Introduction

Obsessive-compulsive disorder (OCD) is a debilitating psychiatric disorder, with lifetime prevalence estimates of 2.6% (Ruscio et al. 2010). Obsessions are intrusive thoughts or ideas that are often associated with compulsive, or repetitive and rigid, behavior (American Psychiatric Association 2013). On factor analyses of OCD symptoms, several symptom dimensions emerge, with varying content of obsessions and/or compulsions (O/C) (Table 1). The symptoms of OCD are time-consuming, distressing, and result in significant interference in occupational and social function (American Psychiatric Association 2013). Since the publication of the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), OCD is included with body dysmorphic disorder, hoarding disorder, trichotillomania, and excoriation disorder in a single diagnostic chapter on ‘Obsessive-Compulsive and Related Disorders’ (OCRDs; American Psychiatric Association 2013). OCD is not only often co-morbid with conditions classified in the OCRD cluster, but also with mood and anxiety disorders (Torres et al. 2016).

Table 1 Common OC symptom dimensions. Adapted from ( Abramovitch and Cooperman 2015 ; Markarian et al. 2010 )
Full size table

Animal models may be useful for investigating the underlying etiopathology of psychiatric illness. Not only can they contribute to understanding underlying neurobiology, but they may also allow investigation of novel treatments (Fineberg et al. 2011). In the past, a number of animal models of OCD have been proposed. Given that obsessions are difficult to demonstrate in animals, models of OCD have focused on repetitive and rigid behavior that is reminiscent of the symptoms of OCRDs. Such behavior includes excessive lever-pressing (Joel 2006) and nest building behavior (Hoffman and Rueda Morales 2009; Li et al. 2006; Wolmarans et al. 2016a; Greene-Schloesser et al. 2011), rigid locomotive patterns (Szechtman et al. 1998; Wolmarans et al. 2013; Yadin et al. 1991), aberrant grooming and hair pulling behavior (Greer and Capecchi 2002; Kinnear et al. 2000; Welch et al. 2007), compulsive-like chewing (Chou-Green et al. 2003a, b), compulsive-like marble burying (Greene-Schloesser et al. 2011) and hoarding (Andersen et al. 2010). Together, these models conceivably represent different OC-like phenotypes that may contribute to our understanding of the mechanisms underlying different symptoms. Animal models of OCD can be based on naturalistic or conditioned behavior, pharmacological challenges, or genetic manipulation (Alonso et al. 2015). Whereas naturalistic and conditioned models may provide more insight into the behavioral triggers and course of OC-like behavior, pharmacological and genetic models may provide targeted frameworks for studying specific neurobiological and genetic mechanisms underlying O/Cs (Alonso et al. 2015).

The deer mouse (Peromyscus maniculatus bairdii) model of OCD (Korff et al. 2008) can be regarded as a naturalistic model characterized by spontaneous OC-like behavior (Hoffman 2011; Wolmarans et al. 2013). Over the past decade, our group has published a number of investigations of this model. The current paper will review progress to date.

The deer mouse model of OCD

Peromyscus maniculatus (deer mouse) and congeneric species are the most common mammals native to the North-American continent (Shorter et al. 2012). As opposed to rats, dogs, cows, sheep, and laboratory mice (Mus) where selective breeding has led to genomic alterations (Vrana 2007), deer mice represent a true wild-type mammalian model system, although bred in captivity at the Peromyscus genetic stock center of the University of South Carolina (Joyner et al. 1998). Wild-type strains differ from inbred strains in being genetically more diverse (Yang et al. 2011), and deer mice have proven to be useful in the study of genetic variability and epi-genetic influences underlying different behavioral phenotypes. One subspecies, viz. the tall grass prairie P. maniculatus bairdii strain (hereafter only referred to as deer mice), of which stock animals have been derived from 40 wild-type ancestors caught in Washtenaw County, Michigan, has been used in studies of spontaneous repetitive behavior (Hadley et al. 2006; Presti et al. 2004; Shorter et al. 2014).

