|Home | About | Journals | Submit | Contact Us | Français|
Several neurodegenerative diseases are typified by intra-neuronal α-synuclein deposits, synaptic dysfunction and dementia. While even modest α-synuclein elevations can be pathologic, the precise cascade of events induced by excessive α-synuclein and eventually culminating in synaptotoxicity is unclear. Towards this, we developed a quantitative model-system to evaluate evolving α-synuclein-induced pathologic events with high spatial and temporal resolution, using cultured neurons from brains of transgenic mice over-expressing fluorescent-human-α-synuclein. Transgenic α-synuclein was pathologically altered over time and over-expressing neurons showed striking neurotransmitter release deficits and enlarged synaptic vesicles; a phenotype reminiscent of previous animal-models lacking critical presynaptic proteins. Indeed several endogenous presynaptic proteins involved in exo- and endo-cytosis were undetectable in a subset of transgenic boutons (‘vacant synapses’) with diminished levels in the remainder; suggesting that such diminutions were triggering the overall synaptic pathology. Similar synaptic protein alterations were also retrospectively seen in human pathologic brains, highlighting potential relevance to human disease. Collectively the data suggest a previously unknown cascade of events where pathologic α-synuclein leads to a loss of a number of critical presynaptic proteins, thereby inducing functional synaptic deficits.
Pathologic intra-neuronal deposits of α-synuclein are seen in diverse neurodegenerative diseases including Parkinson’s disease (PD), dementia with Lewy bodies (DLB), and a subset of Alzheimer’s disease (AD), and are often accompanied with dementia (Mukaetova-Ladinska and McKeith, 2006). Under physiologic conditions α-synuclein localizes to presynapses, and in pathologic states α-synuclein accumulations are seen in cell bodies, neuronal processes as well as synapses (Roy, 2009). Duplications/triplications of the WT-human-α-synuclein (h-α-syn) gene is seen in familial forms of PD/DLB (Singleton et al., 2003; reviewed in Cookson and van der Brug, 2008), where a mere two fold increase in WT-protein levels lead to profound clinical and pathologic phenotypes (Farrer et al., 2004; Fuchs et al., 2008; Ikeuchi et al., 2008). These data underscore that not only is excessive h-α-syn pathologic, but that even quite modest protein elevations are deleterious to neurons. Synaptic dysfunction and associated neurodegeneration/dementia are established components of α-synuclein-induced pathology as well (McKeith and Mosimann, 2004; Aarsland et al., 2008); collectively indicating that excessive h-α-syn induces a chain of pathologic events that eventually culminate in synaptic dysfunction and dementia.
Though it is evident that excessive α-synuclein has an overall deleterious role in human disease, the precise cascade of early pathologic events resulting from elevations of h-α-syn, and culminating in eventual synaptic dysfunction and dementia are unclear (Cookson and van der Brug, 2008). A prevailing hypothesis is that excessive h-α-syn leads to deficits in vesicular transport/trafficking. The initial clues supporting this notion emerged from studies in yeast model-systems over-expressing h-α-syn tagged to green fluorescent protein (GFP), which implicated several genes involved in vesicular trafficking pathways (Outeiro and Lindquist, 2003; Gitler et al., 2008; Soper et al., 2008). Subsequent studies in phechromocytoma (PC12) and chromaffin cell-lines over-expressing h-α-syn support the hypothesis, showing that α-synuclein over-expression leads to deficits in vesicular mobilization and release from the cytoplasm of mammalian secretory cells (Larsen et al., 2006).
Though these studies provide important clues into the consequences of h-α-syn over-expression, neuronal correlates of such trafficking defects or their relevance to the synaptic dysfunction seen in human diseases have been difficult to construe. A recent study in cultured neurons over-expressing h-α-syn elegantly demonstrated deficiencies in the presynaptic apparatus (Nemani et al. 2010); however the pathologic events leading up-to these deficits, or the relevance of such deficits to human disease is still unclear. Other studies have related losses of endogenous mouse α-synuclein to deficits in vesicular trafficking/exocytosis (Abeliovich et al., 2000; Chandra et al., 2004; Chandra et al., 2005); however the precise pathologic events induced by α-synuclein over-expression within neurons and their relevance to human disease has not been systematically explored.
