Neuropathology in Mice Expressing Mouse Alpha-Synuclein

α-Synuclein (αSN) in human is tightly linked both neuropathologically and genetically to Parkinson's disease (PD) and related disorders. Disease-causing properties in vivo of the wildtype mouse ortholog (mαSN), which carries a threonine at position 53 like the A53T human mutant version that is genetically linked to PD, were never reported. To this end we generated mouse lines that express mαSN in central neurons at levels reaching up to six-fold compared to endogenous mαSN. Unlike transgenic mice expressing human wildtype or mutant forms of αSN, these mαSN transgenic mice showed pronounced ubiquitin immunopathology in spinal cord and brainstem. Isoelectric separation of mαSN species revealed multiple isoforms including two Ser129-phosphorylated species in the most severely affected brain regions. Neuronal Ser129-phosphorylated αSN occured in granular and small fibrillar aggregates and pathological staining patterns in neurites occasionally revealed a striking ladder of small alternating segments staining either for Ser129-phosphorylated αSN or ubiquitin but not both. Axonal degeneration in long white matter tracts of the spinal cord, with breakdown of myelin sheaths and degeneration of neuromuscular junctions with loss of integrity of the presynaptic neurofilament network in mαSN transgenic mice, was similar to what we have reported for mice expressing human αSN wildtype or mutant forms. In hippocampal neurons, the mαSN protein accumulated and was phosphorylated but these neurons showed no ubiquitin immunopathology. In contrast to the early-onset motor abnormalities and muscle weakness observed in mice expressing human αSN, mαSN transgenic mice displayed only end-stage phenotypic alterations that manifested alongside with neuropathology. Altogether these findings show that increased levels of wildtype mαSN does not induce early-onset behavior changes, but drives end-stage pathophysiological changes in murine neurons that are strikingly similar to those evoked by expression of human wildtype or mutant forms.


Introduction
Disorders collectively referred to as the a-synucleinopathies include a number of clinically diverse neurodegenerative diseases that constitute a critical biomedical problem. Prevalent a-synucleinopathies include idiopathic Parkinson's disease (iPD), dementia with Lewy bodies (DLB) (7-30% dementia in elderly), the Lewy body variant of Alzheimer's disease (LBVAD) with rare forms in some familial forms of PD (fPD), the familial form of AD and Down syndrome, multiple systems atrophy (MSA), Hallervorden-Spatz disease (HSD), neurodegeneration with brain iron accumulation type-1 (NBIA-1), Niemann-Pick Type C Disease (NPC), parkinsonism-dementia complex of Guam (PDC-Guam), diffuse neurofibrillary tangles with calcification (DNTC) and pure autonomic failure [1]. The common neuropathological hallmarks in neurons and glia are microscopic proteinaceous inclusions, composed mainly of aggregated fibrillar alpha-synuclein (aSN). aSN is an abundant presynaptic protein in the brain. Its 140 amino-acid sequence is highly homologous across human, rat and mouse (for review see [2]). Initially, aSN microscopic aggregates were postulated to play a key role in the pathophysiology of a-synucleinopathies. Neurotoxicity findings implicate aSN protofibrils, soluble aSN protein complexes, posttranslationally modified forms of aSN (in particular nitrosylated), phosphorylated at serine 129 (Ser129), as well as mono-and di-ubiquitinated aSN forms [3]. In DLB brains more than 90% of the insoluble aSN is phosphorylated at Ser129 compared to about 4% phosphorylated at Ser129 in brains of normal individuals. Furthermore, Ser129 phosphorylated aSN is targeted to mono-and di-ubiquitination in a-synucleinopathy brains [4]. Extensive phosphorylation at Ser129 and/or its mono-and di-ubiquitination are critical events in the pathophysiology of aSN. However, direct experimental evidence supporting this notion is lacking and it is still debated whether these molecular forms of aSN are on the critical pathophysiological path rather than representing molecular epiphenomena of the disease process.
As multiple toxic mechanism have been proposed for aSN, it is important to determine which of its molecular forms are on the critical pathophysiological path. One main hypothesis of aSN toxicity is based on its capability to form toxic oligomers. Familial forms of Parkinson's disease possess mutant forms of aSN A53T and A30P (and E46K) that form oligomers more rapidly than wildtype aSN. In idiopathic forms of a-synucleinopathies that lack heritable aSN mutations, it is speculated that compromised handling of aSN and/or specifically modified forms are hampering aSN catabolism as well as that of other proteins. Oxidative damage of aSN could change aSN into toxic forms that trigger such a pathophysiological cascade [3].
It is unclear how critical to the disease process are some of the differences in aSN amino-acid sequence between human, rat and mouse. There is no solid evidence for endogenous mouse aSN coaggregating with human aSN expressed in transgenic rodent models [5,6,7,8,9]. Furthermore, non-fibrillar aSN neuropathology in brain regions of human aSN transgenic mice is prominent also in regions where neurons express little or no endogenous mouse aSN [6,7]. Some transgenic mouse models develop human-like fibrillar aSN structures and this may to a large extent depend on the transgene expression cassette that is used [7]. Thus, it appears that aSN pathology in transgenic species varies and is influenced by a number of experimental and endogenous factors. Knowing these factors could shed more light on genetic and environmental risk factors associated with diseases involving aSN. In an attempt to resolve some of these questions we generated transgenic mice overexpressing murine wildtype aSN driven by the Thy1 regulatory sequences enabling a direct comparison with previous human aSN transgenic lines generated previously in our laboratory [6].

