Overexpression of human α-synuclein (α-syn) using recombinant adeno-associated viral (rAAV) vectors provides a novel tool to study neurodegenerative processes seen in Parkinson's disease and other synucleinopathies. We used a pseudotyped rAAV2/5 vector to express human wild-type (wt) α-syn, A53T mutated α-syn, or the green fluorescent protein (GFP) in the primate ventral midbrain. Twenty-four adult common marmosets (Callithrix jacchus) were followed with regular behavioural tests for 1 year after transduction. α-Syn overexpression affected motor behaviour such that all animals remained asymptomatic for at least 9 weeks, then motor bias comprising head position bias and full body rotations were seen in wt-α-syn expressing animals between 15 and 27 weeks; in the later phase, the animals overexpressing the A53T α -syn, in particular, showed a gradual worsening of motor performance, with increased motor coordination errors. Histological analysis from animals overexpressing either the wt or A53T α -syn showed prominent degeneration of dopaminergic fibres in the striatum. In the ventral midbrain, however, the dopaminergic neurodegeneration was more prominent in the A53T group than in the WT group suggesting differential toxicity of these two proteins in the primate brain. The surviving cell bodies and their processes in the substantia nigra were stained by antibodies to the pathological form of α-syn that is phosphorylated at Ser position 129. Moreover, we found, for the first time, ubiquitin containing aggregates after overexpression of α-syn in the primate midbrain. There was also a variable loss of oligodendroglial cells in the cerebral peduncle. These histological and behavioural data suggest that this model provides unique opportunities to study progressive neurodegeneration in the dopaminergic system and deposition of α-syn and ubiquitin similar to that seen in Parkinson's disease, and to test novel therapeutic targets for neuroprotective strategies.
Parkinson's disease is mainly characterized by loss of dopamine (DA) cells in the substantia nigra (SN) and the presence of filamentous intracytoplasmic inclusions called Lewy bodies (LB) (Forno, 1996; Dunnett and Bjorklund, 1999). After the description of an autosomal dominant form of Parkinson's disease, due to a point mutation alanine to threonine at position 53 (A53T) in the α-synuclein (α-syn) gene, it was found that α-syn was a major component of the LBs seen in familial and sporadic Parkinson's disease, and the glial inclusions seen in multiple system atrophy (MSA) (Polymeropoulos et al., 1997; Spillantini et al., 1998; Galvin et al., 2001). The importance of α-syn in Parkinson's disease was further supported with identification of two more point mutations (A30P or E46K) (Kruger et al., 1998; Zarranz et al., 2004). Additional compelling evidence that α-syn could be toxic to DA neurons came from the discovery that a multiplication in part of chromosome 4 including the coding region for α-syn was linked to Parkinson's disease, confirming that overexpression of normal α-syn can induce dopaminergic cell death (Singleton et al., 2003; Chartier-Harlin et al., 2004; Ibanez et al., 2004).
In Parkinson's disease brains, the surviving neurons show abnormal accumulations of various cellular proteins, possibly secondary to oxidative modifications or as a consequence of impairments in the degradation pathways (review in Cookson, 2005). Nitrated and phosphorylated forms of α-syn have been identified in LBs in post-mortem Parkinson's disease brain samples (Giasson et al., 2000; Fujiwara et al., 2002). While DA neurons might be particularly vulnerable to α-syn toxicity, they are not the only affected cells in the brain as evidenced by the fact that abnormal α-syn-containing inclusions are also found in the glial cells in MSA patients (Wenning and Jellinger, 2005).
The manipulation of α-syn levels in animals [e.g. through over-expression of wild-type (wt) or mutant forms of the human protein] provides a useful model of Parkinson's disease comprising the key pathological features, of α-syn-containing intracytoplasmic inclusions, DA cell and fibre loss and motor impairments. Genetically modified flies expressing human wt and mutant α-syn show age-dependent loss of DA neurons, cytoplasmic α-syn-containing inclusions and motor deficits, and provide a very useful tool for drug-screening in vivo (Feany and Bender, 2000). Although several lines of transgenic mice overexpressing full-length human α-syn have been developed, motor deficits related to dopaminergic neurodegeneration were found only in one of these lines, while in the other mouse strains behavioural outcome appeared to be related to degeneration in other areas, as no pathology was found in the DA system (see Fernagut and Chesselet, 2004 for review). Recently, Tofaris and collaborators showed that overexpression of a C-terminal truncated form of the human α-syn under the tyrosine hydroxylase promoter results in pathological accumulations in the nigral neurons and a decrease in striatal DA content; this leads to abnormal motor activity suggesting that neuropathology in DA neurons can be induced by the truncated α-syn species (Tofaris et al., 2006). While transgenic mouse lines remain as the gold-standard for overexpression studies in vivo, it has been difficult to express high levels of disease causing proteins specifically in nigral DA neurons. However, this limitation has been overcome by targeted overexpression of α-syn using intracerebral injection of viral vectors unilaterally in the midbrain. Using either recombinant adeno-associated viral (rAAV) or lentiviral vectors, it was possible to get selective yet progressive neuronal pathology in DA cells and fibres and α-syn-containing inclusions in the rat (Kirik et al., 2002; Klein et al., 2002; Lo Bianco et al., 2002; Yamada et al., 2004). In addition, a key advantage of the viral vector approach is that it can be scaled up to larger species including the primate. In fact, we have earlier provided proof-of-principle that overexpression of human full-length α-syn in the primate brain results in formation of α-syn-containing cytoplasmic inclusions and granular deposits and loss of DA cells in the SN (Kirik et al., 2003).
In the present study, we report detailed characterization of the long-term behavioural and pathological consequences of α-syn overexpression in marmoset monkeys. For this purpose, the animals were injected with rAAV vectors expressing either human wt or A53T α-syn, or green fluorescent protein (GFP) unilaterally in the ventral midbrain and were assessed behaviourally prior to surgery and for a period of 1 year after transduction.
Material and methods
Experimental procedures were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986. Thirty laboratory-bred adult common marmosets (Callithrix jacchus), 17 females and 13 males, were used. The monkeys weighed 309–497 g and were between 21 and 66 months old (mean ± SEM = 39.1 ± 3.0 m) at the start of the experiment. All animals were housed in pairs; where the pairs were of opposite sexes the male was vasectomized.
The goal of this study was to determine the effect of sustained unilateral expression of wt or mutated α-syn in the SN of marmoset monkeys following intracerebral injection of appropriate viral vectors. There were three groups of 10 monkeys, each group injected with one of three vectors: (i) rAAV2/5-CBA-α-syn encoding for human wt α-syn (WT group); (ii) rAAV2/5-CBA-A53T-α-syn encoding for the A53T mutated form of human α-syn (A53T group); (iii) rAAV2/5-CBA-GFP encoding for GFP group used as control. In order to define the profile of transgene expression two animals from each group were killed 3 weeks after surgery. In order to demonstrate the progression of behavioural changes, the remaining monkeys (n = 8/group) were assessed behaviourally in two batteries of tests prior to the surgery and following transduction with viral vectors over a period of 1 year. Details of the experimental time-line are illustrated in Fig. 1.
Time course of the experiment. All monkeys received injection of a rAAV2/5 vector coding for human wt or A53T α-syn, or the GFP protein. They were monitored during 1 year on several behavioural tests classified in two batteries. A first battery of tests was based on observation of animals’ motor coordination that did not require training, and included head position bias, full body rotations, cage crosses and motor coordination, and head tilt (see above the time-line). The second battery consisted of motor performance tasks that required training of the animals, and included the six-tube search task, two-tube choice task, and Hill and Valley versions of the staircase task (see below the time-line). After killing at 52 weeks, ex vivo high-resolution 3D MR scans were performed followed by histological processing.