Notably, the repetitive and stereotypic behavior observed in these animals, i.e. jumping, backward somersaulting and pattern running, is expressed in varying frequencies across the population in laboratory settings, suggesting that these behaviors are influenced by a combination of genetic and environmental factors (Shorter et al. 2014). While repetitive motor patterns are not necessarily indicative of pathology (Eilam et al. 2006; Langen et al. 2011), the seemingly purposeless and time-consuming stereotypies of varying forms and intensity in deer mice, have been proposed to resemble the repetitive and rigid symptomology of OCD (Korff et al. 2008) and autism (Lewis et al. 2007). That said, given that stereotypic behavior is seen in deer mice in naturalistic settings also (Shorter et al. 2012, 2014), it is likely that this may be an adaptive response.

Spontaneous stereotypy in deer mice as a phenotype of persistent, but compulsive behavior

Jumping, backward somersaulting, and pattern running in deer mice were originally studied based on their resemblance to stereotypical movement disorders, in particular the motor manifestations of autism (Powell et al. 1999). Compared to animals housed in standard cages, individuals maintained in enriched cages express lower levels of stereotypy (most notably pattern running), characterized by a delayed onset and a lower incidence (Bechard et al. 2016; Hadley et al. 2006; Powell et al. 1999). However, it has been suggested that up to 62% of deer mice housed in standard laboratory cages developed stereotypy.

In early work, behavioral categorization was performed by means of visual observation (Powell et al. 1999), with each animal observed twice weekly for 5 min. Subsequently, the same group introduced automated screening (Presti et al. 2004), and animals were classified into low- (L) and high- (H) stereotypical groups based on the mean stereotypy score, i.e. the number of distinct stereotypical movements, generated over an eighteen-hour-long session (Presti et al. 2004).  Further, to obtain accurate results from neurochemical investigations, only animals expressing the lowest and highest levels of stereotypy were included in follow-up studies, excluding a grey margin of animals and that yielded a more distinct separation of the two behavioural phenotypes. This approach has since been adopted by others (Wolmarans et al. 2013, 2016a, b, 2017a).

Korff et al. (2008) classified deer mice into non- (N), low- (L), and high- (H) stereotypic groups on the basis of a mean stereotypy score obtained during three individual one-hour long behavioral screening sessions, each one week apart. Deer mouse stereotypy was variable within the population with 45% of animals classified as H, 41% as L, and 14% as N, irrespective of sex. As deer mice are nocturnal animals, follow-up investigations measured the time spent executing stereotypy over 12-h during the dark cycle (Wolmarans et al. 2013); this allowed us to demonstrate that the stereotypy frequency and intensity vary across subjects and between different assessments, and that H deer mice express time-consuming stereotypy during specific bouts of the dark cycle only (Fig. 1, two of the normal 5 baseline stereotypy trials shown). This is arguably reminiscent of the waxing and waning nature of OC symptomology (American Psychiatric Association 2013; Wolmarans et al. 2013).

Fig. 1: Varying intra- and inter-trial expression of stereotypy.
figure 1

Data represents vertical activity counts of one animal generated across 2 respective 12-h dark cycles, each divided into 24 30-min time-intervals. T1 – T2 : Trial 1–2, spaced one week apart

Full size image

A central question to modeling OCD in animals is whether it is possible to characterize stereotypy not simply as a motoric phenomenon, but rather as representing an underlying cognitive-affective alteration (Tanimura et al. 2009, 2008; Wolmarans et al. 2017a). In this regard, previous findings must be considered. Repeated administration of psychostimulants induces stereotypy and facilitates the transition of goal-directed to habitual responses (Burguière et al. 2015; Graybiel 2008). However, since deer mouse stereotypy is not subject to amphetamine-induced behavioral sensitization (Tanimura et al. 2009), it is likely that the stereotypical phenotype described in deer mice differs from a purely habitual phenomenon.

