Translational approaches to obsessive-compulsive disorder: from animal models to clinical treatment

Abstract

Obsessive-compulsive disorder (OCD) is characterized by obsessions (intrusive thoughts) and compulsions (repetitive ritualistic behaviours) leading to functional impairment. Accumulating evidence links these conditions with underlying dysregulation of fronto-striatal circuitry and monoamine systems. These abnormalities represent key targets for existing and novel treatment interventions. However, the brain bases of these conditions and treatment mechanisms are still not fully elucidated. Animal models simulating the behavioural and clinical manifestations of the disorder show great potential for augmenting our understanding of the pathophysiology and treatment of OCD. This paper provides an overview of what is known about OCD from several perspectives. We begin by describing the clinical features of OCD and the criteria used to assess the validity of animal models of symptomatology; namely, face validity (phenomenological similarity between inducing conditions and specific symptoms of the human phenomenon), predictive validity (similarity in response to treatment) and construct validity (similarity in underlying physiological or psychological mechanisms). We then survey animal models of OC spectrum conditions within this framework, focusing on (i) ethological models; (ii) genetic and pharmacological models; and (iii) neurobehavioural models. We also discuss their advantages and shortcomings in relation to their capacity to identify potentially efficacious new compounds. It is of interest that there has been rather little evidence of ‘false alarms’ for therapeutic drug effects in OCD models which actually fail in the clinic. While it is more difficult to model obsessive cognition than compulsive behaviour in experimental animals, it is feasible to infer cognitive inflexibility in certain animal paradigms. Finally, key future neurobiological and treatment research areas are highlighted.

LINKED ARTICLES

This article is part of a themed issue on Translational Neuropharmacology. To view the other articles in this issue visit http://dx.doi.org/10.1111/bph.2011.164.issue-4

Introduction

Obsessive-compulsive disorder (OCD) is a common neuropsychiatric disorder, affecting roughly 2% of the adult population worldwide (Zohar, 1999), and responsible for considerable psychosocial morbidity (Bolton et al., 1995; Hanna, 1995; Hollander et al., 2010). The illness is characterized by obsessions (recurrent unpleasant thoughts) and/or compulsions (repetitive unwanted behaviours that the individual is unable to resist) (APA, 2000). Obsessions and compulsions are experienced as unpleasantly insistent and intrusive. OCD patients usually report a wide range of obsessive-compulsive (O-C) symptoms in a number of overlapping behavioural domains, some of which naturally cluster together, which may also vary over time (Mataix-Cols et al., 2005; Katerberg et al., 2010). The disorder usually emerges in childhood or early adulthood (earlier in males) and runs a lifelong, fluctuating course (Skoog and Skoog, 1999). Existing treatment with drugs and cognitive behavioural forms of psychotherapy are usually only partially effective and roughly one third of cases do not achieve an adequate clinical response (Fineberg and Gale, 2005).

OCD is commonly complicated by the co-occurrence of other ‘comorbid’ mental disorders, notably depression which supervenes in up to two-thirds of clinical cases (Fineberg et al., 2005a; Peris et al., 2010). Other disorders that are characterized by O-C behaviours, such as O-C personality disorder, body dysmorphic disorder, trichotillomania (repetitive hair-pulling) and Tourette's Syndrome (Hollander, 2008; Hollander et al., 2010; Phillips et al., 2010) also tend to cluster with OCD, occurring either within the same individual or within close family members and implying the possibility of shared pathophysiological mechanisms. Indeed, this so-called ‘obsessive-compulsive spectrum disorders (OCSDs)’ is characterized by considerable phenotypic heterogeneity and overlap. Moreover, there is evidence of overlap in the treatment-response across some disorders. Thus, the complexity and clinical morbidity associated with OCD is high (Phillips et al., 2010) and new and better treatments are sorely needed.

Multiple layers of evidence link OCD with dysregulation of fronto-striatal neuro-circuitry and associated monoamine systems. However, attribution of cause and effect may easily be confounded by the multiplicity of associated symptom domains that occur within such a complex mental disorder. Translational research investigates underlying mechanisms, and may therefore be more able to pinpoint neural contributions driving specific aspects of mental disorders. Convergent evidence from translational studies, largely in human subjects, suggests that a tendency towards behavioural disinhibition (Chamberlain et al., 2005), presumably resulting from failures in ‘top–down’ cortical control of fronto-striatal neural circuits, or alternatively from over-activity within striatal ‘habit’ circuitry, may underpin aspects of behavioural compulsivity that are found in OCD and related disorders (Fineberg et al., 2010; Padhi et al., 2010). These abnormalities represent key targets for existing and novel treatment interventions.

As our understanding of the behavioural, cognitive, neural and genetic substrates of OCD and related OCSDs advances, the search and evaluation of appropriate animal models that can be used to test out the efficacy of potential new treatments and their mechanisms of action in early-phase studies becomes increasingly relevant. To date, no universally accepted animal model for OCD exists. However, models of repetitive habits and inhibitory control problems represent possible equivalents of aspects of compulsive behaviour in OCD patients, and may offer potential for enhancing our understanding of the pathophysiology and treatment of OCD and spectrum disorders. Such models need to be validated, in terms of being seen to accurately represent the human condition, and to have the capability to be reliably generated. This review aims critically to evaluate existing animal models of relevance for OCD and related disorders against these criteria. Several studies have provided important information regarding the neural and neurochemical substrates of OCD that could be used to endorse any such model. In addition, the availability of somewhat effective pharmacological treatments, such as clomipramine and the selective serotonin reuptake inhibitors (SSRIs) (Fineberg and Gale, 2005), provides an ancillary criterion for model validation, although their precise mode of action in OCD remains incompletely characterized. In this paper, we focus on (i) ethological models; (ii) genetic and pharmacological models; and (iii) neurobehavioural models. We discuss their advantages and shortcomings with examples of compounds that are of clinical benefit. It is of interest that there has been rather little evidence of ‘false alarms’ for therapeutic drug effects in OCD models which actually fail in the clinic. Those cases that have arisen may result from an incomplete pharmacological characterization of the model.

Validation criteria for animal models

No single animal model can account for the entire psychiatric syndrome it purports to represent. Therefore, the validation criteria that each model is expected to fulfil in order to demonstrate its validity are, for practical purposes, largely determined by the objective of the model and its intended use (McKinney and Bunney, 1969; Matthysse, 1986; Willner, 1991; Geyer and Markou, 1995; Mckinney, 2000; Geyer and Markou, 2002). According to the well-known classification by Willner 1984), refined by Geyer and Markou (Geyer and Markou, 1995, Geyer and Markou, 2002), the criteria for assessing animal models are grouped into those used to establish face validity (phenomenological similarity between inducing conditions and specific symptoms of the human phenomenon), predictive validity (the extent to which an animal model allows accurate predictions about the human phenomenon based on the performance of the model, for example, similarity in response to pharmacological or behavioural treatment) and construct validity (similarity in underlying physiological or psychological mechanisms). Reliability, on the other hand, requires that the behavioural outputs of the model are robust and reproducible between laboratories. Geyer and Markou recommend that the evaluation of animal models in neurobiological research should principally rely on reliability and predictive validity, with face and construct validity, which tend to be, respectively, highly subjective or dependent upon assumptions and inferences, reserved as secondary criteria. Thus, in order to have the capacity to predict the response of a mental disorder such as OCD to a new pharmacological treatment, a proposed animal model needs to produce a specific, measurable behaviour reliably, which is pharmacologically analogous with the clinical disorder. On the other hand, predictive validity can be unduly restrictive and lead to the generation of ‘me-too’ compounds (as is the case for antidepressants) as opposed to the enhanced understanding and capacity for innovation which can be attained via construct validity.