Results
High expression of maSN mRNA and protein levels in the Thy1-maSN transgenic mouse line Three transgenic C57BL/6 mouse lines were produced that express different levels of the Thy1-mouse aSN (maSN) transgene ( Figure 1A). Two lines 1S14 and 1S16 had comparable low levels whilst the third line 1S13 expressed transgene mRNA ( Figure 1B upper part) and protein ( Figure 1B lower part) levels in brain that were up to 6-fold above endogenous aSN in wildtype mice as shown by quantification in Figure 1C. Transgene mRNA levels in 1S13 mouse line were comparable to those in two lines described previously, expressing the A53T fPD and wildtype form of human aSN (haSN) [6]. Our analyses described here focus on the line 1S13 (named Thy1-maSN hereafter).
Similar to the Thy1-haSN mouse lines [6], expression of Thy1-maSN transgene mRNA and maSN protein in Thy1-maSN mouse brain was widespread. This is illustrated by in situ hybridization ( Figure 1D) and aSN protein immunohistochemistry ( Figure 1E) in low-magnification sagittal brain sections from Thy1-maSN (aSN knock-out (KO) mouse brains served as a negative control). The overall expression pattern of the transgene in Thy1-maSN was also very similar to those reported for the two lines expressing haSN under the control Thy1 regulatory sequences [6]. Interestingly there was no apparent weight loss in Thy1-maSN mice until 6 months of age ( Figure 1G) in contrast to mice over-expressing haSN with an early-onset weight loss ( Figure S1A). Not until around 6-7 months of age Thy1-maSN mice stopped gaining weight and in addition start to display severe motor deficits. This is again in sharp contrast to Thy1-haSN mice that showed early-onset impairments of motor performance ( Figure S1 and [6]). Furthermore we observed increased mortality in Thy1-maSN mice compared to control wildtype (wt) littermates ( Figure 1H).

Overexpression of wildtype murine aSN leads to mild impairment of motor performance
We performed different behavioral studies to determine motor function. Thy1-maSN mice showed no difference in the open field paradigm. Neither velocity (Figure 2A) nor total activity ( Figure 2B) was changed. Furthermore, no difference could be detected in forelimb grip strength ( Figure 2C). Motor coordination was assessed using the accelerated rotarod task starting at two months of age. During the first four weeks, Thy1-maSN mice showed impaired motor learning but by 12 weeks of age and after a number of training sessions, the performance of Thy1-maSN mice was indistinguishable from wt mice up to the age of six months ( Figure 2D). From 6-7 months onwards, a steady and rapid decline in rotarod performance in Thy1-maSN mice became obvious ( Figure 2D). Interestingly no difference in light/dark cycle activity, assessed by an actimeter for 48 h, could be detected between Thy1-maSN and wt mice ( Figure 2E). In order to determine the anxiety of Thy1-maSN mice we performed dark-light box and elevated plus maze experiments ( Figure 2F,G). We observed similar latencies and total time spend in the lit compartment between wt and mutants in the dark-light box ( Figure 2F), suggesting no impact on anxiety. This was fortified using the elevated plus maze ( Figure 2G). It is remarkable that Thy1-maSN mice displayed a late-onset and much less pronounced motor impairment than transgenic mice expressing the haSN transgene with early-onset (already at 5 weeks of age) and steady decline in motor control ( Figure S1 and [6]).