In addition, we have noted differences in sociability between H and N deer mice, both within- and between cohorts (Wolmarans et al. 2017a), with changes in sociability after administration of escitalopram. This again suggests that stereotypic behavior in deer mice is not merely a motoric phenomenon, but reflects more broad-spread mechanisms. By investigating group interactions between three animals in different social paradigms, viz. HHH, NNN, HHN and NNH, we demonstrated that H animals group together in the presence of an N conspecific (HHN paradigm), while being marginalized by animals of the N cohort (NNH paradigm). Our data relating to the sociability of deer mice has key implications. Notably, P. maniculatus bairdii engage in fewer social interactions compared to P. polionotus subgriseus, another species within the genus (Shorter et al. 2014). However, stereotypical behavior in P. polionotus subgriseus is negligible. Taken together, results from the Shorter et al. (2014) and Wolmarans et al. (2017a) investigations may indicate that the sociability of P. maniculatus bairdii is modified by the level of stereotypy displayed by conspecifics, although it is premature to draw strong conclusions about a causal relationship between stereotypy and altered social competence. This is consistent with clinical evidence regarding the social behavior of OCD patients and their social experiences in the presence of healthy peers (Berrocal et al. 2006; Kim et al. 2012; Rosa et al. 2012; Storch et al. 2006).

Taken together, deer mouse stereotypy resembles OC behavior in that 1) it is repetitive, persistent and time consuming, 2) it manifests as a narrow range of phenotypes, i.e. jumping, pattern running and backward somersaulting that can possibly be differentiated at a neurobiological level, 3) it demonstrates within- and between-subject variance in frequency and intensity, 4) it is expressed in a waxing and waning pattern, 5) it is resistant to behavioral sensitization and so may represent a form of abnormally regulated goal-directed behavior, rather than a habit, 6) it is characterized by social deficits, and 7) it is influenced by both environmental and genetic factors.

Parallels between the treatment response of deer mouse stereotypy and OCD

Response to chronic, but not sub-chronic selective serotonin reuptake inhibitors (SSRIs)

During the initial stages of investigating deer mouse stereotypy as a putative animal model of OCD, Korff et al. (2008) demonstrated that chronic (21-day) intraperitoneal treatment with a high dose SSRI, i.e. fluoxetine 20 mg/kg/day, but not the noradrenaline reuptake inhibitor (NRI), desipramine, attenuated stereotypy intensity in H and L animals without affecting normal locomotion. As chronic and high dose SSRI treatment is the first-line pharmacological treatment for OCD, while noradrenergic compounds are ineffective (Fineberg and Craig 2007), these findings accredit deer mouse stereotypy with valuable predictive validity as an animal model of OCD. Subsequent work extended these initial findings. To account for individual fluctuations in stereotypy over the course of a single dark cycle and considering that the distinct forms of stereotypy do not demonstrate any significant association with one another, a focus was established on time and stereotypy intensity (Fig. 2) (Wolmarans et al. 2013). This classification system also allows for genetic predisposition and epigenetic influence in the stock colony as it appraises deer mouse behavior on a continuum from normal to severe manifestations of stereotypy. As opposed to the use of cut-off criteria, the balance may shift in any direction without changing the fundamental goal of the investigation, viz. comparing normal and stereotypic animals.

Fig. 2: Stereotypy intensity versus time spent engaging in high stereotypical activity over 12 h.
figure 2

(a) Mean of highest individual daily vertical activity scores across the first five baseline behavioral trials versus time spent engaging in H activity. (b) Mean of highest individual daily cage revolution scores over the first five baseline behavioral trials versus time spent engaging in H activity. Sex symbols indicate selected male and female H animals

Full size image

To exclude the effect of injection stress on the manifestation of stereotypy, oral dosing with high dose escitalopram (50 mg/kg/day), a highly potent selective serotonin reuptake inhibitor (Owens et al. 2001), was introduced, together with a comparison between chronic (4 weeks) and sub-chronic (1 week) treatment. Subsequently, while adjusting H bouts to periods of normal rodent activity, we demonstrated that chronic but not sub-chronic escitalopram treatment reduced the time spent executing stereotypy (Wolmarans et al. 2013). Notably however, such animals still engaged in bouts of spontaneous H behavior (Wolmarans et al. 2013), arguably not unlike that observed in OCD (Overduin and Furnham 2012). These observations are important as we have previously suggested that by increasing the number of bouts of normal rodent activity, escitalopram engenders control over the urge to engage in H behavior (Wolmarans et al. 2013). Instead of persistently expressing less severe compulsive-like behavior, H animals can engage in normal rodent activities for a greater part of their wake cycle, although not being able to abstain entirely from OC bouts.