Recombinant AAV vector preparation
The three vectors used in this experiment were constructed and titered at the University of Florida Gene Therapy Center Vector Core. The rAAV2/5-CBA-α-syn vector contains the coding sequence for the human α-syn gene under the control of the synthetic chicken beta actin (CBA) promoter (Xu et al., 2001; Kirik et al., 2002; Kirik et al., 2003), and the human bovine growth hormone poly A site, flanked by AAV2 ITRs. The rAAV-CBA-A53T-α-syn mutant has the same features as the WT vector except for the mutated amino acid (an alanine to threonine substitution at position 53). The rAAV2/5-CBA-GFP vector contains the coding sequence for humanized GFP (Zolotukhin et al., 1996) under the control of the synthetic CBA promoter (Xu et al., 2001) and the SV40 polyadenylation signal followed by the neomycin resistance gene under the control of the mutant polyoma virus enhancer/promoter (PYF441) and the human bovine growth hormone poly A site, flanked by AAV2 ITRs. Pseudotyped rAAV2/5, (rAAV2 vectors packaged in AAV5 capsids) was purified by iodixanol step gradients and Sepharose Q column chromatography, as described in detail elsewhere (Zolotukhin et al., 2002). Vector titres were determined by dot blot assay as previously described (Zolotukhin et al., 1999). The final titre for vectors encoding wt-α-syn, A53T-α-syn and GFP genes were 2.1 × 1013, 4.9 × 1013 and 2.9 × 1013 genome copies/ml, respectively as determined by dot blot.
For intracerebral vector injections, monkeys were premedicated with 0.05 ml ketamine (Vetalar; Schering-Plough Ltd, Welwyn Garden City, UK; 100 mg/ml, intramuscularly) and anaesthetized with 0.5 ml alphaxalone–alphadolone (Saffan; Schering-Plough Ltd, Welwyn Garden City, UK; 12 mg/ml given intramuscularly). A supplementary dose of 0.2–0.3 ml Saffan was administered intramuscularly during surgery if necessary. The monkeys were administered 0.03 ml of the analgesic carprofen prior to surgery (Rimadyl; Pfizer, Sandwich, UK; 50 mg/ml given subcutaneously).
The injections of viral vectors were made with a 28 g 10 µl Hamilton needle at a rate of 0.5 µl/min. In each animal two injections of 2 µl each were made unilaterally into the SN, via a small cranial burr hole, in either the left or the right of the brain at anterioposterior (AP), mediolateral (ML) and dorsoventral (DV) coordinates: (i) AP 5.5, ML 2.5 and DV 6.6; (ii) AP 4.5, ML 2.5 DV 6.6, according to the stereotaxic atlas (Stephan et al., 1980). For each group, half the animals were injected in the left hemisphere and half in the right. After each injection, the needle was kept in place for an additional 5 min before it was withdrawn and the skin sutured. Three different sets of instruments, one for each vector, were used to eliminate the risk of cross contamination from the needles.
Behavioural data were collected by two observers (A.E. and R.M.C.) who were unaware of the group identity of the vector treated monkeys. Inter-rater reliability between the two observers was checked on two of the subjective measures and the correlations were high (for head turn bias, r = 0.985, and for motor error scores, r = 0.99). Two batteries of behavioural tests were performed. The first battery was based on the observation of the animals’ general motor coordination that did not require training or interaction with the animal. These tests were repeated every 6 weeks throughout the experiment. The second one included behavioural tasks that required training of the animal, and were performed prior to and post-surgery at time points as indicated below and shown in Fig. 1.
General motor-coordination tests
The observer recorded the position of the monkey's head with respect to the rest of its body (i.e. looking left, right or straight ahead) each second for three separate 1 min intervals to the beat of a metronome (1 beat/s). Recordings were made for 3 (usually consecutive) days. For each monkey the daily score consisted of the average number of seconds per day spent with the head turned ipsilesionally minus the number of seconds spent with the head turned contralesionally. The final score was an average across days. Head turn is known to be sensitive to unilateral loss of DA in marmoset monkeys (Annett et al., 1992).
In order to quantify this novel behaviour, the observer recorded the tilt of the animal's head, i.e. when one eye was higher than the other, each second for 1 min to the beat of a metronome (1 beat/s). Recordings were made for 3 min. The score consisted of the average number of seconds per minute when the ipsilesional eye was higher than the contralesional eye minus the number of seconds per minute when the contralesional eye was higher than the ipsilesional eye. Head tilt and head turn (see aforementioned) could occur simultaneously but be recorded separately.
Cage crosses and motor coordination
Prior to the observation periods, all toys, etc., were removed from the cage and two wooden bars were placed from the top back to the lower front of the cage, crossing in the middle. The monkeys were observed and their behaviour captured on videotape for 15 min. The level of spontaneous activity was determined by counting the number of times the animal crossed from the front of the cage to the back. Motor coordination was determined by counting the number of motor errors for each hand and foot. These errors were defined as: the hand or foot slipping off the bar when the animal was climbing; the hand or foot being held in the air in an unnatural position while the monkey was scent marking or chewing the bar; the hand or foot dangling in the air while the monkey was resting on the bar or perch; the hand or foot slipping while the animal was climbing up or down the cage front or around the cage and the hand or foot being dragged while the monkey moved around the cage. Errors consisting of an overall loss of balance, which was not attributed to an error due to an individual hand or foot, were also counted. (Note that an error of balance as a consequence of a foot slip was only counted as a foot slip.)
Full body rotations
The number of full rotations made by each monkey in each direction was counted during the 15 min tape that was used for motor-coordination and motor errors as well. The final score consists of the number of ipsilesional rotations minus contralesional rotations as previously described (Milton et al., 2004).
Motor performance tasks
Six-tube search task
This task measures the unconstrained monkey's exploration of its peripersonal space (Marshall et al., 2002). For each trial, the monkey was required to search for, and retrieve a food reward (a small piece of marshmallow) that was placed inside only one of the six tubes, which were presented to the monkey in a horizontal array as illustrated in Fig. 2D. The tubes were black 35 mm film canisters, open at the top, fixed side by side on a wooden base. The task consisted of a total of 30 trials, with the location of the reward changing in a fixed random order, such that each tube was baited a total of five times. The time to retrieve the reward was recorded with a maximum of 30 s for each trial. Monkeys were deemed to have reached the pre-surgery criterion once they were able to find and remove each reward in under 6 s. Normal animals tend to look into and retrieve rewards from the centrally located tubes more quickly than from the outer tubes. The presurgery score consisted of their performance on a subsequent assessment. In this test, hand preference was determined by measuring the number of trials the monkey used the contralesional hand to remove the food reward across tubes.
Behavioral impairments in the lateralized tests. Time-course of rotational asymmetry in non-drug conditions revealed three phases characterized by: (A, B) initial asymptomatic period (Phase I, Weeks 0–9); development of ipsilateral bias in the WT group (Phase II, weeks 15–27); and appearance of contralateral rotation in the A53T group (Phase III) [one-way ANOVA followed by Tukey–Kramer HSD post-hoc test; hash symbol: different from other groups, F(2, 22) = 5.80, p = 0.01; dagger symbol: different from WT group, F(2, 22) = 5.39, p = 0.01]. (C) During these phases, the animals also showed a similar bias in head position [one-way ANOVA followed by Tukey–Kramer HSD post-hoc test; asterisk: different from GFP control group, F(2, 22) = 6.79, p < 0.01]; dagger symbol: different from WT group, F(2, 22) = 4.36, p < 0.05]. The 6-tube search task was performed prior to vector injection and at 29 (Phase II) and 42 weeks (Phase III) post transduction. Time taken to find the single hidden reward was averaged across all tubes (because performance between tubes did not differ). Statistical comparison between groups showed that the A53T group was slower to complete the task asterisk: different from GFP control group; hash symbol: different from other groups. The contralesional hand use (total of 5 trials for each tube) was normal in the GFP group as they tended to use the ipsilesional hand on the ipsilesional side and the contralesional hand on the contralesional side. (E, F) Both A53T and WT groups rarely used the contralesional hand at 29 and 42 weeks asterisk: different from GFP p < 0.05. The hill and valley versions of the staircase tests were performed at 42 weeks post surgery, as illustrated in inset in panels G and H. All animals completed the task equally well using the intact arm (G). Time to clear the food baited stairs with the contralesional (impaired) arm was different in the A53T group both when the animals were expected to reach the ipsilateral or the contralateral sides (H) [one-way ANOVA followed by Tukey–Kramer HSD post-hoc test, asterisk: different from GFP group F(2, 22) = 4.60, p < 0.05; hash symbol: different from other groups, F(2, 22) = 9.89, p = 0.001].