Perikaryal and neuritic accumulation of maSN
Similar to earlier observations in mice expressing haSN forms [6] we found maSN expressed in many neurons in telencephalon, hippocampus, brainstem, cerebellar nuclei and spinal cord ( Figure 1E). The maSN expression in the hippocampus showed an increase in perikaryal and neuritic immunostaining for aSN and cerebellar nuclei respectively ( Figure S2). In a substantial neuronal subset expression of the transgene was sufficient for perikaryal and neuritic maSN accumulation, which did not change over time ( Figure S2). This is further demonstrated by maSN immunostaining of hippocampal neurons in mice expressing the Thy1-maSN transgene on a mouse genetic background with a disrupted endogenous aSN gene (aSN KO) ( Figure S3D). The specificity of the aSN immunostainings is illustrated by the very low levels of background staining in aSN KO mouse brain sections ( Figure S3C).

Prominent development of maSN pathology, axonal degeneration and breakdown of myelin sheaths in spinal cord and brainstem
Like Thy1-haSN mice [6], the Thy1-maSN mouse developed a pronounced aSN pathology in spinal cord around the age of 6 months. We located prominent perikaryal and neuritic aSN staining in sections through the anterior horn ( Figure 3A,B) and in addition strong ubiquitin immunoreactive motor neurons with spindle-shaped dilated proximal dendrites ( Figure 3C). Using an antibody specific for the serine 129 phosphorylated form of aSN (P-Ser129aSN), we found immunolabeling of motor neuron cell bodies and presynaptic boutons in transgenic ( Figure 3D) but not wt mouse spinal cord (not shown). The P-Ser129aSN antibody recognizes specifically a form of aSN that is phosphorylated at Ser129 and is abundant in a-synucleinopathy lesions in the diseased human and aSN transgenic brain, but not in normal mouse brains [10].
Until recently aSN axonal pathology was grossly underestimated although it is now documented in several of the transgenic animal models [6,7,11,12,13]. In this current study, we found that motor neuron pathology was accompanied by axonal pathology in spinal cord white matter ( Figure 3E-H). Immunostaining for ubiquitin ( Figure 3E), and Holmes-Luxol staining ( Figure 3F) revealed axonal degeneration in long white matter tracts of the spinal cord with breakdown of myelin sheaths into rows of myelin ovoids. Many axons in the cord and spinal roots were immunolabeled with aSN antibody ( Figure 3G). Central axons were often enlarged and a subpopulation immunolabeled with anti-P-ser129aSN ( Figure 3H) which consistently left unstained the same tissue in wt mice (not shown). Unlike in the transgenic lines expressing haSN [6], ubiquitin immunopathology was detected in every single Thy1-maSN mouse aged 6 months (see below).
The aSN histopathology in brainstem, cerebellum and spinal cord was accompanied by prominent astrogliosis ( Figure S4A-E), microgliosis (IBA1-positive cells; Figure S4F-H) and axonal degeneration (Campbell silver stainings; Figure S4I-K). Notably, other brain areas including hippocampus, cortex, striatum and thalamus showed little or none of these histopathological hallmarks despite many neurons showing high aSN transgene expression in these brain areas.

