Gene–environment interactions: key to unraveling the mystery of Parkinson’s disease

2.1 Familial aggregation and twin studies of PD

Approximately 10–30% of patients with PD report a first-degree relative with parkinsonism (Farrer, 2006; Rocca et al., 2004; Sveinbjornsdottir et al., 2000). Such familial aggregation of PD used to be considered as a consequence of shared exposure to a common environment. PD was traditionally considered a non–genetic disorder of ‘sporadic’ origin, since heritability estimates in twin studies revealed conflicting results that were heavily weighted against a genetic basis for this disease. For instance, multiple studies show low concordance in both monozygotic and dizygotic twins (Duvoisin et al., 1981; Marttila et al., 1988; Tanner, 2003; Ward et al., 1983; Wirdefeldt et al., 2004). However, a higher risk ratio for concordance in monozygotic versus dizygotic twins was reported in early-onset PD patients (Tanner et al., 1999). Striatal ¹⁸F-dopa positron emission tomography (PET) scan, a presymptomatic detection marker of PD, show three times higher concordance rate for PD in monozygotic twins (55%) than in dizygotic twins (18%), implying a significant genetic contribution (Piccini et al., 1999). By minimizing methodological problems in ascertainment and recall bias, twin studies can achieve more consistent results and advance our understanding of genetic influence on PD.

2.2 Monogenically inherited PD

Recent advances in molecular genetics have unveiled that 5–20% of PD patients from different geographical areas have monogenic forms of PD. Numerous genetic defects (~82% simple mutations and ~18% copy number variations) in more than 13 loci and 9 genes have been identified in inherited PD patients and families (Dawson et al., 2010; Lesage and Brice, 2009; Nuytemans et al., 2010). To date, a causal role for PD has been confirmed with five genes: SNCA (PARK1) encoding α-synuclein, Parkin (PARK2) encoding Parkin, PINK1 (PARK6) encoding PTEN-induced putative kinase 1 (PINK1), DJ-1 (PARK7) encoding DJ-1, and LRRK2 (PARK8) encoding Leucine-rich repeat kinase 2 (LRRK2). Extensive mutation screening of these five causative genes has discovered approximately 330 confirmed or possible pathogenic mutations in over 1900 families (Nuytemans et al., 2010).

2.3 Dominantly inherited mutations

2.4 Recessively inherited mutations

Loss-of-function mutations in genes encoding Parkin, PINK1, and DJ-1 cause early-onset autosomal recessive PD. All these forms of PD are relatively rare and are L-DOPA responsive. Simple mutations (e.g. missense, nonsense and splice site mutations), exonic rearrangements, small insertions and deletions, as well as copy number variations of the promoter region and exon(s) have been identified in these recessive genes. While homozygous or compound heterozygous mutations in these recessive genes have a complete penetrance, the pathogenic effect of a single heterozygous gene defect in these genes is far less clear. The higher prevalence of the heterozygous variants in PD patients than in control individuals (Lesage et al., 2008b; Marongiu et al., 2008) implies that a heterozygous recessive mutation may predispose the carriers to PD. To date, no neuropathological studies have been reported for patients with PINK1 and DJ-1 mutations.

2.5 Crosstalk among PD-associated genetic pathways

Striking similarities in the phenotypes among familial PD patients carrying various PD-related gene mutants and sporadic PD patients strongly suggest shared molecular pathways and genetic interactions in the pathogenesis of both familial and sporadic PD (Figure 1). In support of this notion, PINK1, Parkin and DJ-1 have been reported to functionally interact to maintain mitochondrial integrity and functionality and to protect against adverse effects of multiple stressors (Deng et al., 2008a; Exner et al., 2007; Narendra et al., 2010; Park et al., 2006; Poole et al., 2008; Vives-Bauza et al., 2010b) (Figure 1). Transgenic expression of Parkin attenuates PINK1 loss-of-function phenotypes in flies, whereas transgenic expression of PINK1 had no effect on Parkin loss-of-function phenotypes (Clark et al., 2006; Park et al., 2006). PINK1 mediates the translocation of Parkin to mitochondria, which directs mitochondria to autophagy (Vives-Bauza et al., 2010a). Furthermore, PINK1 also interacts with mitochondrial serine protease HtrA2 (PARK13); PINK1-dependent phosphorylation of HtrA2 might modulate the proteolytic activity of HtrA2, thereby contributing to an increased resistance of cells to mitochondrial stress (Plun-Favreau et al., 2007). Coexpression of PINK1, Parkin or DJ-1, but not PD-associated mutants of these genes, rescues cells from mitochondrial fusion elicited by α-synuclein overexpression (Kamp et al., 2010). It has been proposed that the functional interaction among different PD-associated genes does not necessarily result from a direct physical interaction among corresponding proteins; alternatively, the convergence at the level of mitochondrial integrity of different pathways appears, at least partially, responsible for the overall outcome of such interactions. Moreover, DJ-1, Parkin, and PINK1 have been reported to form a ubiquitin E3 ligase complex promoting unfolded protein degradation (Xiong et al., 2009). In double transgenic Drosophila, the coexpression of PINK1, DJ-1 or Parkin alters various phenotypes of the eye and dopaminergic system elicited by LRRK2 expression in a complex fashion (Venderova et al., 2009). In α-synuclein/LRRK2 double mutant mice, overexpression of LRRK2 promoted and knockout of LRRK2 suppressed the abnormal aggregation and somatic accumulation of α-synuclein as well as the progressive loss of cortical and striatal neurons of the forebrain induced by overexpression of Ala53Thr α-synuclein (Lin et al., 2009). These findings reveal an important functional interplay between LRRK2 and α-synuclein and suggest these two dominant PD gene products may share common pathogenic mechanisms. Certainly, further investigation of detailed molecular cascades of genetic interactions will glean many fresh insights into PD pathogenesis (Figure 1).

