Inactivation of the cholinergic M4 receptor results in a disinhibited endophenotype predicting alcohol use

The muscarinic cholinergic M4 receptor subtype (M4 mAChR) is densely expressed in brain areas known to be involved in the reinforcing effects of drugs of abuse and we were the first to show that mice lacking M4 mAChRs exhibit elevated operant responding for alcohol and reduced capacity to extinguish this alcohol-seeking behaviour. Here we explore possible underlying determinants of this phenotype. We subjected M4 mAChR knockout mice and their littermate wildtype controls to tests of spontaneous activity, learning and memory, novelty seeking, as well as anxiety and examined the relationship of a newly discovered "disinhibited" endophenotype of these mice with voluntary alcohol consumption and relapse. We found a positive correlation between "disinhibited" behaviour on the plus maze and alcohol preference as well as relapse to alcohol drinking after a period of abstinence. Taken together, these data point to M4 mAChRs as a potential target for improved treatment strategies for alcohol use disorder. This receptor should be further investigated for its involvement in modulating behavioural inhibition in relation to loss of control over consumption of alcohol.


Introduction
Despite remarkable progress in the understanding of susceptibility factors contributing to the development of alcohol use disorder, the exact mechanisms behind this phenomenon remain to be elucidated. Today we know that consumption of alcohol is linked to the activity of an individual's brain reinforcement system [1]. The ventral tegmental area (VTA), the nucleus accumbens (NAc) and the prefrontal cortex (PFC) constitute the main sites of the brain reinforcement system, with the effects of alcohol being particularly linked to dopamine activity in the NAc [1][2][3][4][5]. Alcohol-induced increase of extracellular dopamine in the NAc is, in turn, dependent on cholinergic activity within the VTA, the posterior pedunculopontine nucleus (PPN) and the laterodorsal tegmental nucleus (LDT) [6]. Nicotinic acetylcholine receptors in the VTA were previously shown to be involved in the control of alcohol drinking behaviour and alcohol-induced dopamine overflow in rodent models [6][7][8][9]. However, until recently, the G protein-coupled muscarinic acetylcholine receptor subtypes (M 1 -M 5 ) were not investigated in this regard. Since the M 4 receptor subtype is largely expressed in the PPN, the LDT and the NAc, it may constitute an important target for regulation of alcohol-induced dopamine activity and alcohol drinking behaviour [10][11][12][13]. Indeed, we have previously reported that M 4 − /− mice, in which the M 4 receptor is inactivated, display an increased operant response for alcohol and a reduced capacity to extinguish their alcohol-seeking behaviour [14]. Also, other laboratories have shown that the M 4 receptor is downregulated in some brain regions in individuals with alcohol use disorder and in rats after long-term alcohol consumption and that positive allosteric modulators of M 4 receptor signaling reduced both home cage and operant alcohol self-administration in rats [15][16][17]. However, the mechanisms by which this is established remain to be fully understood. Possibly the M 4 receptor is involved in mediating the reinforcing value of drugs of abuse or it may be involved in the regulation of behaviours such as anxiety and novelty seeking often referred to as vulnerability markers for the development of substance use disorders. Such behavioural traits have previously been associated both with midbrain dopamine function and drug-taking behaviour [18,19] and M 4 − /− mice have previously been referred to as displaying a dopamine-hyper reactive behavioural phenotype [20][21][22][23]. Differences in novelty-seeking behaviour have also been associated with alcohol and substance use disorders in humans [24][25][26][27]. The role of anxiety in the development of these disorders is less clear; however, anxiety-like behaviours have been linked to both changes in dopamine function and drug self-administration behaviours [28,29] and M 4 − /− mice were previously found to exhibit decreased burying behaviour in the shock-probe burying model of anxiety-like behaviour, while they did not differ from wild type mice in the light-dark transition test [30,31]. Addiction may also be viewed as a disorder of learning and memory, as converging evidence points to addiction representing a pathological usurpation of normal mechanisms of learning and memory [32] and the cholinergic system is clearly implicated in these functions [33]. Consequently, it is plausible that the M 4 receptor influences drug taking behaviour via this avenue. Indeed, it has been reported that a positive allosteric modulator of the M 4 receptor enhances some memory functions [34][35][36], while M 4 − /− mice were found to have normal memory in various models [30,31,37]. However, M 4 − /− mice have not previously been assessed on the Barnes maze.
To further our understanding of the role of M 4 receptors in the regulation of alcohol consumption we assessed M 4 − /− mice and their littermate WT controls in tests of learning and memory, anxiety-like behaviour, spontaneous activity and novelty seeking, and examined the relationship of a newly discovered "disinhibited" endophenotype of these mice with voluntary alcohol consumption and relapse.