Synaptic defects in the neuromuscular junction
Aggregates and/or soluble forms of aSN are present in neuronal somata and dendrites under pathological conditions in human and aSN transgenic mouse brains as well as in cultured neurons [14]. This contrasts with aSN being mainly presynaptic under normal circumstances. With respect to adverse effects on the neuronal cell in its entirety, it remains unclear whether pre-or post-synaptic changes are compromised first. Interestingly, muscles contained small angulated fibers reminiscent of neurogenic muscular atrophy ( Figure 3I). In addition, we found that neuromuscular synapses showed signs of presynaptic degeneration although less pronounced as reported previously in lines expressing haSN [6]. a-Bungarotoxin staining patterns for postsynaptic acetylcholine receptors were not different between wt and Thy1-maSN mice (age 6 months) ( Figure 3J,N), neither in soleus (slow-twitch, Figure 3J-Q) nor extensor digitorum longus (EDL, fast-twitch) muscles (not shown). Also, we detected little or no changes between wt and transgenic mice in presynaptic synaptophysin staining ( Figure 3K,O). In contrast, staining of presynaptic neurofilaments differed dramatically. The neuromuscular junctions in Thy1-maSN mice showed thinning or absence (not shown) of presynaptic neurofilament staining ( Figure 3L,P). In summary, neuromuscular junctions in Thy1-maSN mice showed degeneration that was independent of muscle fiber type and similar, as reported for mice expressing haSN transgene [6].

Phosphorylated aSN is expressed in hippocampal neurons that lack ubiquitin pathology
Strong aSN immunoreactivity could be observed in perikarya and dendrites, mainly in the CA3 region of the hippocampus ( Figure 4A). Immunostaining with anti-P-Ser129aSN showed the abundant presence of the phosphorylated form of aSN in many neurons and throughout brain regions where the transgene is expressed ( Figure 4B,C). In the hippocampus, a significant subset of neurons showed perikaryal and dendritic accumulation of aSN ( Figure 4C). The more striking observation was that in CA1 and CA3 hippocampal neurons, when co-immunolabeled for aSN and P-Ser129aSN, the latter was prominently localized in the nucleus ( Figure 4C). Similar to a previous study [6], the hippocampus lacked neurons immuno-positive for ubiquitin (not shown). Also in cortex only very few neurons showed ubiquitin pathology and/or a perikaryal accumulation of aSN ( Figure 4E,H).

Non-overlapping phosphorylated aSN and ubiquitin puncta along neuronal processes
Double labeling for ubiquitin and P-Ser129aSN was carried out in paraffin sections of neurons in regions such as the cortex, where only very few cells stained for ubiquitin ( Figure 4D-I) and additionally in regions with pronounced ubiquitin pathology such as brainstem, colliculus and spinal cord (not shown). This revealed an extraordinary staining pattern, in particular, in processes. As shown, P-Ser129aSN and ubiquitin immunopositive stretches in processes alternate and did not overlap ( Figure 4J). In contrast, in cell somata, the distribution patterns of P-Ser129aSN and ubiquitin were strikingly similar and overlaped to a high extent ( Figure 4D-I).
Thy1-maSN mice displayed abnormal mitochondria, demyelination, axonal loss and non-fibrillar amorphous aggregates Enlarged mitochondria are a sign of cells trying to compensate for energy deficits, reflecting local increases in the need for energy, vacuolization and/or loss of inner-outer membrane integrity of the mitochondria. We found grossly enlarged mitochondria with an abnormal high number of cristae but without any obvious vacuolization in spinal cord dendrites of Thy1-maSN mice ( Figure 5A). Spinal cord axonal degeneration was also evident by an accumulation of pathological organelles (including mitochondria) in the axoplasma ( Figure 5B) with disappearance of the axon, loosening of the myelin wraps and vesicular disruption of the myelin sheath ( Figure 5C). Interestingly, the axon showed pronounced beading with focal anti-P-Ser129aSN staining on a background of diffuse aSN immunostaining ( Figure 5D). Immunoelectron microscopy (10 nm immunogold) showed P-Ser129aSN antibodynegative immunostained neurofilaments with side-branches protruding from the filaments in Thy1-maSN ( Figure 5E). We found that aSN over-expression results in short, thick, and less well oriented filaments of approximately 10 nm in diameter. They were devoid of side-branches, focally decorated by gold particles and coincide with non-fibrillar amorphous aggregates ( Figure 5E,F) similar as the granular aggregates found in Thy1-haSN mice [6].
To provide biochemical evidence for the observed aggregation of aSN in Thy1-maSN mice, we perfomed solubility assays (see [13]). Brainstems from 3-5 mice were pooled, yielding matched starting tissue wet weight of 0.2 g. Tissue was homogenized in Tris buffer, and the buffer-insoluble material was dissolved in 1% Triton X-100. Such fractions were immunoblotted and probed with anti-aSN. The endogenous aSN from wt mice was mostly recovered in the buffer-soluble fraction, as expected ( Figure 6). Transgenic mice showed the increased expression of aSN in the buffer-soluble fraction. In addition, Thy1-maSN mouse tissues contained also buffer-insoluble aSN ( Figure 6). Importantly, and consistent with the age-dependent aggravation of neuropathology (see above), the amount of buffer-insoluble aSN increased when comparing 2-3 months with 5-6 months old mice ( Figure 6A). Thus, the increase of insoluble aSN in these mouse brains was not simply due to higher total aSN expression, but seems to indicate a shift towards insolubility with age. The amounts of soluble and insoluble aSN in 5-6 months old heterozygous Thy1-maSN Various post-translational modified aSN isoforms expressed in neurons of Thy1-maSN mice Isoelectric focusing Western blotting using several antibodies were performed to characterize the aSN isoforms expressed in the brain of Thy1-maSN mice (Figure 7). We found a novel aSN isoform specific to colliculus and brainstem, the two regions with extensive ubiquitin pathology ( Figure 7A,B,D). Importantly the novel P-Ser129aSN isoform is not detected using aSN antibodies targeting the C-terminus (Fig. 7C,E) and additionally not present in previously characterized mouse lines expressing haSN [6] ( Figure 7F).
Summarizing the expression of maSN isoforms in mice showed pronounced ubiquitin immunopathology in spinal cord including a novel aSN isoform. Additionally, we observed a strong aSN pathology in the spinal cord accompanied with axonal degeneration. These findings were followed by signs of presynaptic degeneration with reduced neurofilament staining in neuromuscular junction synapses. Interestingly, hippocampal neurons showed strong aSN accumulation but no ubiquitination in contrast to spinal cord motor neurons. Furthermore, we showed that few neurons in the cortex display an intriguing staining pattern of ubiquitin and phosphorylated maSN, suggesting that these posttranslational modifications play a role in trafficking and localization of aSN.