A comprehension of the sequence of pathologic events induced by excessive h-α-syn is clearly critical to our understanding of the mechanistic basis of these diseases. Here we illustrate a multi-faceted approach that we took to address this issue; combining contemporary quantitative cell-biology with electrophysiology, ultrastructural studies and neuropathology. Our studies suggest a surprising cascade of pathologic events that may underlie the h-α-syn-induced synaptic dysfunction seen in these diseases.
The PDGF-h-α-syn:GFP mice (C57/B6 background) used in this study have been described previously (Rockenstein, 2005). Hippocampal neurons were obtained from brains of heterozygous post-natal (P0–P2) α-synuclein:GFP transgenic pups. Pups were screened using a “GFP flashlight” (Nightsea, Bedford, MA) that made the GFP+ pups glow. Non-transgenic littermates were used as controls. For all cell biology experiments, dissociated cells were plated at a density of 100,000 cells/cm in poly-D-lysine coated glass-bottom culture dishes (Mattek, Ashland, MA) and maintained in Neurobasal/B27 media (Invitrogen, Carlsbad, CA) supplemented with 0.5mM glutamine. All animal studies were performed in accordance with University of California guidelines. Immunofluorescence studies were performed as previously described (Roy et al., 2008). Briefly, cultured neurons were fixed with paraformadehyde/120mM sucrose, rinsed several times and stained with the appropriate antibodies. Alexa 488, 594 and 647 dyes (Invitrogen, Carlsbad, CA) were used as secondary antibodies.
Endogenous mouse VAMP, Piccolo, synapsin and amphiphysin was detected using a mouse monoclonal anti VAMP-2, a rabbit polyclonal anti-Piccolo, a rabbit polyclonal antibody to amphiphysin (all from Synaptic systems, Goettingen, Germany) and a rabbit polyclonal antibody to synapsin-I (Invitrogen, Carlsbad, CA, USA). Total (mouse + human) synuclein was detected using an in-house guinea-pig α-synuclein antibody (GPSYN) that was generated in Virginia Lee’s laboratory, University of Pennsylvania. Human α-synuclein in tissue-sections was detected using a rabbit polyclonal antibody (Millipore, Billerica, MA). Other antibodies used were a mouse monoclonal MAP2 antibody (gift from Dr. Virginia Lee, University of Pennsylvania), a mouse monoclonal anti-PSD-95 antibody (Calbiochem, Darmstadt, Germany), and the human-specific α-synuclein antibodies (LB509 and syn211, both from Abcam, Cambridge, MA, USA). All chemicals were from Sigma unless otherwise noted.
Images were acquired using an Olympus inverted motorized epifluorescence microscope equipped with a Z-controller (IX81, Olympus, Center Valley, PA) and a motorized X–Y stage controller (Prior Scientific), attached to a ultra-stable light source (Exfo exacte, Ontario, Canada) and CCD cameras (Coolsnap HQ2, Photometrics, Tucson, AZ). All images were acquired and processed with Metamorph software (Molecular Devices, Sunnyvale, CA). To capture the majority of synaptic profiles in a given field, Z-stack images were obtained using procedures similar to those used in previous studies of synaptic proteins in cultured neurons (Custer et al. 2006). Briefly, a z-series of images was collected at a resolution of 0.2µm, deconvolved, and saved as a single projection. Subsequent processing for all images was as performed in three steps as described below, largely based on Krueger et al., 2003. (1) Background subtraction-Background fluorescence was calculated as the mean plus two times the standard deviation of 10–20 3-pixel diameter regions of peri-boutonic/axonal fluorescence, and this value was subtracted from respective images. (2) Bouton-size criteria: Background corrected puncta measuring less than 3px (0.48uM) in width and height were not included in the analysis. (3) Colocalization criteria: GFP+ve/α-synuclein+ve puncta were considered to be colocalized with other synaptic proteins if their fluorescence overlapped by >5 pixels in area. For each data-set, a sampling of this analysis was manually counted to confirm the integrity of analysis. Experimental groups were statistically analyzed using a non-parametric Student’s t-test and expressed as mean ± standard error of mean (SEM); all experiments were repeated at least 2–3 times, on separate culture-sets. A p value of <0.05 was considered significant.