Animals and housing conditions
Male M 4 − /− mice were generated as previously described [22] and bred at the Panum Institute, University of Copenhagen in a fully AAA-LAC accredited facility. Founder mice of a mixed genetic background (129SvEv/CF1) were backcrossed to the C57BL/6Ntac strain for 13 generations, and genotyping was performed on mouse-ear DNA using the polymerase chain reaction. Male M 4 +/+ littermates were used as controls. After weaning, mice were housed with littermates in groups of 4-8 in Makrolon cages (20 ×35×15 cm 3 , Tecniplast, Varese, Italy). They were provided pelleted feed (Altromin 1319; Brogaarden, Gentofte, Denmark) and acidified tapwater ad libitum. Cages were lined with aspen chips (Tapvei Oy, Kortteinen, Finland) and enriched with nest building material (Lilico, Horley, UK), bite blocks (Tapvei Oy) and cardboard tubes (Lilico). The animal room was kept at a constant temperature (22-24 ℃) and the light regimen was a 12/12 h dark/artificial light cycle with 30 min of "twilight" and the lighting period starting at 7:00 AM. After being transferred from the breeding facility, the animals were allowed to acclimatize to the experimental facility for at least 1 week prior to initiation of any experiments. All behavioral experiments were performed during the light cycle between 8:00 AM and 4:00 PM. Separate cohorts of mice were used for the different behavioural tests and mice continued to be group housed with littermates unless otherwise indicated.

Ethics
The animal experiments were approved by the Animal Experiments Inspectorate under the Danish Ministry of Food, Agriculture and Fisheries (license number 2012-15-2934-00038). All procedures were performed in accordance with the EU directive 2010/63/EU in a fully AAALAC accredited facility under the supervision of a local animal welfare committee. All efforts were made to minimize pain or discomfort as well as the number of animals used during the experiments.

Drugs
Alcohol (96%, Sigma Aldrich, Denmark) was diluted with tap water to produce solutions of 2-16% of alcohol.

Spontaneous locomotor activity
Assessment of spontaneous locomotor activity was conducted in locomotor activity cages (approximately 100 lx inside the cages) (Ellegaard Systems, Denmark) equipped with 5 × 8 infrared light sources plus photocells. The light beams crossed the cage 1.8 cm above the bottom of the cage. The recording of a motility count required the interruption of two adjacent light beams, thus avoiding counts generated by stationary movements of the mice. All experiments were conducted in a clean cage with a scant lining of bedding material. Mice were transported to the test room and allowed to acclimatize for 30 min before being placed individually into the apparatus and activity was recorded for 60 min (N = 12 for both genotypes).