Discussion
Transgenic animals are considered excellent preclinical models to study a-synuclein (aSN) disease pathophysiology and test therapeutic strategies. Here we show that wildtype murine aSN can induce pathological changes in mouse brain closely resembling those observed in post-mortem human PD and DLB brains. These transgenic mice are very similar to those over-expressing human wildtype or the familial PD point-mutated A53T aSN [6]. Van der Putten et al. used the same Thy1 promoter to drive comparable aSN levels in similar brain regions compared to our Thy1-maSN transgenic mice. This is very intriguing since for the first time we show that murine aSN as well as human aSN can be pathogenic in neurons in vivo.
Interestingly, profound neuropathological changes could be detected solely in spinal cord, brainstem and cerebellum (after six months of age). Forebrain areas were always histopathologically unaffected despite a strong murine aSN over-expression. We show that neurons in the forebrain (like CA1 pyramidal cell) displayed the same strong somatic aSN staining as cells in the brainstem and cerebellum. Thus, this abnormal somatic accumulation of aSN does not account for the difference of histopathology observed for the different brain areas leading to the conclusion that rather the endogenous 'normal' aSN expression pattern might be responsible for the different neuronal vulnerability. For example, in wildtype (wt) animals, aSN is highly expressed in forebrain regions compared to brainstem and cerebellum. This suggests that neurons with low endogenous aSN levels are more sensitive to over-expression of aSN at a certain age. Unfortunately, the short life expectancy of Thy1-maSN mice hindered a possible histopathology in the forebrain later in development. A forebrain-specific expression (e.g. aCamKII) could unravel this issue.
In solution aSN is natively unfolded, whereas the acidic phospholipids are a-helical in nature. In cells, the structure of aSN is highly dynamic forming equilibrium between oligomers, bsheets and large aggrosomes that is context dependent [15,16]. We observed in brains of Thy1-maSN (,6 months of age) several isoforms of maSN whereas one of it was specifically restricted to brainstem and colliculus, areas of late-onset neuropathology. This specific isoform was Ser129-phosphorylated (P-Ser129aSN) and notably, could be detected before the appearance of any immunopathology. Moreover, this specific isoform could not be detected with antibodies targeting the C-terminus, which might be due to the heavy phosphorylation of this isoform that hindered antibody binding. Brain areas without any late-stage histopathology were devoid of this isoform, while the abundance of other isoforms was unchanged. The identification of the sequence of this isoform might give more insights about its function and its involvement in the neuropathological process.
However, we could not observe any brain area specific isoform in mice over-expressing haSN which could merely be related to the antibodies used. Thus, the P-Ser129aSN isoform could be attributed to the brain area-specific late-onset neuropathology observed in Thy1-maSN mice. Soluble forms of aSN, aSN fibrils and protofibrils, soluble protein complexes of aSN with 14-3-3 protein, phosphorylated, nitrosylated, and ubiquitinated aSN species have all been implicated in neurotoxicity [2,17].
There is a cause-and-effect relationship between Ser129phosphorylation of aSN and the disease. Noted in several animal models was an accumulation of Ser129-hyperphosphorylated, nitrosylated, ubiquitinated, and/or fragmented aSN species [6,7,8,9,13,18]. P-Ser129aSN yields more insoluble sediments and oligomers as compared to its non-phosphorylated counterpart. Moreover, the aSN pathology in human brain, transgenic mouse brain and transgenic fly neurons are enriched in Ser129hyperphosphorylated aSN [10,13,19,20]. Our current findings suggest that Ser129-phosphorylation of aSN by itself is not sufficient to cause aSN pathology in neurons in vivo. We found that over-expression of wildtype maSN greatly enhanced levels of P-Ser129aSN in different brain regions, but only some of these regions show cellular hallmarks of aSN pathology. Posttranslational modification of Ser129 by phosphorylation seems therefore part, but not the whole reason to convert the aSN molecule into toxic entities. Transgene expression of a phosphorylation-defective Ser129 substitution mutant Ala129 is now needed to experimentally confirm this hypothesis.
