Sex‐associated differences in incentive salience and drinking behaviour in a rodent model of alcohol relapse

Abstract

The ability of environmental cues to trigger alcohol‐seeking behaviours is thought to facilitate problematic alcohol use. Individuals' tendency to attribute incentive salience to cues may increase the risk of addiction. We sought to study the relationship between incentive salience and alcohol addiction using non‐preferring rats to model the heterogeneity of human alcohol consumption, investigating both males and females. Adult rats were subjected to the alcohol deprivation effect (ADE) paradigm, where they were given voluntary access to different alcohol solutions with repeated interruptions by deprivation and reintroduction phases over a protracted period (five Alcohol Deprivation Cycles). Before each Alcohol Deprivation Cycle, rats were tested in the Pavlovian Conditioned Approach (PCA) paradigm, which quantifies the individual salience toward a conditional cue and the reward, thus allowing us to trace the process of attributing incentive salience to reward cues. During the final Alcohol Deprivation Cycle (ADE5), animals were tested for compulsive‐like behaviour using quinine taste adulteration. We investigated sex differences in drinking behaviour and PCA performance. We observed thatb females drank significantly more alcohol than males and displayed more sign‐tracking (ST) behaviour in the PCA, whereas males showed goal‐tracking (GT) behaviour. Furthermore, we found that high drinkers exhibited more ST behaviour. The initial PCA phenotype was correlated with later alcohol consumption. Our findings indicate a complex relationship between incentive salience and alcohol addiction and emphasize the importance of considering both sexes in preclinical research.

The ADE model was used to investigate the contribution of PCA to the development of alcohol addictive and compulsive behavior. We found that females drank significantly more alcohol than males throughout all ADE phases and exhibited more ST behavior in the PCA, whereas males showed GT behavior. PCA phenotypes were stable throughout the experimental time. Together, our results show sex effects in both incentive salience and alcohol drinking behavior. Thus, our findings indicate a relationship between incentive salience and alcohol addiction.

1. INTRODUCTION

Alcohol addiction is the result of a complex interplay between social, psychological and biological factors ¹ , ² and has devastating health and economic consequences. Harmful alcohol consumption accounts for 8.6% and 1.7% of the global disease burden in men and women, respectively. While numerous environmental and genetic factors associated with problematic alcohol‐drinking and addiction have been identified, ³ a clarification of how these factors differ by sex is needed. ⁴

It has been proposed that the development of alcohol dependence is mediated by the degradation of executive control over behaviour and a transition from goal‐directed to more habitual and compulsive drug‐seeking. ⁵ Furthermore, drug‐related stimuli can acquire significance when presented and induce excessive emotional and motivational responses. ⁶ The tendency to attach excessive motivational importance to stimuli may predispose individuals to develop addictive behaviour and is observed to vary between different types of drinkers, for example, heavier drinkers show a more substantial attentional bias toward alcohol‐related cues than low drinkers. ⁷ Over time and after repeated alcohol exposure, neutral cues become associated with reward drive and reinforce reward‐seeking behaviour by attributing incentive‐motivating properties. ⁸ Previously neutral cues can trigger craving and drug‐seeking behaviour, eventually leading to relapse. ⁹

Understanding how incentive salience and relapse processes are related may provide essential insights into addiction mechanisms. A paradigm to assess the process of attributing salience to reward stimuli is the Pavlovian conditioned approach (PCA). ¹⁰ , ¹¹ , ¹² , ¹³ The PCA paradigm allows to trace the process of attributing incentive salience to reward cues and to identify different phenotypes. While some animals interact exclusively with unconditioned stimuli (so‐called ‘goal‐trackers’ (GT)), others tend to engage with conditioned stimuli (so‐called ‘sign‐trackers’ (ST)). ST animals appear to assign higher incentive salience to stimuli associated with reward and are thought to be more prone to developing drug addiction. ⁸ , ¹⁴ , ¹⁵

