of the access period (Fig. 2A). There were no other significant effects or interactions (p’s = 0.10 – 0.78). Therefore, we can once again conclude that the two subgroups within our alcohol access rats differed in their level of alcohol consumption. As in the previous experiment, the main effect of Injection and the interaction effects including it were not significant, suggesting there were no pre-existing differences between animals that received different treatments during fear conditioning.

Figure 2.

Data from Experiment 2. A. Weekly alcohol consumption (mean±SEM) across the 6 weeks of pre-training access during adolescence and early adulthood. Gray circles indicate low alcohol drinkers (< 2 g/kg/24-h) and black circles/bars indicate high alcohol drinkers (> 2.5 g/kg/24-h). *# = p < 0.05 significant interaction between Alcohol Consumption (high vs. low drinkers) and Week (the 6 weeks of consumption). B. Average lever-presses/min (mean±SEM) during the precue period during training (left) and testing (right) for alcohol versus water-only groups (collapsing across high and low drinkers in the alcohol group). C. Average suppression ratio (mean±SEM) during the cue during training (left) and testing (right) for alcohol versus water-only groups (collapsing across high and low drinkers in the alcohol group). D. Average lever-presses/min (mean±SEM) during the precue period during training (left) and testing (right) for high vs. low alcohol drinkers. E. Average suppression ratio (mean±SEM) during the cue during training (left) and testing (right) for high- vs. low-drinkers. For B-E, black bars indicate rats that received pre-training LY2456302 injections and white bars indicate rats that received pre-training vehicle injections.

3.2.2. Behavioral Results

3.2.2.1. Group Differences Analyses

We compared water-only vs. alcohol access groups (collapsing across high and low drinkers) to determine if prior alcohol access affects LY2456302’s effects. Neither alcohol consumption nor LY2456302 affected precue lever-pressing or suppression ratios during fear training.

None of the groups differed in precue lever-pressing on the first training trial or across the conditioning session. Precue lever-pressing on the first trial of fear conditioning did not differ between groups (lever-presses/min): Alcohol-LY2456302 = 14.01 ± 1.77, Water-LY2456302 =13.58 ± 2.10, Alcohol-Vehicle = 14.08 ± 0.96, and Water-Vehicle = 13.18 ± 1.68. A 2x2 ANOVA of the trial 1 precue lever-press rates with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle) found no significant effects or interaction (p = 0.69 - 0.92). Next, we ran a 2x2 ANOVA on the average precue lever-press measure across all trials of fear training with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.45 - 0.94) (Fig. 2B, left).

None of the groups differed in conditioned suppression of lever-pressing on the first training trial or across the conditioning session. Suppression ratio on the first trial of fear conditioning did not differ between groups: Alcohol-LY2456302 = 0.43 ± 0.08, Water-LY2456302 = 0.32 ± 0.10, Alcohol-Vehicle = 0.52 ± 0.06, and Water-Vehicle = 0.46 ± 0.10. A 2x2 ANOVA of the trial 1 suppression ratios with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle) found no significant effects or interaction (p’s = 0.17 - 0.82). Next, we ran a 2x2 ANOVA on the average suppression ratio measure across all trials of fear training with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.15 - 0.68) (Fig. 2C, left).

Neither alcohol consumption nor LY2456302 administration affected precue lever-pressing or suppression ratios during testing. We ran a 2x2 ANOVA on the precue lever-press measure during testing with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.08 - 0.94) (Fig. 2B, right). Next, we ran a 2x2 ANOVA on the suppression ratio measure during testing with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.82-0.94) (Fig. 2C, right).

3.2.2.2. Individual Differences Analyses

When we compared the high versus low alcohol consumption groups during fear training, there was no relationship between alcohol drinking group or LY2456302 treatment and the measures of precue lever-pressing or suppression ratio.