In several transgenic models aSN causes an impairment of neuronal function and the development of aSN micro and/or macro aggregates (amorphous, granular and/or fibrillar). Some have been shown to be proteinase-resistant aSN aggregates [13]. The pathology has been observed in synapses, axons, dendrites and neuronal cell somata. Occasionally, there was also inclusion formation in glia. However, in the majority of transgenic rodent models aSN aggregates appeared non-fibrillar, granular and/or amorphous at the electron microscopic level [13]. This is in line with our observations of non-fibrillary amorphous aggregates. Biochemical extractions did show an age-dependent shift into buffer-insoluble fractions, but no detergent resistance of aSN in Thy1-maSN mouse brains. Rarely, investigators have reported fibrillar aSN structures similar to filamentous aSN observed in human samples [7,13]. Altogether the findings seem to suggest that fibrillar, as well as other types of aSN aggregates, are associated with pathophysiological effects of aSN in vivo [6,8,9,18].
Phosphorylated mono-and di-ubiquitinated aSN forms exist in human brains with a-synucleinopathy suggesting that phosphorylated aSN is targeted to mono-and di-ubiquitination [4]. UCH-L1 and/or Parkin mediated ubiquitination of aSN could control its function, catabolism/stability, localization and interaction with other molecules and levels of toxic protofibrils, although knock-out of UCH-L1 in Thy1-maSN had no effect on aSN metabolism and localization (unpublished observation). Ubiquitination in Thy1-maSN transgenic mice was evident along with the other histopathologies in every mouse that we analysed and was restricted to brainstem, cerebellum and spinal cord. This is in contrast to mice over-expressing the human form of aSN showing ubiquitination only sporadically [6]. The amino-acid sequence of human and mouse aSN are very similar and differ only at seven positions. This includes position 53, where in wildtype mouse aSN the amino-acid threonine instead of alanine is present. The human pathogenic mutation A53T hence naturally exists in mouse and thus, cannot account for the difference observed in ubiquitination in mice over-expressing wildtype mouse and human A53T pointmutated aSN. This suggests that rather the six other amino-acid differences between human and mouse that are located downstream of position 53 guide the ubiquitin pathology. In human brains, ubiquitinated structures represent mainly classical Lewy bodies and Lewy neurites [21]. However, the brain areas positive for strong ubiquitination in our mice did not display any enhanced degree of aSN aggregation. This indicates that ubiquitination might not be required for the formation aSN inclusions as described elsewhere [22].
Mice over-expressing murine wildtype aSN showed no early motor deficits aside from minor motor learning impairment. Motor impairments are not obvious until 6 months of age that coincide with a rapid decline in health, resulting in death of the animal. This is in sharp contrast to mice over-expressing human aSN displaying early-onset motor deficits. Hence, murine aSN in 6-fold higher dose unlike human aSN somehow might not interfere with normal neuronal function in early development.
All Thy1-maSN animals analysed developed alongside these motor deficits degeneration of the NMJ, astrogliosis, microgliosis, axonal and ubiquitin pathology. Classical Lewy body-like structures were not observed. Unfortunately, the Thy1 promoter fails to express in dopaminergic neurons (data not shown) and thus, no pathology in the substantia nigra pars compacta could be observed. Although, these cells in the human brain are most sensitive to develop Lewy pathology, and the resulting nigral lesions are primarily involved in the clinical symptoms of PD [23,24,25,26], extranigral Lewy pathology is very common in PD and LBD brains [23,27,28,29].
Moving towards disease-modifying therapies requires a general understanding of the role of (epi)genetic and environmental factors. Moreover, insights into the presymptomatic/symptomatic changes and of the molecular identity of the culprit(s) and pathway(s) that drive disease process are necessary. We also need diagnostic tools, biomarkers and translational animal models that mimic aSN-induced pathophysiological changes and allow testing of the effects of drugs, antibodies, genes and RNAi that halt and/ or reverse disease. Human mutant, human wt, and mouse wt aSN drive disease pathophysiology and loss of neuronal cell function. These transgenic mice display many hallmark features of human pathology and provide means to address fundamental aspects of disease pathophysiology, explore surrogate markers, test therapeutic strategies with behavioural and biochemical read-outs and provide a good model for extra-nigral a-synucleinopathy.