To evaluate the mechanisms underlying alcohol addiction, we focused on the alcohol deprivation effect (ADE) paradigm, a validated preclinical model of relapse to alcohol. ¹⁶ , ¹⁷ In this model, animals are given long‐term access to alcohol, with repeated periods of deprivation. When alcohol is reintroduced after a deprivation period, it leads to a transient increase in consumption (the ADE), accompanied by an eventual shift in preference toward more concentrated alcohol solutions. ¹⁷ , ¹⁸ Further, investigations in alcohol‐preferring rats have found that after several deprivation periods, rats show compulsive‐like drinking behaviour and continued alcohol consumption despite the presence of taste‐aversive stimuli like quinine, suggesting the loss of control and behavioural inflexibility. ¹⁷ , ¹⁹ , ²⁰ , ²¹

Both PCA and ADE studies have often been conducted in male, alcohol‐preferring animals, which may provide an incomplete picture and limit the generalizability and translational value of findings. Drinking patterns in human males and females differ, with females drinking less than males. ²² In contrast, in both rats ²³ and mice, ²⁴ alcohol intake is higher in females than in males. Thus, it is important to conduct research on females to examine the attribute incentive salience, the consumption of alcohol and the development of compulsive drinking behaviours. A longitudinal approach will give a better understanding of the sex‐dependent developmental trajectories of drinking behaviour and associated biomarkers.

In the present study, we aimed to assess whether early behavioural markers were associated with addiction‐like behaviour, and whether they inform later developmental trajectories of drinking behaviour in alcohol non‐preferring rats. In both male and female nonalcohol‐preferring rats, we used the PCA and ADE experimental paradigms to investigate the development of incentive salience attribution and the loss of control over alcohol consumption. In addition, the relationship between performance on the PCA task and drinking behaviour over time was examined.

2. MATERIALS AND METHODS

2.1. Animals

We followed EU guidelines (European Communities Council Directive 2010/63/EU), and all procedures were approved by the local state authority (Landesamt für Gesundheit und Soziales [LaGeSo], Berlin) guidelines. Experiments were conducted using PND60 female (alcohol‐drinking n = 13, controls n = 8) and male (alcohol‐drinking n = 15, controls n = 8) RccHan Wistar rats (Envigo, Netherlands). Sample size was predetermined by power analyses. All animals were single housed in a 24 h light/dark cycle (lights on at 6 am, lights off at 6 pm). Single housing of animals allows the accurate recording of high‐resolution alcohol consumption at the level of the individual. Room temperature and humidity were kept constant (temperature: 20–24°C, relative humidity: 45%–65%). Standard laboratory rodent chow (Ssniff, Soest, Germany) and water were provided ad libitum throughout the experimental period. All efforts were made to reduce animal suffering and the number of animals used.

2.2. ADE procedure

On postnatal day 60 (PND60), all rats were provided ad libitum access to tap water and 5%, 10% and 20% ethanol solutions (v/v, prepared from 96% ethanol [TechniSolv, VWR Chemicals, France] diluted in tap water) in a 4‐bottle free‐choice paradigm. Special bottle caps (Zoonlab GmbH, Germany; bent by infra e motion: Fritz Kutschera) were used to minimize spillage and evaporation. The positions of the bottles were changed regularly to avoid location preferences.

Rats underwent five Alcohol Deprivation Cycles. After an initial 8 weeks of continuous voluntary alcohol access, the first 2‐week deprivation period was introduced, followed by the reintroduction of alcohol. Over the next 11 months, rats underwent subsequent deprivation (3–4.5 weeks) and reintroduction phases (6–8 weeks) (Figure 1A). Ethanol consumption and animal weights were measured daily for 5 days before and after each deprivation phase. Baseline ethanol intake (BL) was calculated as the daily average across the 5 days (grams of total pure ethanol consumed per day and kg of body weight (g/kg/day) preceding a deprivation period). Upon reintroduction of alcohol, consumption was measured daily to compare with baseline consumption. For the first 24 h of the 5th alcohol‐reintroduction phase, quinine was added to the three different alcohol solutions (5%, 10%, 20%) at a concentration of 0.1 g/L. Note that quinine is in the bottles only on Day 1 (D1), with increased consumption from Day 2 as a result of the removal of quinine. No method was used for randomization to allocate animals to experimental groups.

FIGURE 1.