Overlaps between deer mouse behavior and treatment resistance in OCD

Treatment resistance remains a clinical obstacle in approximately 30% - 50% of OCD patients who remain unresponsive to SSRI monotherapy (Fineberg et al. 2006). The treatment of refractory OCD may include an increase in the dose of the SSRI and a longer duration of treatment (Bejerot and Bodlund 1998), or switching treatment to another SSRI (Fineberg et al. 2006). A third strategy is to augment SSRI therapy with a low dose D2 blocker (Erzegovesi et al. 2005; Hollander et al. 2003; Ipser et al. 2006; McDougle et al. 2000). Recently, we investigated marble burying (MB) behavior in deer mice as a measure of anxiety- and/or compulsive-like behavior (Dixit et al. 2014; Kedia and Chattarji 2014). We identified persistent burying behavior in 11% of deer mice; such behavior was found independent of stereotypy levels or sex (Wolmarans et al. 2016b). Moreover, in contrast with previous work in different species (Ichimaru et al. 1995; Li et al. 2006), chronic high dose oral escitalopram (50 mg/kg/day) failed to attenuate this behavior. We therefore speculate that MB in deer mice may be useful in modeling treatment resistant OCD; however, this remains to be proven.

Together with findings that backward somersaulting, but not jumping and pattern running, may reflect a behavior that is resistant to change (Tanimura et al. 2008), it is possible that some phenotypes of deer mouse behavior may be representative of different underlying psychobiological processes that may respond differentially to OCD treatment strategies. Although backward somersaulting does respond to escitalopram treatment, it may be useful to establish whether it demonstrates differential response to augmentation treatment strategies, compared to vertical jumping and pattern running. There is clinical evidence to suggest that different psychobiological mechanisms are variably affected in different patients with OCD, perhaps influencing treatment outcomes (Mataix-Cols et al. 1999; Rufer et al. 2006). To date, little work on animal models of OCD has focused on the issue of treatment-resistance, and this may be a useful focus for future investigation.

Biobehavioral overlaps between deer mouse stereotypy and OCD

Cortico-striatal-thalamic-cortical (CSTC) circuitry in OCD

OCD may reflect underlying disruption in cortico-striatal-thalamic-cortical (CSTC) signaling (Ahmari 2016; Tanimura et al. 2008; Wolmarans et al. 2013). Abnormal regulation of goal-directed behavior may be central to OCD symptomology (Gillan et al. 2011). Brain areas implicated in OCD mediate goal-directed behavior. These include the prefrontal cortex, striatum and thalamic nuclei which communicate with each other via different pathways (Evans et al. 2004; Nambu 2008). The CSTC circuit (Fig. 3) is organized in such a manner that the anterior cingulate cortex (ACC), via innervation of the ventral striatum, exerts feedback through the thalamus to the orbitofrontal cortex (OFC) (Haber and Knutson 2010). Consisting of a direct (behaviorally activating) and an indirect (behaviorally inactivating) pathway, the CSTC circuit is important for planning, executing and terminating complex motor responses and to facilitate reward learning (Morein-Zamir et al. 2014; Stocco et al. 2010). A relative bias in favor of the direct over the indirect pathway may underlie OC symptomology (Saxena and Rauch 2000; van den Heuvel et al. 2005). This not only results in an overactive OFC, resulting in dysfunctional reward processing (Haber and Knutson 2010), but also increases the activity in the CSTC circuit as a whole (Bartz and Hollander 2006; Saxena and Rauch 2000).