Two-tube choice task
This task assesses hemispatial neglect unconfounded by any motor retardation because it measures untimed choice behaviour (Marshall et al., 2002). Two film canisters were fixed side by side on a wooden base and both tubes were baited with a food reward. The monkey had free choice as to which to remove first and the choice of side was noted. The tubes were randomly presented to the front, left or right side of the monkey, with 10 trials in each position. There was no time limit for each trial. Normal monkeys take reward from either tube when the two tubes are presented centrally but from the nearer, innermost tube when the pair of tubes is presented on the left or right. Monkeys were deemed to be sufficiently familiar with the task, presurgery, when they had removed the reward from the innermost tubes 8 out of 10 times when the tubes were presented on the left or right side of the monkey. The presurgery score consisted of their performance on a subsequent assessment.
Hill and Valley staircase tasks
In these tasks, each monkey was required to reach through vertical slots in a Plexiglas screen attached to the front of the cage in order to retrieve marshmallow rewards, which were placed on the steps of two staircases as illustrated in Fig. 2G and H. In the Hill Task, the staircases rose towards the centre and the slots were positioned laterally so that the monkey had to reach with the left arm into left hemispace and with the right arm into right hemispace. In this task, motor impairment in the contralesional arm and perceptual impairment in contralesional hemispace would be compounded. In the Valley Task, the staircases rose away from the centre and there was one centrally placed slot so that the monkey had to reach with the left arm into right hemispace and with the right arm into left hemispace. In this task, motor impairment in the contralesional arm would be dissociated from perceptual impairment in contralesional space. Each monkey received three trials on each task in a random order. The monkey was allowed 3 min to retrieve all the rewards on each trial. On each trial, the time taken to pick up the first reward on each side (time-to-contact) and the time taken to clear all the rewards, once contact had been made (time-to-clear), of each staircase were recorded. Performance on these two staircase tasks were analysed together in order to contrast and compound hemispatial and hemi-motor deficits.
Post-mortem MRI imaging
The fixed monkey brains were mounted in a gel-holder and imaged at 7 T using a 210 mm bore magnet (Magnex Scientific, Oxford, UK) equipped with Bruker BioSpec Electronics (Bruker BioSpin MRI GmbH, Ettlingen, Germany) and a quadrature birdcage radio frequency coil (50 mm i.d.) (Rapid Biomedical, Rimpar, Germany). An MSME sequence (TR/TE = 1233/10 ms FOV 30 mm, 52 sagittal slices with an isotropic resolution of 380 µm, 80 NEX giving a total scanning time of 2 h) was used to collect the main data for the statistical parametric mapping (SPM) analysis. Second, an MDEFT sequence was used as a mask leaving the brain at full intensity and the gel with low or no intensity (MDEFT Steady State 3D sequence, TR/TE = 15.6/2 ms, flip angle 20°, FOV 30 mm, acquisition matrix 32 × 32 × 32 giving an isotropic resolution of 940 µm and a scanning time of 15 min). The reconstructed data sets from the MSME acquisition were band-pass filtered using the ImageJ software (US National Institutes of Health, Bethesda, MD, USA) with upper and lower limits 40 and 2 pixels respectively. This allowed reduction of noise in the images. In order to create a template, one MSME scan, from the control group, was manually rotated and proportionally scaled to roughly match the marmoset atlas coordinates (Stephan et al., 1980). Then, two copies of all MSME scans (original and L-R flipped) were co-registered to the first, matched scan using the linear, mutual information algorithm in the Statistical Parametric Mapping 2 software (SPM2, Wellcome Dept. Cogn. Neurology, London, UK). The template was created by first averaging all MSME scans, then an average of the MDEFT sequences was used to mask the signal from the areas outside the brain from the brain tissue and finally smoothed using a 1 mm FWHM Gaussian kernel. After this, all MSME sequences (both original and L-R flipped) were normalized to the template using the non-linear mutual information algorithm (2 mm FWHM kernel) and corrected for interscan intensity bias, in the SPM2 package.
Under pentobarbital anaesthesia the animals were perfused through the ascending aorta with physiological saline, followed by 4% ice-cold paraformaldehyde. The brains were postfixed in the same solution for 2 h, transferred to 30% sucrose. All MRI scans were done within the 48 h and then brains were sectioned on a freezing microtome at 35 µm in the coronal plane. Immunohistochemical stainings were performed on free-floating sections using antibodies raised against TH (rabbit polyclonal antibody; 1 : 500; Pel-Freez, Rogers AS, USA), GFP (rabbit polyclonal antibody, 1 : 20 000, Abcam Cambridge, UK), and α-syn (syn-211 and LB509, mouse monoclonal antibodies, 1 : 2000 and 1 : 500, respectively, courtesy of Dr Virginia M. Lee, University of Pennsylvania; AB5336P and AB5038P, sheep and rabbit polyclonal antibodies; 1 : 1000, Chemicon, Temecula, CA USA; 610787, mouse monoclonal antibody, 1 : 500, BD Biosciences; AB6176, rabbit polyclonal antibody, 1 : 750, Abcam), PSer 129 α-synuclein (rabbit polyclonal antibody, 1 : 300, courtesy of Dr Iwatsubu, University of Tokyo), ubiquitin (rabbit polyclonal antibody, 1 : 1000, Chemicon), CNPase (rabbit polyclonal antibody, 1 : 200, Sigma-Aldrich, Sweden), APC (mouse monoclonal antibody, 1 : 200, Chemicon) and Iba-1 (rabbit polyclonal antibody, 1 : 500, Wako, Osaka, Japan). Sections were rinsed three times in potassium phosphate buffer (KPBS) between each incubation period. All incubation solutions for free floating sections contained 0.25% Triton X-100 in KPBS. The sections were quenched for 10 min in 3% H2O2/10% methanol. One hour of pre-incubation with 5% normal serum (NS) was followed by incubation with the primary antibody in 2% NS at room temperature overnight and incubation with 1 : 200 dilution of an appropriate secondary antibody (Vector Laboratories, Burlingame, CA) in 2% NS, followed with avidin–biotin–peroxidase complex (ABC Elite; Vector Laboratories, Burlingame, CA), and visualized using 3,3-diaminobenzidine as a chromogen, mounted on chrome-alum-coated glass slides, and coverslipped.
Double immunohistochemical stainings (α-syn/TH and α-syn/APC) were performed using the primary antibodies as described earlier. α-Syn staining was performed using a biotinylated secondary antibody followed by streptavidin-Alexa488 while the secondary antibody for APC was Cy3 conjugated goat anti-rabbit, and Cy5 conjugated goat-anti-rabbit secondary antibody was used for visualization of the TH. All sections were coverslipped using DABCO, examined and digitally photographed using confocal laser-scanning microscopy (Leica TCS).
