Cognitive dysfunction and depression in Parkinson’s disease: what can be learned from rodent models?

6-hydroxydopamine (6-OHDA)

In 1968, Ungerstedt showed that unilateral injection of 6-OHDA into the nigrostriatal pathway of rats resulted in degeneration of DAergic neurons in the SNpc, accompanied by marked postural bias and turning asymmetry towards the side of the lesion (Ungerstedt, 1968). More than four decades later, this model remains the most widely used rodent model of PD in terms of assessing motor functions, largely due to its simple automation and its reproducibility, and because its anatomical substrate(s) can be dissected by systematic variation of the injection site. Although attempts have been made in modelling neuropsychological symptoms in rats with a unilateral 6-OHDA lesion (Dowd & Dunnett, 2004, 2005; Perez et al., 2009; Eskow Jaunarajs et al., 2010), the accompanying severe motor impairment in this model is a strong confounding factor in many tests. In addition, in more complex tasks requiring higher cognitive functioning, the intact hemisphere may compensate for the loss of function on the lesioned side thus masking potential deficits. On the other hand, extensive bilateral 6-OHDA lesions produce profound debility which require special husbandry (Ungerstedt, 1971; Zigmond & Stricker, 1972), especially if the toxin is administered into the lateral ventricles or the medial forebrain bundle to achieve maximal levels of depletion. However, when 6-OHDA is injected in the striatum it results in more restricted partial lesions, which have proven to offer a feasible model for studying non-motor symptoms (Dunnett & Iversen, 1982; Turle-Lorenzo et al., 2006; De Leonibus et al., 2007, 2009; O’Neill & Brown, 2007; Branchi et al., 2008; Tadaiesky et al., 2008; Santiago et al., 2010). There are reports of pronounced recovery after partial 6-OHDA lesions due to the significant degree of plasticity in residual DAergic neurons (Zigmond et al., 1990); this can compromise use of this model in long-term studies. In addition, although partial bilateral lesions do not result in overt motor impairment as compared with a complete unilateral lesion, sensitive tests for motor function can reveal significant changes that may influence the performance in certain tasks for neuropsychological symptoms (Tillerson et al., 2002).

MPTP

1-Methy-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) is another toxin specific to DAergic neurons; it was first discovered through its capacity to induce severe PD-like symptoms in illegal drug users (Langston et al., 1983). MPTP also causes DAergic cell death after peripheral administration in primates (Burns et al., 1983), and this has subsequently proved the most widely used model of PD for studying potential therapeutics and interventions in large mammal species. However, its applicability in rodents has been more limited as only mice are sensitive to its toxic effect after peripheral injections due to the relative restriction of catecholamine metabolism to the monoamine oxidase (MAO)-B isoform of MAO in the rat (Cannon & Greenamyre, 2010), whereas in the mouse, primate and humans MPTP is converted to its toxic metabolite, MPP+, via an MAO-A-dependent pathway. In line with this, MPTP toxicity in both primates and rodents is not restricted to DAergic neurons but can also profoundly affect the serotonergic and noradrenergic systems, which also express the monoamine oxidase enzyme (Perez-Otano et al., 1991; Fukuda, 2001). Multiple intraperitoneal injections of MPTP in mice can reproduce the classical histological features of PD, including degeneration of DAergic neurons and α-synuclein inclusions as well as motor impairments, making it the most extensively used model in mouse (Meredith et al., 2008). In contrast, the motor impairment is not always evident in the acute MPTP model (Donnan et al., 1987; Arai et al., 1990), but this model has instead been used to study aspects of the neuropsychological symptoms seen in PD (Dluzen & Kreutzberg, 1993; Tanila et al., 1998; Vuckovic et al., 2008; Schintu et al., 2009; Deguil et al., 2010). Although rats are conspicuously insensitive to systemic injections of MPTP, intracranial or intranasal administration of the toxin results in DAergic degeneration accompanied by motor impairment as well as the appearance of neuropsychiatric impairments (Ferro et al., 2005; Prediger et al., 2009).). Moreover, rats are sensitive to the MAO-independent active toxic metabolite of MPTP, MPP+, when it is centrally injected into the SNpc (Sirinathsinghji et al., 1988). One concern about use of MPTP as an experimental toxin is that it can prove equally harmful to experimenters handling the compound (Langston et al., 1983), so that it must be handled within a highly controlled safety environment in any experimental use.