Long‐term drinking over the experimental period measured by daily total ethanol (EtOH) intake (g/kg/d) before and after an EtOH withdrawal period of 2 to 4 weeks. (A) Experimental design showing the time of alcohol and deprivation periods as well as Pavlovian conditioned approach (PCA), defined as the alcohol deprivation cycle. At the end the quinine test was performed (B) alcohol deprivation Cycle 1 and 4. (C) Quinine test (alcohol deprivation Cycle 5). BL intake is the mean of the last 5 days of baseline drinking. Note that quinine is only in the bottles on D1, with increased consumption from D2 likely due to its removal. Error bars indicate SEM. *p < 0.05; **p < 0.01; ***p < 0.001 indicate statistically significant differences between males and females. Abbreviations: ADE5D1, quinine; ADE, alcohol deprivation effect; BL, baseline; D, day; EtOH, ethanol; females, orange; males, purple.

2.2.1. PCA

PCA tests were performed at the end of each deprivation phase to determine incentive salience phenotypes (Figure 1A). Rats' individual performances in PCA test indicate three different behavioural phenotypes: sign‐trackers (ST), which interact with conditioned stimuli (CS), attributing incentive salience to cues; goal‐trackers (GT), which interact with the reward directly; and intermediate trackers (IT), which show both behaviours. ²⁵ , ²⁶ Control groups not exposed to the ADE paradigm (n = 16, 8F/8M) also completed the PCA procedure in parallel.

Operant conditioning chambers

Four operant conditioning chambers (20.5 × 24.1 cm floor area, 29.2 cm high; IMETRONIC, Pessac, France) at the Animal Behavioural Phenotyping Facility (ABPF, Charité Berlin Mitte, Germany) were used for PCA training. Each chamber had a food pellet magazine containing an infrared sensor. A single retractable lever was located at 2.5 cm either right or left from the pellet magazine. The position of the lever on either side of the food receptacle was counterbalanced across boxes to eliminate an effect of side bias. Operation of a pellet dispenser delivered 45‐mg banana‐flavoured food pellets (BioServe, #F0059, Frenchtown, NJ, USA) into the food magazine. Two days before starting PCA procedures, 25 pellets were placed in each home cage to familiarize the animals with the food. Each operant conditioning chamber was located in a sound‐attenuating cabinet.

PCA—Procedure

PCA sessions were conducted between 06:00 AM and 06:00 PM. The week before starting the PCA paradigm, animals were acclimatized to the behavioural facility. Each PCA block lasted six consecutive days, with 1 day of PCA pretraining followed by 5 days of PCA training (one session per day). ²⁵

PCA—Pretraining

During the first 5 min of pretraining, rats were habituated to the conditioning boxes. Afterward, 25 food pellets were delivered on a variable interval (VI) 60‐s schedule (average interval: 30 s, range: 0–60 s). Pretraining sessions lasted approximately 17.5 min. All rats consumed all food pellets.

PCA—Training

Each training session consisted of 25 trials. Each trial consisted of a 1‐min waiting period, followed by the presentation of the lever (conditioned stimulus; CS) into the chamber for 8 s and simultaneous illumination with an LED light. Retraction of the lever was immediately followed by the response‐independent delivery of one food pellet (unconditioned stimulus; USA). The CS was presented on a random interval 90‐s schedule (average interval: 90 s, range: 30–150 s). Each session lasted approximately 40 min.

PCA—Behavioural characterization

PCA index scores were calculated for each Alcohol deprivation cycle to assign animals the different phenotypes (ST, IT, GT). ²⁵ Briefly, for each session, a PCA score was calculated based on the number of times animals interacted with the lever or waited at the pellet dispenser for the reward. Output parameters (number and latency of lever presses and magazine entries) were recorded using POLY software (Imetronic, Pessac, France) and used to pre‐calculate Response Bias, Latency Score and Probability Difference for each animal and eventually to determine the PCA scores. PCA scores from the last 2 days were averaged to obtain the final PCA Index Score.

Animals whose scores ranged from 0.5 to 1.0 were classified as STs; those whose scores ranged from −0.5 to −1.0 were categorized as GTs. Intermediate responders ranged between −0.5 and 0.5 (for more details, see previous works ²⁵ ). The term ‘performance’ refers to how often the animal engages with either the lever or the pellet dispenser to be classified as ST or GT.

2.3. Statistical analyses

Statistical tests were performed using GraphPad Prism software (Version 9.1, La Jolla CA, USA, 2020). A probability p‐value of <0.05 was considered significant.

