Instrumental learning in a mouse model for obsessive-compulsive disorder: Impaired habit formation in Sapap3 mutants

Abstract

It has been hypothesized that maladaptive habit formation contributes to compulsivity in psychiatric disorders such as obsessive-compulsive disorder (OCD). Here, we used an established animal model of OCD, Sapap3 knockout mice (SAPAP3^−/−), to investigate the balance of goal-directed and habitual behavior in compulsive individuals and if altered habit formation is associated with compulsive-like behavior.

We subjected 24 SAPAP3^−/− and 24 wildtype littermates (WT) to two different schedules of reinforcement in a within-subjects design: a random-ratio (RR) schedule to promote goal-directedness, and a random-interval (RI) schedule, known to facilitate habitual responding. SAPAP3^−/− acquired responding under both schedules, but showed lower response rates and fewer attempts to collect food pellets than WT, indicative of altered reward processing. As expected, WT were sensitive to sensory-specific satiety (outcome devaluation) following RR training, but not RI training, demonstrating schedule-specific acquisition of goal-directed and habitual responding, respectively. In contrast, SAPAP3^−/− were sensitive to outcome devaluation after both RR and RI training, suggesting decreased engagement of a habitual response strategy. No linear relation was observed between increased grooming and behavior during the outcome devaluation test in SAPAP3^−/−.

Together, our findings demonstrate altered reward processing and impaired habit learning in SAPAP3^−/−. We report a diminished propensity to form habits in these mice, which albeit inconsistent with the predominant idea of excessive habit formation in OCD, nonetheless points at dysregulation of behavioral automation in the context of compulsivity. Thus, the habit hypothesis of compulsivity should be updated to state that an imbalance of habitual and goal-directed responding in either direction can contribute to the development of compulsive behavior.

Introduction

All behavior is eventually a unique compromise of actions based on intentions, knowledge, well-established routines, and environmental necessities. Behavior is guided by divergent response styles referred to as goal-directed or habitual response strategies. Employment of a goal-directed strategy is typically assumed when behavioral responses are sensitive to outcome value and to situational action-outcome contingencies (Balleine & O’Doherty, 2010). Habitual responding, on the other hand, is conceptualized as automated, inflexible behavior that is predominantly evoked by stimulus-response (S-R) associations with low levels of experienced contingency between an action and its consequences (Dickinson, 1985). While goal-directed behavior requires cognitive deliberation, habitual actions rely on repetition and generally demand less attention. However, optimal behavioral adaptation depends on flexible shifting between habitual and goal-directed response strategies.

It has been hypothesized that overreliance on habitual strategies could underlie compulsive behavior observed in several psychiatric disorders, including obsessive-compulsive disorder (OCD), as habits and compulsions share some similarity in the composition of their tightly structured actions (Fineberg et al., 2014, Gillan et al., 2011, Gremel and Costa, 2013). So far, only a small number of studies investigated appetitive habit formation and compulsivity (Gillan et al., 2011, Snorrason et al., 2016, Voon et al., 2015). These studies, showed a link between increased habitual responding and compulsive symptoms, but are limited by practical issues (pharmacological treatment of OCD patients, non-clinical study subjects) and by uncertainty whether the appetitive tasks, originally developed in animal studies, are valid paradigms to assess habits in human patients (Ceceli and Tricomi, 2018, Foerde, 2018, Watson and de Wit, 2018).

The relation between habit formation and compulsivity is understood insufficiently. Therefore, we set out to test habit formation in Sapap3 knockout mice (SAPAP3^−/−), a validated genetic mouse model of OCD, that exhibits compulsive-like self-grooming (Burguiere et al., 2013, Pinhal et al., 2018, van den Boom et al., 2019, Welch et al., 2007, Yang and Lu, 2011). We tested these mice in an outcome devaluation paradigm, the most prominent method for exploring appetitive habit formation (Adams and Dickinson, 1981, Gremel and Costa, 2013, Gremel and Costa, 2013, Hilário et al., 2007). The outcome devaluation paradigm relies on the observation that different reinforcement schedules promote acquisition and selection of either goal-directed or habitual response strategies. These strategies can be discerned in an outcome devaluation test, in which the value of the reinforcer is diminished and sensitivity to this devaluation is probed in an extinction test (Gremel & Costa, 2013). Random ratio schedules (RR) bias towards goal-directed responses because they provide high action-outcome contingency, sensitivity towards changes in outcome value, and flexible adaptation of behavior DeRusso et al., 2010, Moore et al., 2009. Random-interval schedules (RI) on the other hand, are characterized by a weak correlation between actions and outcomes, causing an insensitivity to outcome devaluation, denominated as habitual response strategy (Dickinson, 1985).

Previous research with mice trained in the outcome devaluation paradigm showed that goal-directed and habitual strategies are mediated by cortico-striatal brain circuits (Gremel & Costa, 2013). These cortico-striatal circuits are compromised in OCD patients (Graybiel and Rauch, 2000, Saxena et al., 1998) and in SAPAP3^−/− Burguiere et al., 2013, Pinhal et al., 2018, Welch et al., 2007, suggesting that this mouse model is well-suited to test hypotheses regarding the relationship between aberrant habit formation and compulsivity. Following the hypothesis that obsessions and compulsions in OCD patients are based on overactive neural circuits involved in habit formation (Graybiel & Rauch, 2000), we tested if SAPAP3^−/− exhibit an imbalance in acquisition and expression of habitual and goal-directed action strategies. In addition, we explored if a positive relation exists between compulsive grooming (duration and/or frequency) during the outcome devaluation test and the imbalance between habitual and goal-directed response strategies. Such a relation would suggest a common neural substrate for regulation of habitual, goal-directed, and compulsive behavior.