Rotenone and LPS

Although it has been hypothesized before, there is an increasing amount of recent evidence that environmental toxins such as pesticides may be associated with an increased risk of PD (Jenner, 2001; Tanner et al., 2011; Wang et al., 2011). For example, rotenone is a pesticide with similar mechanisms as MPTP and has therefore received particular interest as a potential model of PD. However, peripheral rotenone administration in rats has shown a strong variability in phenotype, ranging from pronounced nigral cell loss and α-synuclein accumulation (Cannon et al., 2009) to minor loss of tyrosine hydroxylase (TH) reactivity in striatum and nonspecific neuropathology resulting in variable motor impairments (Fleming et al., 2004; Zhu et al., 2004; Klein et al., 2011).

Lipopolysaccharide (LPS) is a bacterial endotoxin that causes glial activation and subsequent immune reaction when it is injected into the SNpc. This results in an inflammatory DA neurodegeneration to a variable extent depending on the dosing regimen (Dutta et al., 2008), but the accompanying motor impairment has not been reproducible between studies (Liu et al., 2008; Choi et al., 2009; Ariza et al., 2010). Only one study has compared the effect of bilateral 6-OHDA, MPTP, rotenone and LPS lesion of the SNpc in terms of neuropsychological function (Santiago et al., 2010), and there are a couple of studies in which depressive-like behaviours have been assessed after systemic LPS injections in mice, although these mice lack specific PD pathology (Zhu et al., 2010; Dobos et al., 2012).

Genetic models of PD

Genetic models of PD, based on different genetic mutations and risk factors, have been developed over the last decade to create models that would more closely mimic the progressive degeneration of DAergic neurons as well as the extranigral pathology evident in the human disorder. The first mutation identified in familial PD was in the gene encoding the vesicular protein α-synuclein (Polymeropoulos et al., 1997; Kruger et al., 1998) and the subsequent discovery of α-synuclein inclusions in the majority of all PD patients, as a component of the key pathological feature of the disease also indicated an involvement of α-synuclein in sporadic PD (Spillantini et al., 1997). Several lines of α-synuclein-overexpressing mice have been engineered, using both different promoters and transgenes, and most recapitulate some, although not all, aspects of PD (Magen & Chesselet, 2010). These lines of mice can be categorized according to their α-synuclein pathology and expression pattern: mice with prominent cortical and hippocampal expression recapitulating dementia with Lewy Bodies; and mice with predominant cortical and subcortical limbic α-synuclein pathology, which is more similar to idiopathic PD (Magen & Chesselet, 2011).

Subsequent mutations in several other genes have been identified in familial PD; these include DJ-1 (Bonifati et al., 2002), Parkin (Kitada et al., 1998) and PTEN-induced kinase 1 (PINK1; Valente et al., 2002), and the effects of each of these genes have been modelled in mice using different genetic manipulations. The autosomal-recessive mutation in the Parkin, PINK1 and DJ-1 gene is usually modelled using knockout strategies. Common features of these mouse lines are the lack of overt nigrostriatal pathology and the fact that they often exhibit only mild motor impairments. However, they do exhibit chronic alterations in DA release and synaptic function in the striatum, and are therefore useful tools for investigating early preclinical stages of PD. As certain non-motor symptoms often precede the motor impairments in PD, these mice models could also be advantageous for modelling of symptoms not related to motor deficits. However, only a couple of studies have addressed cognitive and depressive-like behaviour in these lines of mice (Zhu et al., 2007; Pham et al., 2010).