Statement on Animal Health
All experiments were carried out in accordance with authorization guidelines of the Swiss Federal and Cantonal veterinary offices for care and use of laboratory animals. Studies described in this report were approved by the Swiss Cantonal veterinary office and performed according to Novartis animal license number 2063.

Northern blot analysis
Northern blot analysis was carried out with total brain RNA (TriZol method; 10 mg loaded per gel lane). Blots were probed with a full length (449 bp) mouse a-synuclein cDNA using standard procedures.

In Situ hybridization
The spatial distribution pattern of transgene versus endogenous a-synuclein expression was determined by in situ hybridization as described before [6] using cRNA derived from a 449 bp complete coding cDNA template of mouse a-synuclein.
For solubility assays, pooled frozen brainstems (matching 0.2 g starting wet weight) were homogenized in 10 volumes buffer A (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol) using a Teflon potter. The homogenate was cleared by centrifugation for 10 min at 10006 g. Cleared homogenates were centrifuged for 20 min at 350,0006 g. The buffer-insoluble pellets were resuspended in buffer A+1% Triton X-100, followed by another centrifugation for 20 min at 350,0006 g. All steps were carried out on ice. Preparation of purifed recombinant aSN control protein was described elsewhere [31]. Samples were boiled in Lä mmli buffer and subjected to denaturing 12.5% polyacrylamide gel electrophoresis. Western blots were prepared and probed with monoclonal anti-aSN (1:250; Transduction Laboratories). Ponceau red staining of the polyvinylidene fluoride membranes confirmed equal loading.
Immunostaining and confocal analysis of neurofilament and synaptophysin at neuromuscular junctions (NMJs) were as follows. Mice were killed by anaesthesia. Extensor digitorum longus (EDL) and soleus muscle were stained with Alexa Fluor 488-labeled abungarotoxin (1:200, Molecular Probes) for 30 min, washed with PBS (3615 min) and fixed with 1.5% paraformaldehyde for 10 min. Muscles were teased into approximately 20 thin bundles and permeabilized with 1% Triton X-100 in PBS for 1 h. Bundles were treated with 100 mM glycine in PBS, followed by ''blocking solution'' of 1% BSA in PBS, for 30 min. Then, they were incubated overnight at 4uC with a mixture of primary antibodies against synaptophysin (1:200, DAKO) and neurofilament (1:1000, MAB 1621, Chemicon) in blocking solution and washed three times 1 h in blocking solution. The bundles were incubated with a mixture of Cy3-labeled goat anti-mouse IgG (1:1000, Jackson ImmunoResearch) and Cy5-labeled goat anti-rabbit IgG (1:500, Jackson ImmunoResearch) in blocking solution for 45 min at rommtemperature. After washing three times for 1 h with blocking solution, bundles were mounted on glass slides using Citifluor (Plano), and examined with a confocal microscope (Leica TCN NT).