None of the groups differed in precue lever-pressing on the first training trial or across the conditioning session. Precue lever-pressing on the first trial of fear conditioning did not differ between groups (lever-presses/min): High-LY2456302 = 11.33 ± 1.28, Low-LY2456302 = 16.85 ± 3.07, High-Vehicle = 13.97 ± 1.28, and Low-Vehicle = 14.18 ± 1.43. A 2x2 ANOVA of the trial 1 precue lever-press rates with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle) found no significant effects or interaction (p’s = 0.14 - 0.99). We performed a 2x2 ANOVA on the average precue lever-press measure across all trials of fear training with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.54 - 0.66) (Fig. 2D, left).

None of the groups differed in conditioned suppression of lever-pressing on the first training trial or across the conditioning session. Suppression ratios on the first trial of fear conditioning did not differ between groups: High-LY2456302 = 0.36 ± 0.12, Low-LY2456302 = 0.49 ± 0.10, High-Vehicle = 0.57 ± 0.09, and Low-Vehicle = 0.48 ± 0.08. A 2x2 ANOVA of the trial 1 suppression ratios with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle) found no significant effects or interaction (p’s = 0.27 - 0.85). Next, we ran a 2x2 ANOVA on the average suppression ratio measure across all trials of fear training with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.20 – 0.64) (Fig. 2E, left).

Similar to our findings during training, there was once again no relationship between the level of alcohol consumed or treatment and the measures of precue lever-pressing or suppression ratio during testing. We ran a 2x2 ANOVA on the precue lever-press measure during testing with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.36 – 0.73) (Fig. 2D, right). Next, we ran a 2x2 ANOVA on the suppression ratio measure during testing with the between-subjects factors of Consumption (High vs. Low) and Injection (LY2456302 vs. Vehicle). There were no significant effects or interactions (p’s = 0.12 – 0.81) (Fig. 2E, right).

3.2.2.3. Correlations

We examined whether there was a relationship between the amount of alcohol a rat consumed with conditioned fear during training and testing. We correlated the average alcohol consumed (g/kg/24-h) across the 6 weeks of access or during weeks 5 and 6 of access with suppression ratios during training or testing. We performed separate correlations for rats that received LY2456302 or vehicle. We found a significant correlation between the amount of alcohol consumed across the 6 weeks of access by rats that received LY2456302 and conditioned fear during testing, r(35) = −0.34, p < 0.05, but not conditioned fear during training. There were no other significant correlations (r = −0.32 – −0.11, p’s = 0.06 – 0.97, Tables 3 and 4).

Table 3.

Experiment 2 correlations during training.

		6 week		2 week
Consumption	Injection	r	p	r	p
Alcohol	LY	−0.01	0.97	−0.01	0.96
Alcohol	Veh	−0.11	0.50	−0.01	0.94

Table 4.

Experiment 2 correlations during testing. Significant effects are in bold.

		6 week		2 week
Consumption	Injection	r	p	r	p
Alcohol	LY	−0.34	0.04	−0.32	0.06
Alcohol	Veh	0.11	0.49	0.04	0.80

4. Discussion

We found increased conditioned fear with pre-training naltrexone administration, but no effect of pre-training LY2456302 administration. All groups that received pre-training naltrexone showed increased fear during training and testing, with no differences between rats that consumed different levels of alcohol or water-only. We also found no differences in baseline fear in animals that received pre-training vehicle injections. This latter finding is in opposition to our previous findings in which high drinkers showed relatively low levels of fear and low drinkers showed relatively high levels of fear. Possibly related to this lack of individual differences in baseline fear, we also did not observe individual differences in reactivity to the opioid receptor antagonists. We discuss possible explanations below.