Insolubility of the human α-syn aggregates was illustrated utilizing a modified version of the proteinase K (PK) digestion and immunohistochemistry protocol as described (Chu and Kordower, 2006). For this purpose, the sections both from WT and A53T groups were first mounted on coated glass slides (Super Frost Plus, Menzel GmbH & Co KG, Braunschweig, Germany) and dried at 55°C overnight. They were then rehydrated in Tris buffer containing 0.05% Tween-20 (TBST). Digestion of the sections was carried out at 55°C by incubation in TBST containing 50 µg/ml PK (Invitrogen) for a period of 90, 135 or 270 min. Control sections were incubated in TBST alone for 270 min. Following fixation of the digested sections in 10% cold formalin for 10 min and a series of washes, all sections were quenched with 3% H2O2/10% methanol and processed for immunohistochemistry using either the syn211 antibody that recognizes the human α-syn protein or AB6176 antibody that shows the endogenous marmoset α-syn well. The colour reaction was developed using DAB as described earlier.
Stereological estimation of nigral TH-positive cell numbers
The unbiased stereological estimation of the total number of TH-positive cells in SN was done using the optical fractionator principle. This sampling technique is not affected by tissue volume changes and does not require reference volume determinations. The random systematic sampling of counting areas was done using the Olympus CAST system version 2.0 (Olympus, Denmark A/S, Albertslund, Denmark). A low power objective lense (4X, SPlan) was used to delineate the borders of the SN at all levels in the rostocaudal axis. In the anterior sections, the pars compacta cells were the only TH-positive cells visible. In the central and caudal sections the border between the TH-positive cells in the ventral tegmental area and the pars compacta region was just lateral to the roots of the third nerve. At the most caudal level, the pars compacta was defined as the dense TH-positive cell group ventral to the medial lemniscus and the retrorubral area. The TH-positive cells in the pars retriculata were included in the cell counts. This definition led typically to eight sections per structure being measured. A counting frame (1600 μm2) was placed randomly on the first counting area and systematically moved through all counting areas until the entire delineated area was sampled. The sampling frequency was chosen by adjusting the XY-axis step length so that about 100–200 TH-positive cells were counted in each specimen. Actual counting was done using a high power objective lens (40× DApo UV, NA 1.30 oil). The estimates of the total number of neurons were calculated according to the optical fractionator formula and coefficient of error <0.10 was accepted (Gundersen and Jensen, 1987; West, 1999).
One-way or two-way ANOVA with repeated measures were performed where appropriate, followed by a Tukey–Kramer HSD post hoc test when P < 0.05 using JMP Statistic Software Package v5.01 (SAS Institute Inc., Cary, NC, USA). For the MRI data, voxel-wise t-tests were performed in SPM2 comparing difference maps calculated for each animal (normalized lesioned side minus normalized contralateral side) in three groups. The resulting parametric map was thresholded using a false discovery rate (FDR) value of 0.05 and an extent threshold of 20 voxels, leaving only truly significant differences.
In order to characterize α-syn overexpression mediated nigrostriatal pathology, 3 equal groups of monkeys (24 total) received injections of either rAAV-wt-α-syn, rAAV-A53T-α-syn or rAAV-GFP. All monkeys tolerated the surgery well and were returned to their home cages within 24 h. One monkey from the A53T group was killed 19 weeks later because of unrelated illness. Post-mortem examination revealed a chronic active colonic ulcer. The brain from this monkey was processed for histological verification of transgene expression, but the behavioural data from this animal were excluded from analysis.
Overexpression of α-synuclein leads to progressive impairments in contralesional motor behaviour
In order to reveal motor impairment or hemispatial neglect that has previously been found following 6-OHDA lesions leading to loss of nigral DA neurons and thus the DA input to the caudate nucleus and the putamen (Eslamboli et al., 2003; Milton et al., 2004), we tested spontaneous motor bias without administration of any drugs. During weeks 3 and 9, none of the groups showed rotation suggesting a presymptomatic period of at least 2 months. Between weeks 15 and 27, the WT group displayed a significant rotation towards the ipsilesional side. This ipsilesional rotation in the WT group diminished, while in the A53T group a contralesional rotation appeared starting from week 33 (Fig. 2A). Therefore, we have interpreted the changes in full body rotations as an indicator of three distinct behavioural phases (depicted as Phases I–III in Fig. 2A and B). Head turn bias recorded at the same time points showed a similar pattern supporting this interpretation (Fig. 2C).
We also employed tests of hand use such as the six-tube search task where animals were expected to find a food reward placed in one of the six tubes presented to them. The task was performed before and at 29 and 42 weeks after vector transduction. Prior to surgery, the groups did not differ in the time taken to find the reward and/or the hand used for retrieval. At 29 and 42 weeks there was a significant group difference in retrieval time [three-way repeated measures ANOVA, effect of group F(2, 120) = 23.09, P < 0.0001]. Because the slowness of the response was the same for all tube positions [three-way repeated measures ANOVA, groups × tube interaction F(10, 120) = 0.09, P = 0.36] the data obtained from all the tubes were averaged to produce the overall score for each animal at each time-point (Fig. 2D). Subsequent analysis revealed that the A53T group was significantly slower to complete the task compared with the GFP group at 29 weeks [one-way ANOVA, F(2, 22) = 4.19, P < 0.05], and different from both GFP and WT group at 42 weeks [one-way ANOVA, F(2, 22) = 6.39, P < 0.01]. The monkeys in groups WT and A53T used their ipsilesional (unimpaired) hand to make almost all retrievals from all tubes at weeks 29 and 42 even though the GFP group used their contralesional hand for the contralesional tubes (Fig. 2E and F).
The impairments seen in the six-tubes search task suggested that the deficit was mainly motor and did not have a major component of contralesional neglect. This finding was supported by the data obtained in the two-tubes choice task. In this task, both tubes are rewarded with food, and hemispatial neglect, unconfounded by motor impairment, is indicated by the choice of tube, irrespective of the time or hand used. When the tubes are presented to either side of the monkey, normal animals usually choose the tubes closer to the midline but monkeys with hemispatial neglect reach to the further-away tube when the tubes are presented in the ipsilesional side. Results confirmed the absence of neglect in the α-syn over-expressing monkeys as all three groups of animals performed the two-tube task in a similar way, which did not differ from normal behaviour at any time [one-way ANOVA, week 29: F(2, 22) = 0.54, P = 0.59; week 42, F(2, 22) = 1.24, P = 0.31].
We performed the staircase tasks (both Hill and Valley versions; see supplementary methods for further details) at 42 weeks when behavioural impairments in other tests were obvious. The animals were expected to collect food reward from visible stairs behind the opening in the front panel of the cage (see insert drawing in Fig. 2G and H). Time to initiate the first contact with a reward was not different in any of the groups confirming the absence of hemispatial neglect (data not shown). Time to complete the task, however, was impaired in the α-syn groups; this difference was observed when the animals collected rewards when using the contralesional arm from the ipsilesional or contralesional side. This was most pronounced and statistically significant in the A53T group (Fig. 2H).
Phase III was characterized by a worsening in general motor coordination
By 9 weeks, the contralesional limbs (hand and foot) of some monkeys in the WT and A53T groups were seen to slip off the perches. Inspection of videotapes revealed that monkeys in both α-syn groups but not the GFP group made repeated errors with their contralesional limb (Fig. 3A). Quantification of these errors revealed a significant number of errors in A53T animals during Phase II (up to week 27) (Fig. 3A and B). Phase III was characterized by an increase in the motor errors in the A53T group (Fig. 3A and B). At the end of Phase III (51 weeks) four of the seven animals in the A53T group made substantial number of errors (20–100 errors in 15 min). Another striking feature of Phase III was the occurrence of persistent head tilt (down on the contralesional side) that was first noted at 35 weeks post-surgery. This behaviour was observed in three of the seven animals of the A53T group and one monkey in the WT group (Fig. 3C at week 45). The groups did not differ statistically on the number of front-to-back cage crosses at any time during the 1-year follow-up, indicating that overall locomotor activity was not changed [two-way repeated measures ANOVA, effect of group F(2, 20) = 2.38, P = 0.12]. Note that, preference tests are much more sensitive than disability tests so it is quite possible for an animal to develop a turning bias without this affecting the extent to which it moves about the cage.