Immunoelectron and electron microscopy
For immunoelectron microscopy, transgenic and wildtype C57Bl/6 mice were perfused transcardially with a mixture of 1.5% picric acid, 0.1% glutaraldehyde, and 4% paraformaldehyde in 0.1 Mphosphate buffer, pH 7.4. Vibratome sections were stained free-floating with antibody to P-Ser129 a-synuclein (1:2000, anti-P-Ser129 a-synuclein, WAKO) dehydrated in ascending series of ethanol and acetone, and flat-embedded between glass slide and coverslip in Embed-812 (Electron Microscopy Sciences). Fragments of the spinal cord were then dissected out and ultra-thin sections were cut from the tissue surface, and these were mounted on copper grids and analyzed with a microscope (EM900, Zeiss). For conventional electron microscopy, mice were anesthetized and perfused transcardially with cold saline, followed by 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M cacodylate buffer. Small tissue blocks were cut out from brainstem and spinal cord, immersion-fixed for 12 h at 4uC in the same buffer, and epoxy-embedded, and ultrathin sections were prepared and placed on 200-mesh copper grids for staining with uranyl acetate and lead citrate.

Behavior
Rotarod. To measure motor coordination mice were placed on a computerized treadmill (TSE rotarod system, Germany). The rotarod program consists of accelerating running speed from 5 rpm to 36 rpm in 5 min. Rotarod performance was assessed by evaluating the two best trials out of three performed in one day. The 3-step rotarod consists of a modified rotarod program of three different running speeds (12 rpm, 24 rpm and 36 rpm) each for 30 sec with intervals of acceleration lasting for 10 sec. Starting speed is 4 rpm. Rotarod performance was assessed by evaluating the two best trials out of three performed in one day.
Grip strength. To measure forelimb grip strength, mice are allowed to grasp a handle connected to a force-measuring device (San Diego Instruments, USA) and then pulled back with their tails until they release the handle. The best out of four consecutive trials is evaluated.
Open field. To measure exploratory behavior (pattern and activity), mice were placed in an open field box (70 cm670 cm, height of walls: 30 cm) subdivided into nine quadrants with one middle quandrant. The horizontal distance travelled during 5 min was recorded by an EthoVision 3.0 system (Noldus, The Netherlands). In addition the number of rearings was determined by visual inspection.
Actimeter. Recordings were made in automated circular corridors (Imetronic, France) for 48 h. These corridors with a radius of 4.5 cm and a width of 5.3 cm were equipped with 4 photocells, equidistant of 7 cm and 45u from each other, connected to an electronic interface, itself connected to a computer. Motor activity corresponds to the number of photocell interruption per time unit (20 min) and locomotor activity corresponds to the number of quarter turns corresponding to the successive interruption of two photocells. The dark and light phases lasted 12 h each.
Dark/light box. The dark/light box consists of a dark and a bright compartment. Mice were placed in the bright compartment and given the opportunity to move to the dark box for 5 min. Parameters measured by EthoVision 3.0 (Noldus, The Netherlands) were the time spent in the bright compartment and the latency of first entry to the dark compartment. The number of transitions and the latency of first exit back to the bright compartment were measured visually.

Maintenance
The animals were housed in a temperature-controlled room that was maintained on a 12 h light/dark cycle. Food and water were available ad libitum.