2.6 Genome-wide association study (GWAS) of PD

Multiple GWASs conducted in PD patients and controls have independently identified polymorphisms in the SNCA (4q22), MAPT (17q21.1) and LRRK2 (12q12) regions as risk factors for PD, confirming the known associations of these genes with PD (Edwards et al., 2010; Maraganore et al., 2005; Pankratz et al., 2009; Satake et al., 2009; Simon-Sanchez et al., 2009; Tan et al., 2010). A GWAS and two replication studies conducted in a total of 2,011 PD patients and 18,381 controls from Japan have detected two new risk loci: 1q32 (PARK16) and 4p15 (BST1) (Satake et al., 2009). While SNCA polymorphisms as a major risk locus across populations have been confirmed by several GWASs, BST1 and MAPT as risk loci for PD show population differences (Satake et al., 2009; Simon-Sanchez et al., 2009). Thus, an unequivocal role for common genetic variants in the etiology of typical PD implies population-specific genetic heterogeneity in PD. Additionally, a recent GWAS of 2,000 PD patients and 1,986 controls from the NeuroGenetics Research Consortium Common has identified an association of genetic variation in the HLA region with late-onset sporadic PD (Hamza et al., 2010), which will be discussed in detail later (see Section 4.3). Certainly, GWASs in a population sample may prove to be productive in identifying association of genetic variability with disease risk. However, association does not imply causation and can not be used to predict disease risk of an individual for a certain disease. The pathogenic effects of specific mutations and polymorphisms must be investigated.

3.1 Overexpression of α-synuclein

A variety of transgenic animals overexpressing either wildtype or mutant human α-synuclein variably recapitulate neurodegeneration and α-synuclein pathology (e.g. aggregation, truncation, phosphorylation and ubiquitination) in different brain regions (Farrer, 2006). For example, α-synuclein transgenic Drosophila (Feany and Bender, 2000) and virus-infected mice, rats, and primates (Kirik et al., 2002; Klein et al., 2002; Lo Bianco et al., 2002; Yamada et al., 2004) revealed loss of dopamine neurons and formation of Lewy body-like inclusions. However, of the many α-synuclein transgenic mouse models, a few showed alterations in the nigrostriatal dopamine system (Masliah et al., 2000; Richfield et al., 2002; Rockenstein et al., 2002; Tofaris et al., 2006), and only very few of them developed overt loss of dopamine neurons in the SN. For instance, when human wildtype or doubly mutated α-synuclein (A53T and A30P) was expressed in mice by using rat tyrosine hydroxylase promoter, the number of nigral dopamine neurons declines with age (Thiruchelvam et al., 2004). Although the reason for the lack of overt loss of nigral dopamine neurons in most human α-synuclein transgenic mice is unclear, we speculate the following reasons may be partially responsible for such a phenomenon. Firstly, while A53T mutant α-synuclein is pathogenic in humans, the threonine at position 53 is the wildtype amino acid of rodent α-synuclein; this difference and additional five amino acid differences between mouse and human α-synuclein may affect the oligomerization and fibrillogenesis as well as consequent pathogenicity of α-synuclein. Secondly, the coexistence of human α-synuclein and mouse α-synuclein is postulated to influence the phenotypes of human α-synuclein transgenic mice (Cabin et al., 2005; Rochet et al., 2000). Thirdly, abnormalities in genetically engineered mice carrying PD-related genetic defects (e.g. mutant SNCA and LRRK2), probably at the very best only mimic the earliest dysfunction of the nigrostriatal dopaminergic system and the initial preclinical stage of PD. Fourthly, the short lifespan of mice (< 3 years) makes it strictly difficult, if not impossible, to replicate the decades-long neurodegenerative process of PD patients in these animals. Fifthly, the presence of neuromelanin in dopamine neurons in humans and non-human primates, but not in rodents, may render primates more vulnerable than rodents to diverse PD-associated insults (e.g. genetic mutations and environmental toxins). Sixthly, one single genetic defect is not sufficient to produce overt loss of nigral dopamine neurons, and an environmental trigger may be required for PD neurodegeneration. These potential reasons except the first two may be also partially responsible for the paucity of the full-spectrum of PD in other genetic mouse models.

