Anorexia nervosa (AN) is defined as an eating disorder yet up to 80% of individuals also exercise excessively. This is clearly maladaptive and is associated with increased suicide risk, worse symptoms, poorer treatment outcomes, and higher rates of relapse. Initial drivers of dieting and exercise in AN are often motivated by various goals for thinness, for control, and/or for perceived “health” – that is, these behaviours are classically “goal-directed”. At some point then, in some individuals, a shift from goal-directed to compulsive behaviour occurs and this process could be targeted with early intervention strategies, if it could be reliably identified.
We used the activity-based anorexia (ABA) model in combination with chemogenetic, fiber photometric and operant learning paradigms to understand the role of the dorsal striatum in the development of compulsive exercise in the context of pathological weight loss. Female Sprague-Dawley rats underwent stereotaxic surgery to deliver excitatory designer receptors exclusively activated by designer drugs (i.e., Gq-DREADDs) or control virus (i.e., DREADDs negative), into the dorsomedial striatum (DMS) (n=9/group). Rats were then exposed to ABA conditions and were assessed for goal-directed learning using an outcome devaluation task. In a subgroup of rats (n=4), dual-colour fiber photometry recordings of dopamine (DA) and acetylcholine (ACh) release in the DMS were performed during instrumental learning before and after exposure to the ABA paradigm.
We show that DMS activation during early stage ABA suppresses running activity in comparison to control rats. Contrastingly, DMS activation had the opposite effect on running at late stage ABA with an exacerbation of hyperactivity, suggesting that the DMS is differentially engaged as wheel running becomes compulsive. In terms of goal-directed action, exposure to ABA conditions intensified responding during instrumental learning on the operant task and impaired discrimination following outcome devaluation, even after weight recovery, suggesting a persistent effect of ABA on compulsive action selection. Additionally, fiber photometry revealed that DA and ACh release dynamics in the DMS were anti-correlated during instrumental learning, with peaks and troughs of DA aligning with reward anticipation and receipt, respectively. Of note, the peak of ACh release at reward receipt was specifically blunted following exposure to ABA conditions.
Taken together, these studies provide the first direct evidence that the dorsal striatum is involved in the development of food-restriction evoked hyperactivity in the ABA model and point toward a mechanism involving the interaction between DA and ACh responsible for the development of compulsive exercise.