For these experiments, GFP+ or WT littermate neurons were plated at very low densities (1000 cells/cm2) with an astrocyte feeder layer suspended above them (Kaech and Banker, 2006) to maximize apposition of transgenic boutons to GFP+ soma/dendrites within a coverslip. Under these conditions, 52±3% of the boutons apposed to h-α-syn-GFP+ primary and secondary dendrites were transgenic (10 neurons, 3 coverslips, 546 boutons analyzed). Whole-cell voltage-clamp recordings were performed with an Axopatch 200B amplifier from α-synuclein:GFP-expressing neurons and neurons from wild-type littermates; recordings were obtained after switching to DIC optics. Coverslips were placed into a perfusion chamber (Warner Instruments). The external recording solution was (mM): 155 NaCl, 3.5 KCl, 2 CaCl2, 1.5 MgSO4, 10 glucose, 10 HEPES (pH 7.4), 320 mosM. The pipette solution contained (mM): 140 K-methanesulphonate, 5 KCl, 10 HEPES, 2 MgCl2, 0.5 EGTA, 2 ATP, 1 GTP (pH 7.3). Miniature excitatory post-synaptic currents (mEPSCs/"minis") were recorded in the presence of 1 µM tetrodotoxin (TTX) and 50 µM bicuculline at − 60 mV holding potential. The solution was allowed to perfuse into cells during whole-cell recording for 3 min, prior to recording spontaneous postsynaptic currents for 10–20 min. Only cells with stable access resistance were included in the data analysis. The electrode resistance after back-filling was 3–5 MΩ. Recordings were digitized using a Digidata 1320A interface and the Pclamp9 (Axon Instruments, Union City CA) software package. Data for mEPSC analysis were sampled at a rate of 20 kHz and filtered at 3 kHz. Synaptic events were detected and analyzed (amplitude, kinetics, frequency) off-line using a peak detection program (Mini Analysis program, Synaptosoft, Decatur NJ). The mean frequency (number of events/duration) and amplitude of the synaptic events were computed after automatic detection of a series of at least 500 events from continuous recording stretches lasting 10–20 minutes.
For the proteinase-K experiments, DIV-21 hippocampal neurons were fixed with paraformadehyde/120mM sucrose for 10 minutes and extensively rinsed with PBS. 1µg/ml proteinase-K was added to the cultures for 2 hours, after which the reaction was quenched by adding serum. Images were then processed for immunostaining as described above. All FM4–64 experiments/imaging were performed on living neurons maintained at ≈37°C inside a live-cell environment chamber (Precision Control) mounted on the microscope. Neurons were first rinsed in a “live imaging media” containing Hibernate E and supplements (Roy et al., 2008) and the antagonists (10µM CNQX and 50µM DLAP5, Tocris Biosciences). They were then transferred to a stimulating media (5mM HEPES, 31.5mM NaCl, 1mM MgCl2, 2mM CaCl2, 30mM glucose and 90mM KCl) containing 15µM FM4–64 and antagonists for 2 minutes. This protocol is known to label mainly the rapidly recycling synaptic vesicles (Deak et al., 2004; Gaffield and Betz, 2006). Immediately thereafter, cells were washed for three times in the “live imaging media” with antagonists and incubated for two 10 minute periods (with rinses) to remove off excess dye on cell surface. Living neurons were then imaged as described above. To analyze the temporal kinetics of FM release, transgenic or WT neurons were loaded with FM4–64 as described above and then incubated for an additional two minutes to allow maximal endocytosis (Deak et al., 2004). FM-loaded neurons were washed extensively and destained using the high K+ solution. The temporal release of the dye from boutons was monitored by live-imaging and corrected for photobleaching. Decay curves from transgenic (green) and WT (red) boutons were calculated as mean±SEM F/F0 values for all boutons analyzed.