Fig. 3: The cortico-striatal-thalamic-cortical (CSTC) circuit.
figure 3

Solid lines , no cortical activation of pathways; dotted lines , cortically activated pathways; crosses , no considerable neurotransmitter release; minus signs , GABAergic inhibition; plus signs , disinhibition of target / glutamatergic activation; GPi / SNr , globus pallidus interna/substantia nigra pars reticulata; GPe , globus pallidus externa; STN , subthalamic nucleus; GABA , gamma-amino butyric acid

Full size image

The striatum mainly consists of GABAergic projections that divide into two subgroups, i.e. striato-nigral (SN) neurons of the direct pathway (projecting to and inhibiting the GPi/SNr), and striato-pallidal (SP) neurons constituting the indirect pathway (projecting to and inhibiting the GPe) (Rymar et al. 2004; Yelnik et al. 1991). Further, both pathways are tonically inhibited under resting conditions (Wilson and Groves 1981). However, upon initiating a behavioral action, signaling in both the activating (direct) and inactivating (indirect) pathways are triggered. This functional antagonism is resolved by the substantia nigra pars compacta (SNc) that modulates both pathways via dopaminergic signaling. Whereas the SN neurons of the direct pathway express Gs associated dopamine-1 (D1) receptors, the SP neurons of the indirect pathway express Gi associated D2 receptors (Tanimura et al. 2010). As such, stimulation of D1 elevates cAMP and increases GABA release, resulting in the activation of the direct pathway (Fig. 3). Conversely, activation of the SP D2 receptors decreases cAMP concentrations and inhibits the release of GABA, thereby inhibiting the indirect pathway (Gerfen et al. 1990; Tepper and Bolam 2004). Striatal dopamine release will therefore initiate motor behavior by shifting the executive balance to the direct pathway (Beiser et al. 1997; Chesselet and Delfs 1996).

Hyperactivity of the CSTC circuitry in OCD is hypothesized to be related to deficits in reward processing (Ferreira et al. 2017; Figee et al. 2011; Palminteri et al. 2012; Pinto et al. 2014), a process that is closely correlated with cortico-striatal dopaminergic signaling (Ljungberg et al. 1991; Mirenowicz and Schultz 1994; Schultz et al. 1993). Briefly, initial anticipation of a possible reward activates nearly 75% of the dopaminergic neurons in the basal ganglia (Ljungberg et al. 1991). However, repetitive exposure to the same stimulus facilitates a process of reward conditioning that enables the brain to evaluate future confrontations with the same set of factors it was conditioned to (Romo and Schultz 1990). It has been shown that the basal ganglia code differences between predicted and actual rewards with ‘reward prediction errors’ (Schultz et al. 1997; Schultz 2002). These are important for reward- and punishment based learning, as it is applied to implement sensorimotor changes to either keep experiencing the same reward in the case of a positive error, or achieving a better outcome in the case of a negative error. Therefore, the relative lack of a significant dopaminergic response after the manifestation of a fully predicted reward may account in part for inadequate closure after task completion in patients with OCD (Figee et al. 2011). It can thus be hypothesized that a dysfunctional reward system and altered dopaminergic signaling plays a role in OCD (Denys et al. 2004; Gillan et al. 2011; Husted et al. 2006).

Although dopamine plays a prominent role to facilitate and maintain motor behavior, it is drugs that target serotonergic and not dopaminergic signaling that have proved most useful in the first-line treatment of OCD (Fineberg and Craig 2007). The fact that behavioral inhibition has been associated with serotonergic neurotransmission (Cools et al. 2008; Daw et al. 2002) may be relevant. In a review of the opponent interactions between serotonin and dopamine (Daw et al. 2002), the term ‘opponency’ describes a paradigm in which more than one system codes for different affective events. While it is known that the dopaminergic system codes rewarding stimuli, serotonin is activated during the experience of aversive stimuli (Fletcher 1995; Fletcher and Korth 1999; Fletcher et al. 1999; Kapur and Remington 1996). Indeed, by enhancing serotonergic signaling, both conditioned behaviors (such as lever pressing for food) and unconditioned behaviors (such as feeding) normally associated with dopaminergic signaling, are inhibited. Consequently, the opposite effect is achieved when antagonizing serotonin or stimulating dopamine. This corresponds with data demonstrating that serotonin antagonizes the effects of dopamine in the SN neurons of the direct pathway (Daw et al. 2002; Kapur and Remington 1996). Therefore, with respect to OCD, it is possible that the balance between reward seeking behavior and aversive reactions is related to the balance between dopaminergic and serotonergic signaling (Ferreira et al. 2017).