Mixed‐effects models were used to analyse longitudinal data, specifying a compound symmetry covariance matrix assuming equal variance and covariance and using Restricted Maximum Likelihood (REML) estimation, with the Geisser–Greenhouse procedure to correct for violations of the assumption applied when required. Sidak tests were performed to correct for multiple comparisons.

Pearson correlation analyses were conducted for the whole sample and stratified by sex. No animal was excluded from our dataset. The investigator who ran the analysis was blind to the group allocation during the experiments and when analysing the data.

2.3.1. Drinking behaviour

To test for the presence of ADEs (a significantly higher alcohol intake after reintroduction compared to baseline), alcohol consumption (g/kg) was specified as the dependent variable, with sex (M/F) and day (BL and ADE D1) as fixed factors and individual rat and residuals as random factors. The same model was specified for all Alcohol Deprivation Cycles (including the final ADE + quinine). Pearson correlations were also calculated to examine the relationship between BL and ADE drinking. Alcohol solution preference (5%, 10% and 20%) during cycles was calculated using a preference index (PI) following:

$$\mathrm{PI}=\left(5\%\times \text{vol5}\%\right)+\left(10\%\times \mathrm{vo}10\%\right)+\left(20\%\times \mathrm{vol}20\%\right)/\text{vol5}\%+\mathrm{vol}10\%+\mathrm{vol}20\%$$

A mixed model was used to test for effects of time points and sex on preferences.

2.3.2. PCA

Proportions of different PCA phenotypes over time were examined. To investigate the development of PCA index scores over time, PCA score was specified as the dependent variable, with sex (M/F), time point (Alcohol Deprivation Cycle during which the PCA test was conducted) and the interaction between sex and time point as fixed factors and individual rat and residuals as random factors.

2.3.3. Drinking and PCA performance

To examine the relationship between PCA performance and drinking, Pearson correlations between drinking (during BL and ADE phases) and corresponding PCA performance within the same cycle (e.g., the correlation between PCA1 and BL1/ADE1) were calculated.

PCA performance results were defined as the average of the last 2 days of PCA training during the respective PCA block. To examine the ability of initial PCA performance to predict later drinking outcomes, we also tested the correlation of PCA score at the first cycle (PCA1) and later drinking (BL4/ADE4, BL5/ADE5). We define ‘initial’ as the first PCA phenotype that emerged during the first PCA test (phenotype is calculated from five consecutive training days and described as PCA1).

3. RESULTS

3.1. Drinking behaviour

Descriptively, the ADE was most pronounced at ADE1 in both males and females and decreased over subsequent Alcohol Deprivation Cycles (Figure 1B). During the second and third cycles, no changes were observed in PCA performance compared to the first PCA block, and ADEs did not reach statistical significance (data not shown).

ADEs were observed during the first day of ADE1 (main effect of day, F[1.26] = 36.5, p < 0.001). During ADE4, the magnitude of the ADE approached but did not reach significance (F[1.25] = 3.80, p = 0.06).

Female rats drank more ethanol than male rats (main effect of sex, ADE1: F(1.28) = 37.2, p < 0.001; ADE4: F(1.26) = 24.2, p < 0.001).

No significant interaction effects between sex and day were observed during ADE1 (F(1.26) = 0.206, p = 0.65) or ADE 4 (F[1.25] = 0.688, p = 0.41).

Preference scores of ethanol concentrations did not significantly differ over time points (F[1.94; 27.2] = 2.09, p = 0.14), or according to sex (F[1.00; 14.0] = 0.091, p = 0.77).

3.2. PCA performance

PCA performance was established before ADE measurements (the first cycle) and did not change significantly over time (F[2.11, 55.0] = 1.42, p = 0.25). Figure 2B illustrates phenotype proportions throughout the experiment. All other rats that started as either sign‐trackers (predominantly female rats) or goal‐trackers (only males) maintained the phenotypes across the 5 cycles (5 PCA). None of the female rats were classified as GT at any time point. Only a single male rat began as an IT (PCA1 Index Score: 0.36) and showed ST behaviour during the remainder of the experiment (Figure 2B). No interaction effect between sex and timepoint was observed (F[3, 78] = 0.148, p = 0.93) (Figure 2A).

FIGURE 2.