For ultrastructural analysis of the synapses expressing α-synuclein:GFP, we performed immuno-EM using a mouse monoclonal anti-GFP antibody (Chemicon) to unequivocally identify transgenic boutons, and compared them with WT boutons. For these experiments, cultured neurons were fixed with 2.5% glutareldehyde and 2% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) at 4°C for 30 minutes and post-fixed with 1% osmium tetroxide, 0.8% potassium ferrocyanide, 3µM calcium chloride in 0.1M cacodylate buffer, pH 7.4 and embedded in epoxy. Cells were washed and immunostained with a primary antibody to GFP detected using 5nm gold particles with silver enhancement (Electron Microscopy Sciences, Hatfield, PA). Blocks were sectioned with an Ultracut E ultramicrotome (Leica, Nussloch, Germany) and analyzed with a Zeiss EM10 electron microscope (Carl Zeiss, Oberkochen, Germany). Vesicle sizes were assayed by partitioning each synaptic bouton into four quadrants and then measuring the diameter of all vesicles within a quadrant. Average vesicle sizes were calculated as mean±SEM for vesicles within each bouton. Average bouton density was calculated as the total number of vesicles within a bouton quadrant divided by the area analyzed. To determine the range in the vesicular sizes within synapses, about 30 vesicles were randomly selected from each WT or transgenic bouton, and their diameters were measured. These data were then pooled and plotted to compare the variability in the sizes of boutons between the WT and transgenic groups. To analyze human brain tissue, prefrontal cortices of two controls (aged 77 and 85) and two LB dementia (aged 79 and 85) cases were obtained from the Alzheimer’s Disease Research Center brain bank at UCSD. Free-floating vibratome sections were immunolabeled with human α-synuclein and synapsin antibodies as described previously (Rockenstein et al., 2005), and imaged/quantified as noted above.
To evaluate the evolution of pathology in a setting of modest α-synuclein over-expression, we used cultured hippocampal neurons from transgenic mice over-expressing h-α-syn tagged at the C-terminus to enhanced-GFP (h-α-syn:GFP) (Rockenstein et al., 2005). The h-α-syn in this mouse-model is driven by a platelet-derived growth factor (PDGF) promoter. Unlike some other mouse models, the α-synuclein expression is largely restricted to the limbic and cortical areas, and the overall expression-levels are much more modest (Rockenstein et al., 2002; Rockenstein et al., 2005). Thus the spatial and quantal distribution of α-synuclein in these mice may more reasonably approximate dementing α-synuclein-related illnesses like DLB. Behavioral testing of 6 month old PDGF-h-α-syn:GFP mice revealed striking deficits of spatial memory in the Morris water-maze test as shown in Supplementary fig. 1, further underlining the suitability of this mouse model in the investigation of α-synuclein-related dementias. Brain homogenates from these transgenic mice show a single band with the expected molecular weight of the h-α-syn:GFP fusion protein (Supplementary fig. 2A), and no proteolyic products were seen, unlike a previous report with transient transfections of C-terminally-tagged α-synuclein (McLean et al., 2001). Our results are similar to data from other research groups showing that C-terminus tagging of h-α-syn does not appear to alter the biophysical properties of the protein (Nakamura et al., 2008). Moreover, transgenic h-α-syn was robustly and unequivocally enriched at synapses, (Supplementary fig 2B, and see data below) as expected for a presynaptic protein.
The overall experimental strategy is shown in Figure 1A. First we identified heterozygous GFP positive pups by optical screening using a “GFP flashlight” (methods) and obtained cultured neurons from the hippocampi of transgenic animals or their WT littermates. The transgenic, presynaptically enriched α-synuclein was easily identifiable within the neurons by their native GFP fluorescence, allowing us to follow the transgenic neurons and rigorously explore the evolving pathology over time. Using these methods, we achieved modest and consistent levels of h-α-syn over-expression (see below) in large numbers of neurons and were able to analyze thousands of over-expressing boutons with relative ease (figure 1B). In these heterozygous cultures, ≈40% of the boutons were GFP positive, and we designed quantitative algorithms to isolate the “green” boutons and analyze the progression of synaptic pathology specifically in these transgenic boutons, including their colocalization with other synaptic elements (see methods). In our hands, this strategy was far more reliable than transient transfections (used in all other α-synuclein studies in cultured neurons to date), where cell to cell variations and low transfection efficiencies (typically <5%) precluded consistent and reliable inferences from large synapse populations.
The GFP-tag allowed us to precisely identify living h-α-syn positive cells/boutons and perform functional studies on living neurons. As expected, the vast majority of the over-expressed h-α-syn localized to presynaptic boutons. Figure 2A shows two representative fields with h-α-syn:GFP boutons apposed to microtubule associated protein (MAP2), a somato-dendritic marker and the post synaptic density marker, PSD-95. Next, we quantitatively determined h-α-syn over-expression levels in our model-system. We performed most of our studies at 21 days in-vitro (DIV-21), a time when these neurons are mature with well-developed presynaptic boutons and dendritic spines. We found that the average α-synuclein fluorescence in transgenic boutons at DIV-21 was only ≈2.5 times higher than that of WT littermates (fig. 2B), providing a reasonable approximation of elevated α-synuclein levels in human disease. Importantly, over-expression of h-α-syn in this setting did not result in an overt loss of synapses (fig. 2C), allowing us to investigate early/evolving pathologic events, before any obvious synapse/cell loss.