Clinical profile of OCD

The hallmark symptoms of OCD involve the unwanted and needless repetition of thoughts and actions. Based on factor analysis, OCD symptoms have been split into four phenotypic categories (Leckman et al., 1997; Summerfeldt et al., 1999; Cavallini et al., 2002): (i) aggressive sexual and religious obsessions with checking compulsions; (ii) symmetry obsessions with compulsions of classification, sorting and repetitiveness; (iii) obsessions of contamination with cleaning compulsions; and (iv) hoarding. Many of these symptoms resemble normal childhood behaviour that disappears during development. Such behaviour appears habit-driven and may be evolutionarily conserved, inasmuch as it is expressed across species (e.g. hoarding, grooming, sorting) and may also fulfil an adaptive role under conditions of privation (Leckman et al., 2010). The maturation of the prefrontal cortex and its subcortical connections may result in the natural suppression of these habitual acts, in favour of more adaptive, goal-directed behaviours (Gillan et al., 2010), and thus, by inference, OCD may arise at least in part as a result of a relative failure of this ‘top-down’ suppression. There is some evidence that in OCD patients, these symptom clusters differ in terms of constituent temporal and spatial dimensions of the behaviour (Zor et al., 2010), treatment response (Black et al., 1998; Mataix-Cols et al., 1999; Winsberg et al., 1999, Mataix-Cols et al., 2002), co-morbidity with other psychiatric disorders (Samuels et al., 2002), imaging profile (Mataix-Cols et al., 2004; van den Heuvel et al., 2005) and genetic predisposition (Leckman et al., 2003; 2010; Katerberg et al., 2010). Thus, in OCD, the heterogeneity of observed symptomatology may be underpinned by subtly differing pathophysiological mechanisms.

Traditional learning theory applied to OCD proposes that an increase in anxiety occurs when an obsessive thought (e.g. of committing an aggressive act) is experienced and that this anxiety subsequently drives the urge to perform a neutralizing ritual (e.g. checking for harm) that has a negatively reinforcing effect, leading to a vicious cycle of obsession and compulsion (Drummond and Fineberg, 2007). In line with this theory, OCD has been categorized as an Anxiety Disorder in the major diagnostic classificatory systems such as the Diagnostic and Statistical Manual IV (APA, 2000), implying a key role for anxiety dysregulation in its aetiology. However, the role of anxiety in the pathophysiology of OCD has always been controversial, and its nosological status is currently under review (Hollander, 2008; Hollander et al., 2008; Phillips et al., 2010). Whereas OCD symptoms do generally appear to worsen under psychosocial stress, the expression of psychological and physical anxiety symptoms as part and parcel of the syndrome is unreliable and emotions other than anxiety, such as horror or disgust, may be more prominent (Sprengelmeyer et al., 1997). Depressive symptoms are also common in OCD. The depressive syndrome is associated with relatively increased worry and rumination and less vegetative disturbance compared with major depressive disorder (Fineberg et al., 2005a). Moreover, the depressive symptoms respond to pharmacological treatment in tandem with the OCD (Hoehn-Saric et al., 2000), suggesting that they are integral to this disorder.

Individuals with OCD also commonly present with movement disorders, most notably tics, that can vary in severity from the relatively rare, explosive actions associated with Tourette's syndrome to commoner mild, barely perceptible facial twitches that are focused around the eyes, nose and mouth and that appear similar to ‘neurological soft signs’ (Hranov and Fineberg, 2010). The presence of co-morbid tics in children and adolescents with OCD predicts a positive outcome in adulthood, whereas primary hoarding symptoms are associated with persistent OCD (Bloch et al., 2009). Indeed, the OCSDs may, to a greater or lesser extent, represent ‘formes-frustes’ of OCD. Of these disorders, body dysmorphic disorder (BDD), most closely resembles OCD symptomatically (Phillips et al., 2010). In BDD, an obsessional preoccupation with irregularities of bodily appearance leads to compulsive checking and remediative acts. In contrast, in trichotillomania, rumination is less prominent and the repetitive hair-pulling may produce positively reinforcing, soothing effects (Chamberlain et al., 2007c; 2009).

Neurobiological substrates of OCD

The essential features of OCD and related spectrum disorders most readily captured by animal models are the maladaptive and perseverative behavioural and cognitive outputs (Boulougouris et al., 2009). For example, the repetitive rituals in OCD, or recurrent hair-pulling in trichotillomania. These behaviours are thought to be mediated by dysfunctional nodes within the fronto-striatal circuitry, possibly mediated by glutamate neurotransmission, under modulation by altered dopaminergic or serotoninergic influences. In OCD, neuroimaging studies have implicated in particular the orbitofrontal cortex (OFC) and the caudate nucleus, and cingulotomy has had a limited therapeutic success (Baxter, 1999). Moreover, there may be grounds for considering OCD spectrum disorders as reflecting impaired functioning of several distinct fronto-striatal ‘loops’ (Graybiel, 1997; Jog et al., 1999; Graybiel and Rauch, 2000; Chamberlain et al., 2005; Nakao et al., 2005; Whiteside et al., 2006; Menzies et al., 2008).

Pharmacological profile of OCD

OCD responds to a characteristically narrow range of pharmacological treatments. According to a considerable body of evidence from randomized controlled clinical trials, drugs with potent inhibitory effects on the synaptic reuptake of serotonin, such as the non-selective tricyclic clomipramine and the more highly selective serotonin reuptake inhibitors (SSRIs), are reasonably effective in approximately two thirds of cases. The treatment effect develops slowly and gradually over weeks and months, and higher SSRI doses, and extended treatment duration appear to produce greater effect sizes (Fineberg and Gale, 2005). Importantly, treatments found to be effective in other anxiety and affective disorders, such as antidepressants that act via noradrenergic mechanisms, benzodiazepines and mood stabilizers are not effective in OCD. There is very little evidence-based treatment available for SSRI-resistant illness. Positive results from a small number of randomized controlled trials show limited extra benefit from adjunctive first and second generation antipsychotics taken in low or modest doses (Fineberg et al., 2005b; 2006a), and lesser evidence from a randomized trial (Ninan et al., 2006) supports increasing the dose of SSRI above formulary limits (Fineberg et al., 2006b; Pampaloni et al., 2010). Tic-related OCD may respond less well to SSRI monotherapy and prefentially to adjunctive antipsychotic (Fineberg et al., 2006c). Antipsychotics represent first-line treatment for Tourette's Syndrome and it is therefore, interesting that their combination with SSRIs shows greater efficacy in tic-related OCD (Bloch et al., 2006). Compulsions associated with autistic disorders may also respond to low-dose SSRI and to antipsychotics (Kolevzon et al., 2006). Trichotillomania may respond to SRIs and to antipsychotics, though confirmation in well-powered controlled studies is required (Chamberlain et al., 2007c). Trichotillomania has also been shown to respond to treatment with the glutamate modulator N-acetyl cysteine in a randomized placebo-controlled trial (Grant et al., 2009). Promising results from experimental open-label treatment of small numbers of OCD cases with alternative adjunctive glutamatergic compounds, such as riluzole and memantine, remain to be validated in controlled clinical trials.