General coordination tests. (A) Starting at 9 weeks post transduction animals in the A53T group were observed to have slips off the perches. (B) These motor coordination errors were present and stable throughout Phase II. However, in Phase III, the A53T group became progressively worse. One-way ANOVA followed by Tukey–Kramer HSD post hoc test; asterisk: different from GFP group, Phase II: F(2, 22) = 7.40, P < 0.01, and Phase III: F(2, 22) = 4.20, P < 0.05]. (C) During the final phase some of the animals also displayed head tilt down on the contralateral side. This was particularly prominent in three of the monkeys in the A53T group.
Pseudotyped rAAV2/5 vectors transduce neurons and glia in the primate midbrain
Pseudotyped rAAV2/5 vectors (rAAV2 vectors packaged in an AAV5 capsid) were used to over express human wt or A53T mutated α-syn, or the GFP marker protein as control. We expected this capsid would produce a transduction profile different from vectors carrying the AAV2 capsid for two reasons. First, the fact that these two capsids are known to recognize different cell surface proteins for attachment and internalization (Walters et al., 2001; Burger et al., 2004), and secondly, AAV5 capsid allows a more efficient concentration of the final vector preparation (to >1013 genome copies/ml), influencing spread of the vector and the repertoire of successfully transduced cell types.
A small number of monkeys (n = 2/transgene) were killed for histological analysis 3 weeks after injection of the rAAV vectors to evaluate the early phase of transgene expression. At this time the transgenic protein was mainly located at the level of cell bodies and proximal processes (Fig. 4). The remaining monkeys (n = 7–8/transgene) were killed for histology following behavioural assessment for 1 year (Fig. 5). In the vental midbrain several cell groups were transduced (Fig. 4A and E; Fig. 5C, G and K). Large numbers of cells clustered in compact layers were observed in the SN pars compacta (SNc) (Fig. 4B and F; Fig. 5D, H and L). In addition, some transduced cells were located in the SN pars reticulata (SNr) (Fig. 4C and G). Similarly, dorsal to the SNc some GFP-positive cell bodies that morphologically resembled astrocytes were found in mesencephalic tegmentum (data not shown). Interestingly some cells in the cerebral peduncle (CPd) were also transduced. They had small cell bodies with extensive processes covering large areas in the white matter (Fig. 4D and H).
Photomicrographs show expression of the GFP transgene (A–D), or human α-syn (E–H) 3 weeks after transduction. Transgenic proteins were expressed in cells located in the SNc (B and F), dorsally in the mesephalic tegmentum and ventrally in the SNr (C and G). In the CPd scattered cell bodies and ramifications of processes reminiscent of oligodendrocytes were transduced also (D and H). The observations in WT and A53T groups were comparable. In panels E–H a representative example from the WT group is shown. Scale bar in A and E: 1 mm, and scale bar in B and F: 100 µm and apply to B–D and F–H, respectively. CPd: cerebral peduncle, SNc: substantia nigra pars compacta, SNr: substantia nigra pars reticulata, VM: ventral mesencephalon.
After 1 year, the GFP staining was more intense overall and the transduced cells, including their distal axon terminals were entirely filled with the GFP protein. The most prominently transduced neuronal projection system was the nigrostriatal DA neurons. The entire caudate–putamen (CPu) was densely innervated with GFP-positive fibre terminals (Fig. 5A and B). GFP-positive fibres also extended into ventral areas including nucleus accumbens and olfactory turbercle (Fig. 5A). In the striatum, some medium spiny neurons and scattered astroglial cells positive for GFP were also observed (Fig. 5B, see inset). In the dorsal and dorsolateral cortical areas, single scattered GFP positive neuronal and astroglial profiles were observed (data not shown). In histological specimens obtained from animals killed at 3 weeks, none of these fibres was seen, suggesting a slow build up of the transgenic protein in the transduced cells. Taken together, these observations argue that the synthesis and accumulation of the transgenic protein in primate cells is a slow process. This is different from observations in the rodent brain using essentially the same viral vectors where the cells and all the processes including the distal axon terminals are filled with GFP protein by 3 weeks (Bjorklund et al., 2000; Burger et al., 2004).
Photomicrographs show expression of GFP (A–D), human wt-α-syn (E–H), or human A53T mutant-α-syn (I–L) transgenes 1 year after injection of the rAAV2/5 vector. (A) Intense GFP staining was observed throughout the caudate nucleus and the putamen, lateral olfactory tubercle, and less homogeneously in medial olfactory tubercle and nucleus accumbens. (B) A high power photo from caudate nucleus illustrating dense GFP-positive fiber staining. In addition, few scattered GFP-positive cells with astroglial or neuronal morphologies were seen. (C) At this time-point, i.e., 1 year after transduction, the GFP-positive staining covered the entire SN. (C and D) GFP was expressed in the vast majority of the neurons in SNc, (C) in cells dorsally located in the mesephalic tegmentum and ventrally in cells of the SNr, and lateral part of the VTA. (E, F, I and J) In the WT and A53T groups, the intensity of staining at the level of striatum was low. Both in the putamen (F) and in the caudate nucleus (J), there were many fewer fibers as compared with the GFP group, and numerous small and large α-syn-positive aggregations were observed in the surviving fibers (arrows in F and J). (G) Overall staining at the level of SN was similar to 3 weeks time point in the WT group. Numerous human α-syn-positive cells were observed in the SNc. (K) In the A53T group, large areas in SN appeared to be cleared from α-syn staining, (L) leaving fewer surviving cell bodies expressing the transgene. (M–T) Presence of insoluble aggregates were analyzed by PK digestion of the sections for varying times: 90 mins (N and R), 135 mins (O and S), or 270 mins (P and T). Upper panel (M–P) shows high power images from midbrain sections at the level of SN stained using the syn211 antibody specific for human α-syn protein, while the lower panels (Q–T) shows corresponding sections stained with AB6176, which recognizes endogenous monkey α-syn and shows staining in axon terminals in the dentate gyrus of the hippocampal formation known to be rich in α-syn protein. Note that nearly all endogenous α-syn is digested already at 90 mins incubation with PK (R), by constrast, the human A53T mutated protein found in inclusions persisted in abundance both at 90 and 135 mins (N and O). (P and T) Prolonged treatment for 270 mins lead to loss of specific staining and disintegration of the tissue. Scale bar in A and C: 1 mm apply to A, E and I, and C, G and K respectively. Scale bar in B and D: 100 μm and apply to B, F and J, and D, H and L, respectively. Scale bar in T: 25 μm and applies to M–T Ac: nucleus accumbens, cc: corpus callosum, CN: caudate nucleus, CPd: cerebral peduncle, Ctx: cortex, IC: internal capsule, Hpc: Hippocampus, NIII: third nerve root, Put: putamen, Sep: Septum, SNc: substantia nigra pars compacta, SNr: substantia nigra pars reticulata, VTA: ventral tegmental area.
