MINIREVIEW Parkinson’s disease: genetic versus toxin-induced rodent models Mu ¨ gen Terzioglu 1 and Dagmar Galter 2 1 Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden 2 Department of Neuroscience, Karolinska Institutet, Stockholm, Sweden Introduction Parkinson’s disease (PD) is a common neurodegenera- tive disease with a complex etiology resulting from genetic factors, environmental exposures, or a combi- nation of both. The clinical key symptoms are motor dysfunctions such as bradykinesia, resting tremor and muscle rigidity combined with postural instability, but many patients also suffer from autonomic and cognitive dis- turbances. Selective degeneration of dopamine neurons in the substantia nigra (SN) causes the major PD symptoms, but there is often widespread neurodegener- ation and pathology in other regions of the brain, including the proteinaceous inclusions called Lewy bodies (LBs) and dystrophic neurites called Lewy neurites. By the time clinical manifestations appear, about 60–70% of the dopamine fibers in the caudate Keywords 6-OHDA; conditional knockout mice; DAT-cre; dopamine system; Engrailed; intracellular aggregates; mitochondrial dysfunction; MPTP; PARK genes; progressive neurodegeneration; a-synuclein Correspondence D. Galter, Department of Neuroscience, Karolinska Institutet, Retzius va ¨ g8, 171 77 Stockholm, Sweden Fax: +46 8 32 37 42 Tel: +46 8 524 87018 E-mail: dagmar.galter@ki.se (Received 23 October 2007, revised 17 December 2007, accepted 7 January 2008) doi:10.1111/j.1742-4658.2008.06302.x Parkinson’s disease (PD), a common progressive neurodegenerative disor- der, is characterized by degeneration of dopamine neurons in the substantia nigra and neuronal proteinaceous aggregates called Lewy bodies (LBs). The etiology of PD is probably a combination of environmental and genetic factors. Recent progress in molecular genetics has identified several genes causing PD, including a-synuclein, leucine-rich repeat kinase 2 (LRRK2), Parkin, DJ-1 and PTEN-induced kinase 1 (PINK1), many of them coding for proteins found in LBs and ⁄ or implicated in mitochondrial function. However, the mechanism(s) leading to the development of the disease have not been identified, despite intensive research. Animal models help us to obtain insights into the mechanisms of several symptoms of PD, allowing us to investigate new therapeutic strategies and, in addition, pro- vide an indispensable tool for basic research. As PD does not arise sponta- neously in animals, characteristic and specific functional changes have to be induced by administration of toxins or by genetic manipulations. This review will focus on the comparison of three types of rodent animal models used to study different aspects of PD: (a) animal models using neurotoxins; (b) genetically modified mouse models reproducing findings from PD link- age studies or based on ablation of genes necessary for the development and survival of dopamine neurons; and (c) tissue-specific knockouts in mice targeting dopamine neurons. The advantages and disadvantages of these models are discussed. Abbreviations 6-OHDA, 6-hydroxydopamine; cre, cre-recombinase; DA, dopamine; DAT, dopamine transporter; En, Engrailed; IR, immunoreactive; LB, Lewy body; L-dopa, L-3,4-dihydroxyphenylalanine; LRRK2, leucine-rich repeat kinase 2; MAO-B, monoamine oxidase B; MPP + , 1-methyl-4- phenyl-2,3-dihydropyridium ion; MPTP, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; PD, Parkinson’s disease; PINK1, PTEN-induced kinase 1; ROS, reactive oxygen species; SN, substantia nigra; TFAM, mitochondrial transcription factor A; TH, tyrosine hydroxylase; VMAT, vesicular monoamine transporter; VTA, ventral tegmental area. 1384 FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS putamen and at least 50% of the dopamine neurons in the SN are already lost. Although slow in most cases, progression of the disease is irreversible, and different drug treatments ameliorate symptoms without arrest- ing or slowing down the pace of neurodegeneration. In order to understand the underlying mechanisms and to develop new drugs or therapies for PD, it is important to have available animal models that reca- pitulate key symptoms and the slow progression of the disease as accurately as possible. Because the disease is not known in any animal species, except perhaps mild parkinsonism in aged nonhuman primates in captivity, different models have been developed in several species together with specific behavior tests to assess motor dysfunctions. This review focuses on rodent animal models and compares the most recently available tissue-specific knockout mouse models with older genetic and toxin- induced