The pharmacological mechanisms underpinning the anti-obsessional treatment response remain poorly understood. The superior efficacy of SSRI in doses higher than those needed to completely inhibit the serotonin transporter suggest that other receptor mechanisms in addition to increased intra-synaptic serotonin concentrations may be relevant. The extended development of the treatment effect implies the recruitment of adaptive processes such as neurotransmitter receptor modulation, perhaps focussed at neurones within relevant neuro-circuitry such as the OFC and/or caudate nucleus, that may take days or weeks to develop (Blier and de Montigny, 1998). Evidence from pharmacological challenge studies in which the non-selective serotonin receptor agonist m-chlorophenylpiperazine (mCPP) induced OCD symptoms that were blocked by pre-treatment with clomipramine or SSRI (Zohar et al., 1988; Hollander et al., 1991), implicate serotonin receptors in the pathophysiology and the treatment response. There is evidence from ligand-based positron emission tomography studies of striatal and cortical alterations in 5-HT2A receptors and in striatal D2 receptors in OCD (Denys et al., 2004b; Westenberg et al., 2007). Interactions between serotonin and dopamine systems have also been inferred. In rats, co-administration of quetiapine with fluvoxamine, a combination with established efficacy in OCD (Fineberg et al., 2006c), robustly increased dopamine release in the prefrontal cortex (Denys et al., 2004a) and this effect has been suggested to play a possible role in the treatment response.

Lucey et al. (1993) suggested the involvement of the cholinergic system in OCD; compared with normal subjects, OCD patients exhibited an increased growth hormone response after pyridostigmine administration, providing evidence of cholinergic hypersensitivity (Lucey et al., 1993). Glutamate has also been implicated in OCD symptomatology; administration of substances that act upon glutaminergic receptors caused exacerbation of compulsive behaviour in a genetic model of OCD and Tourette syndrome (Mcgrath et al., 2000). Moreover, there have been attempts to treat OCD using d-cycloserine, which is active at the glycine site of the NMDA receptor. In addition, during the last years neuropeptides and gene steroids have been implicated in OCD pathophysiology (Lochner et al., 2004a,b). It should nevertheless be noted that, although all these data are limited, they do not contradict the prevailing theory of OCD pathogenesis as resulting from a dysmodulation of orbitofrontal-striatal circuitry via serotoninergic and dopaminergic mechanisms.

Validating animal models for OCD

Animal models of OCD spectrum disorders have generally fulfilled the criteria of face validity, but have sometimes been based on psychological theorizing, thus attempting the deeper level of modelling ‘construct validity’. In a seminal experiment, Solomon et al. (1953) paired electric shocks with a light to condition dogs to become anxious and escape when the light bulb was switched on (Solomon et al., 1953). This escape behaviour was conceptualized as being close to a compulsive ritual in that it led to immediate relief. By preventing escape (‘response prevention’) when the light bulb was turned on, Solomon subsequently induced extinction of the conditioned anxiety and of the compulsive urge to escape. In translating aspects of this model to the human condition, early behaviourists such as Meyer (1966) developed exposure and response prevention as an effective form of psychotherapy for OCD (Meyer, 1966). In this approach, Meyer (1966) exposed patients to anxiety-evoking stimuli and constant staff supervision to prevent compulsions. Predictive validity can also be employed to some extent in OCD models, given the known, but largely unexplained, efficacy of the SSRIs and other less widely evaluated candidate treatments such as D1 receptor antagonists and specific 5-HT receptor agents.

Animal models of OCD

Existing animal models for OCD may be grouped into naturally occurring ethological models and laboratory-based genetic, pharmacological and neurobehavioural models.

Ethological models (Table 1)

Table 1.

Animal models of Obsessive-Compulsive Disorder (OCD)

	Model	Modeled behaviour (Face validity)	Neuroanatomical/neurochemical substrate (construct validity)	Predictive validity
Ethological models	Tail-chasing, acral lick dermatitis in dogs, psychogenic alopecia (hair pulling) in cats, feather picking in birds, cribbing in horses, schedule induced polydipsia, food-restriction-induced hyperactivity	++	?	The effects of selective serotonin reuptake inhibitors have been tested and compared with the effects of drugs ineffective in OCD e.g. remediating effects of clomipramine on canine lick dermatitis
Genetic models	Hoxb8 mutant mice	++ (trichotilomania)	++ (gene expression in areas implicated in OCD)	?
	D1CT-7 mice	++ (OCD/TS)	++ (transgene expression in neural systems hyperactive in human OCD)	?
	DAT KD mice	+/− (mimics behaviours relevant to other disorders as well)	+ (dopaminergic involvement)	?
			++ (basal ganglia are implicated in grooming and OCD)
	5-HT2c KO mice	+/− (mimics behaviours relevant to other disorders as well)	+ (5-HT2c receptors involvement in OCD)	?
			+ (functional abnormalities in neural substrates of OCD)
	Slitrk5−/− mice	++ (OCD/trichotillomania)	++ (overactivation of brain areas implicated in OCD)	++ (response to fluoxetine)
			+ (alterations in glutamate receptor composition)
	Sapap3 gene mice	++ (trichotillomania and OCD)	++ (gene expression in areas implicated to OCD)	++ (response to fluoxetine)
Pharmacological models	Quinpirole-induced compulsive-checking	+++ (OCD)	+ (dopaminergic involvement in OCD)	+ (response to clomipramine.)
	8-OHDPAT-induced spontaneous alternation	+/− (motor perseveration apparent in other disorders as well)	+ (5-HT1a receptor involvement in OCD)	++ (response to fluoxetine and clomipramine, but not desipramine)
	mCPP-induced directional persistence in Reinforced Spatial Alternation	++ (OCD)	+ (5-HT2c receptor involvement in OCD)	++ (response to fluoxetine, but not to diazepam or desipramine)
Behavioural models	Barbering	+++ (trichotillomania)	++ (spontaneous development)	?
	Marble burying	+ (OCD)	?	+++ (response to SSRIs)
				− (response to anxiolytics)
				+ (no response to desipramine)
	Signal attenuation	++ (OCD)	+ (deficient psychological process implicated in OCD)	+++ (response to fluoxetine, but not diazepam, desipramine or haloperidol)
			++ (involvement of brain areas implicated in OCD)
Other possible behavioural	Reversal learning	+ (OCD)	++ (brain areas implicated in OCD)	?
			+++ (5-HT involvement in OCD)
	Attentional set-shifting	−/+	+ (brain areas implicated in OCD)	?
			− (no involvement of 5-HT)
	Extinction	−/+	?	?
	Habit-learning	−/+	+ (brain areas implicated in OCD)	?
	Stop-signal reaction time task (SSRT)	−/+	++ (brain areas implicated in OCD)	?

Animal models of Obsessive-Compulsive Disorder (OCD). Each column estimates the extent to which a model meets each criterion (+, ++ or +++, model does well; −, model does badly; ?, there are no relevant data). 5-HT, serotonin; 5-HT2c KO mice, 5-HT2c receptor knockout (KO) mice; 8-OHDPAT, 8-hydroxy-2-(di-ni-popylamino)-tetralin hydrobromide, 5-HT1A agonist; ADHD, Attention deficit/hyperactivity disorder; ALD, Acral lick dermatitis; D1CT mice, transgenic mice expressing a neuropotentiating protein (cholera toxin A1 subunit) within a cortical-limbic subset of dopamine D1-receptor expressing (D1+) neurones; DAT KD mice, dopamine transporter (DAT) knockout (KD) mice, expressing 10% of wild-type DAT levels and exhibit elevated extracellular dopamine concentration; mCPP, meta-chlorophenylpiperazine, non-selective serotonin agonist; OFC, orbitofrontal cortex; SSRIs, selective serotonin reuptake inhibitors; SSRT, Stop-signal reaction time task; TS, Tourette's syndrome.