Pavlovian conditioned approach (PCA) paradigm. (A) Index Scores and phenotypes for ctrl (up) and ADE (below) animals during the experiment. (B) The percentage distribution of phenotypes averaged over the final 2 days of PCA training for ctrl (up) and ADE (below) animals stratified by sex. Error bars represent ±S.E.M. *p = 0.05; **p < 0.05; ***p < 0.001; ST Phenotype Score ranged from score ≥0.5 (red) and GT Phenotype Score ≤0.5 (green). Abbreviations: ADE, alcohol drinker; ctrl, control; F, females; GT, goal tracker; IT, intermediate tracker; M, males; PCA, Pavlovian conditioned approach; ST, sign tracker.

In controls not exposed to alcohol, a similar effect of sex was observed (F[1.14] = 4.37, p = 0.047), and PCA performance also did not change over time.

3.3. PCA behaviour and drinking behaviour

Table S1 shows correlations between PCA index scores and drinking behaviour within the same Alcohol Deprivation Cycle in the whole group and stratified by sex. A significant positive correlation was observed between PCA4 and ADE4 drinking behaviour for all animals (r[27] = 0.511; p = 0.005) (see Figure 3A and Table S1). Interestingly, when stratified by sex, no significant associations between PCA performance and drinking behaviour were observed (during any Alcohol Deprivation Cycle), suggesting that differences between males and females drive the associations.

FIGURE 3.

Correlation between Pavlovian Conditioned Approach (PCA) and ADE. (A) Alcohol deprivation Cycle 4 showed a significant correlation between PCA and its respective ADE, for example, PCA4 and ADE4D1 for the whole group (males = purple; females = orange). There was no correlation when restricted to within sexes. (B) Between the initial PCA index score and later ADE, there was a significant positive correlation between initial PCA and ADE4, for example, for PCA1 and ADE4D1 for the whole group. Data represent alcohol intake (g/kg) and comprise PCA index scores: STs (score ≥0.5), IRs (−0.5 < score <0.5), as well as GTs (score ≤−0.5). Pearson's correlation coefficient (r) between drinking conditions at the same and different cycles for the whole group of animals. The stripped lines indicate two confidence bands (95% confidence). The solid line shows the best fit line. Abbreviations: ADE, alcohol deprivation effect; D, day; PCA, Pavlovian conditioned approach.

3.4. Initial PCA and later drinking

When looking at the whole group of rats, we found a significant positive correlation between PCA1 and later relapse drinking ADE4D1 (r[27] = 0.489; p = 0.008) (Figure 3B; Table S2). When looking at each sex independently, we found no association between PCA1 and drinking during reintroduction.

3.5. Quinine

Considering the whole group of animals, no ADEs were observed during quinine adulteration; instead, a reduction of ADE5D1 drinking compared to baseline (BL5) was observed (main effect of day, F(1.26) = 28.7, p < 0.001). Female rats drank more than male rats (the main effect of sex, F(1.26) = 35.2, p < 0.001), and quinine had a larger effect on female drinking (interaction of sex and timepoint, F(1.26) = 10.3, p = 0.003) (Figure 4A), reducing consumption to a greater degree. Post hoc Sidak tests showed that the reduction of alcohol intake in females was significant (p < 0.001). Drinking was also reduced in males, but the difference was not significant (p = 0.660) (Figure 4A).

FIGURE 4.

Characterization of compulsive drinking by quinine adulteration test. Compulsive drinking, BL5 with ADE5D1_Quinine (A) stratified by sex and (B) individual analyses for each animal per group, females (left panel) and males (right panel). Straighter, darker lines in each graph represent animals with increased drinking during ADE5D1_Quinine vs. BL5. Abbreviations: ADE5D1, Quinine; ADE, alcohol deprivation effect; BL, baseline; D, day; EtOH, ethanol; F, females (orange); M, males (purple).

At the group level, compulsive‐like drinking was not observed. At the individual level, we observed increased drinking during reintroduction in one female and four male rats (Figure 4B).

Additionally, we observed no association between PCA1 score and ADE5D1 drinking (r[28] = 0.030, p = 0.883) in the whole sample. When stratifying by sex, we found a significant negative correlation between PCA1 and ADE5D1 drinking (r[13] = −0.569; p = 0.042) in females (Figure 5, Table S2), but not in males (r[15] = 0.106, p = 0.707).

FIGURE 5.