animal models (Fig. 1). Toxin-induced animal models Early animal models developed for PD research used neurotoxins specific for the dopamine (DA) system such as 6-hydroxydopamine (6-OHDA) and 1-methyl- 4-phenyl-1,2,3,6 tetrahydropyridine (MPTP). For a recent review on classic toxin-induced animal models, see also Schober [1]. The hydroxylated derivative of the neurotransmitter DA was first used in sympathetic heart denervation, and soon after in the central nervous system [2]. When the drug is stereotactically injected into the striatum, the median forebrain bundle or the SN, it induces fast and irreversible DA depletion through reactive oxygen species (ROS) formation and toxic quinines [3]. The relative specific toxicity of 6-OHDA for catecholamine neurons results from its uptake by DA and noradrena- lin transporters. Most widely used is the unilateral lesion of the DA system in rats, where a quantifiable circling behavior is induced after injection of DA receptor agonists or amphetamine. In addition, several other behavioral assessments, such as fine motor skill tasks and the cylinder test, have been developed to measure striatal DA loss [4]. Furthermore, rodent ani- mal models for dyskinesia are mostly based on unilat- eral intracerebral injections of 6-OHDA followed by chronic l-3,4-dihydroxyphenylalanine (l-dopa treat- ment [5,6,6a]. In 1982, an analog of the narcotic drug meperidine was accidentally discovered to be a potent dopamine neurotoxin when young drug addicts developed irre- versible and severe PD symptoms following self-admin- istration of what they hoped to be synthetic heroin [7]. The highly lipophilic substance that they had synthe- sized, MPTP, crosses the blood–brain barrier easily after systemic administration and is converted into the active toxic metabolite 1-methyl-4-phenyl-2,3-dihydro- pyridium ion (MPP + ) by the enzyme monoamine oxi- dase B (MAO-B), located mainly in serotoninergic neurons and astrocytes. The metabolite MPP + is selec- tively taken up into dopamine neurons by the DA transporter (DAT), and irreversibly inhibits complex I ABC Fig. 1. Schematic illustration of different rodent models of PD. (A) Toxin-induced models: the four different toxins penetrate dopamine neu- rons either specifically via DAT (6-OHDA and MPP + ) or through diffusion (rotenone and paraquat) and inhibit complex I of the mitochondrial electron transfer chain (consisting of complex I to complex V), leading to mitochondrial intoxication with enhanced production of ROS and reduced production of ATP. Although all toxins do not exclusively act on dopamine neurons, they induce PD symptoms and key pathology, indicating an increased susceptibility of the DA system to mitochondrial dysfunction. (B) Genetic models: on the basis of the PD-linked genes a-Synuclein, Parkin, Pink1, DJ-1 and LRRK2, several mouse models have been generated in which all cells of the organism are affected where the genes are active. Protein aggregations, altered protein handling and mitochondrial deficits have been detected in these mouse models, mainly in the DA system. (C) Dopamine neuron-specific knockout models: using DAT promoter driven cre expression, three mice models with targeted deletion of floxed genes have been generated to date: deleting the GDNF receptor Ret, the RNA-cleaving enzyme complex Dicer and TFAM from dopamine neurons. M. Terzioglu and D. Galter Animal models of Parkinson’s disease FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS 1385 of the mitochondrial respiratory chain. As the suscepti- bility to MPTP depends on the MAO-B activity, dif- ferent mouse strains react very differently to the toxin. Rats are relatively resistant to MPTP, whereas humans are in danger of intoxication at quite low doses. In mice, systemic MPTP treatment induces bradykinesia, rigidity and posture abnormalities combined with a depletion of dopamine neurons [8]. Continuous MPTP infusion by minipumps has been reported to induce development of intracellular inclusion bodies [9], although these inclusions are not similar to the LBs typically found in human disease [10]. More recent toxin-induced animal models make use of agents with a general toxicity: mitochondrial func- tion inhibitors such as the herbicide paraquat and the insecticide rotenone, or proteasomal inhibitors such as epoxomicin. Paraquat is structurally similar to MPTP, but has no selectivity for DAT and does not accumulate in dopamine neurons after systemic administration. Nev- ertheless, it induces a specific, although modest, loss of tyrosine hydroxylase (TH)-positive neurons of the SN pars compacta [11,12]. Rotenone, produced in the roots and stems of tropical plants, inhibits