We next asked whether there were any pathologic modifications of the transgenic h-α-syn in DIV-21 neurons over time, as seen in human diseases. A well-recognized pathologic alteration of α-synuclein in human pathologic brains and animal models is the accumulation of proteinase-K (PK) resistant α-synuclein. In human DLB brains, PK-resistant α-synuclein has been shown to accumulate at presynaptic terminals (Neumann et al., 2002; Kramer and Schulz-Schaeffer, 2007), in addition to the more obvious somatic α-synuclein accumulations. Synaptic PK-resistant α-synuclein aggregates are also seen in animal α-synuclein over-expression models (Neumann et al., 2002; Periquet et al., 2007) and are thought to be pathologic correlates of synaptic dysfunction and dementia in DLB (Kramer and Schulz-Schaeffer, 2007).
To test if PK-resistant α-synuclein accumulated in our transgenic synapses, hippocampal neurons from h-α-syn:GFP transgenic mice (or their WT littermates) were cultured for three weeks (days in vitro-DIV-21), treated with 1µg/ml proteinase-K as previously described (Periquet et al., 2007; see methods for specifics) and stained with an antibody to total (mouse + human α-synuclein- GP-SYN) as well as a human-specific α-synuclein antibody (LB509) to reveal transgenic boutons (note that it is not appropriate to use the GFP as a read-out in these experiments as the fluorophore is digested by PK-treatment- unpublished observations). We first qualitatively compared total (mouse + human) α-synuclein levels in WT and transgenic neurons. We noticed that while the α-synuclein staining in WT neurons was uniformly diminished, focal areas within transgenic boutons retained high levels of α-synuclein as shown in the panels in figure 3. This suggests that while the PK treatment was sufficient to digest the vast majority of the mouse α-synuclein, a subset of transgenic (perhaps the highest over-expressors) were resistant to PK-treatment. Indeed quantification of human α-synuclein levels in transgenic boutons with/without PK extraction showed that a larger fraction of higher-α-synuclein over-expressing boutons are present in the post-PK data-set; suggesting preferential retention of h-α-syn in those boutons.
Next we asked if the over-expressed h-α-syn in our DIV-21 neurons acquired disease-associated post-translational modifications, namely phosphorylation at Ser-129 residue. Ser-129 phosphorylation is an established pathologic component of α-synuclein inclusions in human disease, as well as animal models of the disease (Fujiwara et al., 2002; Kahle et al., 2002; Chen and Feany, 2005; Anderson et al., 2006; Waxman and Giasson, 2008), and it is generally recognized that such phosphorylation contributes to the pathology in these diseases, though the exact role of such modifications is unclear. In these experiments we stained DIV-21 h-α-syn:GFP over-expressing boutons with an antibody that specifically recognizes α-synuclein when it is phosphorylated at the Ser-129 residue (pSer-129/81A antibody) (Waxman and Giasson, 2008) and compared them to boutons from WT littermates expressing endogenous mouse α-synuclein (fig. 3B). We found that while about half of the transgenic boutons also contained phospho-α-synuclein, it was virtually absent in WT boutons (fig 3B, right). Collectively the proteinase-K resistance and the phospho-h-α-syn accumulation indicate that the transgenically expressed h-α-syn in DIV-21 neurons is pathologically altered.
Next we examined the functionality of h-α-syn over-expressing boutons. Towards this we analyzed synaptic transmission in DIV21 h-α-syn over-expressing neurons using electrophysiology. Live GFP-positive neurons plated at low densities were identified (see methods), and miniature excitatory currents (“minis”) were recorded from perikarya receiving inputs from h-α-syn over-expressing boutons. Cultured neurons from WT littermates were used as controls. All analyses were performed in parallel transgenic and WT culture-sets, to eliminate any possible culture-specific artifacts. Mini frequencies correspond to spontaneous glutamate release from presynaptic boutons and are a classical measure of neurotransmitter exocytosis. There was a striking decrease in the frequencies of the minis in h-α-syn over-expressing neurons, as shown in the representative tracings (fig. 4A). There was no significant change in the amplitudes or the time constants of decay in these experiments (fig. 4B), suggesting that the functionality of the post-synaptic receptors is not significantly altered by h-α-syn over-expression. These data are quantified in figure 4C. In the absence of overt synapse loss (see fig. 2C), these data indicate that the overall synaptic release apparatus is operating at a lower efficiency in h-α-syn over-expressing boutons.