Ethological models focus on spontaneous persistent behaviours that resemble OCD or more likely trichotillomania. They represent a source of naturalistic stereotypies that may be informative about OCD spectrum disorders (Stein et al., 1994). In general, they have good face validity (in being repetitive and superficially resembling common human compulsions) and some show predictive validity in terms of their response to drug treatment, but low practicality and reliability. Such behaviours are often elicited in veterinary contexts and may be attributed to stressful environments, for example, psychogenic alopecia in cats (Swanepoel et al., 1998), cribbing in horses (Luescher et al., 1998) and repetitive pacing in several species. Other such disorders include tail-chasing (Brown et al., 1987), fur-chewing and acral lick dermatitis (paw licking) in dogs (Rapoport et al., 1992); feather-picking in birds (Grindlinger and Ramsay, 1991); wheel-running and allogrooming (or ‘barbering’, akin to trichotillomania) in mice (Garner et al., 2004a,b). For behaviours that represent natural responses under stress, some degree of construct validity for compulsions is also inferred inasmuch as the compulsive behaviours are performed in states assumed to correspond to anxiety. These include marble-burying in mice (the use of bedding material to bury noxious/harmless objects), which may be induced by basic fear avoidance mechanisms (Ichimaru et al., 1995), displacement behaviour in the face of the thwarting of goal-directed activities including ‘schedule induced polydipsia’ (Robbins and Koob, 1980; Woods et al., 1993) and food-restriction-induced hyperactivity (Altemus et al., 1996). Both stereotypies and schedule-induced polydipsia have been considered as ‘coping responses’ that hypothetically reduce stress, akin to compulsions. This hypothesis, however, has proven difficult to test experimentally and may well not apply to all forms of stereotypy.

Some of these models have tested the effects of SSRIs in comparison to drugs ineffective in OCD (Winslow and Insel, 1991; Rapoport et al., 1992; Woods et al., 1993; Altemus et al., 1996; Nurnberg et al., 1997). For example, the efficacy of clomipramine in OCD and trichotillomania was predicated by observations of its remediating effects on canine lick dermatitis (Swedo et al., 1989; Rapoport et al., 1992). In addition, the SSRIs, fluvoxamine and citalopram, clomipramine and a selective, non-peptidergic NK(1) receptor antagonist (RP67580) were all observed to block marble-burying in mice (Millan et al., 2002; Wolinsky et al., 2006). Although the biological bases of this behaviour remain unclear, these observations hint that NK(1) receptors may be implicated in compulsive disorders. However, it is possible that these models relate more to anxiety and the behavioural response to stress than to OCD per se. Agomelatine, a mixed melatonin agonist and 5-HT_2C antagonist with established antidepressant and anxiolytic effects in clinical populations, has also been shown to reduce stress-induced marble burying in mice (Hamon et al., 2005), suggesting potential efficacy in OCD that is in need of validation in a clinical population.

Genetic models of OCD (Table 1)

Several studies indicate that the pathogenesis of OCD has a genetic component. Three genome-wide linkage studies of OCD have so far been published (Hanna et al., 2002; 2007; Shugart et al., 2006).

So far, only single-nucleotide polymorphisms in the glutamate transporter gene SLC1A1, on chromosome 9p24, have been found to be associated with OCD. This transporter is widely expressed in neurones and also involved in cysteine transport. Moreoever, sequence variations in SLC1A1 are also associated with susceptibility to atypical antipsychotic-induced O-C symptoms (Kwon et al., 2009). According to other association studies, several candidate genes have been found as possible risk factors for OCD, including those that involve the serotonergic (e.g. serotonin transporter 5-HTTLPR, 5-HT_2A receptor, 5-HT_1D receptor (Zohar et al., 2004), TPH2 (Mossner et al., 2006), dopaminergic (e.g. DRD4, COMT) (Pooley et al., 2007) and glutamatergic system (e.g. SLC1A1) (Wendland et al., 2009). One murine model of autism, in which a genetically engineered 6.3 Mb duplication of the human 15q11-13 chromosome leads to increased anxiety and impaired reversal learning associated with increased transmission at 5-HT_2C receptors, appears to have some relevance to compulsive behaviour (Nakatani et al., 2009).

Animal models for OCD were not created on the basis of a known mutation in humans that was found to be related to OCD. These models rely on genetic manipulations on mice and are largely based on face validity and behavioural similarity, that is, the behaviour of genetically modified mice resembles in some specific respects that of OCD patients. Some of these responses show clear superficial parallels to the compulsive grooming that characterizes trichotillomania, and perhaps more obliquely to the more elaborate rituals of OCD, including those related to cleaning and checking. It seems likely that these examples of stereotyped behaviour are mediated by basal ganglia, given the known role of the caudate-putamen in stereotyped behaviour produced by psychomotor stimulant drugs such as amphetamine (Creese and Iversen, 1975) and in normal grooming sequences (Aldridge and Berridge, 1998).

Greer and Capecchi (2002) reported that mice with mutations of the Hoxb8 gene (expressed in the OFC, the striatum and the limbic system, all of which are implicated in OCD pathophysiology) groomed excessively to the point of hair removal and skin lesions compared with their control counterparts (Greer and Capecchi, 2002). These mutant mice also excessively groom their wild-type cage mates, suggesting that the excessive grooming behaviour is centrally generated. Evidence suggests that in the mouse brain, the only detectable cells derived from Hoxb8 cell lineage are microglia (Chen et al., 2010), and the far-reaching role of such microglia in the regulation of the brain's immune activity is becoming increasingly apparent. Normal bone marrow transplantation into lethally irradiated Hoxb8 mutant mice rescues the excessive grooming behaviour. Thus, pathological grooming behaviour observed in Hoxb8 mutant mice may originate from defective microglia within OCD-relevant neurocircuitry, and the HoxB8 model offers a paradigm for exploration – at the molecular genetic and cellular levels – of the mechanism by which perturbation of immune function may lead to O-C symptomatology. The HoxB8 model is promising in that excessive grooming has superficial similarity to the symptomatology of OC spectrum disorders and may involve neural systems similar to the ones implicated in OCD, yet, it lacks predictive validity in terms of drug treatment. Immunological dysfunctions such as these are becoming widely linked to many psychiatric disorders including OCD and autism (Leonard and Swedo, 2001; Ashwood et al., 2010).

In the D1CT-7 mouse model, genetic manipulation of dopamine (D1) receptor function using a neuro-potentiating cholera toxin, expressed in the pyriform cortex and amygdala, produces perseveration and repetitive jumping. These effects are probably ultimately mediated via striatal mechanisms (Campbell et al., 1999a,b,c). The repetitive jumping behaviour may be exacerbated by the administration of yohimbine, an anxiogenic drug with antagonist actions at alpha-2 adrenergic receptors inter alia (Mcgrath et al., 1999). Although the D1CT-7 model is promising in the sense that some of the behaviours exhibited by the mice bear similarities to those observed in OCD, and again implicates common neural systems as in OCD, the pharmacological isomorphism of the model with OCD is necessary for strengthening the model's relevance to OCD. To date, only the effects of dopaminergic (i.e. cocaine, and D1 and D2 antagonists) and noradrenergic (clonidine) agents have been assessed (Campbell et al., 1999b; Nordstrom and Burton, 2002), and thus again, predictive validity is absent.