the transfer of electrons from complex I to ubiquinone in the mito- chondrial electron transfer chain. Rotenone interferes with mitochondrial function at the same site as MPP + , but is only mildly toxic for humans and highly unstable, with a short half-life in the environment. In rodents, particularly in rats, chronic infusion can induce a slowly progressing neurodegeneration of dopamine neurons associated with intracellular multi- form a-synuclein immunoreactive (IR) aggregates, occurrence of widespread oxidatively modified DJ-1, and proteasomal impairment [13]. However, the rote- none model has low reproducibility, and many animals die from acute toxicity, unrelated to central nervous system involvement. A further rodent toxin-induced model has been proposed that uses systemic administration of the proteasomal inhibitor epoxomicin [14]. In this PD model, rats reproduced most of the key features of PD pathology, including reduced amounts of dopamine fibers in the striatum, and degeneration of dopamine neurons in the SN accompanied by inflammation and intracellular aggregates with a-synuclein- and ubiqu- itin-like immunoreactivity. However, in a further inde- pendent study, systemic administration of epoxomicin failed to be effective in rats or monkeys [15], although intracerebral injection of epoxomicin and other prote- asomal inhibitors blocked MPP + - or rotenone-induced dopamine neuron death in rats and induced round a-synuclein IR inclusions in dopamine neurons [16]. In summary, most toxins used in PD animal models inhibit mitochondrial function and reveal a greater susceptibility of dopamine neurons to mitochondrial dysfunction and ROS production. Genetically modified mouse models Although the majority of PD cases are sporadic, sev- eral mutations in genes causing familial forms of PD have been recently discovered, and many susceptibility genes have also been identified, leading to new approaches to the study of mechanisms leading to dis- ease. Many animal models are based on genetically modified mice with null mutations, an extra gene copy, or point mutations of genes located in different PARK loci [17,18]. For the recessively inherited loss-of-function muta- tions in Parkin, DJ-1 and PINK1, all of which cause early-onset PD, genetic mouse models can easily be made by null mutation of such genes (knockout mice). For the dominantly inherited gain-of-function muta- tions such as in a-Synuclein and leucine-rich repeat kinase 2 (LRRK2), transgenic mouse models have been created in which extra copies of the gene are intro- duced into the mouse genome or delivered by lenti- or adeno-associated virus. Several mouse strains have been created for a-Synuclein, where either the human wild-type gene is overexpressed under various heterolo- gous promoters, to reproduce the gene duplication and triplication detected in PD families, or the PD-causing a-Synuclein mutations A30P or A53T are expressed in transgenic mice [19]. High levels of mutated a-synuc- lein expression under the mouse prion protein pro- moter induced, for example, a progressive phenotype with intraneuronal inclusions, degeneration and mito- chondrial DNA damage in the neurons [20]. Although no PD key symptoms were detected, this model is valuable for understanding the relationship of a-synuc- lein-positive protein depositions and neuronal damage. Data from mouse models with mutant or wild-type LRRK2 overexpression or null mutation have not yet been published. None of the genetic models based on PD-linked genes recapitulate the key symptoms of the disease, such as loss of dopamine neurons, but more subtle effects on the DA system have been detected, such as a small decrease in DAT binding and slightly reduced DA levels in the striatum, abnormal response to DA agonists, including apomorphine and amphetamine, and motor disturbances, including decreased spontane- ous activity together with protein-handling defects [17]. In several genetic models the MPTP-induced toxicity for dopamine neurons has been analyzed and found to Animal models of Parkinson’s disease M. Terzioglu and D. Galter 1386 FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS be modified: a-Synuclein knockout mice were reported to be less sensitive to MPTP, whereas a-Synuclein transgenic mice and Dj-1 knockout mice were reported to be more susceptible to the toxin [21]. Studies of the effects of toxins in genetic PD models can provide important clues, because the etiology and progression of the disease can be due to a combination of genetic factors and environmental exposures. Moreover, many genes implicated in PD are directly or indirectly involved in mitochondrial function: PINK1, DJ-1 and