3.2 Overexpression of LRRK2

Drosophila and C. elegans overexpressing mutant LRRK2 display death of dopamine neurons and increased sensitivity to rotenone and paraquat (Liu et al., 2008; Ng et al., 2009; Saha et al., 2009; Venderova et al., 2009). However, the short lifespan and the lack of endogenous α-synuclein expression render it potentially problematic to study mechanisms of LRRK2-mediated neurodegeneration in Drosophila and C. elegans, since a prominent feature of PD caused by LRRK2 mutations is α-synuclein pathology. Bacterial artificial chromosome (BAC) transgenic mice overexpressing Arg1441Gly mutant LRRK2 show diminished striatal dopamine, axonal pathology of nigrostriatal dopamine pathway, and age-dependent, levodopa-responsive motor impairment (Li et al., 2009). BAC transgenic mice overexpressing Gly2019Ser mutant LRRK2 exhibited an age-dependent striatal dopamine deficiency but not degeneration of nigrostriatal dopamine neurons or terminals at 12 months of age (Li et al., 2010). Conditional expression of wildtype and Gly2019Ser mutant LRRK2 in dopamine neurons or knockin of the Arg1441Cys mutant in mice failed to produce dopaminergic neurodegeneration (Lin et al., 2009; Tong et al., 2009). So far, the most consistent phenotypes in various LRRK2 transgenic mice are abnormalities in the nigrostriatal dopamine neurotransmission. These abnormalities probably indicate early dysfunction of the nigrostriatal dopamine system. However, the lack of substantial neurodegeneration in LRRK2 transgenic mice limits their utility in testing neuroprotective agents. Interestingly, co-overexpression of either wildtype or Gly2019Ser LRRK2 with A53T α-synuclein promoted abnormal somatic accumulation and aggregation formation of α-synuclein and dramatically accelerated neurodegeneration in cortical and striatal neurons of the forebrain in the double-transgenic mice (Lin et al., 2009). However, the expression of the transgene in less than 5 percent of midbrain dopaminergic neurons and the occurrence of forebrain neurodegeneration in these double-transgenic mice limit their relevance to PD pathogenesis. Crossbreeding of inducible dopamine neuron-specific α-synuclein transgenic mice with LRRK2 mutant mice may provide an innovative and valuable mouse model to investigate how LRRK2 interacts with α-synuclein in the modulation of dopaminergic function and degeneration (Tong and Shen, 2009).

3.3 Knockout of α-synuclein or LRRK2

α-synuclein knockout mice exhibit a subtle electrophysiological phenotype but no neurodegenerative pathology (Perez and Hastings, 2004). Knockout of LRRK2 in mice is not neurotoxic per se and does not exacerbate MPTP-elicited dopaminergic neurodegeneration (Andres-Mateos et al., 2009). Deletion of LRRK homolog in Drosophila causes dopaminergic impairment and locomotive deficit (Lee et al., 2007). Knockdown of LRRK homolog in C. elegans does not affect on baseline viability but enhances their vulnerability to rotenone toxicity (Saha et al., 2009).

3.4 Knockout of DJ-1

Drosophila possesses two orthologs of DJ-1: DJ-1a and DJ-1b. Knockout of both DJ-1 homologs in Drosophila does not alter fertility, lifespan, or dopamine neurons but sensitizes Drosophila to PD-associated environmental toxins, paraquat and rotenone (Meulener et al., 2005). Selective knockdown of DJ-1 by RNAi in dopamine neurons leads to age-dependent loss of dopamine neurons in the dorsomedial cluster in Drosophila (Yang et al., 2005). DJ-1 knockout mice exhibit abnormal dopamine neurotransmission in the nigrostriatal pathway and motor deficits but no observable neurodegeneration (Andres-Mateos et al., 2007; Chen et al., 2005b; Goldberg et al., 2005; Kim et al., 2005).

3.5 Knockout of Parkin or overexpression of mutant Parkin

Parkin-knockout Drosophila reveals male sterility, reduced lifespan, and severe defects in flight and climbing abilities. These phenotypes may be associated with mitochondrial defects and consequent apoptotic cell death in muscle and sperm. Aged Parkin-null flies show dopamine-responsive locomotion deficits and degeneration of a subset of dopamine neurons (Greene et al., 2003; Whitworth et al., 2005). Parkin-null mice display either subtle alterations in the nigrostriatal dopamine circuit or no overt dopaminergic or behavioral abnormalities (Goldberg et al., 2003; Itier et al., 2003; Perez and Palmiter, 2005). Interestingly, overexpression of mutant human Parkin in Drosophila and mice causes progressive degeneration of dopamine neurons (Lu et al., 2009; Sang et al., 2007; Wang et al., 2007).