In addition to specific disease-causing mutations, genetic risk factors for PD have also been probed as a rationale for engineering transgenic mice. For example, loss of vesicular monoamine transporter 2 (VMAT2) immunoreactivity is evident in post-mortem PD brains (Miller et al., 1999). Accordingly, a model based on depletion of VMAT2 has been used to model several non-motor symptoms (Caudle et al., 2007; Taylor et al., 2009). Further examples are provided by the transcription factors PITX3 and Nurr1, which, as well as being implicated in various forms of PD, are important for the differentiation and survival of midbrain DAergic neurons during development (Le et al., 2003; Fuchs et al., 2009; Bergman et al., 2010). However, neither the Pitx3-deficient aphakia nor the Nurr1-deficient mouse models have been thoroughly studied in terms of neuropsychological function, but a few studies have explored other aspects of non-motor symptoms in these mouse lines (Ardayfio et al., 2008; Vuillermot et al., 2011; Glasl et al., 2012).

The alternative approach is to insert mutant forms of dominant genes. For example, transgenic rat models of PD involving extra copies of, or mutations in, the α-synuclein and leucine-rich repeat serine/threonine kinase 2 (LRRK2) genes have recently been developed (Lelan et al., 2011; Zhou et al., 2011). These rat models could be more advantageous than transgenic mice for tests of cognitive and affective aspects in PD considering the well established behavioural paradigms existing for rats. There is no evidence of nigrostriatal pathology or motor impairments in these two transgenic rat models, suggesting a more preclinical phenotype (Lelan et al., 2011; Zhou et al., 2011), but their use as models in preclinical PD research needs to be further evaluated.

Tests of executive function

The dysexecutive syndrome in PD patients is characterized by deficits in attention, working memory, rule learning, reversal and decision-making (Shallice & Burgess, 1996; Kehagia et al., 2010), which are all features that can be dissected and analysed in tasks suitable for rodents.

Lateralized stimulus-response tasks

There is considerable evidence that the striatum and DA are important in stimulus–response performance in a lateralized choice–reaction time task in the nine-hole operant chambers, as nigrostriatal lesions profoundly impair the performance in this task (Carli et al., 1985; Dowd & Dunnett, 2004, 2005). In the task, the rats are trained to hold their nose in the central stimulus hole before responding to a brief stimulus of light in a lateral hole. As well as measuring accurate target decision (response accuracy), measures of latency provide some overall information about motor performance and motivational status.

6-OHDA

Although bilateral striatal lesions using 6-OHDA often result in more partial lesions, ranging from 40 to 70% in different studies, they are often accompanied by some degree of motor impairment, which may be a confounding factor in many cognitive tests. Tadaiesky et al. (2008) showed impaired working memory in a cued version of the Morris water maze in rats with a bilateral 6-OHDA lesion of the dorsal striatum. These rats displayed a 60% neuronal loss in the SNpc, accompanied by reduction in striatal levels of serotonin (5-hydroxytryptamine; 5-HT) and NA (Tadaiesky et al., 2008). Ferro et al. (2005) demonstrated similar findings, in which a bilateral nigral lesion using either MPTP or 6-OHDA resulted in impaired stimulus–response learning and spatial working memory deficits also assessed in the Morris water maze. In contrast, another study did not see any deficits in spatial working memory after a bilateral 6-OHDA lesion (Branchi et al., 2008). In both studies demonstrating working-memory deficits, motor impairment was also evident and this may have influenced the latency to find the platform in the water maze. This suggests that the observed deficit may also be related to motor function and not just to impaired working memory. In addition, it has been demonstrated that bilateral 6-OHDA lesions result in impaired navigation in water-maze tasks (Whishaw et al., 1987), and this may also have had a confounding effect on the results.

Similar results have also been seen in mice: a 70% reduction in striatal DA levels following striatal bilateral 6-OHDA lesions in CD1 mice was associated with impaired discrimination of a visuospatial change in an object–place association task (De Leonibus et al., 2007), a deficit which was ameliorated by administration of a metabotropic glutamate receptor 5 antagonist (De Leonibus et al., 2009). In line with these findings, Mura & Feldon (2003) demonstrated that rats with a bilateral 6-OHDA lesion were impaired in the selection and maintenance of behavioral strategies in a spatial navigation task using the Morris water maze.