4.1. Effect of mu-and kappa-opioid receptor antagonism on conditioned fear

Our experimental results replicate and extend previous demonstrations that a preferential MOR antagonist can increase conditioned fear using conditioned suppression of lever-pressing as a fear measure. Previous demonstrations that a preferential MOR antagonist would increase conditioned suppression used naloxone rather than naltrexone and found effects during different phases of the experiment (ex: extinction tests) and/or in time-periods other than during the cue presentations (ex: in the post-cue period) during ongoing fear conditioning [49, 50]. Ceiling effects were a likely cause of the inability to observe effects during cue presentations during ongoing conditioning in these previous studies, rather than a true non-effect of naloxone during ongoing conditioning. Our findings that naltrexone increased conditioned suppression, measured during the cue presentation during ongoing fear conditioning, extend these previous findings. Our findings on MOR antagonism, that pre-training naltrexone administration increased conditioned fear during training and increased fear memory during a test two days later, were expected. Pre-training administration of naltrexone or other MOR antagonists has been shown to augment fear learning (as measured with conditioned freezing) [30, 31, 33], so our findings are in accord with the prior literature. While naltrexone is preferentially selective for MORs, it also acts as a lower affinity KOR antagonist, so the effects could be mediated by KOR antagonism. However, as we found no effects of the KOR-selective antagonist LY2456302, these effects of naltrexone appear to be mediated by naltrexone’s effects on MORs rather than KORs.

While the experimental design in Exp. 1 was not designed to parse the processes affected by naltrexone, previous research from other laboratories suggest that the effects we found in Exp. 1 may be due to disruption of a conditioned analgesia-based error correction mechanism. In several experiments, researchers have shown that endogenous opioid activity appears to act as a mechanism to limit the aversiveness of shocks as they become more accurately predicted, a form of error correction. For example, in the Rescorla-Wagner model [61] asymptotic conditioned behavior is hypothesized to occur when the expected outcome of a trial (based on all cues present during that trial) is equal to the outcome actually delivered on that trial, and there is no additional learning when the predicted and actual outcomes are equal. Fanselow [62] suggested that, for fear conditioning, the conditioned release of endogenous opioids was the biological basis of this error correction mechanism. According to this hypothesis, endogenous opioids would be released in greater amounts under circumstances in which there was greater conditioned fear and these endogenous opioids would act to decrease pain sensitivity to the upcoming shock. If so, then continued excitatory conditioning would lead to a state where the predicted shock was equal in intensity to the experienced shock, and the conditioned release of endogenous opioids on conditioning trials would be sufficient to minimize reactivity to the shock experienced and make it an ineffective unconditioned stimulus.

Notably, evidence from several conditioning experiments suggest that this conditioned analgesia-based error correction may be the mechanism underlying the effects of naltrexone/naloxone on conditioned fear. Fear conditioning with a low intensity shock unconditioned stimulus (0.4-0.6 mA) typically leads to lower conditioned fear than stronger intensity shocks. This pattern may be explained by lower levels of conditioned opioid release being required to offset the aversive value of the lower intensity shocks, but naloxone injections caused relatively weak shocks to lead to the same asymptotic level of fear as stronger shocks in rats [63]. In a fear conditioning study in humans, naloxone prevented the decreased skin conductance responses to shock presentation that developed across the experiment in participants without naloxone injections [51]. In this study, naloxone also led to increased conditioning (operationalized as faster reaction times to the CS+ for shock) and prevented the development of conditioning-induced anticipatory antinociception responses in a number of brain areas associated with this function (measured with fMRI) [51]. For another source of evidence, the blocking phenomenon, in which an outcome which is fully predicted by one cue (Cue A) in a compound leads to little or no conditioning to the other cue in the compound (Cue B) even if the compound is repeatedly paired with the outcome, is often hypothesized to occur because the outcome is fully predicted by Cue A and there is no prediction error to cause learning to Cue B [61]. Naloxone injections have also been shown to prevent the blocking effect [50, 64], suggesting that it may disrupt a conditioned analgesia-based error correction mechanism. Other error correction-based fear conditioning-related phenomena (ex: the overexpectation effect) are also disrupted by naloxone injections [50, 64], suggesting that opioid receptor antagonists increase conditioned fear by blocking the endogenous opioids’ role in error correction during fear conditioning (with a possible role beyond conditioned analgesia, which may include a role in inhibitory learning).