Deer mouse stereotypy and aberrant CSTC signaling

A strong body of evidence confirms cortico-striatal involvement in deer mouse stereotypy. First, a bias in favor of the direct SN pathway has been demonstrated by findings that the phenotypic expression of deer mouse stereotypy, but not normal patterns of motor behavior, can be inhibited via selective blockade of striatal D1 and N-methyl-D aspartate (NMDA) receptors (Presti et al. 2003). Moreover, this is supported by a significantly higher SN-dynorphin / SP-enkephalin ratio (Presti and Lewis 2005) as well as reduced activity in the subthalamic nuclei (STN) of H, compared to N animals (Tanimura et al. 2010). There is also evidence that such striatal dysfunction underlies an association between deer mouse stereotypy and deficits in cognitive ability (Bechard et al. 2016; Tanimura et al. 2008). As alluded to earlier, rearing deer mice in EE cages improves procedural learning ability in jumpers and pattern runners, but not in backward somersaulters, implicating a possible role for different psychobiological mechanisms underlying unique forms of stereotypy. Importantly, these cognitive changes were linked to striatal rather than hippocampal mechanisms, the former normally associated with age-related deficits in learning ability (Frick and Fernandez 2003; Frick et al. 2003). In addition, in line with findings demonstrating increased EE-induced neuronal firing in the indirect SP pathway (Bechard et al. 2016), a positive correlation was found between the rate of stereotypy and cognitive rigidity in animals that benefited from EE. Thus, improvements in procedural learning ability occurred in parallel with striatally-mediated adaptations in expression of stereotypy. Although CSTC-associated deficits in learning ability have been demonstrated in OCD patients (Eng et al. 2015; Olley et al. 2007), these are not specific to OCD (Colomer et al. 2017; Lewis et al. 2007; Palminteri et al. 2009). While behavioral rigidity in jumpers and pattern runners responds to intervention, backward somersaulting may represent a behavior that is resistant to change (Tanimura et al. 2008).

Dopamine in deer mouse stereotypy

Although dopamine is known to be a pivotal role player in the CSTC circuitry, its role in the pathogenesis of deer mouse stereotypy remains to be clarified. Most parameters of dopaminergic signaling remain unaltered when comparing H vs. N mice, including striatal D1 and D2 receptor density (Powell et al. 1999) and regional brain levels of dopamine (Güldenpfennig et al. 2011; Powell et al. 1999; Presti et al. 2004), 3,4-dihydroxyphenylacetic acid (DOPAC), and homovanillic acid (HVA) (Güldenpfennig et al. 2011; Powell et al. 1999). Also, while systemic and intrastriatal administration of the D1/D2 receptor agonist, apomorphine, elicits typical rodent stereotypies, e.g. hyperlocomotion, gnawing and excessive grooming, it fails to exacerbate the characteristic deer mouse stereotypies, i.e. jumping, pattern running and backward somersaulting (Presti et al. 2002).

Further, deer mouse stereotypy seems to be unrelated to selective interference by D1 or D2 receptor modulators, which neither trigger nor exacerbate its expression (Korff et al. 2008; Presti et al. 2004). Korff et al. (2008) demonstrated attenuation of stereotypical behavior following administration of the selective D2 agonist, quinpirole (5 mg/kg/day × 4 days), while Presti et al. (2004) demonstrated no significant behavioral alteration. While Korff et al. (2008) used a 4-day intraperitoneal dosing and administration schedule, Presti et al. (2004) administered quinpirole intra-striatally at a dose of 5 μg/side over 60 s, which may account for the disparate results. However, findings from both these investigations seem inconsistent with the quinpirole compulsive-like checking model of OCD (Szechtman et al. 2001, 1998). Still, given important phenotypical differences between pharmacologically induced and spontaneous stereotypy in deer mice (Presti et al. 2002) and considering that neither the Korff et al. (2008), nor the Presti et al. (2004) investigations administered quinpirole for a duration comparable to that of the Szcechtman group (1998, 2001), both these models may be useful in understand the range of mechanisms that may underlie OC-like behavior (Szechtman et al. 2017). The dopaminergic system may play a role in processing context, salience, or reward which differ across these models. Indeed, recent work related to different OC phenotypes and its association with context related deficits in reward and punishment processing (Ferreira et al. 2017; Figee et al. 2011; Palminteri et al. 2012; Pinto et al. 2014), have to be considered.