possibly Parkin and LRRK2 are at least partly local- ized to mitochondria; in a-Synuclein transgenic mice, mitochondrial pathology has been detected; in mice lacking the mitochondrial protease HtrA2 ⁄ Omi, motor impairment due to striatal cell loss has been reported [22], and in PolG, the mitochondrial DNA polymerase, genetic variants associated with PD have been identi- fied [23]. These genetic findings together strongly sug- gest mitochondrial involvement in the etiology of PD. Detection of genes that are critical for the develop- ment and survival of dopamine neurons has led to additional mouse models, such as the spontaneously occurring Pitx3-aphakia mouse or the Engrailed (En) double-knockout mouse model. En1 and En2 homeobox transcription factors are expressed as early as embryonic day 8, and they play a role in the development of the midbrain and cerebel- lum. Later in development they have additional func- tions, such as being survival factors for mesencephalic dopamine neurons, where the two genes can compen- sate for each other. Heterozygous knockout mice of En1 and homozygous knockout of En2 (En1 + ⁄ ) ; En2 KO) have adult onset of PD-like features [24]. During the first 3 months after birth, the number of dopamine neurons in the SN declined by about 70%, and DA levels in the striatum were reduced by 40%, but the degeneration abated at this level for the next 15 months. The mice slowly developed reduced loco- motor activity and other motor deficits, but further investigations are needed to clarify whether the altered motor behavior is related to the loss of dopamine neuron function or is caused by other cells deprived of En, such as cerebellar neurons, a subset of interneu- rons in the spinal cord or Bergman glia. The aphakia mouse, a recessive phenotype that occurred spontaneously, is characterized by small eyes that lack a lens, caused by a deletion in the promoter region of Pitx3. The gene expression of this homeobox transcription factor is restricted to the developing eye and to midbrain dopamine progenitor cells from embryonic day 11 to adult life. Adult aphakia mice develop SN-specific dopamine neuron loss combined with a severe reduction of DA levels in the dorsolateral striatum, whereas ventral tegmental area (VTA) dopa- mine neurons are spared overall [25]. No intracellular aggregations or LB-like inclusions have been detected. The motor deficits include reduced rearing and sensori- motor impairments, and repeated l-dopa treatment induces dyskinesia in this genetic model [26]. Tissue-specific knockout mouse models Recently, a new type of rodent animal model for PD has been established, using conditional knockout strat- egies in order to disrupt the expression of genes of interest in a region- or neuron-specific manner. For this purpose, mice that express cre-recombinase (cre) under the control of the DAT promoter are predomi- nantly used to target postmitotic dopamine neurons in the midbrain [27–30]. Other mouse strains targeting wider populations of neurons are also available: cre expression driven by the TH promoter targets all cate- cholamine neurons in the central and peripheral ner- vous system [31,32]; Mice in which cre expression is driven by the En1 promoter [33] or by wingless-1 [34] target the early stages of the developing DA system, although these mice are less convenient as PD models because neither transcription factor is exclusively expressed by dopamine neurons of the midbrain. To generate mice with specific deletion of a particu- lar gene, one of these cre-expressing mouse strains is bred with mice homozygous for a floxed gene; that is, both chromosomal copies of the gene are flanked by LoxP recombination sites. Examples of floxed genes used in conditional mouse models are: the mitochondrial transcription factor A (TFAM) [35], the microRNA enzyme Dicer [36], and the receptors for neurotrophic factors Ret [for glial- cell-line derived neurotrophic factor (GDNF)] and TrkB [for brain-derived neurotrophic factor (BDNF)] [37]. Deletion of TFAM, Ret or Dicer in dopamine neurons induces progressive motor dysfunctions such as slowness and pauperism of movements and limited rearing at different ages: at a few weeks for Dicer,at several months for TFAM and at more than 1 year for Ret conditional knockout mice. In MitoPark mice, which have respiratory chain-deficient dopamine neu- rons due to cell-specific ablation of TFAM, the motor impairments are ameliorated by l-dopa administra- tion, a common treatment for PD patients. Moreover, MitoPark mice respond differently to the same dose of l-dopa, depending on the progression of the symp- toms, very similar to PD patients: in younger mice, as in less