Barnes maze
The Barnes maze consisted of a brightly illuminated (280 lx) circular white-coated platform 100 cm in diameter and elevated 72 cm above the ground. Sixteen 3.9 cm-diameter holes were evenly distributed around the perimeter, 2 cm from the edge. One of those holes allowed the mouse to enter the escape box, a dark plastic storage box (12 ×8×6 cm) located under the escape hole and containing bedding material from the home cage of the mouse being tested. All trials were recorded by a camera mounted above the maze and analysed by the video tracking program EthoVision (Noldus Information Technology, Wageningen, The Netherlands, version 11). The mice underwent three days of shaping to enter the escape box before training started. On day 1 and 2 of shaping, the mice were released in close proximity of the escape hole and confined there for 3 min by a circular semitransparent beaker (diameter 20 cm) surrounding mouse and escape hole. Mice that had not entered the escape hole after the 3 min had elapsed were gently guided there by the experimenter. After 30 s the escape box was removed, and the mouse was transported in the escape box to its home cage. This was repeated 2 times on day 1 and 3 times on day 2. On day 3 of shaping two 40 cm high wooden walls delineated a wedge-shaped corridor from the middle of the maze to the escape hole and its two adjacent holes. The mice were now released in the center of the maze and allowed 3 min to find their way into the escape box. Mice that had not entered the escape box when the 3 min had elapsed were gently guided there by the experimenter. After 30 s the escape box was removed, and the mouse was transported in the escape box to its home cage. This was repeated 2 times.Training consisted of two trials a day for 6 consecutive days. For each trial, the mouse was released in the middle of the maze oriented in a random direction and allowed to explore the maze freely until it entered the hidden escape box or until 3 min had passed. Mice that did not find the escape box within the 3-minute trial were guided there gently by the experimenter. Twenty four hours after the last training session a probe trial was conducted, where the escape box was removed, the mouse was released in the middle of the maze, and allowed to explore freely for 3 min. The video tracking system recorded the number of errors (number of visits to other holes before visiting the escape hole for the first time) and distance travelled (distance before the first visit to the escape hole) N = 8 M 4 +/+ and N = 15 M 4 -/mice were used for this experiment.

Novel object exploration
An open black plastic arena measuring 77 × 56×41 cm was used and illuminated indirectly (80 lx). Behaviour was recorded by a video camera mounted vertically above the test arena and analysed using EthoVision. First a mouse was allowed to explore the empty arena for 30 min. At that point it was briefly removed, while a novel object was placed in the middle of the arena. Then the mouse was re-introduced and allowed to explore for another 5 min [38]. Latency to approach the novel object and time spent exploring the novel object was recorded. N = 8 M 4 +/+ and N = 10 M 4 -/mice were used for this experiment.

Novel environment exploration
The apparatus consisted of two equally sized (19 ×19×43 cm) compartments with distinct visual cues, separated by a wall with a centrally placed door (3.5 ×3.5 cm) that was closed during training and open during testing and illuminated indirectly (80 lx). The compartment used for training was counter-balanced between genotypes and each mouse was allowed to explore its training compartment for 15 min. At that point it was briefly removed, while the door was opened. Then the mouse was re-introduced to its training compartment and allowed to explore for another 5 min. The time spent (s) in each compartment, the total distance moved and the latency to enter (s) the new compartment was recorded.

The elevated plus maze
The elevated plus maze consisted of two opposing open arms (21 × 8 cm) connected by a central square (8 × 8 cm) to two opposing enclosed arms of the same size with 32 cm high walls. The apparatus was elevated 50 cm above the floor in a large room with an ambient light intensity of 80-100 lx. Behaviour was recorded by a video camera mounted vertically above the maze and analysed using EthoVision. For testing, the animal was placed in the centre of the maze and behaviour was recorded

Voluntary alcohol consumption in the two-bottle free-choice model
A new cohort of mice was tested in the elevated plus maze as described above and went on to have continuous free access to increasing concentrations of an aqueous alcohol solution and water in the home cage. For this purpose, the mice were housed singly.
For the first seven days, access was given to two bottles containing tap water, and intake was examined for possible side preference (which was not detected). Alcohol was then provided in one of the bottles, placed randomly, and faded in (2% and 4%, vol/vol, for 5 days each), after which consumption was measured at 8%, 10%, 12% and 16%. Intake was assessed by weighing the bottles each day between 8 and 10 AM, with occasional omissions, obtaining 5-6 data points at each concentration over the course of 8-9 days. The amount of alcohol ingested was expressed as alcohol preference (% alcohol consumed of total fluid intake) and intake (g/24 h/kg body weight). N = 11 M 4 +/+ and N = 11 M 4 -/mice were used for this experiment. However, one M 4 -/mouse got sick and had to be euthanized during the course of the experiment.