Correlation between initial Pavlovian conditioned approach (PCA) index score and drinking. (A) For females, there was a significant negative correlation in alcohol deprivation Cycle 5 between PCA1 and ADE5D1. (B) No significant correlation was found between the initial PCA index score and drinking for males. Data represent alcohol intake (g/kg) and PCA index scores: STs (score ≥0.5), IRs (−0.5 < score <0.5), as well as GTs (score ≤−0.5)—Pearson's correlation coefficient (r) between initial PCA (PCA1) and drinking condition. The stripped lines indicate two confidence bands (95% confidence). The solid line shows the best fit line. *p < 0.05; **p < 0.01; ***p < 0.001. Abbreviations: ADE5D1, quinine; ADE, alcohol deprivation effect; D, day; PCA, Pavlovian conditioned approach.

4. DISCUSSION

The present study examined the relationship between incentive salience and alcohol consumption in a longitudinal preclinical model of alcohol relapse in both male and female non‐preferring RccHan Wistar rats. We observed clear sex differences in alcohol‐drinking and found that females drank more alcohol compared to males. Regarding incentive salience behaviour, female rats showed sign‐tracking behaviour, while male rats exhibited more goal‐tracking behaviour. In general, drinking patterns and PCA phenotypes were established from the beginning of the experiment and remained constant throughout the experiment. We did not find evidence for compulsive‐like drinking in response to the quinine challenge.

We observed sex differences in alcohol consumption throughout the experiment, both at baseline and reintroduction, with female rats drinking significantly more alcohol than males. This is in line with previous research in both alcohol‐preferring and non‐preferring rats. ²⁴ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ In fact, female rats were shown to be more susceptible to the reinforcing effects of alcohol than male rats. ³² , ³³ In analogy, human studies have observed that women have a higher risk of relapse‐like behaviour than men. ³⁴ In most human epidemiological studies, men drink more than women, but what is generally not taken into account is that women have lower body mass than men. In rodent studies, including ours, this is taken into account as alcohol consumption is quantified in g/kg. Interpretation of results from alcohol research in humans is also further complicated by sociocultural factors. ³⁵ , ³⁶ It should also be noted that the levels of alcohol consumption in women are increasing faster than in men, underscoring the need for research including female animals. ³⁷

In contrast to alcohol‐preferring rats, ¹⁷ , ¹⁸ the RccHan rats used in our experiments did not transition to increased preference for higher concentration solutions with higher alcohol concentrations, suggesting that they may overall be less susceptible to alcohol addiction. We did not find the development of stronger ADEs over time, but instead a weakening, which may mirror behaviour in people who consume alcohol but do not become addicted or who do not lose control over alcohol consumption.

PCA performance was largely unchanged throughout the experiment, with different behavioural phenotypes already emerging at the first testing. There was a marked sex difference in PCA phenotypes, with females showing ST behaviour and males showing more GT behaviour. These findings are in line with results reported for Sprague‐Dawley rats ²⁵ , ³⁸ , ³⁹ and for Wistar rats. ⁴⁰ , ⁴¹ Previous studies found that STs attribute more salience to drug‐paired cues and are more susceptible to cue‐induced reinstatement of drug‐seeking behaviour than GTs. Those studies were mainly conducted with male Sprague‐Dawley rats, ¹¹ , ¹⁴ , ⁴² , ⁴³ , ⁴⁴ but PCA in females indicated that they exhibit more and faster ST behaviour. ²⁷ , ⁴⁵

Research previously showed that ethanol‐associated cues (e.g., auditory stimuli) can lead rats treated with ethanol to seek rewards that are not associated with ethanol. ⁴⁶ Drug consumption can reinforce nondrug‐related reward behaviour. ⁴⁷ , ⁴⁸ , ⁴⁹ Thus, environmental stimuli may elicit maladaptive responses under some circumstances.

We observed more ST behaviour at higher levels of alcohol intake after reintroduction. Together with the fact that PCA phenotypes were established early and did not change over time, these phenotypic characteristics might serve as early markers of eventual alcohol‐drinking/relapse. However, when stratified by sex, the correlation between PCA index score and drinking behaviours was not significant, suggesting that sex differences drive the associations when considering the whole sample (i.e., females drank more and had more sign‐tracking behaviour than males). It was expected that STs would develop addictive behaviour differently than GTs and that PCA phenotypes could be used as a type of marker. However, we found that PCA performance was already established at the outset and did not change over the experiment and that sex was a strong determinant of both drinking behaviour and PCA performance.