Overexpression of human α-synuclein leads to neuropathology and neurodegeneration
One year after transduction the distribution of wt-α-syn protein was similar to the distribution of the GFP protein such that the transduced cells in the SN and the distal axonal terminals were filled with the transgenic α-syn protein (Fig. 5E–H; see also Supplementary Fig. 1 for comparison of α-syn staining using five different antibodies recognizing the transgenic protein). There was, however, a notably lower density of fibre staining for α-syn in the WT group than for GFP in the GFP group. Aggregates containing α-syn were seen in the CPu (arrows in Fig. 5F). This finding was even more pronounced in the A53T group, where only a sparse network of α-syn positive striatal fibres was seen (Fig. 5I and J). As we observed an intense fibre staining within the striatal complex in the monkey killed at 19 weeks after injection of rAAV-A53T-α-syn, this would argue that initially α-syn protein is efficiently transported along the axonal processes, but pathological forms of α-syn gradually accumulate over time leading to neurodegeneration. In fact, at 1 year degenerating axons appeared thicker and had abnormal beaded formations reminiscent of dystrophic neurites, some of which reached the size of a cell body in diameter in both α-syn groups (see insets in panels F and J in Fig. 5). In order to determine if the α-syn positive aggregates we have seen in the histological specimens are indeed insoluble inclusions, we performed PK digestion prior to staining (Fig. 5M–T). Sections were digested for 90, 135 or 270 min and stained either for human α-syn using the syn211 antibody as shown at the level of SN (Fig. 5M–P) or for monkey α-syn using the AB6176 (Fig. 5Q–T; see Supplementary Fig. 1 for overview pictures using this antibody) as shown at the level of dentate gyrus of the hippocampus from an animal in the A53T group. Consistent with the notion that the inclusions are resistant to digestion, abundant human α-syn staining was seen in control specimen as well as after digestion for 90 and 135 min. Endogenous monkey α-syn was digested much more rapidly as the specific staining was almost completely lost at 90 min (compare Fig. 5Q with Fig. 5R–T). By and large α-syn containing inclusions were labelled with TH, and were not associated with propidium iodide, which labels nucleic acids, suggesting that the inclusions were localized in axons of DA cells (Supplementary Fig. 2). In the A53T group the human α-syn protein appeared to have cleared from the SN, and there were fewer surviving α-syn-positive cell bodies compared with other groups (Fig. 5K and L). It should be noted here that these data might support the interpretation that the initial damage takes place in the axon terminals and leads to a retrograde degeneration of the cell bodies, although it does not provide conclusive evidence for it.
The status of the nigrostriatal DA projection neurons was evaluated on sections stained with an antibody against the TH enzyme. At the level of the forebrain, TH-positive fibres originating from SNc densely innervate the caudate nucleus and the putamen. In the GFP group, the innervation of these two target nuclei was essentially intact, as shown for two cross-sections at pre- and post-commissural levels (Fig. 6A–F). In the WT and A53T groups, however, the fibre network was less dense when examined at low power. Closer inspection of these areas revealed that there was not only a substantial loss of TH-positive fibre terminals in both structures but there were also numerous pathological TH positive accumulations, suggesting that some of the affected but surviving cells were nonetheless dysfunctional (Fig. 6G–L for WT, and Fig. 6M–R for A53T group).
Photomicrographs showing sections of striatum stained with an antibody against the TH enzyme at a rostral pre-commissural (A–C, G–I and M–O) and a caudal post-commissural level (D–F, J–L and P–R). In the GFP group, dense TH-positive fibres covered the entire putamen and the caudate nucleus (A and D), and appeared to have a fine calibre mash-like axon terminal network, as would be expected in the normal brain. In the WT and A53T groups, however, there was a general reduction in the intensity of the fibre density (G–R) that was more striking and widespread in the A53T group. Inspection of these sections with high-power microscopy showed that some of the surviving fibres were coarse and appeared dystrophic with several aggregates (H, I, K, L, N, O, Q and R). Scale in A: 1 mm applies to A, D, G, J, M and P. Scale in B: 100 µm and applies to B, C, E, F, H, I, K, L, N, O, Q and R. Ac: nucleus accumbens; cc: corpus callosum; CN: caudate nucleus; Ctx: cortex; GP: globus pallidus; IC: internal capsule; LV: lateral ventricle; OT: optic tract; Put: putamen; Sep: septum; Thal: thalamus.
At the level of the midbrain, the effects of human wt and A53T α-syn were clearly different (Fig. 7A–I). Over-expression of the wt-α-syn induced cell death in only two of the animals in this group (35 and 45% cell loss, respectively), while the remaining six animals did not have a substantial cell loss. Although the cell number on the injected side was different from the intact side, it failed to reach significance when compared with the injected side of GFP group (Fig. 7J). In the A53T group, the SN was atrophic with distorted fibres in the SNr and the residual surviving cell bodies in the SNc (Fig. 7F and I). In this group, there was a clear and consistent neurodegeneration in the injected side (six of the seven animals showed ≥40% cell loss), which was significantly different when compared with the injected sides of both the GFP and WT groups (Fig. 7J).
Photomicrographs showing sections at the level of midbrain stained with an antibody against the TH enzyme. In the GFP group, the nigral TH-positive cells and their dendritic profiles on the intact side (A) and the injected side (B) were not different from each other. Numerous nigral neurons were seen in the SNc (G). In the WT group (C and D), the general appearance was similar to GFP in six of the eight animals, where numerous TH-positive cells were seen in the SNc (H). In the A53T group, however, the gross morphology of the ventral midbrain was altered. The cross sectional area of the SNc was smaller, and the dendritic TH-positive fibre network in the SNr was disorganized (F), as compared with the intact side (E). The surviving TH-positive cells in the SNc were shrunken, and in some areas, cell bodies were rather scarce (I). Stereological quantification of TH-positive cells in the SN (J) revealed a clear and consistent cell loss in the A53T group, while the cell numbers on the WT injected side group were not significantly different from the GFP group. Hash symbol: different from other injected sides [one-way ANOVA F(2, 22) = 9.27, P = 0.0014 followed by Tukey–Kramer HSD post hoc test]; asterisk: different from intact side (P < 0.05 paired t-test). Scale in A: 1 mm applies to A–E. Scale in G: 100 µm applies to G–I. CPd: cerebral peduncle; NIII: third nerve root; SNc: substantia nigra pars compacta; SNr: substantia nigra pars reticulata; VTA: ventral tegmental area.
An antibody that specifically recognizes α-syn phosphorylated at serine residue 129 (PSer129; a kind gift from Dr Iwatsubo) was used to determine the presence of this post-translational modification in the brains of monkeys overexpressing human α-syn at the 1-year time point. Some of the neuronal profiles in SN, stained with this antibody, appeared normal (Fig. 8A). Other cells were atrophic with shrunken cell bodies (Fig. 8B and F), or had dystrophic dendrites, some with beaded aggregations (Fig. 8C, arrowhead in E). Interestingly, in several cases PSer129 positive staining was localized to the cell nucleus (Fig. 8D). The latter profile was largely confined to the SNc. In addition, some astroglial cells located in SNr (Fig. 8 bottom cell in E) and oligodendroglial profiles in the CPd (Fig. 8H) stained with the PSer129 antibody. Furthermore, on the injected side, several cell bodies in the CPu were PSer129 positive (Fig. 8G). The latter finding was seen in all animals in the A53T group but was observed in only 1 of the animals in the WT group (corresponding to one of the two animals with cell loss in this group). The number of PSer129 positive neurons in the striatum was greater than that seen with the antibody that was used to recognize specifically the human α-syn protein (both A53T and wt). This could in part be due to phosphorylation of the endogenous α-syn (as the antibody would equally well recognize the phosphorylated monkey protein) as a response of the host brain to neurodegeneration but in order to make a conclusive statement this issue should be further investigated. No staining was observed on sections from animals injected with the control vector suggesting that there was no cross-reactivity of the PSer129 antibody to normal α-syn under these staining conditions (Supplementary Fig. 3). Another important finding was the appearance of ubiquitin-positive inclusions in four of eight animals in WT and two of seven animals in A53T group. These inclusions were seen mostly in cell bodies in SN and CPd (Fig. 8I–L), but some inclusions were not associated with a cell body (arrowhead in Fig. 8K). Colabelling with α-syn and ubiquitin antibodies suggested that the ubiquitin-containing aggregates were partly overlapping with α-syn positive cell bodies. Other ubiqutin stained profiles that were not associated with intact cell bodies appeared as dispersed but similar aggregates, which might be interpreted as remnants of degenerated cells. None of the animals in the GFP group had any ubiquitin-positive inclusions; only a homogeneous staining throughout the cytoplasm was seen in some cells (data not shown).