severe PD patients, l-dopa treatment results in a greater locomotor response than in older mice and M. Terzioglu and D. Galter Animal models of Parkinson’s disease FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS 1387 Table 1. Summary of advantages and disadvantages of selected rodent models of PD. Scoring of dopamine neuron: slight loss (< 30%); loss (30–70%); massive loss (< 70%). The construct validity of a model refers to the degree to which the rodent model reproduces known PD etiology (low = no findings in PD patients indicate a role in PD etiology for the toxin or genetic modification that the model is based on; poor = some findings point to a role in PD etiology; good = findings in PD patients indicate a causative role for genetic modifications repro- duced in the model). KO, knockout. Model PD symptoms PD pathology Advantages Disadvantages 6-OHDA Motor impairments after bilateral lesion Easily quantifiable turning behavior after unilateral lesion Reduced DA levels in the striatum Massive loss of dopamine neurons No intracellular aggregates Works in mice, rats, and monkeys Well characterized Used in dyskinesia models Does not pass the blood–brain- barrier (needs intracerebral injection, which increases variability) Fast, massive neurodegeneration Poor construct validity MPTP Motor impairments Reduced DA levels in the striatum Massive loss of dopamine neurons With chronic administration, formation of aggregates with little LB resemblance Lipophilic Systemic administration Works mainly in mice Well characterized Good construct validity Highly toxic to humans (dangerous to administer) Reduced reliability Paraquat Motor impairments Reduced DA levels in the striatum Loss of dopamine neurons in the SN No aggregate formation Systemic administration Toxic for the whole organism Not well characterized Low construct validity Rotenone Motor impairments Reduced DA levels in the striatum Massive loss of dopamine neurons No aggregate formation Systemic administration Works only in rats Toxic for the whole organism Low construct validity Dj-1 KO, Pink1 KO, Parkin KO Little motor impairment Only slight DA pathology Good construct validity Slight DA pathology a-Synuclein wild-type and A53T, A30P overexpression Little motor impairment Little DA pathology Intracellular aggregates with little LB resemblance Good construct validity Slight DA pathology En1 + ⁄ ) , En2 KO Some motor impairment Reduced DA levels in the striatum Massive loss of dopamine neurons only in the SN during the first 3 months No aggregate formation Slow neurodegeneration Poor construct validity Other cell groups affected in the central nervous system No progression of degeneration after 3 months Pitx3-aphakia Motor impairment Reduced DA levels in the striatum Massive loss of dopamine neurons in the SN only Slow neurodegeneration Poor construct validity Other cell groups affected in the central nervous system MitoPark (DAT-cre, Tfam lox ⁄ lox) Motor impairment Reduced DA levels in the striatum Massive loss of dopamine neurons, predominantly in the SN Intracellular aggregates with little LB resemblance Adult onset of symptoms Slow symptom development Good construct validity Complex breeding scheme Animal models of Parkinson’s disease M. Terzioglu and D. Galter 1388 FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS PD patients with severe motor dysfunctions [35]. In parallel with behavioral changes in all three models, the degeneration of dopamine nerve terminals in the striatum and a progressive loss of dopamine neurons specifically in the SN pars compacta occur. In con- trast, VTA neurons appear to be more resistant to the ablation of TFAM and Ret, because the loss of nerve terminals in the ventral striatum and cell loss in VTA occur later and are less pronounced than in the SN pars compacta, similar to the pathological devel- opment in PD [35,37]. Ablation of Dicer induces a similar degree of degeneration in dopamine neurons of the SN pars compacta and the VTA [36]. The dopamine cell loss results in reduced DA levels in the corresponding parts of the nigrostriatal system in middle-aged MitoPark mice together with a marked increase of DA turnover, as is typically seen in PD and animal models with DA deficiency [35]. In Ret- deficient mice, dopamine cell and nerve terminal losses are less than 40%, even in 24-month-old mice, and DA levels in the striatum are unaltered. How- ever, evoked DA release after electrical stimulation reveals a significant drop in 1-year-old mice and a further reduction in older mice, consistent with pre- symptomatic development in PD patients. Dopamine neuron-specific ablation of TrkB, in contrast, did not affect the motor