Knock-down of the dopamine transporter in mice (DAT KD mice) produces ‘sequential super-stereotypy’ with the perseverative performance of complex chains of grooming behaviour (Berridge et al., 2005). Likewise, a knock-down of the 5-HT_2C receptor leads to perseverative ‘head-dipping’ or the excessively orderly chewing of screen material (Chou-green et al., 2003), a compulsive behaviour which – together with stereotypic locomotion and excessive self-aggressive grooming – has also been shown in rats following chronic lesions of the median raphé nucleus (Hoshino et al., 2004). However, the data obtained from this genetic preparation do not match with other data investigating the same receptor, possibly because of unspecified compensatory processes that may develop in the transgenic preparation, as recent pharmacological data indicate the opposite finding that 5-HT_2C receptor activation is associated with increased compulsivity (Tsaltas et al., 2005; Boulougouris et al., 2008).

The Slitrk family of transmembrane proteins are highly expressed in the nervous system (Zuchner et al., 2006). The function of Slitrks during development of the nervous system has yet to be clearly defined, though they are thought to regulate axon outgrowth during development. A recent study in Slitrk5^−/− mice demonstrated that loss of the neurone-specific transmembrane protein, SLIT and NTRK-like protein-5 (Slitrk5), leads to OCD-like behaviours, which manifests as excessive self-grooming and increased anxiety-like behaviour, and is alleviated by the SSRI, fluoxetine. The knockout mice also show selective over-activation of the OFC, abnormalities in striatal anatomy and cell morphology and alterations in glutamate receptor composition, which contribute to deficient corticostriatal neurotransmission. Slitrk5 may be an essential molecule for normally functioning corticostriatal synapses and its knock-down may provide a new mouse model of OCD-like behaviours with a degree of predictive, as well as construct and face validity (Shmelkov et al., 2010).

The Sapap 3 gene is responsible for synaptic scaffolding and migration of glutamate nerve cells from the caudate to the orbitofrontal cortex. A recent study (Welch et al., 2007) found that mice with a deletion of the Sapap3 gene groomed themselves excessively, exhibited increased anxiety-like behaviour, and had corticostriatal synaptic defects, all of which were preventable with lentiviral-mediated expression of Sapap3 in the striatum. The behavioural abnormalities were also reversible with fluoxetine. Further experimentation showed that variation within the human Sapap3 gene was associated with grooming disorders (pathologic nail biting, pathologic skin picking and/or trichotillomania), suggesting that Sapap3 is a promising functional candidate gene for human grooming disorders (Bienvenu et al., 2009).

In summary, although the genetic models of OCD offer good face validity and behavioural similarities with human OCD, there is often a lack of evidence on the pharmacological isomorphism of these models with OCD. However, if these models were to fulfil the criterion of predictive validity, they may provide great insight for our understanding of the neurochemical mechanisms of OCD.

Pharmacological animal models of OCD

Pharmacological models tend to be based on dopamine-induced stereotypy produced by high doses of stimulant drugs such as d-amphetamine and cocaine (Lyon and Robbins, 1975) and appear superficially appealing as models for OCD. Stereotypies in rodents typically consist of gnawing and licking with repetitive sideways movements of the head, which may represent vestiges of orienting behaviour. They can be elaborated in such models to include grooming (including allogrooming) (Sahakian and Robbins, 1975) and perseverative operant behaviour in which rats may continue to work for food they do not eat (Robbins and Sahakian, 1983). However, there is sufficient evidence to question whether stimulant induced stereotypies represent a true correlate of compulsive behaviour. For example, DICT-7 mice that experience D1 receptor potentiation exhibit reduced stereotypy after treatment with cocaine, suggesting that drug-induced stereotypy and the stereotypies produced by enhanced D1 receptor over-expression do not lie on the same continuum (Campbell et al., 1999b). Furthermore, in clinical studies, single doses of d-amphetamine have been shown to ameliorate OCD symptoms (Insel et al., 1983; Joffe et al., 1991). Notwithstanding, Szechtman et al. (1998) have shown that the D2/D3 DA receptor agonist quinpirole leads to behaviour that can be analysed as a form of repetitive ‘checking’ in rats (Szechtman et al., 1998). Specifically, following drug administration (quinpirole 0.5 mg·kg⁻¹ or saline, twice weekly for 5 weeks), rats were placed individually into an open field with four objects at fixed locations and their activity was recorded for 55 min. Compared with saline, quinpirole-treated rats visited two locations more frequently than controls and at these sites exhibited a ‘ritual-like’ set of motor activities. Moreover, this behaviour is reduced by treatment with clomipramine, consistent with it being a plausible model for OCD.

Perseveration is a term that can be applied to a variety of behavioural outputs ranging from relatively simple to complex. In ‘simple’ cases, a motor output is performed repetitively, whereas ‘complex’ perseveration includes activities such as repetitive approach towards a specific, perhaps moving, goal or persistence in complex sequences of operant behaviour, for example in which rats persist in lever-pressing for food they do not eat. In this type of model, administration of a 5-HT_2A receptor antagonist had only weak effects on compulsive lever pressing (Flaisher-Grinberg et al., 2008). In addition, both spontaneous (Yadin et al., 1991) and reinforced delayed alternation behaviour (Tsaltas et al., 2005) can become perseverative if the animal continues to make the previous choice. In the rewarded T-maze alternation rat model, Tsaltas et al. (2005) found that administration of mCPP, a mixed serotonin agonist with potent 5-HT_2C agonist effects, increased persistence or compulsivity of responding, whereas chronic pre-treatment with an SSRI (fluoxetine), but not a benzodiazepine or desipramine, abolished the effects of mCPP. These results mirror those seen in OCD patients, where acute pharmacological challenge with mCPP exacerbated OCD symptomatology, and this effect was attenuated by pretreatment with fluoxetine (Hollander et al., 1991) and clomipramine (Zohar et al., 1988). These factors suggest the rewarded T-maze alternation rat model may represent a plausible proxy for OCD. Challenge with the 5-HT_1B receptor agonist naratriptan had no effect on compulsivity within this model (Tsaltas et al., 2005), suggesting a specific function for the 5-HT_2C receptor in OCD, which may be down-regulated by chronic SSRI treatment. Activation of the 5-HT_2C receptor has also been shown to induce self-grooming in rats, further supporting the hypothesis that selective stimulation of central 5-HT_2C receptors exacerbates OCD symptoms (Graf, 2006). Consistent with these findings, Boulougouris et al. (2008) found that a 5-HT_2C receptor antagonist improved perseverative responding in rats during reversal learning (see below) (Boulougouris et al., 2008). The same effect was observed in a spatial alternation model of OCD: 5-HT_2C, but not 5-HT_2A, receptor antagonism blocked the mCPP-induced directional persistence. The novel antidepressant agomelatine, which shows a degree of selectivity for 5-HT_2Creceptor antagonism, suppressed stress-induced glutamate release in the prefrontal cortex of stressed rats, in addition to reduced stress-induced marble burying (see above) suggesting a possible role in anxiety and OCD (Tardito et al., 2010). Blockade of stress-induced increase of glutamate release in the rat prefrontal/frontal cortex by agomelatine involves synergy between melatonergic and 5-HT_2C receptor-dependent pathways.

Neurobehavioural models of OCD

Various aspects of human behaviour can be successfully modelled in animals. Neuropsychological and brain imaging studies in OCD patients and their unaffected first-degree relatives revealed performance and/or neural processing deficits in several forms of cognitive or behavioural flexibility and in inhibitory response control: (i) reversal learning, (ii) extra-dimensional attentional set-shifting and (iii) motor impulsivity; these different forms of behaviour are normally modulated in humans and other animals including rats and monkeys by (i) serotonin, (ii) catecholamine (i.e. dopamine and noradrenaline) and (iii) noradrenaline respectively. These impairments are thought to contribute towards the development of ‘cognitive inflexibility’ and ‘motor impulsivity’ as endophenotypic traits within OCD families (Chamberlain et al., 2005; 2007b; Chamberlain and Menzies, 2009).