3.6 Knockout of PINK1

PINK1-null Drosophila and Parkin-null Drosophila share marked phenotypic similarities including male sterility, behavioral deficits, fewer dopamine neurons, mitochondrial defects, and apoptosis in muscle (Clark et al., 2006; Gautier et al., 2008; Park et al., 2006). In the Drosophila loss-of-function models of PINK1 and Parkin, the dopaminergic system is mildly affected, while the most dramatic phenotypes occur outside the nervous system. PINK1 knockout mice show subtle deficits in nigrostriatal dopamine neurotransmission (Gautier et al., 2008; Gispert et al., 2009; Kitada et al., 2007). Because deficiency in Parkin, PINK1, or DJ-1 in mice does not cause substantial nigrostriatal pathology, these mice at the very best only mimic the earliest dysfunction of the nigrostriatal dopaminergic system. Thus, usage of current autosomal-recessive animal models to study mechanisms of dopamine neurodegeneration and to screen neuroprotective agents is potentially problematic or impossible. Overall, gene-based Drosophila and C. elegans models offer powerful tools to rapidly screen genetic and pharmacologic modifiers of known PD-associated genes, but the challenge is to verify the evolutionary conservation of these potential modifiers in human PD.

4.1 Pesticides and other environmental toxins

Although modern genetics uncovers important genetics basis in familial PD, it is worth emphasizing that only about 10%–20% of PD cases result from genetic causes. The majority of PD cases are sporadic and idiopathic. The accidental use of 1-methyl-4-phenyl-1,2,3-tetrahydropyridine (MPTP), a by-product of illicit heroin synthesis, leads to a Parkinson syndrome that is clinically indistinguishable from PD (Langston et al., 1983). This milestone finding has stimulated the search for environmental factors as potential causes of PD. The structural similarity between paraquat (1,1′-dimethyl-4,4′-bypyridinium), a common herbicide, and 1-methyl-4-phenylpyridinium ion (MPP⁺), the active metabolite of MPTP, prompted speculation that paraquat might be a dopaminergic neurotoxicant and exposure to paraquat may be related to the development of PD. Epidemiological studies have reported that exposure to pesticides including paraquat correlates with increased incidence of PD (Ascherio et al., 2006; Liou et al., 1997; Rajput and Uitti, 1987). Case reports have also associated development of Parkinsonism with exposure to maneb (a widely used fungicide) and organophosphorus insecticides (Bhatt et al., 1999; Liu et al., 2003; Meco et al., 1994). Elevated levels of residual dieldrin (an organochlorine pesticide) and polychlorinated biphenyls (widely used industrial chemicals) have been detected in the brains of PD patients (Corrigan et al., 1998; Corrigan et al., 2000; Fleming et al., 1994). Chronic occupational exposure to high levels of manganese in manganese miners causes accumulation of this metal in the basal ganglia, resulting in tremors, rigidity and psychosis that resemble PD (Mergler and Baldwin, 1997). However, elevated levels of certain pesticides or chemicals in the brains of PD patients do not indicate causal relationship between toxin exposure and PD development. Experimental exposure of animal models to several classes of pesticides, such as paraquat, dithiocarbamate-based fungicides (e.g. maneb), insecticide rotenone, and organochlorine pesticides (e.g. dieldrin) leads to dopaminergic neurotoxicity. These pesticides have hence been proposed as potential PD risk factors in humans (Liu et al., 2003). Furthermore, epidemiological and case-control studies have suggested rural residence, well-water consumption, pesticide use, and certain occupations (e.g. farming, mining and welding) are associated with an increased risk of PD (Dhillon et al., 2008; Elbaz et al., 2009; Engel et al., 2001; Gorell et al., 1998; Semchuk et al., 1992). Several specific pesticides including dieldrin, maneb, paraquat, and rotenone have recently been implied to increase PD risk by recent epidemiological studies (Brown et al., 2006; Dhillon et al., 2008; Kamel et al., 2007; Ritz et al., 2009). However, overall epidemiological evidence is fragmentary and insufficient to establish a specific association between exposure to specific pesticides/toxicants and increased risk for PD.

4.2 Brain trauma

Numerous case reports have associated episodes of traumatic brain injuries with the occurrence of an acute-phase Parkinsonism or later-life development of PD (Doder et al., 1999; Nayernouri, 1985). Several epidemiological studies including an analysis of the medical histories of a large group of World War II veterans have identified antecedent head injuries as a major risk factor for the development of PD in later life (Goldman et al., 2006; Stern, 1991; Taylor et al., 1999). However, contradiction epidemiological results suggest no such a correlation (Goetz and Pappert, 1992; Spangenberg et al., 2009; Williams et al., 1991). Boxers who have professionally related higher chance of suffering head injuries appear to have a higher incidence of Parkinson-like movement disorders (Friedman, 1989; Guterman and Smith, 1987).