Altered serotonergic signaling and deer mouse behavior

Deer mouse stereotypy is associated with significant evidence for serotonergic involvement, namely selective response to an SSRI but not an NRI, and significantly reduced striatal SERT density in H-animals (Korff et al. 2008; Wolmarans et al. 2013). Regional brain analysis of the cyclic adenosine monophosphate (cAMP) - phosphodiesterase type 4 (PDE4) cascade in deer mice may assist in learning more on receptor signaling in the CSTC in these animals, and possibly in OCD. The intensity of stereotypy expressed in deer mice is positively correlated to frontal-cortical cAMP concentrations, while being inversely related to PDE4 activity (Korff et al. 2009). Furthermore, chronic SSRI treatment attenuated this response (Korff et al. 2009). This not only supports frontal-cortical dysfunction in deer mouse stereotypy at a neurobiological level (Evans et al. 2004), but suggests involvement of the adenylate cyclase-cAMP-PDE4 cascade. The inverse correlation between PDE4 activity and the intensity of stereotypy may indicate that H-associated increased cAMP is related to increased post-synaptic 5HT1A adenylate cyclase (AS)-cAMP activity (Korff et al. 2009). This hypothesis is strengthened by the demonstration that stimulation of pre-synaptic 5HT1A/1B/1D auto-receptors induce perseverative locomotor paths (Shanahan et al. 2011; Yadin et al. 1991), while their desensitization is thought to mediate some of the ameliorative effects of the SSRIs (Blier et al. 1996). Desensitization of frontal-cortical 5HT1A/1B/D auto-receptors results in increased release of serotonin which in turn is associated with anti-compulsive effects (El Mansari and Blier 2006; Goddard et al. 2008). This may have relevance to earlier studies describing the attenuation of deer mouse stereotypy to meta-chlorophenylpiperazine (mCPP), a non-selective serotonergic agonist (Korff et al. 2008). Although some of the human literature on mCPP indicates an association with exacerbation of OC-symptoms (Aouizerate et al. 2005), mCPP attenuates quinpirole-induced compulsive checking (Tucci et al. 2013; Tucci et al. 2015) providing congruence across at least two animal models of OCD. However, that selective PDE4 inhibition with rolipram decreases methamphetamine-induced stereotypy (Iyo et al. 1995) hints at a possible causal role for disordered PDE4 activity in deer mouse stereotypy.

Disturbances in SERT are well-described in OCD (Hesse et al. 2005; Reimold et al. 2007; Zitterl et al. 2008), while this protein represents an important biological target for the SSRI group of drugs (El Mansari and Blier 2006; Fineberg and Craig 2007). To test the hypothesis that hyposerotonergic signaling underlies OC behavior, we determined frontal-cortical and striatal serotonin transporter (SERT) densities in H and N deer mice (Wolmarans et al. 2013). In line with the theory of behavioral opponency between dopamine and serotonin (Daw et al. 2002) and consistent with findings that deer mouse stereotypy involves a relative bias in favor of the direct SN pathway, we found a significant reduction in striatal but not frontal-cortical SERT density in H, compared to N animals (Wolmarans et al. 2013). This is consistent with clinical (Hesse et al. 2005) and pre-clinical (Vermeire et al. 2012) literature and supports the hypothesis that the biobehavioral effects of a relative increase in SN dopaminergic signaling in deer mice (Presti and Lewis 2005; Presti et al. 2003) are not sufficiently countered by serotonin (Daw et al. 2002).