Reversal learning

Reversal learning refers to the reversal of reinforcement contingencies in a two-choice discrimination paradigm, such that the response to a previously rewarded stimulus is now punished and vice versa. Impairments of such reversal learning may reflect perseveration in responding to a formerly reinforced stimulus, even though its spatial position is shifted over trials. OCD patients and their unaffected relatives have both been shown to exhibit a reduced blood oxygen level-dependent response in the OFC during visual reversal learning, suggesting a possible neuroendophenotype for OCD (Chamberlain et al., 2008). In marmoset monkeys, impaired visual (object) reversal learning is induced not only by orbitofrontal lesions (Dias et al., 1996), but also by 5-HT depletion, specifically within the prefrontal cortex (Clarke et al., 2004) and in later studies when restricted to the OFC (Clarke et al., 2005; 2007). This behaviour appears to be selectively perseverative (rather than resulting from excessive avoidance of the previously non-reinforced stimulus) in nature, insofar as the reversal learning returns to normal if the previously rewarded stimulus is substituted by a novel one (Clarke et al., 2007). It is important to realize that this perseverative behaviour does not simply represent enhanced resistance to extinction; in fact, OFC 5-HT loss does not enhance responding in the extinction of a visual discrimination, although such animals are biased in their responding to the formerly reinforced stimulus. By contrast, selective dopamine depletion from the OFC causes no such bias, but leads to great persistence in responding in extinction (Walker et al., 2009). It is important for this model that involvement of the striatum is also confirmed. The OFC projects to the medial striatum and nucleus accumbens in marmosets (Roberts et al., 2007). Moreover, excitotoxic lesions of the medial striatum also lead to enhanced perseverative behaviour during reversal (Clarke et al., 2008). Thus, specific orbitofrontal-striatal loops are implicated in this form of cognitive rigidity.

Overall, although there is evident translation from rodent to monkey and humans, including OCD patients for the neural substrates of reversal learning deficits, there has been little attempt thus far to remediate reversal learning deficits in these studies of non-human primates, although reversal learning (though of the spatial rather than visual object reversal type) has been used to assess 5-HT agents in rats (Boulougouris et al., 2008). Hence, predictive validation of the reversal learning model has been limited. However, a recent study has shown an obvious relationship among reversal, 5-HT and effects of stress in rats, with elevated stress being associated with impaired reversal learning (Lapiz-Bluhm et al., 2009). This may demonstrate an important relationship between rigidity induced by a cortical lesion in conjunction with effects of stress (presumably leading to anxiety), mediated by the ascending 5-HT system.

Extra-dimensional (ED) set shifting

In clinical studies, another common form of perseverative responding involves derangement of attentional set-shifting, exemplified by perseveration of a learned rule or stimulus category/dimension (such as ‘sort by the perceptual category shape’) in the Wisconsin Card Sort Test. Such impairment may occur as a result of frontal lobe damage (see below). In work by Chamberlain and colleagues, patients with OCD and their unaffected first-degree relatives showed impaired extra-dimensional set-shifting on the CANTAB intra-dimensional – ED task (where subjects are impaired in shifting attention from one perceptual dimension (or aspect) of a complex stimulus to another (Chamberlain et al., 2006; 2007b). OCD patients with concurrent OCPD were significantly more affected (Fineberg et al., 2007) and groups of patients with BDD (Jefferies et al., 2010), and schizophrenia with OCD (Patel et al., 2010), as well as schizophrenia without OCD (Pantelis et al., 1999), have also shown ED impairment compared with a suitably matched control group. Thus, impaired ED-shifting may represent a hallmark of compulsive responding associated with cognitive inflexibility. As opposed to reversal learning, this form of attentional set-shifting modelled in the marmoset is impaired by lateral frontal – but not orbitofrontal – cortex lesions and by catecholamine – but not 5-HT – depletion (see Robbins, 2005 for review). It is also of importance that, whereas the impaired ED-shifting in OCD patients is also seen in their unaffected first-degree relatives (Chamberlain et al., 2007a), this is not the case for schizophrenia (Ceaser et al., 2008), suggesting that this form of cognitive inflexibility may be an endophenotype for OCD but not for schizophrenia.

Signal attenuation

Another neurobehavioural model with confirmed predictive validity invokes ‘signal attenuation’ as a mechanism for compulsive responding. According to this model, OCD results from deficient feedback associated with the completion of goal-directed responses. Normal functioning of such feedback prevents pointless repetitions of responses once their goal has been attained. The goal-directed behaviour of this model is instrumental lever-pressing for food. The feedback for a successful response is a compound stimulus of light and tone. The ‘feedback deficit’, assumed to underlie compulsive behaviour is induced in the model by means of attenuation of the ‘signalling property’ of this compound stimulus (repeated presentation without food in the absence of lever-pressing opportunity). The behavioural control condition for this attenuation process is termed ‘regular extinction’, and is identical in training and testing sequence, apart from the omission of the ‘stimulus devaluation’ (assumed to be equivalent to ‘signal attenuation’) stage. The effects of ‘signal attenuation’ on lever-press responding are assessed under extinction conditions through comparisons to the effects of ‘regular extinction’. Regular extinction and, to a lesser extent, extinction after signal attenuation, both produce excessive lever-presses (ELP) followed by magazine entry (ELP-Completed, ELP-C). Extinction after signal attenuation additionally produces excessive lever-presses not followed by magazine entry (ELP-Uncompleted, ELP-U). According to the authors, ELP-C reflects rats' response to non-reward while ELP-U reflects response to the encounter of an attenuated signal and constitutes the model's focal behaviour (surplus lever pressing). Arguably, Joel and Avisar (2001; Joel et al., 2004) have developed this model more comprehensively than any other model of OCD (Joel and Avisar, 2001; Joel et al., 2004). The instrumental lever-pressing has a perseverative quality which is sensitive to reductions produced by virtually all of the drugs used therapeutically in OCD, but not to those which are less effective, such as diazepam or desipramine. This behaviour is also enhanced by lesions of the rat OFC and sensitive to manipulations of the medial striatum, to which the OFC projects. Joel and colleagues have thus established many of the validating criteria for a successful model of OCD, although the exact theoretical explanation in terms of signal attenuation may perhaps be queried.

Signal attenuation appears to resemble a special form of extinction in which Pavlovian associations of a conditioned stimulus are extinguished differentially with respect to instrumental responding. The perseveration in instrumental behaviour arises because the terminal links in the response chain leading to food are extinguished. Extinction itself also depends on an inhibitory process which suppresses associations which in fact remain intact (Rescorla, 2001). Another example of this form of perseveration has been reported in the performance of an attentional task for rats, namely the five-choice serial reaction time task (5CSRTT). The 5CSRTT (Robbins, 2002) is conducted in operant chambers equipped with an arc of nine holes, four of which are occluded and five exposed. Animals are required to initiate the trial by nose-poking in the food magazine, detect a target visual stimulus presented for 0.5 s randomly in one of the five exposed holes, and then make a nose-poke response to the hole where the light appeared (rewarded response). Perseverative nose-poking possibly caused by a failure to detect response feedback cues can arise from lesions to the OFC in rats (Chudasama and Robbins, 2003).