behavior and nor did it induce any PD-like neuropathology [37]. MitoPark mice display an additional pathological hallmark of PD: affected dopamine neurons contain cytoplasmic proteinaceous aggregates. However, unlike LBs, these intracellular inclusions lack a-synuclein immunoreactivity and they can also form in MitoPark mice with a null mutation for a-synuclein, which develop a progressive PD-like phenotype similar to that seen in MitoPark mice with functional a-synuclein genes. All other conditional mouse models for PD described so far lack cytoplasmic inclusion bodies. Conclusions Regardless of whether a PD model is based on toxins or on genome modifications, no single rodent model for PD created to date reproduces all key symptoms of the disease: slowly progressing motor disturbances com- bined with loss of striatal dopamine fibers, and dopami- ne cell loss in the SN accompanied by LB pathology. Although toxin-induced models, particularly those using drugs with a high specificity for dopamine neurons, induce many of the key features of PD, they are of lesser value in studies addressing PD etiology, because only a few PD cases are caused by intoxication with poisons (see also summary in Table 1). On the other hand, genetic models based on genomic modifica- tions found in PD patients have good construct validity but show only rudimentary PD pathology. Those trans- genic mouse models for a-synuclein exhibiting a more pronounced PD phenotype have often used heterolo- gous promoters (PDGFb, Thy1) that induce nonphysi- ological high expression levels in restricted areas of the brain. Interestingly, in some studies, genetic models are combined with PD-specific toxins to analyze the effect of the genetic modification on toxin susceptibility. The two genetic PD models based on ablation of the transcription factors En and Pitx3 display many of the key features of PD. Their drawbacks are low construct validity, because few studies point to an involvement of DA system development in PD etiology, and the fact that many different cell populations in the brain are affected in En or Pitx3 knockout mice as well as dopamine neurons, making it difficult to interpret the findings. Tissue-specific knockout models for PD based on cre expression directed by the DAT promoter combine the advantages of the earlier models: (a) only dopa- mine neurons are targeted like in toxin models with mainly DA-specific neurotoxicity (6-OHDA and Table 1. (Continued). Model PD symptoms PD pathology Advantages Disadvantages DAT-cre, Ret lox ⁄ lox No motor impairment Slight loss of dopamine neurons in the SN Slow loss of TH-IR fibers No reduction in DA levels Reduced DA release No aggregate formation Very slow progression (preclinical model) Low construct validity Complex breeding scheme DAT-cre, Dicer lox ⁄ lox Motor impairment Massive loss of TH-IR fibers in the striatum Massive loss of dopamine neurons in the SN and VTA No aggregate formation Possibility of studying the role of post-transcriptional mechanisms in PD Fast and early onset of degeneration Complex breeding scheme Low construct validity M. Terzioglu and D. Galter Animal models of Parkinson’s disease FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS 1389 MPTP); (b) high reliability, due to complete penetra- tion and minimal variability; and (c) depending on the floxed gene, the time course of the development of neuropathology varies, but in all models there is slow and progressive neurodegeneration, in contrast to the acute and violent degeneration seen, for instance, in the 6-OHDA model. Slow progressive dopamine neu- ron degeneration has also been achieved through chronic MPTP administration, albeit associated with high morbidity due to drug toxicity. The construct validity of the models also varies with the floxed gene. To date, genetic studies have not indicated distur- bances of neurotrophic factors or their receptors as causes of PD, reducing the construct validity for the DAT-Ret lox ⁄ lox model. Dopamine neuron-specific dele- tion of Dicer induces decreased expression of the micr- oRNA miR133b, which reproduces a deficiency found in midbrain tissue from PD patients and gives this model good construct validity. There are several indi- cations that mitochondrial dysfunction plays a promi- nent role in the etiology and progression of PD, both from genetic studies (genetic variants of the mtDNA polymerase PolG have been associated with PD, and higher loads of mtDNA point mutations or deletions have recently been found in dopamine neurons from PD patients) and from toxin studies (dopamine neu- rons are more susceptible than other neurons to mito- chondrial toxins such as rotenone or paraquat