To further examine exo- and endo-cytosis in individual presynaptic boutons over-expressing α-synuclein, we used FM4–64- a red styryl dye that is selectively endocytosed into rapidly recycling synaptic vesicles and widely used to evaluate synaptic function. Upon brief stimulation in the presence of FM dyes synapses rapidly endocytose the dye, which can be then released upon a second stimulation (Gaffield and Betz, 2006). Thus the endo/exocytosis of the FM dye is a measure of the functional status of the synaptic apparatus. We first asked if there was a subset of boutons in our transgenic cultures that were ‘presynaptically silent’; a possibility given our electrophysiology data. Such ‘silent synapses’ should presumably fail to uptake the FM dye completely, as the recycling machinery would be inoperable. Indeed we found that the FM dye was undetectable over background in ≈25% of the h-α-syn over-expressing boutons, indicating that these boutons failed to endocytose the FM4–64, compared to WT boutons (fig. 5A, left panel). When we examined the transgenic boutons that did endocytose the FM dye, we found that the overall average fluorescence intensity of FM in these boutons was diminished by ≈35%. Next we examined the capacity of the FM+ve transgenic boutons to exocytose the FM dye. For these experiments, neurons were loaded with the FM dye, rinsed and then unloaded by high K+ stimulation as described in the methods. As shown in figure 5B, the temporal kinetics of FM-release is significantly slower in TR boutons compared to their WT littermates, further implicating a failure of the presynaptic exocytic machinery in these α-synuclein over-expressing boutons.
To gain further insights into the morphologic changes induced by α-synuclein over-expression, we examined the ultrastructure of h-α-syn over-expressing boutons in cultured neurons by electron-microscopy (EM). To unequivocally identify the subset of over-expressing boutons in our α-synuclein transgenic (+/−) cultures (typically ≈40%), we performed immuno-EM using an anti-GFP antibody, and then compared immuno-positive, transgenic synapses to WT littermate boutons. Control labeling in WT (GFP-negative) cultures was non-specific (not shown). Though synaptic vesicles were seen in both WT and transgenic boutons, we noticed striking variations in the sizes of the vesicles located within transgenic synapses. While control synapses were populated with small synaptic vesicles of relatively uniform diameters as expected, vesicles in transgenic boutons were highly variable in size and occasionally enormous, as shown in figure 6A (lower right panel). Overall quantitative analyses showed that there was no change in the overall size of boutons, but there was a reduction in the density of vesicles at synapses (fig. 6A). Quantization of the size of vesicles within synapses showed an increase in the number of vesicular structures over 50nm in transgenic synapses, with a corresponding decrease in vesicles that were <50nm ((6B,6B, above). The overall variation in the sizes of vesicles at α-synuclein transgenic boutons, as well as the presence of enlarged vesicles is also evident in vesicle-diameter scatter-plots of from transgenic and WT boutons (fig. 6B, below). Though we recognize that calling such enormous vesicles as “synaptic” is questionable, we use this term to indicate their location within boutons (which is unequivocal), as well as to highlight this curious morphologic alteration in our α-synuclein over-expressing boutons.