4.3 Inflammation

Early-life occurrence of brain inflammation resulting from either brain injury or exposure to infectious agents may modulate PD risk (Figure 1). Epidemiologic studies pointed to an association between early-life viral infections and postencephalitic PD, while case reports tend to relate viral infections to the development of acute Parkinsonism (Duvoisin et al., 1972; Elizan and Casals, 1991; Ghaemi et al., 2000; Henry et al., 2010; Liu et al., 2003). Experimental exposure of neonatal Fisher 344 rats to Japanese encephalitis virus can induce degeneration of nigral dopamine neurons, gliosis in the SN, and movement disorders resembling human PD (Ogata et al., 1997). Moreover, prospective studies suggest that higher plasma concentration of proinflammatory cytokine interleukin-6 correlates with an increased risk for PD (Chen et al., 2008). Epidemiological and case–control studies on associations between chronic use of nonsteroidal anti-inflammatory drugs and PD risk generate mixed results, with some studies suggesting a reduced incidence and others finding no association (Chen et al., 2005a; Chen et al., 2003; Powers et al., 2008; Ton et al., 2006). DNA polymorphisms of several inflammatory cytokines might become risk factors for PD (Wahner et al., 2007). A recent GWAS of 2,000 PD patients and 1,986 controls from the NeuroGenetics Research Consortium Common identifies an association of genetic variation in the HLA region with late-onset sporadic PD (Hamza et al., 2010). This hypothesis-free GWAS lends strong and independent support to the participation of inflammation in PD pathogenesis (Figure 1). HLA (human leukocyte antigen) region contains a large number of genes related to immune function in humans. HLA class II histocompatibility antigen plays a central role in inflammation by presenting peptide antigens to the immune system. While additional studies are needed to establish a causal role for inflammation in PD pathogenesis, the increased awareness of the involvement of inflammation in PD will stimulate research toward new therapeutic targets. Collectively, environmental risk factors, including toxins, early-life infections in the brain, and traumatic brain injuries may contribute to the cause of PD (Figure 1). However, specific causative environmental toxicants or infectious agents (DNA or antigens) for PD have not yet been identified.

4.4 Protective environmental factors

Inverse associations of PD with cigarette smoking, caffeine intake, and high plasma urate concentrations have been noted (Checkoway et al., 2002; Chen et al., 2010; Hernan et al., 2002; Powers et al., 2008; Weisskopf et al., 2007). However, whether such associations reflect true biologic protection needs further investigation. Interestingly, a few randomized controlled clinical trials suggest beneficial effects of exercise (e.g. improvements in postural stability and balance task performance) in PD patients (Dibble et al., 2009; Gobbi et al., 2009; Morris et al., 2009). Larger number of studies, more outcomes used, and longer term follow-up are needed to confirm that excise-related improvements in PD motor dysfunction are long-lasting. A few prospective cohort studies suggest that moderate to vigorous exercise may protect against PD (Xu et al., 2010). In rodents, exercise reduces the behavioral impairments and the loss of dopamine neurons elicited by the dopaminergic neurotoxins MPTP or 6-hydroxydopamine (6-OHDA) (Gerecke et al., 2010; Zigmond et al., 2009). The trigger of endogenous neuroprotective mechanisms (e.g. increased expression of neurotrophic factors) and the preconditioning (mild stress before more intensive stress) have been proposed to be responsible for beneficial effects of exercise (Zigmond et al., 2009).

5.1 MPTP model

Several toxin-based models of PD have provided the framework for symptomatic improvements in medical and surgical therapies for PD. MPTP model is the most widely used PD model for the study of molecular cascades of death of dopamine neurons and for the screening of neuroprotective strategies. MPTP can cross the blood-brain barrier after subcutaneous or intraperitoneal injection. Once inside the brain, MPTP is converted into MPP⁺ by the enzyme monoamine oxidase B in astroglia. Released MPP⁺ is then taken up by the dopamine transporter and concentrated in dopamine neurons where it inhibits mitochondrial complex I, resulting in ATP depletion, increased free-radical generation, and death of dopamine neurons. MPTP is the only known dopaminergic neurotoxin inducing Parkinsonism in both humans and monkeys (Jackson-Lewis and Przedborski, 2007). Mice and monkeys treated with MPTP develop reliable and reproducible nigrostriatal lesion and gliosis (Jackson-Lewis and Przedborski, 2007). MPTP-injected monkeys revealed increased brain levels of α-synuclein, PD-like modifications of α-synuclein (e.g. aggregation, nitration and phosphorylation), or intraneuronal inclusions (Forno et al., 1986; McCormack et al., 2008; Purisai et al., 2005). However, it seems that unless MPTP is chronically infused by osmotic minipump or is administered with probenecid, a uricosuric agent that is used to block the rapid clearance of MPTP, proteinaceous inclusions will not be produced in dopamine neurons in mice (Fornai et al., 2005; Meredith et al., 2002; Shimoji et al., 2005). A few drawbacks, such as strain-specificity of animals, lack of classical and convincing α-synuclein pathology, relatively rapid kinetics of neurodegeneration and partial lesion recovery, make MPTP mouse models less ideal for studying PD progression and for screening drugs to slow down the disease progression.