than other neurons), conferring good construct validity also to the MitoPark model. What are the disadvantages of the tissue-specific knockout models? There are high costs of animal care, because of the slow development of the phenotype (for MitoPark mice, about 5 months, and for DAT- Ret lox ⁄ lox mice, more than 12 months), and because of complex breeding schemes (only 25% of the offspring in a litter have the affected genotype). In conclusion, several recently generated rodent models of PD reproduce more accurately the time course of key symptoms and neuropathology develop- ment seen in patients and are expected to further our understanding of PD etiologies and help in the devel- opment of new therapeutic strategies. Nevertheless, is it important to keep in mind that several other neuro- nal systems are affected in PD, changes that are not reproduced in these disease models. Acknowledgements This work was supported by The Swedish Research Council, The Swedish Brain Foundation, Swedish Brain Power, the Swedish Parkinson Foundation and Karolinska Institutet Funds. References 1 Schober A (2004) Classic toxin-induced animal models of Parkinson’s disease: 6-OHDA and MPTP. Cell Tissue Res 318, 215–224. 2 Ungerstedt U (1968) 6-Hydroxy-dopamine induced degeneration of central monoamine neurons. Eur J Pharmacol 5, 107–110. 3 Bove J, Prou D, Perier C & Przedborski S (2005) Toxin-induced models of Parkinson’s disease. NeuroRx 2, 484–494. 4 Emborg ME (2004) Evaluation of animal models of Parkinson’s disease for neuroprotective strategies. J Neurosci Methods 139, 121–143. 5 Cenci MA, Whishaw IQ & Schallert T (2002) Animal models of neurological deficits: how relevant is the rat? Nat Rev Neurosci 3, 574–579. 6 Lundblad M, Usiello A, Carta M, Hakansson K, Fi- sone G & Cenci MA (2005) Pharmacological validation of a mouse model of l-DOPA-induced dyskinesia. Exp Neurol 194, 66–75. 6a Santini E, Valjent E & Fisone G (2008) Parkinson’s dis- ease: Levodopa-induced dyskinesia and signal transduc- tion. FEBS J doi:10.1111/j.1742-4658.2008.06296 7 Langston JW, Ballard P, Tetrud JW & Irwin I (1983) Chronic Parkinsonism in humans due to a product of meperidine-analog synthesis. Science 219, 979–980. 8 Sedelis M, Hofele K, Auburger GW, Morgan S, Huston JP & Schwarting RK (2000) MPTP susceptibility in the mouse: behavioral, neurochemical, and histological analysis of gender and strain differences. Behav Genet 30, 171–182. 9 Fornai F, Schluter OM, Lenzi P, Gesi M, Ruffoli R, Ferrucci M, Lazzeri G, Busceti CL, Pontarelli F, Batta- glia G et al. (2005) Parkinson-like syndrome induced by continuous MPTP infusion: convergent roles of the ubiquitin–proteasome system and alpha-synuclein. Proc Natl Acad Sci USA 102, 3413–3418. 10 Maries E, Dass B, Collier TJ, Kordower JH & Steece- Collier K (2003) The role of alpha-synuclein in Parkin- son’s disease: insights from animal models. Nat Rev Neurosci 4, 727–738. 11 McCormack AL, Thiruchelvam M, Manning-Bog AB, Thiffault C, Langston JW, Cory-Slechta DA & Di Monte DA (2002) Environmental risk factors and Parkinson’s disease: selective degeneration of nigral dopaminergic neurons caused by the herbicide para- quat. Neurobiol Dis 10, 119–127. 12 Ossowska K, Smialowska M, Kuter K, Wieronska J, Zieba B, Wardas J, Nowak P, Dabrowska J, Bortel A, Biedka I et al. (2006) Degeneration of dopaminergic mesocortical neurons and activation of compensatory processes induced by a long-term paraquat administra- tion in rats: implications for Parkinson’s disease. Neuroscience 141, 2155–2165. Animal models of Parkinson’s disease M. Terzioglu and D. Galter 1390 FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS 13 Betarbet R, Canet-Aviles RM, Sherer TB, Mastroberar- dino PG, McLendon C, Kim JH, Lund S, Na HM, Taylor G, Bence NF et al. (2006) Intersecting pathways to neurodegeneration in Parkinson’s disease: effects of the pesticide rotenone on DJ-1, alpha-synuclein, and the ubiquitin–proteasome system. Neurobiol Dis 22, 404–420. 14 McNaught KS, Perl DP, Brownell AL & Olanow CW (2004) Systemic exposure to proteasome inhibitors causes a progressive model of Parkinson’s disease. Ann Neurol 56, 149–162. 15 Kordower JH, Kanaan NM, Chu Y, Suresh BR, Stan- sell J III, Terpstra BT, Sortwell CE, Steece-Collier K & Collier TJ (2006) Failure of proteasome inhibitor administration to provide a model of Parkinson’s dis- ease in rats and monkeys. Ann Neurol 60, 264–268. 16 Sawada H, Kohno R, Kihara T, Izumi Y, Sakka N, Ibi M, Nakanishi M, Nakamizo T, Yamakawa K, Shibasa- ki H et al. (2004) Proteasome mediates dopaminergic neuronal degeneration, and its inhibition causes alpha- synuclein inclusions. J Biol Chem 279, 10710–10719. 