Exaggerated habit-learning

A related concept is that of exaggerated habit-learning, where compulsive behaviour is driven by relatively heightened stimulus-response (S-R) associations coupled with a generally weakened influence of the ultimate goal. Compulsivity, in the context of OCD, may depend upon a propensity towards excessive, stereotyped behaviour which is carried out to reduce the likelihood of adverse consequences (APA, 2000; Chamberlain et al., 2009). OCD patients acknowledge that their behaviours are excessive and typically ineffective, yet they are unable to exert adequate control over the drive to perform these compulsive acts. This observation has led to the hypothesis that OCD compulsions may not be under goal-directed control and instead are driven by maladaptive habit learning (Graybiel and Rauch, 2000; Boulougouris et al., 2009). A study on humans with OCD provides the first experimental evidence for a selective impairment in OCD patients in flexible, goal-directed control over behaviour, forcing them to rely instead on S-R habits (Gillan et al., 2010).

Recent neuroscientific investigations implicate a circuit linking ventromedial prefrontal cortex (vmPFC) and caudate in goal-directed action control. Thus, persistent dominant habitual control can be induced in animals by lesioning specific brain areas (see Balleine and O'Doherty, 2010 for review). Animals with lesions to the dorsomedial striatum (DMS) and the prelimbic cortex persist in habitually responding towards food outcomes that are no longer desirable as a consequence of pairing with lithium chloride-induced nausea or specific satiety (Corbit and Balleine, 2003; Killcross and Coutureau, 2003; Yin et al., 2005). More recently, human functional magnetic resonance imaging (fMRI) studies have provided convergent support for this dissociation in homologous brain regions. Studies using instrumental learning tasks have implicated the vmPFC (Valentin et al., 2007; De Wit et al., 2009) and anterior caudate nucleus (DMS in rodents) (Tricomi et al., 2004; Tanaka et al., 2008) in goal-directed response selection. Importantly, habitual control is supported by different neural structures, including specific sectors of the striatum (dorsolateral striatum, probably homologous to the putamen) (e.g. Yin and Knowlton, 2006) and infralimbic cortex in animals (Coutureau and Killcross, 2003; Yin et al., 2004) and the putamen in humans (Tricomi et al., 2009). Thus, in OCD, disrupted goal-directed control may force OCD patients to rely strongly on inflexible, S-R habits which are supported by a parallel corticostriatal pathway, including the putamen and possibly the sensorimotor cortex (Tricomi et al., 2009; Balleine and O'Doherty, 2010). A major unanswered question is how habitual responding is converted into compulsive behaviour, relevant to OCD. One clue may come from the observation that amphetamine sensitization has been shown to enhance habit learning (Nelson and Killcross, 2006). Sensitization is a form of neural plasticity that leads to heightened behavioural responses to the drug, probably mediated by elevated striatal dopamine function, and so this suggests again that dopamine contributes to compulsive behaviour. Additionally, recent observations have shown how stress may also enhance habit learning in rats (Dias-Ferreira et al., 2009), suggesting once again a link between anxiety states in OCD and compulsive behaviour.

Motor response inhibition

In addition to a possible shift in control to habit-based representations, OCD patients also exhibit decreased behavioural and cognitive inhibition in a variety of tasks (Tien et al., 1992; Enright and Beech, 1993; Rosenberg et al., 1997; Bannon et al., 2002; see Chamberlain et al., 2005 for review), in addition to the increased errors they show on the alternation learning task (Abbruzzese et al., 1997; Cavedini et al., 1998). However, motor response inhibition is perhaps most readily investigated using the stop-signal reaction time task, in which it is necessary to stop an already-initiated response on presentation of a stop-signal. The stop-signal reaction time (SSRT) may be calculated in humans by measuring the response latency required to successfully cancel a response in a choice-reaction time procedure (Logan et al., 1984). A recent comparative study of OCD and trichotillomania (Chamberlain et al., 2006) shows an interesting dissociation in which trichotillomania patients had greatly lengthened SSRTs and that OCD patients were also significantly slowed on this measure, as compared with age- and IQ-matched controls. By contrast, OCD patients were significantly impaired on the ED-shift test, whereas trichotillomania patients were not. These data suggest that whereas OCD is accompanied by a general problem in cognitive flexibility, trichotillomania is associated more specifically with a failure to inhibit pre-planned motor activity. Moreover, recent studies of OCD patients and their first-degree relatives (Chamberlain et al., 2007b; Menzies et al., 2007) identified behavioural deficits on these tasks in ‘at risk’ individuals, linked with structural abnormalities of fronto-striatal circuitry.

Studies of human patients with frontal lobe damage have localized one critical zone for SSRT to the right inferior frontal gyrus (Aron et al., 2003) and other data implicate the striatum and subthalamic nucleus in this inhibitory process (Aron et al., 2007, but see also Hampshire et al., 2010). A similar neural network may be implicated in the ED-shift, according to a recent fMRI study (Hampshire and Owen, 2006) and other evidence of common noradrenergic mediation (Lapiz and Morilak, 2006; Robinson et al., 2008b). A method of measuring SSRT in rats has been developed, which is dependent on possibly homologous structures in the lateral OFC and medial striatum (Eagle and Robbins, 2003; Eagle et al., 2007). Intriguingly, however, the SSRT is insensitive to serotoninergic manipulations in both rats and humans (Chamberlain and Sahakian, 2007), but may be amenable to noradrenergic remediation, for example, with methylphenidate or atomoxetine in patients with attention deficit hyperactivity disorder (Chamberlain et al., 2007a; Devito et al., 2009).

Lengthened SSRTs can be interpreted as enhanced impulsivity, supporting the view of functional relationships between impulsivity and compulsivity postulated clinically (Hollander and Rosen, 2000) in animal models of stimulant drug addiction (Everitt and Robbins, 2005; Berlin et al., 2008). In view of these possible links between the two constructs, an intriguing dissociation between premature responding in the 5CSRTT and reversal learning (i.e. impulsivity and compulsivity) has been reported. Specifically, studies utilizing the 5CSRTT have shown that systemic administration of a 5-HT_2C antagonist (SB 242084) exacerbated the enhanced impulsivity normally observed following global 5-HT depletion produced by intra-cerebroventricular administration of 5,7-dihydroxytryptamine; a similar SB242084-related enhancement in impulsivity was seen in sham-operated rats (Winstanley et al., 2004). In contrast, systemic administration of a selective 5-HT_2A receptor antagonist (M100907) had opposite actions, remediating impulsivity in both sham-operated and 5-HT-depleted rats. These contrasting influences of the 5-HT_2A and 5-HT_2C receptor antagonists were mimicked by infusions of the drugs into the nucleus accumbens, but not the medial prefrontal cortex, in intact animals (Robinson et al., 2008a). On the contrary, studies utilizing a simple serial spatial reversal task, shown to be sensitive to orbitofrontal lesions (Boulougouris et al., 2007), showed that systemic administration of the 5-HT_2C receptor antagonist promoted reversal learning, while M100907 had the opposite effect of impairing it. Note that in terms of remediation, this is opposite to what was found for measures of impulsivity. Similar enhancements of reversal learning after treatment with the 5-HT_2C antagonist were also found after infusion into the OFC (Boulougouris and Robbins, 2010).