Deprivation-induced alcohol consumption
Approximately four weeks after the introduction of 16% of alcohol, a deprivation phase was introduced where the alcohol bottles were removed for a time period of two weeks. At the end of the deprivation phase, the 16% alcohol bottles were reintroduced, and the alcohol deprivation effect (ADE)-induced alcohol intake was measured daily over five days.

Determination of blood alcohol concentration
After systemic administration of alcohol (− 30 min; 2 g/kg, i.p.) trunk blood alcohol levels were analyzed using an Analox GL6 instrument (Analox Instruments Ltd, 22 Acton Park Estate, The Vale London). The Analox instrument analyzes blood alcohol levels via alcohol oxidase (AOD) catalysed oxidation of alcohol (ethyl alcohol) to acetaldehyde and hydrogen peroxide (H2O2). Five microliters of EDTA plasma per mouse were injected into the Analox system, then the sample was mixed with air and lyophilized enzyme. The amount of H2O2 generated is directly proportional to the alcohol concentration measured over a 5.0% W/V alcohol standard.

Statistical analysis
The two experimental groups (M 4 − /− and M 4 +/+ mice) were compared by independent samples t-tests, when only one dependent variable was considered and by mixed-model ANOVA in cases where two curves were compared (locomotor activity, learning in the Barnes maze, drinking of increasing concentration of alcohol, relapse-like drinking). When appropriate this was followed by post-hoc Student t-tests. Additionally, the relationship between performance on the elevated plus maze and alcohol preference was assessed by Spearman's rank correlation coefficient (ρ). All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) software version 25.0 (IBM Corporation, Armonk, NY, USA). Statistical significance was defined as p < 0.05. All statistical tests were two sided and data are presented as mean ± SEM.

Barnes maze
Mixed-model ANOVA (with training day as the repeated measure) of the distance travelled on the Barnes maze before entering the escape box showed a significant interaction between training day and genotype (F (5105)= 3.004, p = 0.014; Fig. 1B however, the overall effect of genotype was not significant. Both groups improved their performance significantly over time. Mixed-model ANOVA of the number of errors did not reveal a significant effect of genotype either, nor was the interaction term significant (Fig. 1C). Both groups improved their performance significantly over time. During the probe trial the performance of M 4 − /− and M 4 +/+ mice did not differ, neither with respect to distance travelled nor the number of errors.

Novel object exploration
When the novel object was introduced to the testing chamber M 4 − /− mice displayed a significantly and markedly shorter latency to approach it, compared to their M 4 +/+ littermates (t(16) = 3.034, p = 0.008; Fig. 2A). The total time spent exploring the novel object did not differ between groups (Fig. 2B) and neither did the distance travelled in the testing chamber (data not shown).

Novel environment exploration
When the door between the two compartments of the apparatus was opened, M 4 − /− mice displayed a slightly shorter latency to enter the new environment, compared to their M 4 +/+ littermates, however, this was not significant (t(28) = 1.843, p = 0.076; Fig. 2C). The total time spent exploring the novel environment did not differ between groups (Fig. 2D) and neither did the distance travelled in the testing chamber (data not shown).

The elevated plus maze
The   littermates. Total distance moved, velocity and total number of armentries did not differ between the groups (see supplementary materials).

Voluntary alcohol consumption in the two-bottle free-choice model
Initially, also this cohort of mice was tested in the elevated plus maze. The   Fig. 5A). With respect to alcohol intake, the difference between the genotypes was only significant at the 8% alcohol concentration (t(19) = 2.446, p = 0.024, see Fig. 5B). Water intake did not differ significantly between the two genotypes, although the water intake level changed over the period when the mice were offered increasing concentrations of There was no significant difference in total fluid intake between the two genotypes and the total fluid intake level did not change over the period when the mice were offered increasing concentrations of alcohol (see supplementary materials).