17 Fleming SM, Fernagut PO & Chesselet MF (2005) Genetic mouse models of parkinsonism: strengths and limitations. NeuroRx 2, 495–503. 18 Melrose HL, Lincoln SJ, Tyndall GM & Farrer MJ (2006) Parkinson’s disease: a rethink of rodent models. Exp Brain Res 173, 196–204. 19 Richfield EK, Thiruchelvam MJ, Cory-Slechta DA, Wuertzer C, Gainetdinov RR, Caron MG, Di Monte DA & Federoff HJ (2002) Behavioral and neurochemi- cal effects of wild-type and mutated human alpha-syn- uclein in transgenic mice. Exp Neurol 175, 35–48. 20 Martin LJ, Pan Y, Price AC, Sterling W, Copeland NG, Jenkins NA, Price DL & Lee MK (2006) Parkin- son’s disease alpha-synuclein transgenic mice develop neuronal mitochondrial degeneration and cell death. J Neurosci 26, 41–50. 21 Bohlen und HO (2005) Modeling neurodegenerative dis- eases in vivo review. Neurodegener Dis 2, 313–320. 22 Martins LM, Morrison A, Klupsch K, Fedele V, Moisoi N, Teismann P, Abuin A, Grau E, Geppert M, Livi GP et al. (2004) Neuroprotective role of the Reaper-related serine protease HtrA2 ⁄ Omi revealed by targeted deletion in mice. Mol Cell Biol 24, 9848–9862. 23 Biskup S & Moore DJ (2006) Detrimental deletions: mitochondria, aging and Parkinson’s disease. Bioessays 28, 963–967. 24 Sgado P, Alberi L, Gherbassi D, Galasso SL, Ramakers GM, Alavian KN, Smidt MP, Dyck RH & Simon HH (2006) Slow progressive degeneration of nigral dopami- nergic neurons in postnatal Engrailed mutant mice. Proc Natl Acad Sci USA 103, 15242–15247. 25 Nunes I, Tovmasian LT, Silva RM, Burke RE & Goff SP (2003) Pitx3 is required for development of substan- tia nigra dopaminergic neurons. Proc Natl Acad Sci USA 100, 4245–4250. 26 Ding Y, Restrepo J, Won L, Hwang DY, Kim KS & Kang UJ (2007) Chronic 3,4-dihydroxyphenylalanine treatment induces dyskinesia in aphakia mice, a novel genetic model of Parkinson’s disease. Neurobiol Dis 27, 11–23. 27 Zhuang X, Masson J, Gingrich JA, Rayport S & Hen R (2005) Targeted gene expression in dopamine and serotonin neurons of the mouse brain. J Neurosci Methods 143, 27–32. 28 Backman CM, Malik N, Zhang Y, Shan L, Grinberg A, Hoffer BJ, Westphal H & Tomac AC (2006) Charac- terization of a mouse strain expressing Cre recombinase from the 3¢ untranslated region of the dopamine trans- porter locus. Genesis 44, 383–390. 29 Turiault M, Parnaudeau S, Milet A, Parlato R, Rouzeau JD, Lazar M & Tronche F (2007) Analysis of dopamine transporter gene expression pattern – generation of DAT-iCre transgenic mice. FEBS J 274, 3568–3577. 30 Ekstrand MI, Falkenberg M, Rantanen A, Park CB, Gaspari M, Hultenby K, Rustin P, Gustafsson CM & Larsson NG (2004) Mitochondrial transcription fac- tor A regulates mtDNA copy number in mammals. Hum Mol Genet 13, 935–944. 31 Lindeberg J, Usoskin D, Bengtsson H, Gustafsson A, Kylberg A, Soderstrom S & Ebendal T (2004) Trans- genic expression of Cre recombinase from the tyrosine hydroxylase locus. Genesis 40, 67–73. 32 Gelman DM, Noain D, Avale ME, Otero V, Low MJ & Rubinstein M (2003) Transgenic mice engineered to target Cre ⁄ loxP-mediated DNA recombination into catecholaminergic neurons. Genesis 36, 196–202. 33 Borgkvist A, Puelles E, Carta M, Acampora D, Ang SL, Wurst W, Goiny M, Fisone G, Simeone A & Usi- ello A (2006) Altered dopaminergic innervation and amphetamine response in adult Otx2 conditional mutant mice. Mol Cell Neurosci 31, 293–302. 34 Baquet ZC, Bickford PC & Jones KR (2005) Brain- derived neurotrophic factor is required for the establish- ment of the proper number of dopaminergic neurons in the substantia nigra pars compacta. J Neurosci 25, 6251–6259. 35 Ekstrand MI, Terzioglu M, Galter D, Zhu S, Hofstetter C, Lindqvist E, Thams S, Bergstrand A, Hansson FS, Trifunovic A et al. (2007) Progressive parkinsonism in mice with respiratory-chain-deficient dopamine neurons. Proc Natl Acad Sci USA 104, 1325–1330. 36 Kim J, Inoue K, Ishii J, Vanti WB, Voronov SV, Mur- chison E, Hannon G & Abeliovich A (2007) A Micro- RNA feedback circuit in midbrain dopamine neurons. Science 317, 1220–1224. 37 Kramer ER, Aron L, Ramakers GM, Seitz S, Zhuang X, Beyer K, Smidt MP & Klein R (2007) Absence of Ret signaling in mice causes progressive and late degeneration of the nigrostriatal system. PLoS Biol 5, e39, doi:10.1371/journal.pbio.0050039. M. Terzioglu and D. Galter Animal models of Parkinson’s disease FEBS Journal 275 (2008) 1384–1391 ª 2008 The Authors Journal compilation ª 2008 FEBS 1391 . MINIREVIEW Parkinson’s disease: genetic versus toxin-induced rodent models Mu ¨ gen Terzioglu 1 and Dagmar Galter 2 1. tissue-specific knockout mouse models with older genetic and toxin- induced animal models (Fig. 1). Toxin-induced animal models Early animal models developed for