5.2 6-OHDA model

6-OHDA is the first catecholaminergic neurotoxin that was used to generate animal models of PD. Because of its preferential uptake by dopamine and noradrenergic transporters, 6-OHDA induced relatively selective neurotoxicity to monoaminergic neurons (Ungerstedt, 1968). Inside these neurons, 6-OHDA produces reactive oxygen species and quinones that inactivate biological macromolecules. Being unable to cross the blood-brain barrier, 6-OHDA must be administered by local stereotaxic injection to target the nigrostriatal dopaminergic pathway (Javoy et al., 1976). When injected into the SN or median forebrain bundle that carries ascending dopaminergic and serotonergic projections to the forebrain, 6-OHDA causes acute loss of dopamine neurons within 24 hr (Jeon et al., 1995); when injected into the striatum, it induces retrograde degeneration of nigral dopamine neurons, which lasts for 1–3 weeks (Przedborski et al., 1995; Sauer and Oertel, 1994). To date, no Lewy body-like inclusion has been described in 6-OHDA model. The paucity of classical PD pathology, the relatively acute neurotoxicity, and the requirement of stereotaxic injection make 6-OHDA model less popular in recent years.

5.3 Rotenone model

Rotenone, a mitochondrial complex I inhibitor (Earley and Ragan, 1984), is commonly used as a pesticide and fish poison. Although it remains to be determined whether typical human exposure to rotenone alone is likely to increase risk for PD, rotenone has been employed to model PD. Intracerebral application of rotenone led to damage to the nigrostriatal dopaminergic pathway in rats (Heikkila et al., 1985). Initial attempts to systemically administer rotenone, acutely or sub-chronically, to rodents produced only minimal damage to the nigrostriatal dopaminergic system (Ferrante et al., 1997; Thiffault et al., 2000). Later, continuous systemic administration of rotenone to rats and mice reproduces cardinal features of PD, including the degeneration of the nigrostriatal dopaminergic pathway, the formation of Lewy body-like inclusions containing ubiquitin and filamentous α-synuclein in remaining nigral neurons, and movement impairments (Alam and Schmidt, 2002; Betarbet et al., 2000; Hoglinger et al., 2003; Sherer et al., 2003). Mechanistically, inhibition of mitochondrial complex I activity with consequent ATP depletion and oxidative damage, facilitation of α-synuclein aggregation and inflammation-induced injury may be responsible for the dopaminergic neurotoxicity of rotenone (Betarbet et al., 2000; Gao et al., 2002a; Sherer et al., 2002; Sherer et al., 2003). A high mortality has limited an extensive utility of rotenone model.

5.4 Paraquat model

Intracerebral injection of paraquat resulted in loss of nigral dopamine neurons, depletion of dopamine in the striatum, and an elevated response to apomorphine-induced rotational behavior (Liou et al., 1996). Effects of systemic administration of paraquat on dopamine neurons have been less consistent. Paraquat does not easily penetrate the blood brain barrier (Shimizu et al., 2001), and its distribution in the CNS is not region-specific or cell-type-specific. Mice that underwent a single or repeated systemic administration of paraquat exhibited a selective loss of nigral dopamine neurons with or without degeneration of striatal dopamine fibers or a reduction in striatal dopamine content (Brooks et al., 1999; Manning-Bog et al., 2002). Free radical generation and redox cycling (Bonneh-Barkay et al., 2005; Fukushima et al., 1995; Wu et al., 2005; Yumino et al., 2002), α-synuclein upregulation and fibrillation (Manning-Bog et al., 2002; Uversky et al., 2001), as well as apoptotic cell death (Chun et al., 2001) have been associated with paraquat-induced dopaminergic neurotoxicity. Furthermore, repeated systemic administration of maneb and paraquat to mice induced synergistic nigrostriatal dopaminergic neurodegeneration and motor behavioral abnormalities (Thiruchelvam et al., 2000). Neonatal exposure to both maneb and paraquat sensitized the nigrostriatal dopaminergic system to rechallenge with the same agents during adulthood (Thiruchelvam et al., 2002).