Deprivation induced alcohol consumption
After a two week period of alcohol deprivation, the mice' preference for and intake of the 16% alcohol concentration was reassessed.  Fig. 6A). This was not significant with respect to intake (main effect of genotype: F(1,19)= 4.056, p = 0.058, Fig. 6B). There was also a significant main effect of day on drinking behaviour (F(3,57)= 10.031, p = 0.000 for alcohol preference and F(3,57)= 3.996, p = 0.005 for alcohol intake), reflecting the fact that alcohol consumption declined over days. There was no significant interaction between the effects of genotype and day with respect to deprivation induced drinking. Post-hoc Student t-tests showed that M 4 − /− mice displayed a greater preference for 16% alcohol on day 3 after the end of the deprivation period (t(19) = 3.928, p = 0.001), while the difference between genotypes did not reach statistical significance on the other days. (see Fig. 6A).

Correlation between plus maze behaviour and alcohol consumption
We examined the relationship between time spent on the elevated plus maze open arms and alcohol preference as well as intake (mean over the four alcohol concentrations offered in the two-bottle freechoice model) by calculating Spearman's rank correlation coefficient. We found that mice that consumed larger amounts of alcohol and showed greater alcohol preference had spent more time on the elevated plus maze open arms (ρ(19) = 0.586, P = 0.005 for preference and ρ(19) = 0.490, P = 0.024 for intake, see Fig. 7 A and B).
We also examined the relationship between time spent on the elevated plus maze open arms and alcohol preference as well as intake (mean over the five days) during deprivation induced alcohol consumption by calculating Spearman's rank correlation coefficient. We found that, also during this phase of the experiment, mice that consumed larger amounts of alcohol and showed greater alcohol preference had spent more time on the elevated plus maze open arms (ρ(19) = 0.555, P = 0.009 for preference and ρ(19) = 0.487, P = 0.025 for intake, see   4. In a separate cohort of M4 +/+ (N = 11) and M4 − /− (N = 11) mice, that went on to voluntary alcohol consumption in the two-bottle free-choice model, the finding of decreased anxiety-like behavior in the elevated plus maze was replicated, as indicated by increased time spent in both, the entire open arms (A), the outer parts of the open arms (B) as well as decreased time spent in the closed arms (C) and an increased number of head dips (D). Data are shown as mean + /-SEM, * ** p < 0.001, * p < 0.05.

Blood alcohol concentration
Blood alcohol levels were similar between the M 4 − /− mice and the littermate control M 4 +/+ mice measured thirty minutes after systemic administration of 2 g/kg alcohol (see Fig. 8).