If this conditioned analgesia-based error correction mechanism is applied to our data in Exp. 1, then it would suggest that the effect of naltrexone was to prevent the normal decrease in the effectiveness of the shock as an aversive unconditioned stimulus due to endogenous opioid-induced conditioned analgesia. Notably, our 0.5-mA shock intensity is within the range of shock intensities at which naltrexone was previously shown to increase asymptotic fear levels (between 0.4 and 0.6 mA) [63]. If so, then the effects of naltrexone should not have been apparent until after trial 2 of training, because conditioned analgesia would first decrease the aversiveness of the shock on trial 2 and this would not affect conditioned fear until trial 3. When we performed an analysis of suppression ratios on trial 2, we found no evidence for increased fear on this trial when the water-only groups were compared to the combined alcohol groups: Alcohol-Naltrexone = 0.42 ± 0.10, Water-Naltrexone = 0.41 ± 0.14, Alcohol-Vehicle = 0.33 ± 0.08, and Water-Vehicle = 0.46 ± 0.12. A 2x2 ANOVA with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle) found no significant effects or interactions (all F<1). We also found no evidence for increased suppression ratios on this trial when high drinkers were compared to low drinkers: High-Naltrexone = 0.38 ± 0.15, Low-Naltrexone = 0.46 ± 0.15, High-Vehicle = 0.32 ± 0.10, and Low-Vehicle = 0.33 ± 0.12. A 2x2 ANOVA with the between-subjects factors of Consumption (Alcohol vs. Water) and Injection (Naltrexone vs. Vehicle) found no significant effects or interactions (all F<1). Our pattern, in which naltrexone increased fear during the training session but did not increase it on the first or second conditioning trial, provides support for the idea that naltrexone’s actions acted to increase conditioned fear by preventing the normal decrease in aversiveness of the shock across trials that results from conditioned analgesia. However, there is also some evidence that pre-training systemic naltrexone can augment learning in a spatial learning task that does not include shock [65] and that post-training naloxone can augment memory for passive avoidance learning [66]. As neither of these experiments had the MOR antagonist in the body during shocks (either because the procedure involved no shocks or because the antagonist was given after training), these results suggest that it is possible that naltrexone may have enhanced learning through a mechanism other than blocking conditioned analgesia. Additional research would be needed to assess (and possibly exclude) these possibilities.

In contrast, we did not find any effects of LY2456302 on fear learning (as measured during training or testing) in Exp. 2. This was unexpected, due to existing literature showing that pre-training KOR antagonism decreases fear learning/expression (as measured with conditioned freezing and fear-potentiated startle) [31, 38]. However, there are several aspects of the prior research on KOR antagonists in conditioned fear that may limit the generality of this pattern. Studies that investigate KOR function often use pre-training injections of norBNI or JDTic, which are extremely long-lasting with effects lasting several weeks, so pre-training injections would have effects on kappa receptors during both training and testing phases of an experiment. This makes interpretation of previous findings and isolation of KOR function during training versus testing difficult, as previous reports tended to only report fear measures during test. However, in prior experiments that have examined effects of post-training/pre-test KOR antagonism, effects on conditioned fear are often only observed under a narrow set of parameters (ex: renewal of fear in the original training context, but not fear in the extinction context or a novel context) [53] or are not observed at all [32]. This suggested that, although prior experiments with pre-training injections may have still had the antagonist active during testing, the behavioral effects were due to the antagonist’s effects during training. LY2456302 is shorter-acting (with a half-life of 1.8 h after a 1 mg/kg intravenous injection in rats [42]), so our pre-training injections should have no effect on the receptors during the test. It is unclear whether this means prior effects of pre-training KOR antagonist injections in other experiments were due to effects of the drug during test, or some other procedural difference (such as use of the conditioned suppression as the fear measure) was responsible for our null effect of KOR antagonism. Further research is needed to determine the reason for this discrepancy.