5.5 Inflammation-associated model

Whether bacterial or viral infection can act as an initiating factor in human PD is unclear, but intracranial or systemic administration of bacterial endotoxin or virus into rodents has been used to produce inflammation-associated PD model. Direct nigral injection of lipopolysaccharide (LPS), the active immunostimulant in the cell wall of Gram-negative bacteria, causes robust microglial activation and acute loss of nigral dopamine neurons in rodents (Castano et al., 1998; Kim et al., 2000; Liu et al., 2000; Tomas-Camardiel et al., 2004). Chronic infusion of LPS for 2 weeks into the rat SN triggered a rapid activation of microglia, followed by a selective and gradual loss of nigral dopamine neurons that began at between 4 and 6 weeks and reached 70% by 10 weeks (Gao et al., 2002b). This is the first report that microglial activation induced by chronic exposure to inflammogen was capable of producing a delayed, progressive and selective degeneration of nigral dopamine neurons. Microglia-originated free radicals play a critical role in dopaminergic neurotoxicity in this inflammation-mediated model of PD (Gao et al., 2002b). Additionally, in utero exposure of developing rat fetuses to LPS results in lesions of nigrostriatal dopaminergic system in neonates (Ling et al., 2002). Adult rats with prenatal exposure to LPS developed progressive dopaminergic neurodegeneration following chronic nigral LPS infusion (Ling et al., 2006). Moreover, an intraperitoneal injection of LPS (5 mg/kg) led to significant loss of nigral dopamine neurons in C57BL/6 mice 7 months after the injection; such neurodegeneration progressed further overtime (Qin et al., 2007). These findings together indicate that once reaching certain threshold, the inflammatory process is capable of initiating neurodegeneration. The higher density of microglia in the SN than other brain regions (Kim et al., 2000; Lawson et al., 1990) may be partially responsible for the foremost loss of dopamine neurons in the inflammation-induced neurodegenerative process.

Although humans infected by highly pathogenic H5N1 influenza virus show acute neurological impairments ranging from mild encephalitis to motor disturbances to coma (de Jong et al., 2005; Gambotto et al., 2008), the long-term neurological manifestations of H5N1 infection among surviving hosts remain unknown. A recent study has reported that after intranasal inoculation in C57BL/6 mice, H5N1 virus travelled from the peripheral nervous system into the CNS. Microglial activation and α-synuclein phosphorylation and aggregation persists long after resolution of the infection in infected brain regions. While H5N1 virus was no longer detectable in the CNS 21 days post-inoculation, a significant loss of nigral dopamine neurons was observed 60 days after the virus infection (Jang et al., 2009). This study provides important information on the role of virus-associated inflammation in potential long-term impairment to the CNS. However, the requirement of a biosafety level 3⁺ laboratory will limit the usage of this virus-based PD model.

6.1 LPS and α-synuclein

The fact that only very few of many different α-synuclein transgenic mouse models developed loss of dopamine neurons in the SN reveals the limitations of rodent models in the replication of all aspects of age-related neurodegeneration. Alternatively, such a fact might also suggest that other factors participate in the development of PD. Stereotaxic injection of LPS into the SN of wildtype mice and transgenic mice overexpressing either wildtype or A53T mutant human α-synuclein induced similar nigral inflammatory reactions. Transgenic expression of human α-synuclein renders dopamine neurons more vulnerable to LPS-induced inflammation, which in turn leads to accumulation of insoluble α-synuclein aggregates in nigral neurons and death of nigral dopamine neurons. Nitrated/oxidized α-synuclein was detected in Lewy body-like cytoplasmic inclusions in LPS-injected transgenic mice (Gao et al., 2008). This two-hit (neuroinflammation and overexpression of wildtype or mutant human α-synuclein) animal model not only provides better tool to advance our understanding of the role of neuroinflammation and α-synuclein dysfunction in the pathogenesis of PD, but also underscores synergistic effects of genetic predisposition and environmental exposures in the development of PD. Furthermore, a two-hit, chronic, progressive model has been created by an intraperitoneal injection of a relatively low dose of LPS (3 × 10⁶ EU/kg) into 7-month-old transgenic mice overexpressing A53T mutant human α-synuclein (Gao et al., 2011). Such a LPS injection induced similar systemic inflammation and a comparatively mild neuroinflammatory response in both wildtype and α-synuclein transgenic mice. However, only in LPS-injected transgenic mice, the low-grade neuroinflammation sustains; α-synuclein accumulates and forms aggregates; Lewy body-like neuronal inclusions containing aggregated α-synuclein occur; and dopaminergic neurodegeneration in the nigrostriatal pathway progresses over time. Thus, synergistic effects of low-grade neuroinflammation and α-synuclein dysfunction drive chronic PD neurodegeneration, while neither factor alone is sufficient to cause neuronal death (Gao et al., 2011). This new model mimics PD multifactorial etiology and recapitulates the key features of PD. More importantly, the chronic progressive nature of dopaminergic neurodegeneration makes this model invaluable for the study of mechanisms of PD progression and for the screening of disease-modifying therapeutics.

6.2 LPS and Parkin

Although loss-of-function mutations in the parkin gene cause early-onset familial PD, Parkin-deficient mice do not display nigrostriatal degeneration. The intraperitoneal administration of low-dose of LPS (7.5 × 10⁵ EU/kg) twice a week into Parkin-null or wildtype mice for 3 to 6 months indicated that genetic deficiency in Parkin sensitizes nigral dopamine neurons to inflammation-mediated neurotoxicity. Specifically, although persistent neuroinflammation and oxidative stress responses triggered by LPS are not different between two genotypes, only Parkin-null mice displayed selective loss of nigral dopamine neurons and subtle fine-motor deficits (Frank-Cannon et al., 2008). This model may enable identification of early biomarkers of neurodegeneration and preclinical screening of potential compounds that can delay the progressive degeneration of the nigrostriatal pathway.