Discussion
The findings of the present study indicate that muscarinic acetylcholine M 4 receptors impact alcohol preference and alcohol intake, as well as alcohol deprivation induced relapse-like drinking behaviour by a mechanism involving disinhibited approach behaviour. This is supported by three lines of evidence: First, the M 4 − /− mice displayed disinhibited approach behaviour in a test of novelty preference and a tendency to do so in another, second, the M 4 − /− mice had disinhibited exploratory behaviour on the elevated plus maze, and third this "disinhibited" endophenotype correlated with alcohol preference and alcohol intake, as well as alcohol deprivation induced relapse-like drinking behaviour measured subsequently in the same mice. Thus, the increased propensity to consume drugs of abuse observed previously in M 4 − /− mice may be driven by an altered regulation of approach behaviour [14,40]. or pure129SvEv genetic background, respectively [22,31,44]. We conclude that neither spontaneous, novelty stimulated locomotor activity nor habituation to a novel environment, a non-associative form of learning, is affected in fully backcrossed M 4 − /− mice. Also, M 4 − /− mice did not differ from their littermate M 4 +/+ controls with respect to spatial learning and memory on the Barnes maze, which confirms previous findings with the Morris water-maze in M 4 − /− mice maintained on a pure 129SvEv background [31]. This indicates that the disinhibited phenotype of M 4 − /− mice revealed in the present study is not driven by a deficit in habituation, nor is the higher intake and preference for alcohol related to deficits in learning and memory. When exposed to two different tests of novelty preference, novelobject and novel-environment exploration, the M 4 − /− mice showed "disinhibited" approach behaviour, in as much as their latency to approach the novel stimulus was substantially shorter than that of their M 4 +/+ littermates, while the total time spent exploring the novel stimulus did not differ between genotypes. We suggest that this reflects increased novelty seeking/sensation seeking which has been associated with alcohol use [45]. We thereupon subjected the M 4 − /− mice to the elevated plus maze test, as it has been suggested that time spent in an  environment that mice perceive as more hazardous (here, the open arms of the elevated plus maze) represents a balance between the innate drive to explore novel environments as opposed to anxiety about the dangers of that novel environment [46,47] that this resembles previous findings of a correlation between "disinhibited", sensation seeking, behaviour and the propensity to consume alcohol and other drugs of abuse [48,49] and suggests that the M 4 receptors role in drug seeking behaviour [14,40] might in part be mediated via an effect on the response to novelty. It is well known that exposure to novelty increases midbrain dopamine levels [50] and we have previously shown that M 4 − /− mice exhibit a dopamine "hyperreactive" phenotype [40,43]. The difference between the genotypes with respect to alcohol intake and preference was more pronounced at lower alcohol concentrations and was abolished at the highest (16%) concentration of alcohol. This confirms previous findings with M 4 − /− and M 4 +/+ mice allowed to respond for alcohol on an FR1 schedule of oral operant selfadministration [14] and could reflect an aversive response to the taste of high alcohol concentrations frequently encountered in mice [51,52]. Since alcohol and drug dependences are chronic relapsing disorders, the study of vulnerability to relapse is of particular importance We therefore subjected the mice to a 2-week period of abstinence whereupon we reintroduced the choice to consume alcohol at the 16% concentration. This is known to induce a relapse-like pattern of increased alcohol consumption and has been termed "the alcohol deprivation effect" [53]. And indeed, both genotypes increased their consumption of and preference for the 16% alcohol concentration, but the M 4 − /− mice did so to a higher degree and for longer time than their M 4 +/+ littermates.
This is in line with previous findings in these mice, showing that M 4 -/mice displayed a significantly reduced capacity to extinguish their response for both alcohol and cocaine [14,40]. Our finding of similar blood alcohol concentrations in M 4 − /− and M 4 +/ + mice after systemic administration of a bolus dose makes it unlikely that these findings are due to metabolic differences between the genotypes. A changed pattern of responding during extinction or a changed pattern of relapse-like drinking could reflect an altered learning or memory ability. However, this seems not to be the explanation with respect to the M 4 − /− mice, since they did not show impairment in learning or memory on the Barnes maze, as discussed above. The M 4 − /− mice did show a transient increase in distance travelled during acquisition; however this increase likely mirrors reactivity to the novel environment rather than a change in general learning ability. We therefore suggest that the changed drug-taking behaviour during baseline as well as during relapse-like drinking in the M 4 − /− genotype is associated with disinhibited approach behaviour rather than learningor memory deficits. And indeed, also the alcohol deprivation effect correlated moderately with the time the mice had spent on the open arms of the elevated plus maze many weeks previously. This further supports our contention that the M 4 receptors role in the modulation of various aspects of drug-seeking behaviour might in part be mediated by disinhibited approach behaviour. However, this needs further experimental examination as the causative role of the endophenotype in the drinking pattern cannot be inferred from correlations alone. A limitation to this study is the fact that only male mice were studied, especially since it has been shown that females are more vulnerable to alcohol related harm [54]. However, little is known about potential sex differences in M 4 receptor signaling [15]. Further studies are needed to examine the potential contribution of M 4 receptor signaling in alcohol consumption in females. Another limitation of this study is the use of constitutional knock out mice as this might have given rise to adaptations in other components of the cholinergic system. However, the M 4 − /− mice used here have been reported to display normal levels of M 2 receptors and striatal acetylcholine [22], and double knock-outs of M 2 and M 4 receptor had normal levels of M 1 and M 3 receptors [55]. The M 5 receptor has not been studied in these mice.
In conclusion, we show here for the first time that the muscarinic M 4 receptor modulates drug taking behaviour in a non-operant model of voluntary alcohol consumption and relapse-like drinking. This is in line with previous findings using operant models of drug taking behaviour, for both alcohol and cocaine. We additionally show that this phenotype correlates with disinhibited approach behaviour, providing a potential mechanistic explanation. Taken together, these findings further support the potential usefulness of targeting M 4 receptors in the treatment of substance use disorder.