Regardless of the precise elucidation of mechanism, these data pharmacologically dissociate impulsivity and compulsivity, suggesting that they cannot arise simply from a common process of behavioural disinhibition. This dissociation must be task-dependent as both tasks require response inhibition for efficient performance, suggesting that there is some other aspect of the processes engaged by the task, which differentiates them. These results also imply that impulsivity and compulsivity are functionally separate and reciprocally yoked, lending support to the impulsive–compulsive diathesis model (Hollander and Wong, 1995). They also suggest that impulsivity and compulsivity are neuroanatomically and neurochemically dissociated by selective 5-HT2 receptor agents and may lead to new clinical applications. However, further experimental evidence is required to resolve how these data fit with the consistent finding that OFC 5-HT depleted marmosets show impairments on visual object reversal learning (Clarke et al., 2004; 2005). In addition, it would seem likely that these seemingly opposed effects are mediated through separate neural pathways: in the case of impulsivity, through projections from the infralimbic vmPFC (area 25), an area richly innervated by 5-HT_2A receptors and strongly implicated in affective regulation, towards the shell of the nucleus accumbens (Vertes, 2004) and, in the case of compulsivity, in connections between the OFC and the caudate nucleus (or the dorsomedial striatum in the rat) (Schilman et al., 2008).

Anxiety models

Several other theoretical positions may be especially useful in explaining certain forms of OCD. For example, the theoretical construct that anxiety is the prime trigger of OCD, as posited for example by Rachman and Hodgson (1980), should not be underestimated (Rachman and Hodgson, 1980). Active avoidance behaviour in animals is well known to be very persistent as it so rarely has the opportunity for extinction, and drugs such as d-amphetamine exacerbate this perseverative tendency (Lyon and Robbins, 1975). Thus, behaviour that initially has some adaptive value, for example, that results in avoiding shocks, apparently loses its rationale after thousands of trials in which shock is never presented. A more recent formulation by Szechtman and Woody (2004) suggests that OCD-like activity arises as an aberrant excess of behaviour motivated by the need for security (Szechtman and Woody, 2004). These theories are of obvious clinical interest and will ultimately depend on their validation by the importance assigned to anxiety in producing the persistent symptoms of OCD.

These findings may provide interesting new insights into the clinical understanding of compulsivity in OCD that link habit and goal-directed learning and affective state. Consider this common example. Seeing, or being reminded of a potential contaminant triggers anxiety relating to a potentially catastrophic outcome, and activates the compulsive response of washing one's hands. Although the individual clearly recognizes that this act has little or no bearing on contracting illness, performing the compulsion causes a momentary reduction in anxiety which is experienced as relief. In other words, the act of washing one's hands may not be driven by its direct consequences, but rather by external triggers of compulsive habits that are reinforced by the experience of relief within the general aversive motivational state of anxiety. Although compulsivity, in the context of OCD, is avoidant and not appetitive, it is likely that the same fundamental mechanisms may give rise to reliance on S-R habit reinforcement. In line with this hypothesis, Kim et al. (2006) showed that the OFC is engaged not only when people gain rewarding events, but also when aversive events are successfully avoided (Kim et al., 2006).

Deep Brain Stimulation (DBS)

In patients with severe, treatment refractory OCD, psychosurgery is sometimes considered as a means of alleviating the symptoms (for review see Greenberg et al., 2010). In one such approach – DBS – small electrodes are implanted into the brain guided by imaging techniques, and are subsequentially used to stimulate particular neural nodes. Several pilot patient studies have reported beneficial reductions in OCD symptoms when electrodes have been implanted into such neural regions as the ventral striatum (Greenberg et al., 2008), caudate (Aouizerate et al., 2004), subthalamic nucleus (Mallet et al., 2008) and the nucleus accumbens (Denys et al., 2010). As described previously, most of these regions have been implicated in the neurobiology of OCD per se; however, it should be noted that choice of electrode site has also been guided by what is known of the neurobiology. Several translational studies have explored effects of DBS in animal models of the disorder.

Low- but not high-frequency stimulation of the thalamic nucleus was effective in reducing 8-OHDPAT-induced perservation in rats (Andrade et al., 2009). In the rat, quinpirole-induced repetitive checking model, high frequency stimulation of the subthalamic nucleus reduced compulsive behaviours transiently, as did stimulation of the nucleus accumbens shell and core (Klavir et al., 2009; Mundt et al., 2009; Djodari-Irani et al., 2011). Similar benefits have been reported in the signal attenuation model of Joel and colleagues, with post-training high-frequency stimulation of the subthalamic nucleus, and globus pallidus, leading to anti-compulsive effects (Klavir et al., 2009). Collectively, the available animal studies involving stimulation of specific neural regions show remarkable parallels with findings in human OCD patients, and also suggest potential novel therapeutic anatomical targets.

Conclusions

We are thus intriguingly close to providing useful theoretically motivated models of OCD, particularly with regard to repetitive motoric habits and inhibitory failure. The animal models reviewed above constitute an important vehicle for the investigation of several aspects of OCD. However, every model has its strengths and weaknesses (Table 1) which should be taken into consideration for determining the needs it can serve. An important feature of a model for anti-compulsive activity screening is its predictive validity. Regarding predictive validity, it should be noted that around 40–60% of OCD patients are resistant to SSRI monotherapy (Fineberg et al., 2006b). Therefore, the establishment of a model's predictive validity lies not only on the effectiveness of SSRIs but, more importantly, on the ineffectiveness of drugs known not to be efficacious in OCD as well. Additionally, chronic drug administration might be a good candidate for such a differentiation. The signal attenuation and reinforced spatial alternation models of OCD have good predictive validity, as they have shown pharmacological isomorphism with the treatment of OCD and the lack of effect on the models' focal behaviours of drugs not effective to OCD treatment. However, the signal attenuation model is not suitable for examining the effects of chronic pharmacological treatment, as prolonged drug administration may contaminate the early stages of the procedure. Yet, the genetic and neurobehavioural models previously discussed lack predictive validity, although they look more convincing in terms of construct validity and may have promise for the development and screening of anti-compulsive drugs.

Elucidation of the neurobiological substrates of OCD is amply represented in many animal models, contributing to their construct validity. Additionally, behavioural models such as those based on signal attenuation or reversal learning have already been shown to be sensitive to serotoninergic/dopaminergic systems and orbitofrontal dysfunction, both heavily implicated in OCD. On the other hand, genetic models of OCD involving single gene alterations might be extremely useful for the understanding of certain forms of OCD pathophysiology. It would be of considerable interest to determine whether the more obvious motor manifestations of the other conditions, such as trichotillomania, are associated with structural and/or functional impairments of similar cortico-striatal loops, possibly more at striatal than cortical nodes, or whether, as seems likely, these are associated with impairments in other fronto-striatal pathways: for example, related to the putamen and its role in the control of motor output.

Although none of the animal models reviewed in this paper can account for simulating OCD in its entirety, as presupposed by an ‘ideal’ model, some could potentially be enhanced by further investigation. None of the animal models provide a good model for obsessions, as opposed to compulsive behaviours. Given the heterogeneity and aetiological complexity of OCD, the findings emerging from the combined use of different models may provide insight to the various aspects and aetiology of the disorder and lead to new treatments. Direct comparison of these findings might also elucidate genuine anti-compulsive effects rather than effects limited to a specific model that is not necessarily related to OCD.

Glossary

Abbreviations

5CSRTT

5-choice serial reaction time test

5-HT

serotonin

APA

American Psychiatric Association

BOLD

blood oxygen level-dependent

CS

conditioned stimulus

DBS

deep brain stimulation

DSM

Diagnostic and Statistical Manual

ELP

excessive lever presses

(f)MRI

(functional) magnetic resonance imaging

ID-ED

intra-dimensional – extra-dimensional

mCPP

meta-chlorophenylpiperazine

NMDA

N methyl-d aspartic acid

O-C

Obsessive-compulsive

OCD

Obsessive-Compulsive Disorder

OCSDs

Obsessive-Compulsive Spectrum Disorders

OFC

Orbitofrontal Cortex

SR

stimulus-response

SSRIs

selective serotonin reuptake inhibitors

SSRT

stop-signal reaction time