High-power photomicrographs showing cells stained with an antibody that specifically recognizes α-synuclein protein phosphorylated in serine residue position 129 (A–H), and ubiquitin (I–L). PSer129-positive cells were seen in both WT and A53T groups. In most cases both cell bodies and proximal dendrites were stained (arrowheads in A–C). In some cases, the staining was most intense in the nucleus (D), while in others they were seen as aggregates in processes (E), or as large inclusions with clear cores (F). Several cell bodies with light PSer129 immunoreactivity was observed in the striatum (G), and a few densely stained profiles were also seen in the CPd (H). Ubiquitin containing aggregates were most often seen as small discrete deposits in the cell bodies and in some cases as multiple aggregates (arrowheads in I and J). In few cases, we observed ubiquitin aggregates that were not associated with cell bodies (arrowhead in K). In rare cases ubiquitin-positive aggregates were located in the CPd (L). Scale bars: 20 µm. CPd: cerebral peduncle; CPu: caudate putamen; SN: substantia nigra.
α-Synuclein induced neuropathology can be visualized using MR microscopy
In order to visualize alterations in the brain tissue as a consequence of overexpression of human α-syn protein, we employed high field (7 T) ex vivo MR imaging using a magnetization transfer enhanced 3D MSME (T2-weighted) sequence. MSME scans revealed an increased intensity around the needle tracks and the site of injection that was more evident in the WT and A53T α-syn monkeys than in the GFP monkeys. In order to reveal changes induced specifically by overexpression of α-syn, we generated statistical parametric difference maps. Paired t-tests comparing areas of significant difference between the GFP and WT groups showed that the ventral midbrain overlying the white matter bundles were affected (i.e. increased signal in this mode of acquisition) in the WT group (Fig. 9A–A″). As a reduction in the area of the CPd ventral to the SN was observed in some animals, we performed immunohistochemical staining for the CNPase enzyme localized to oligodendroglial cells and their processes in the white matter tracts (Fig. 9D and E). In four of the eight monkeys in the WT group, we observed areas where there was a clear increase in CNPase staining adjacent to areas devoid of CNPase positive oligodendrocytes (Fig. 9E). Double immunohistofluorescence and confocal analysis showed that at 3 weeks time-point there were several APC-postive oligodendrocyte cell bodies and processes with normal morphologies that were expressing human α-syn. At 1 year time-point, however, most α-syn-positive profiles in the CPd had disrupted morphological profiles as signs of an ongoing degenerative process (Supplementary Fig. 4). In the GFP group, however, there was only a homogeneous low level staining throughout the CPd (Fig. 9D). Luxol blue staining confirmed that loss of CNPase-positive cells and processes in parts of CPd was associated with loss of myelin content (Fig. 9B and C). In the A53T group, these changes were seen in two of the seven animals and were not detected by the SPM analysis (data not shown). Comparison between the WT and A53T groups showed that the changes occurring along the nigrostriatal pathway were more pronounced in the A53T group (Fig. 9F–F″). The difference map showed an area immediately ventral and medial to the internal capsule extending about 4 mm from anterior 6 to 10 relative to the interaural line (Stephan et al., 1980). Interestingly, in this region, namely the medial forebrain bundle, an ongoing pathological process was seen in histological specimens stained for TH, PSer 129 α-syn and Iba1, which is a marker for microglia (Fig. 9G–L).
Association between post-mortem MRI and neuropathology. Statistical parametric maps with a FDR adjusted threshold P < 0.05 (colour coded areas) indicating differences between the GFP and WT (A–A″), and WT and A53T (F–F″) groups were generated from normalized 3D MSME data sets and are shown overlaid on a grey tone image from an animal in the GFP group to localize the statistically significant regions. GFP vs WT group comparison revealed a significant change in the white matter tracts ventral to SN. This was confirmed with myelin (B and C) and CNPase (D and E) staining. Myelin staining with luxol blue revealed loss on myelin content in the transduced side (C) as compared to the contralateral control side (B). Immunostaining for CNPase enzyme showed homogeneous low-level staining in the GFP group (border with the substantia nigra reticulata is depicted by the dotted line) (D). In the WT group, there were areas devoid of CNPase staining, surrounded by a perimeter where staining appeared more intense (B). Similarly, comparison between WT and A53T groups identified a difference at the level of the medial forebrain bundle (F–F′). There was an ongoing pathological process in this region, as visualized by TH staining (compare G and H) in an A53T animal. In addition, the same areas were filled with PSer 129 α-syn positive fibres (compare I and J) (arrow in J, see inset for high power), and increased microglial staining with Iba1 (compare K and L). Note that the white matter tracts indicated in panels I–L correspond to dark grey regions around the colour coded area in the MRI image in panel F. Scale bars: B and D = 80 µm and applies to B, C and D, E; G = 20 µm applies to G and H; I = 200 µm applies to I–L, respectively; scale bar in inset in panel J is 10 µm.
The purpose of this experiment was to characterize the long-term behavioural and neuropathological consequences of α-syn overexpression in the ventral midbrain in marmoset monkeys. Twenty-three monkeys were followed for 1 year after transduction with rAAV vectors encoding human wt or A53T mutant α-syn, or the GFP marker protein.
There were two major findings. First, we were able to define three phases in the post transduction period. Phase I (0–9 weeks) was presymptomatic, because none of the animals displayed any behavioural impairments. We defined Phase II (15–27 weeks) as the period when overt behavioural deficits were seen in α-syn overexpressing animals. This period was characterized by ipsilesional full body rotations and head turn bias in the WT group. Motor coordination errors were seen in A53T group and to a lesser extent in the WT group in Phase II. In Phase III (33–52 weeks), there was a progressive worsening of the motor coordination errors in the A53T group. Importantly, however, not all animals showed a similar level of impairment in one group neither were the two α-syn expressing groups the same. While the variation between the monkeys may be due to individual differences between the animals, the nature of the viral vector mediated approach makes it difficult to conclude definitively. Similarly, due to the differences in onset and peak of transgene expression provided by different vectors or even different serotypes of the same vector [e.g. rAAV5 vectors used in the present work as compared to rAAV2 vectors used in our previous work (Kirik et al., 2002)], the time-line for onset and progression of behavioural symptoms may be modified. Nevertheless, in light of the fact that the human disease presents itself with wide variations, this may not only be a drawback but should be seen as a parallel to the challenge that will eventually have to be faced to prove efficacy of any neuroprotective therapy strategy in the clinics.
Secondly, after 1 year the TH-positive fibre terminals in the striatum were partly lost while those remaining were dystrophic in both the WT and A53T groups. Intracellular aggregates containing α-syn were seen in the axonal and dendritic projections of surviving DA neurons in all animals in both groups. Interestingly, many of these pathological aggregates contained α-syn that was phosphorylated at Ser129, similar to that seen in Parkinson's disease (Fujiwara et al., 2002). Moreover, midbrain ubiquitin-containing aggregates were observed after 1 year of α-syn overexpression, which has not been previously detected in the rodent brain using the same approach (Kirik et al., 2002; Lo Bianco et al., 2002; Kirik et al., 2003). Loss of nigral DA neurons, however, was consistent in the A53T group but was less severe and more variable in the WT group. In the rodent studies, we have confirmed the loss of nigral neurons, as opposed to reduction in the TH enzyme levels, by both use of other general neuronal markers, and also with tracing studies where the nigral neurons were retrogradely filled with Fluorogold and followed after the vector transduction.