6.3 Neurotoxins and α-synuclein

The possible interaction between genetic factors and neurotoxins has been investigated by examining the sensitivity of α-synuclein transgenic mice to secondary neurotoxic insults, such as paraquat, maneb, MPTP, or rotenone. In general, either dopaminergic neurodegeneration or α-synuclein pathology is missing or has not been determined in those neurotoxin-exposed α-synuclein transgenic mice. For instance, intraperitoneal injection twice a week for 3 weeks into transgenic mice overexpressing A53T mutant human α-synuclein with maneb and paraquat, but not with either pesticide alone, drastically exacerbates neuronal α-synuclein pathology (e.g. accumulation of aggregated α-synuclein and formation of α-synuclein-containing inclusions) throughout the CNS. However, no noticeable neuron loss in the SN or other brain regions was observed in transgenic mice after treatment with maneb and/or paraquat (Norris et al., 2007). Whether the exacerbated α-synuclein pathology in pesticide-exposed transgenic mice represents early disease stage that precedes neuronal degeneration remains to be determined. In addition, weekly intraperitoneal injections of paraquat (10 mg/kg, 3 times) into wildtype and transgenic mice expressing human wildtype or A53T mutant human α-synuclein driven by the tyrosine hydroxylase promoter resulted in an upregulation of α-synuclein and the formation of α-synuclein-positive inclusions. However, only wildtype mice, but not transgenic mice, developed nigral degeneration after paraquat injection (Manning-Bog et al., 2003). The mechanism underlying the protective effect of transgenic overexpression of human α-synuclein against pesticide-induced neurodegeneration in transgenic mice warrants further investigation. Interestingly, recent evidence also indicates that transgenic expression of α-synuclein abolishes the lethality and neurodegeneration induced by ablation of cysteine-string protein-α (CSPα, an abundant synaptic vesicle protein with a cochaperone function), whereas deletion of endogenous synucleins exacerbates these phenotypes. Here, transgenic α-synuclein attenuates CSPα-deficiency-elicited inhibition of SNARE-complex assembly, in which binding of α-synuclein to synaptic vesicle phospholipids is requires (Chandra et al., 2005).

Chronic treatment with MPTP (daily intraperitoneal injection of 16 or 30 mg/kg MPTP for 5 consecutive days), but not with rotenone (0.125 mg/kg twice a week for 3 consecutive weeks), led to greater deterioration of the nigrostriatal dopamine system in transgenic mice expressing human A30P mutant α-synuclein than in wildtype littermates (Nieto et al., 2006). Transgenic mice expressing wildtype or a doubly mutated (A53T and A30P) α-synuclein show increased density of the dopamine transporter and elevated vulnerability to MPTP neurotoxicity. The doubly mutated mice exhibit reduced locomotor responses to repeated doses of amphetamine (Richfield et al., 2002). Collectively, although neurotoxin-exposed α-synuclein transgenic mice reveal either exacerbated α-synuclein pathology or greater deterioration of nigrostriatal dopamine pathway, it seems there is dissociation between toxicant-induced α-synuclein pathology and neurodegeneration

6.4 Neurotoxins and LRRK2

The ubiquitous expression of LRRK2 increases lifespan and fertility of the Drosophila but enhances the sensitivity of these flies to rotenone (Venderova et al., 2009). Similar to α-synuclein knockout mice, LRRK2 knockout mice that lack the kinase domain of LRRK2 have no observable neurodegenerative pathology. Different from α-synuclein knockout mice that are more resistant to neurotoxicity induced by LPS or MPTP (Dauer et al., 2002; Gao et al., 2008), LRRK2 knockout mice display the same susceptibility to MPTP as wildtype mice (Andres-Mateos et al., 2009).

6.5 Glial cell line-derived neurotrophic factor (GDNF) and methamphetamine

The importance of GDNF for maintenance of dopamine neurons has been well established. Enhanced dopaminergic cell survival and fiber outgrowth as well as improved behavioral recovery in rodent and monkey models of PD have encouraged the use of GDNF in clinical trials. Safety concerns for CNS delivery of GDNF and mixed results from limited human studies make it impossible to draw a firm conclusion about the efficacy and potential usage of GDNF in PD treatment. The genetic reduction of Gdnf and the reduced availability of GDNF protein in young Gdnf^+/− mice increase the vulnerability of dopamine systems to methamphetamine exposure (Boger et al., 2010). This dual-hit model will extend current knowledge of the impact of GDNF on progressive dopaminergic degeneration and motor dysfunction in PD. Experimental exposure of these Gdnf^+/− mice to more relevant environmental toxins (e.g. rotenone or paraquat) may provide fruitful information on PD etiology and pathogenesis.