Deer mouse stereotypy and oxidative stress

Recent clinical studies have indicated oxidative stress in OCD (Behl et al. 2010; Chakraborty et al. 2009a, b; Selek et al. 2008), as well as effective augmentation of standard SSRI treatment with the glutathione precursor, N-acetyl cysteine (NAC) (Camfield et al. 2011; Lafleur et al. 2006; Sayyah et al. 2010). Consistent with these findings, we have demonstrated that H deer mice present with a disturbed frontal-cortical redox balance, i.e. reduced activity of the glutathione system as evinced by diminished concentrations of reduced (GSH) and oxidized glutathione (GSSG) in frontal cortical circuits with these deficits correlated with stereotypy severity (Güldenpfennig et al. 2011). A positive correlation was found between the intensity of stereotypy and the glutathione redox balance (Güldenpfennig et al. 2011), possibly suggesting a relative protective upregulation of glutathione synthesis as a function of stereotypy. Similar findings were demonstrated in deer mice exposed to low levels of environmental toxins (Wu et al. 2009) perhaps suggesting that P. maniculatus bairdii is able to counter the effects of low to moderate degrees of oxidative stress.

Taken together, these findings possibly indicate that H behavior is associated with levels of oxidative stress akin to that of mild pathology. However, caution must be applied when drawing causal relationships between deer mouse stereotypy and OCD, and it is notable that oxidative stress has been found in a number of psychiatric disorders (Berk et al. 2011; Chauhan and Chauhan 2006; Sarandol et al. 2007; Wang et al. 2009). In addition, the response of OC symptoms to NAC augmentation may be related to its modulation of NMDA receptor signaling, rather than specific effects on oxidative stress (Lafleur et al. 2006). Further study of the association between deer mouse stereotypy and oxidative stress, and of the response of such stereotypy to anti-oxidants is needed.

Peromyscus maniculatus bairdii as a model of heterogeneous OC behavior

OCD is characterized by a narrow range of different symptoms. The most prevalent obsessions are concerns about contamination (55%), inappropriate aggressive and sexual thoughts (50% and 32% respectively), and concerns about symmetry and order (36%). The most common compulsions are ritualistic checking (80%), cleaning and decontamination rituals (46%) and counting (21%) (Abramowitz et al. 2010). Recently, we began to investigate whether P. maniculatus bairdii may express different OC-related phenotypes in addition to spontaneous stereotypy. We studied MB, previously proposed as a measure of compulsive activity (Albelda and Joel 2012) and nest building (NB) (Greene-Schloesser et al. 2011), which represents normal rodent activity but with between- and within-species variance (Jirkof 2014; Smithers 1983). As referred to earlier, high MB behavior was observed in 11% of our deer mouse cohort, independent of stereotypy or sex (Wolmarans et al. 2016a), while 30% of animals, again independent of stereotypy or sex, displayed persistent large NB behavior. Escitalopram had no effect on high MB, but reduced high NB (Wolmarans et al. 2016a, b). Further work is needed to determine whether such observations are analogous to clinical findings (including differential response of different symptom dimensions to stressors and to SSRIs). However, that large NB but not high MB responded to escitalopram, and that neither behavior was associated with a specific stereotypical cohort, suggests that NB and MB in the deer mouse reflect different underlying neurobiological mechanisms. Further, it is possible that such neurobiological differences may be species specific (Greene-Schloesser et al. 2011). While large NB also occurs naturally, mainly in response to environmental change (Jirkof 2014), the persistent and severe nature of large NB in some laboratory houseddeer mice only, suggests that in this sub-group, large NB may not be goal-directed or adaptive. This tentatively establishes a degree of face and predictive validity for large NB as reflecting a different, but also naturalistic OC-like phenotype in deer mice.

Conclusion

The deer mouse (Peromyscus maniculatus bairdii) offers an opportunity to study the neurobiology of OC-like behavior within a naturalistic framework. The species presents with stereotypies that are reminiscent of OCRD symptoms, while such stereotypies appear to share, at least in part, underlying psychobiological mechanisms and treatment response typical of OCD (Table 2). Given that deer mice display a narrow range of stereotypical behaviors, it is possible that some of these may be useful for studying specific symptom dimensions found in OCRDs. Further research on the genetic and epigenetic associations of stereotypies in the deer mouse model may be useful.

Table 2 Summary of key findings from the deer mouse model of OCD
Full size table