One important caveat to discuss (for both experiments) is that we studied the effects of a single dose of naltrexone or LY2456302 in our experiments. Although this was necessary due to the factorial nature of our experimental designs, it does limit the conclusions we can draw based on the data collected here. We were not able to generate a dose-response curve for each of these drugs. Generally, as MOR antagonist dosage increases, so does its efficacy in affecting fear/aversive learning (for example, see [30, 67]), and KOR antagonists exhibit the same pattern (for example, see [38, 53, 68]). Likewise, although these studies did not assess fear/aversive learning, previous research has also shown that the effects of LY2456302 are also dose-dependent [42, 69-72]. Although we picked our dosing was based on the existing literature, it is possible that using different doses in our experiments could lead to different results. For example, a lower naltrexone dose might lead to different effects in different drinking groups if one group was more sensitive to naltrexone’s effects but the current dose was too high and obscured this differential sensitivity. Conversely, it is possible that our LY2456302 dose was too low, and a higher dose would have been effective in altering conditioned fear. Future research examining the effects of these drugs across dose-response curves would be needed to assess these possibilities.

4.2. Effects of alcohol access on conditioned fear and sensitivity to opioid receptor antagonist

In Exp. 1 of the current examination, we observed lower conditioned fear in the alcohol-exposed group during training, but not in the test session that took place two days later. This difference was present on the first conditioning trial (before any shock had been delivered), so it appears to be a difference in unconditioned reactivity to the tone rather than a difference in fear conditioning. Because this effect was not replicated in Exp. 2 (either the effect on the first trial or average responding across the session), this does not appear to be a reliable effect. In addition, our prior research using the same CIA procedures and age of access found no effect of alcohol access on conditioned fear under two sets of fear conditioning procedures (both different than the ones used here) [14, 15]. Some previous experiments have found long-term effects of alcohol exposure on conditioned fear, with these effects including both for increased [11] and decreased [12, 13]. However, the prior demonstration of increased fear used injections of 1.5 g/kg alcohol (which lead to BECs of ~200 mg/dl [73] that are much higher than the estimated 5-15 mg/dl based on prior pilot data from our lab [14]) and led to increased fear two days after the end of alcohol exposure but this effect seemed to have faded if fear was tested 10 days after the end of alcohol exposure [11]. In our experiments, the fear conditioning session did not occur until 15 days after the final alcohol access period. Therefore, even based on the one demonstration of long-term alcohol-induced fear increases, the effect would be likely to fade by the time we performed fear conditioning and testing. We propose that the lack of an effect of alcohol access on fear observed in Exp 2. is more likely representative of the (non-)effect of adolescent/early adult alcohol access on conditioned fear.

We also did not find that the effects of MOR or KOR antagonism on conditioned fear was affected by alcohol access during adolescence/early adulthood. Naltrexone augmented fear in all groups, regardless of alcohol access, whereas LY2456302 had no significant effect on fear in any group. However, these results must be interpreted with caution for several reasons, rather than interpreting them to mean that alcohol access had no effects on the opioid systems. First, we only examined MOR and KOR antagonists, and we did not assess effects on agonists for either opioid receptor or examine the effects of delta-opioid receptor agonists or antagonists. Second, we only assessed behavioral outcomes and did not assess whether there were any alterations in neurobiological measures that may have failed to alter behavior. Notably, other researchers have found that chronic alcohol exposure in rats (using several methods) can lead to altered MOR binding at least 5-21 days after the end of alcohol exposure (although several other experiments found no difference; as reviewed in [74]). In addition, after chronic i.p. alcohol injections (2 g/kg body weight, twice per day for 7 days), KOR antagonism increases dopamine release in the nucleus accumbens 16 h after the final alcohol injection, an effect not observed in saline-treated animals [75]. Likewise, chronic injections of alcohol (2 g/kg twice per day for 13 days) led to increased concentrations of dynophin (an endorphin that activated KORs) in nucleus accumbens 21 days after the final injection (with a trend towards increases after 5 days) and in the periacqueductal gray 5 days after the final injection (with a trend towards increases after 21 days) [76]. Although the experiments reviewed used methods that would lead to higher BECs than the voluntary access methods we used, our results do not exclude the possibility that alcohol exposure can lead to alterations in the opioid system using our alcohol access method or other methods of alcohol exposure.