There are other studies investigating how the sign‐tracker phenotype is related to human behavioural traits such as compulsivity, impulsivity and addiction. The ST phenotype is also associated with high impulsivity, compulsive tendencies and addiction tendencies. ⁵⁰ This suggests a co‐segregation or simultaneous occurrence of these behaviours. Hence, the sign‐tracker phenotype may serve as an indicator or marker of impulsivity, compulsivity and addiction‐related behaviours.

Previous ADE studies have broadly used alcohol‐preferring strains, such as P rats ⁵¹ or Wistar rats, ²¹ , ⁵² while previous PCA studies mostly used Sprague‐Dawley and Wistar rats. ⁵³ The rat strain employed in the present study, namely, RccHan Wistar rats, is less prone to alcohol consumption but shows more variability‐related individual differences in voluntary alcohol‐drinking and is therefore believed to better reflect the human population. ⁵⁴ Indeed, in our study, we did not find the development of stronger ADEs over time, underscoring the value of using RccHan Wistar rats as a translational model. This may even have the potential to examine protective factors against loss of control in future studies.

Overall, compulsive‐like drinking was not observed in the present study. In females, we found that more ST behaviour was correlated with less drinking during quinine adulteration. Previous studies in alcohol‐preferring (P) rat strains have demonstrated that high alcohol intake does not necessarily lead to or predict the development of compulsive alcohol‐seeking ¹⁹ or compulsive drinking. ⁵⁵ It was also shown that habitual alcohol exposure did not predict the transition to compulsivity in alcohol‐preferring rats. ¹⁹ Our results are in line with this, as female rats who drank more throughout the experiment but did not develop stronger compulsivity as male rats.

The present study has several limitations. The sample sizes of the subgroups are small, and larger independent samples will be needed to validate the results. We found correlations between alcohol consumption and specific PCA behaviour in alcohol‐drinking subtypes.

While other studies have observed compulsive‐like behaviour with quinine, we found that drinking in the RccHan rat strain is not as strongly affected by quinine. Different behavioural tests, for example, foot‐shock resistance, ⁵⁶ , ⁵⁷ , ⁵⁸ might be better suited to examine compulsive drinking. However, due to ethical considerations and the German/EU law (2010/63/EU), alternatives need to be found. Another limitation is that we did not determine the exact aversive threshold in this rat strain. However, quinine concentrations of 0.1 g/L were shown to be effective in many studies. Still, we cannot exclude that variability between strains and facilities could have affected our results.

4.1. Conclusion

In this study, we observed sex differences in both incentive salience attribution and alcohol‐drinking behaviour in rats. Humans also differ in their susceptibility to cues that contribute to problematic alcohol use. Thus, incentive salience behaviour may serve as a predictor of later drinking outcomes. Further studies are needed to better understand the sex‐related differences in the processes contributing to drinking behaviour. In addition, the molecular and cellular mechanisms underlying the behavioural changes and sex differences needed to be explored.

AUTHOR CONTRIBUTIONS

Aileen Hakus conducted experiments, analysed data and wrote the manuscript. Jerome Clifford Foo analysed data and wrote the manuscript. Marta Casquero‐Veiga, Franziska Hintz and Asude Zülal Gül contributed to conducting experiments. Marion Rivalan and York Winter provided conditioned boxes. Ravit Hadar conceptualized the project and wrote the manuscript. Josef Priller conceptualized the project. Christine Winter conceptualized the project and supervised Aileen Hakus and wrote the manuscript. All authors read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no competing interests.

Supporting information

Table S1: Pearson correlations of PCA index score and drinking during the same time point in the whole sample and stratified by sex.

Table S2: Correlation of Initial PCA index score and later drinking in the whole sample and stratified by sex.

Hakus A, Foo JC, Casquero‐Veiga M, et al. Sex‐associated differences in incentive salience and drinking behaviour in a rodent model of alcohol relapse. Addiction Biology. 2025;30(1):e70009. doi: 10.1111/adb.70009

DATA AVAILABILITY STATEMENT

All data needed to evaluate the conclusion are present in the paper and/or the Supplementary Materials.