Several studies have dissected out certain aspects of the dysexecutive function evident in PD patients and modelled them in rodent models of PD. For example, attentional deficits have been demonstrated in rats with a bilateral 6-OHDA lesion of the dorsal striatum using the five-choice serial reaction time task (Baunez & Robbins, 1999). In line with this finding, two other studies have shown similar results using a related reaction time task (Turle-Lorenzo et al., 2006; Courtiere et al., 2011) in which the deficit was reversed by L-DOPA and DA D2/D3 agonist treatment (Turle-Lorenzo et al., 2006). Reversal learning is another important feature in flexible adaptation and shifting of response patterns, and has been shown to be impaired in rats with bilateral striatal lesions assessed in both a water T-maze (Haik et al., 2008) and in a simple discrimination task using digging bowls (O’Neill & Brown, 2007).

By pre-training rats on a lateralized choice–reaction time task, Carli et al. (1985) demonstrated that unilateral DA depletion in striatum profoundly impaired accurate responding to the contralateral side as well as increasing the latency to react to the stimulus and withdraw from the center hole (reaction time). In contrast, both the accuracy and the identical withdrawal movement were unchanged on the intact side, as was the latency to execute the lateral movement (Carli et al., 1985). It was suggested that this is a form of intentional neglect and demonstrates that DA is related to the initiation of a movement in contralateral space (Carli et al., 1985, 1989). Similar deficits in contralateral accuracy have also been demonstrated by Darbaky et al. (2003) after nigral 6-OHDA lesions. In a further evaluation of the role of DA in a similar reaction-time task, Dowd & Dunnett (2004, 2005) demonstrated that the lesion deficit in accuracy and reaction time on the contralateral side of the lesion was not apparent on the first couple of days of post-operative testing but rather required several days of testing before emerging. This demonstrated that although DA-depleted rats were motorically impaired, they were capable of performing as well as controls on the contralateral side for the first few days of testing. The authors therefore hypothesized that the core of the lesion-induced impairment in accuracy was not purely motor but could rather be explained as a disruption of the reward-related feedback pathway, which impaired the rats’ ability to maintain the association between the stimulus and the outcome (Dowd & Dunnett, 2007).

MPTP

Studies focusing on cognitive impairment in PD have used both the acute MPTP mouse model in which the mice receive four successive injections of the toxin, and the chronic model in which the mice are injected over 5 weeks at 3.5-day intervals (Meredith et al., 2008). Deguil and collaborators demonstrated impaired memory functions after chronic MPTP treatment in C57Bl/6 mice, both in spatial working memory and in a reference memory version of the Morris water maze task (Deguil et al., 2010). These mice displayed a 50% reduction in the number of TH-positive neurons in the SNpc as well as altered protein synthesis regulation in several cerebral regions (Deguil et al., 2010). The potential motor impairment was not investigated in this study but two other studies using a similar dosing regimen of MPTP showed a pronounced motor deficit using the beam walk, rotarod and pole tests (Pothakos et al., 2009; Luchtman et al., 2012). Both these studies also demonstrated impaired procedural memory (Luchtman et al., 2012) and working memory (Pothakos et al., 2009), assessed in the Morris water maze task. A very recent study showed that impaired working memory, assessed in a novel-object-recognition task (Levin et al., 2011) in an acute MPTP mouse model, was associated with reduced calcium/calmodulin-dependent protein kinase II activity in the hippocampus (Moriguchi et al., 2012). In line with this finding, Tanila et al. (1998) showed impaired performance in the delayed alternation task in a T-maze; this was moderately ameliorated by an acute injection of dexmedetomidine, an α₂-receptor agonist.