4.3. Alcohol group and individual differences

In contrast to our expectations and our previous findings, we did not find individual differences in alcohol consumption associated with fear. It is possible that our discrepant findings may be due to procedural changes made between our current experiments and our prior research. There were two fairly significant changes to the procedures in the current studies compared with our previous studies in which we found that the level of alcohol consumption was associated with the level of conditioned fear during training and testing [14, 15]. First, we bred the rats for the current studies in our animal facility rather than purchasing and having them shipped to arrive on PND 21 (in response to reviewers of previous peer-reviewed papers suggesting that purchasing rats to arrive in our facility on PND 21 might represent early life-stress). This meant that unlike the animals in our previous study who were weaned 1 day early (on PND 20) and shipped across the country overnight, the rats in the current study were weaned on PND 21 and did not experience any potential shipping stress. Prior research on effects of shipping rats at weaning has shown decreased CB1 receptors in several brain areas involved in learning and memory [77] and alterations in the way that adolescent rats respond to injections of haloperidol or Δ9-tetrahydrocannabinol [78]. These effects could also alter how the nervous system responds to alcohol. Second, we altered the fear conditioning procedures from those used in our previous experiments [14, 15] to procedures with shorter sessions (primarily because it was not feasible to test the large number of rats required for these experiments within each day using our longer session length). Our prior experiments utilized 90-min fear conditioning sessions, each containing 10 tone-shock pairings, and a 35-min cue test with 4 tone presentations. The current study utilized a 51-min fear conditioning session with 8 tone-shock pairings and a 25-min cue test with 4 tone presentations. It is possible that these altered experimental parameters obscured the individual differences patterns we observed in previous experiments. Additional research is needed to determine the factors that affect when individual differences in voluntary alcohol consumption during early adulthood is associated with the level of conditioned fear.

In contrast to our expectations, we also did not find that the effects of the MOR and KOR antagonists differed in high versus low drinkers. However, as our prediction that effects of opioid receptor antagonists would differ in high versus low alcohol drinkers were based on our expectation that high versus low drinkers would differ in baseline levels of fear, perhaps it is not surprising that we did not find individual differences in response to opioid receptor antagonism. It is possible that differences in the opioid system were the cause of the different levels of conditioned fear in high versus low drinkers in the experiments in our previous publications we found previously (using different procedures, as described in the previous section) [14, 15]. However, without replicating the pattern of individual differences in baseline fear we previously observed, we cannot assess this possibility. There was some evidence that alcohol consumption assessed across the entire 6-week access period might be associated with conditioned fear using our current procedures. However, this effect was found in an exploratory analysis, and the period of alcohol consumption used to calculate the association (across the 6 weeks versus in the last 2 weeks) and the method of analyzing the association (computing a correlation versus dichotomizing drinking and comparing the two groups) differed from our previous research. Additional research will be needed to assess the effects of opioid receptors antagonists using procedures in which baseline fear differs between high and low alcohol drinkers, to determine if these individual differences are associated with differences in the opioid system.

4.4. Conclusions

In conclusion, we found that pre-training naltrexone administration augmented fear acquisition (observed during training and during a fear retention test) whereas pre-training LY2456302 administration had no effect on fear acquisition or expression. In addition, individual differences in alcohol consumption were not associated with individual differences in reactivity to opioid receptor antagonism, although we did not observe the relationship between alcohol consumption and baseline fear learning or expression that we observed in previous experiments. Therefore, it is difficult to determine whether the individual differences we observed under our previous experimental conditions were due to differences in one of these opioid systems. Further investigation is needed to determine the reasons for the discrepancy between this study and those we conducted previously, and to determine the mechanism for the individual differences in alcohol consumption associated with fear learning in our prior examinations.

Highlights.

• Pre-training naltrexone increased conditioned fear during training and test

• Pre-training LY2456302 (kappa opioid receptor antagonist) did not affect fear

• Alcohol drinking (alcohol vs. water and high vs. low) was unrelated to fear levels