The rAAV vectors carrying the AAV5 capsid transduced not only the DA neurons in the SNc, but also the GABA-ergic neurons and astrocytes ventrally in the SNr and dorsally in mesencephalic tegmentum, in addition to oligodendrocytes located in the CPd. While degeneration of nigral DA neurons was the most prominent pathological alteration, qualitative microscopical analysis indicated that other neurons and astrocytes appeared to be unaffected by the overexpression of α-syn. Interestingly, however, there was white matter shrinkage in the CPd, accompanied by loss of myelin, suggesting that at least some of the transduced oligodendrocytes were functionally impaired or lost due to the presence of α-syn. Some degree of glial pathology is observed in Parkinson's disease patients, as suggested by the appearance of inclusions containing α-syn and ubiquitin in astrocytes and oligodendrocytes (Arai et al., 1999; Hishikawa et al., 2001) From the present experiment, it is not possible to conclude which elements of the behavioural syndrome seen in these monkeys can be ascribed to loss of DA neurons in the SN versus changes in the white matter tracts. However, this needs to be further investigated as the interpretation of the behavioural improvements that could be seen following application of novel protective strategies that might preferentially rescue one or the other cell population will depend on this knowledge.
A53T mutated α-syn was found to possess altered aggregation properties in vitro and to inhibit proteasomal degradation (Conway et al., 1998). Moreover, differential toxicity of A53T-α-syn was seen in cell culture systems where expression in the primary mesencephalic cultures led to profound degeneration, while the wt-α-syn was less toxic (Zhou et al., 2000; Zhou et al., 2002). It has recently been shown that wt-α-syn could be degraded by the lysosomal pathway, while the A53T mutation impairs the lysosomal uptake leading to decrease protein degradation (Cuervo et al., 2004). However we, and others, have reported that overexpression of wt as well as the mutated forms of human α-syn protein in DA neurons leads to a similar neurodegeneration in rat brain (Kirik et al., 2002; Klein et al., 2002; Lo Bianco et al., 2002; Kirik et al., 2003; Yamada et al., 2004). This did not seem to be an artefact of protein overexpression as neither expression of GFP marker protein nor rat wt-α-syn led to a similar cell death. Our findings in this study are in line with the in vitro studies, and differ from the findings in the rats, as we found about 45% cell loss overall in the A53T group whereas, in the WT group, only 2 of the 8 monkeys had detectable cell loss. Taken together, these data suggest that the human α-syn protein is processed differently in rat and human DA neurons so that, in the rodent cellular environment, human wt and A53T α-syn have similar effects. When the same overexpression was achieved in primate DA neurons, on the other hand, wt and A53T mutant variants acted differently. It should be noted here that although vector batches were selected based on a close match in particle titres, as we have not included additional monkeys to determine the total brain levels of the transgenic protein using western blot analysis, we cannot completely rule out a possible variation in the expression levels from these vectors. This aspect is an inherent weakness of the viral vector mediated overexpression studies.
Another important observation was the presence of ubiquitin-containing inclusions in the brains of these monkeys. Although overexpression of the same proteins in rats resulted in formation of numerous aggregates in the affected cells, these inclusions did not contain ubiquitin even after 6 months survival (Kirik et al., 2002; Lo Bianco et al., 2002). Nor were there any ubiquitin positive aggregates in the monkeys at 17 weeks post transduction (Kirik et al., 2003). These findings suggest that accumulation of ubiquitin is a late event in the pathological process, and that the signal for ubiquitination might differ between the primate and the rodent brain. This argument is in line with post-mortem observations in Parkinson's disease brains where only some of the α-syn-positive inclusions were found to be ubiquitinated (Spillantini et al., 1998; Sampathu et al., 2003).
Finally, we employed MRI imaging to investigate if α-syn-induced neuropathology could be detected. The MRI had to be done as ex vivo scans due to unavailability of a scanner at the site. Importantly, although the scans were performed in fixed brains, we adopted parameters that would be suitable for living animals by limiting the scanning time to about 2 h, and the voxel size not sensitive to breathing artefacts. Nevertheless, we found that MRI at high magnetic fields is sensitive enough to detect neuropathological processes and provides a valuable tool that should be further evaluated in the future studies.
From the disease modelling perspective, the viral vector mediated overexpression of α-syn in the primate brain offers an opportunity that is not matched by the MPTP intoxication model. First, although MPTP treatment leads to an increase in expression of α-syn both at the mRNA and protein level (Kowall et al., 2000; Purisai et al., 2005), DA neurons in the MPTP lesioned primates do not exhibit neuropathology involving formation of α-syn and ubiquitin-positive inclusions. Secondly, motor behavioural impairments can be detected in the WT group, where neuropathological changes in the DA system were restricted to projection fibres, and in the A53T group, which showed only a partial cell loss in the SN in addition to the loss of projection fibres. In contrast, near complete DA depletion is required in the MPTP model, since survival of a subpopulation of DA cells might lead to regaining normal function. This would argue that at least some of the α-syn overexpressing neurons in the present model survive for several months after transduction, although they might do so in a functionally compromised state. In our opinion, the progressive nature of the behavioural impairments, damage to the DA system and development of α-syn and ubiquitin proteinopathy in the present model are superior to other available primate models, and thus provide an excellent tool to study novel classes of agents that might be effective in clearing the pathological aggregates in vivo prior to their assessment in clinical trials.
The authors would like to thank Ulla Jarl and Rosalyn M. Cummings for their excellent technical assistance; Bengt Mattson for his help in the graphic work/design of the figures, the Powell Gene Therapy Center Vector Core for production of the vectors used in this study and to Michael Horn at the Center for Mouse Physiology and Bio-Imaging for assistance in the acquisition of the MRI data. This work was supported by grants from the Medical Research Council, UK (R.M.R. and H.F.B.), Michael J. Fox Foundation Protein degradation grant (D.K.), Swedish Research Council (K2003-33SX-14552-018, K2003-33P-14788-018, K2005-33IT-15332-018 and K2005-33X-14552-03A) (D.K.), Parkinsonfonden (M.R.R. and D.K.) and National Institute of Neurological Disorders and Stroke (NS 36302 to R.J.M., N.M., and C.B.). M.R.R. is a Marie Curie fellow of the Framework Programme 5 from the European Union.
↵*Present address: Institute of Medical Biochemistry, Bldg., 1170 University of Aarhus, Aarhus C, DK-8000, Denmark
. Towards a neuroprotective gene therapy for Parkinson's disease: use of adenovirus, AAV and lentivirus vectors for gene transfer of GDNF to the nigrostriatal system in the rat Parkinson model. Brain Res 2000;886:82-98.
. Recombinant AAV viral vectors pseudotyped with viral capsids from serotypes 1, 2, and 5 display differential efficiency and cell tropism after delivery to different regions of the central nervous system. Mol Ther 2004;10:302-17.
. Neurobiol Dis. Vol. 25. 2006. Age-associated increases of alpha-synuclein in monkeys and humans are associated with nigrostriatal dopamine depletion: is this the target for Parkinson's disease?; p. 134-49.
. Pathological changes in dopaminergic nerve cells of the substantia nigra and olfactory bulb in mice transgenic for truncated human alpha-synuclein(1-120): implications for Lewy body disorders. J Neurosci 2006;26:3942-50.
. CMV-beta-actin promoter directs higher expression from an adeno-associated viral vector in the liver than the cytomegalovirus or elongation factor 1 alpha promoter and results in therapeutic levels of human factor X in mice. Hum Gene Ther 2001;12:563-73.
. Overexpression of alpha-synuclein in rat substantia nigra results in loss of dopaminergic neurons, phosphorylation of alpha-synuclein and activation of caspase-9: resemblance to pathogenetic changes in Parkinson's disease. J Neurochem 2004;91:451-61.