Alterations in cognitive functions have also been demonstrated in rats when the MPTP has been administered either directly into SNpc or intranasally. Braga and co-workers showed that MPTP-lesioned rats were impaired in both the retention and acquisition of a delayed alternation task for working memory in a Y-maze (Braga et al., 2005). In addition, a couple of studies have also demonstrated impaired working memory in the novel-object recognition task after bilateral intranigral injection of MPTP (Wang et al., 2009; Sy et al., 2010; Ho et al., 2011; Hsieh et al., 2012); the impairment could be attenuated by an acute MK-801 (NMDA antagonist) injection (Hsieh et al., 2012). Intranasal administration of MPTP has also been effective in causing impaired working memory assessed in a social recognition task, and these memory deficits could be alleviated by lithium or valproate treatment (Castro et al., 2012).

Genetic models

Apart from different types of α-synuclein-overexpressing mice, only a couple of studies have used other genetic mouse models of familial forms of PD, namely Parkin and DJ-1, to study cognitive impairment (Zhu et al., 2007; Pham et al., 2010). Parkin-deficient mice showed a modest impairment in spatial learning in the Morris water maze without any apparent motor deficits, assessed as unaltered locomotor activity in the open-field test (Zhu et al., 2007). It was suggested that the deficit in spatial learning stemmed from an altered DA metabolism in the striatum (Zhu et al., 2007). In DJ-1-knockout mice, the male DJ-1^−/− mice displayed a memory deficit in the object recognition task as well as a modest reduction in the number of TH-positive neurons in the ventral tegmental area (Pham et al., 2010).

Several transgenic mouse lines overexpressing either the wildtype or mutated human α-synuclein have also been used to evaluate potential cognitive impairment: mice overexpressing wildtype α-synuclein under the Thy1 promoter (Rockenstein et al., 2002) showed fewer alternations in a Y-maze and decreased novel-object recognition as well as deficits in reversal learning, evident from ∼4 to 12 months of age (Fleming et al., 2008; Magen & Chesselet, 2011). The time course of the impairments was paralleled by changes in extracellular levels of both DA and acetylcholine (Lam et al., 2011; Magen & Chesselet, 2011), providing possible mechanisms for the deficits seen in these tasks. Similar findings were obtained in a model overexpressing the mutant A30P form of α-synuclein under the same Thy1 promoter. These mice displayed an age-dependent decline in reference memory assessed in the Morris water maze without any reduction in swimming speed or distance travelled (Freichel et al., 2007). In line with these findings, mice overexpressing wildtype α-synuclein under the platelet-derived growth factor β promoter displayed both learning and reference memory impairment associated with cortical and subcortical accumulation of α-synuclein (Masliah et al., 2011). Nuber et al. (2008) created a conditional mouse model of α-synuclein by using the tet-off conditional expression system and these mice were shown to express high levels of human wildtype α-synuclein in midbrain and forebrain regions. The α-synuclein accumulation resulted in modest nigral and hippocampal neuropathology, which was associated with a progressive decline in motor performance as well as a modest impairment in reference memory, the development of which was inhibited following tetracycline administration (Nuber et al., 2008).

6-OHDA

Depressive-like behaviour has been demonstrated after both unilateral and bilateral 6-OHDA lesions, although loss of more specific motor functions influencing the neuropsychiatric findings was rarely examined. A study by Tadaiesky et al. (2008) was one of the first to characterized emotional deficits after bilateral 6-OHDA injections into the dorsal striatum of rats, and they reported decreased sucrose consumption, increased immobility time in the forced-swim test and a reduced percentage of entries into the open arms in the elevated-plus maze. In another study using a similar type of 6-OHDA lesion in rats, Branchi et al. (2008) also showed evidence of increased immobility time in the forced-swim test but, in contrast to the former study, decreased anxiety-like behaviour and no changes in the preference for sucrose. Bilateral nigral injections of 6-OHDA in rats also resulted, in the forced-swim test, in increased immobility time which was strongly correlated with hippocampal reduction in levels of DA, 5-HT and NA (Santiago et al., 2010). Surprisingly, considering the potential contralateral compensation, depressive-like behaviour has also been reported after unilateral 6-OHDA lesions of the medial forebrain bundle. These rats showed increased immobility time in the forced-swim test and increased anxiety-like behaviour, although the motor impairment that inevitably follows that type of lesion may well have influenced these results (Eskow Jaunarajs et al., 2010; Zhang et al., 2011).

MPTP

Depressive- and anxiety-related symptoms have also been demonstrated in mice with acute MPTP lesions (20 mg/kg in four injections 2 h apart) but, as with the motor impairment data for this dose regime protocol (Donnan et al., 1987; Arai et al., 1990), these neuropsychiatric findings have been inconsistent between studies. Mori and co-workers demonstrated increased immobility in the tail-suspension test (mouse version of the forced-swim test; Steru et al., 1985) and acute injections of L-DOPA as well as a D2 agonist ameliorated this increase in immobility time (Mori et al., 2005). In contrast, two other studies using the same dosing regimen of MPTP did not see any depressive-like behaviour assessed in terms of motivational change as decreased sucrose consumption (Vuckovic et al., 2008; Gorton et al., 2010).

By circumventing the periphery, bilateral nigral injections of MPTP in rats have been shown to increase immobility time in the forced-swim test as well as to decrease consumption of sucrose. These rats also showed a decrease in locomotor activity, but loss of more fine motor function was never accounted for (Santiago et al., 2010). Several papers have also demonstrated increased anxiety in rats after intranigral MPTP injection, assessed as less time in the open arms compared to controls in the elevated-plus maze (Wang et al., 2009; Sy et al., 2010; Ho et al., 2011).

Rotenone

Bilateral infusion of rotenone into the SNpc of rats resulted in depressive-like behaviour in the forced-swim test and decreased sucrose preference, accompanied by reductions in 5-HT levels in the hippocampus (Santiago et al., 2010). Again, loss of more fine motor function was never accounted for as a confounding factor in the forced-swim test (Santiago et al., 2010). As a cautionary note, the use of rotenone has been debated due to nonspecific toxicity and motor impairments unrelated to nigral pathology (Fleming et al., 2004; Klein et al., 2011).

Genetic models

Few of the studies using genetic mouse models based on the familiar genetic mutations associated with PD, such as PINK1 and DJ1, have been able to demonstrate depressive-like symptoms or anxiety-related behaviour. In contrast, both neuropsychiatric aspects have been demonstrated in the VMAT2-deficient mouse model (Guillot & Miller, 2009). In these mice, the expression of VMAT2 was reduced by 95% and accompanied by a moderate loss of nigrostriatal neurons, α-synuclein accumulation and loss of DA, NA and 5-HT in striatum, hippocampus and cortex (Caudle et al., 2007; Taylor et al., 2009). These mice showed an age-dependent development of depressive-like behaviour assessed in the forced-swim test (Taylor et al., 2009), which tightly followed the progressive loss of DA, NA and 5-HT (Caudle et al., 2007). These mice also showed an increased anxiety-like behaviour but, in contrast to the depressive-like behaviour, this phenotype was only evident at an early age (Taylor et al., 2009).

Mice overexpressing human wildtype α-synuclein under the Thy-1 promoter (Thy1-αSyn) have widespread overexpression of α-synuclein throughout the brain including cortex, locus coeruleus, SNpc and brainstem (Rockenstein et al., 2002). These mice showed a lower anxiety score than their wildtype littermates (Mulligan et al., 2008), on the assumption that lower anxiety was related to their increase in locomotor activity (Lam et al., 2011). A similar result was found in mice overexpressing the A53T mutant form of α-synuclein, with an increased time spent in the open arms in the elevated-plus maze test (George et al., 2008) and neuropathology affecting selectively the noradrenergic over the DAergic system in the striatum (Sotiriou et al., 2010).

One study has investigated potential non-motor symptoms in Parkin-deficient mice (Zhu et al., 2007). These mice displayed an intact DAergic system but an increased DA turnover in the striatum as adults, accompanied by increased anxiety assessed in a light–dark exploration test paradigm (Zhu et al., 2007).