The above data show that prolonged h-α-syn over-expression in neurons leads to defects in the synaptic vesicular release apparatus, as well heterogeneity and enlargement in the size of vesicles at synapses. We were struck by the similarity of this constellation of synaptic alterations to previous animal models lacking synaptic vesicular proteins, including soluble NSF-attachment protein receptors (SNARE) and active-zone components as well as endocytic proteins that are critical to the synaptic release machinery. Specifically, reduced spontaneous synaptic responses and deficits in neurotransmitter release have been demonstrated in mice (Schoch et al., 2001), C. elegans (Nonet et al., 1998) and drosophila (Deitcher et al., 1998) deleted for a critical vesicular (v-SNARE) protein- synaptobrevin/vesicle associated membrane protein (VAMP or its homologues). Pharmacologic disruption of SNAREs in other studies also led to deficits in neurotransmitter release (Neale et al., 1999; Verderio et al., 1999), and similar alterations have been reported in animal-models lacking active-zone proteins as well (Schoch and Gundelfinger, 2006). Neurons lacking other major presynaptic proteins like synapsin also show defects in neurotransmitter release (Gitler et al., 2004). Intriguingly, EM studies of cultured hippocampal neurons from mice lacking VAMP-2 and dynamin, as well as neurons from flies and worms lacking synaptotagmin and clathrin (various synaptic proteins involved in recycling/endocytosis) all show heterogeneity and increases in the size of synaptic vesicles (Zhang et al., 1998; Nonet et al., 1999; Poskanzer et al., 2006; Ferguson et al., 2007), somewhat similar to our results.
Thus we wondered if the aggregate of synaptic deficits in our h-α-syn over-expressing neurons were simply a consequence of a depletion of critical mouse presynaptic proteins at h-α-syn over-expressing synapses. Accordingly, we quantitatively evaluated levels of the following endogenous mouse presynaptic proteins in h-α-syn over-expressing boutons- a vesicular-SNARE protein (VAMP-2), an active-zone protein (Piccolo), a versatile synaptic vesicle-associated protein thought to play a role in vesicular mobilization and fusion at synapses (synapsin), and amphiphysin- a presynaptically enriched protein that is thought to recruit dynamin to sites of clathrin-mediated endocytosis. For these experiments we fixed DIV-21 h-α-syn:GFP over-expressing neurons, retrospectively immunostained them for the respective endogenous mouse presynaptic proteins (using red secondary antibodies), then quantified co-localization of GFP+ve transgenic boutons to the respective (red fluorescent) bouton containing mouse presynaptic proteins. WT littermate boutons containing endogenous mouse α-synuclein and the respective presynaptic protein under consideration were used as controls.
Surprisingly, we found that all four mouse presynaptic proteins were undetectable over background in subsets of h-α-syn over-expressing boutons (fig. 7A, B), and we started calling them ‘vacant synapses’. We retain this term here to recognize this subset of boutons in our studies. The appearance of such vacant synapses was gradual, increasing in numbers from DIV-7 to DIV-21 (fig. 7C). These temporal changes were likely due to sustained h-α-syn over-expression, and not a result of a sudden escalation in α-synuclein from DIV-7 to 21, as the average levels of over-expressed GFP fluorescence within boutons was practically unchanged during this period (1126±34.56 AFU and 1117±31.27 AFU in DIV-7 and DIV-21 neurons respectively, N=300–600 boutons). Similar diminutions in endogenous presynaptic proteins was also seen in hippocampal and cortical in-vivo tissue sections from h-α-syn:GFP transgenic mice (Supplementary fig. 3). We also considered the possibility (albeit remote) that these ‘vacant synapses’ could represent GABAergic boutons. Though neurons in hippocampal cultures like ours’ are predominantly glutamatergic, a small fraction of GABAergic neurons are present. However this is unlikely to be the case here, as in a set of experiments only 4% of our transgenic boutons were opposed to GABAergic post-synapses (out of 1229 α-synuclein transgenic boutons, only 54 were opposed to post-synaptic terminals immuno-positive for the GABA-A receptor). Thus the vast majority of ‘vacant synapses’ in our system are likely glutamatergic.
We wondered whether the ‘vacant synapses’ in our system represented a unique subset of transgenic boutons that lacked endogenous presynaptic proteins, or whether these vacant boutons were simply on the extreme end of a continuum where α-synuclein over-expression diminished the physiologic presynaptic targeting of these endogenous proteins. To address this, we quantitatively evaluated average fluorescence levels of the four endogenous presynaptic proteins, focusing only on the transgenic boutons that did contain these presynaptic proteins (i.e. the ‘non-vacant synapses’). Levels of all four proteins were diminished in ‘non-vacant’ boutons, as shown in representative pseudo-color (heat-map) panels in figure 8A and the corresponding quantitative analyses in figure 8B. Note that the extent of quantitative diminution of a specific protein generally mirrored the fraction of ‘vacant synapses’ within the given group (compare fig. 8B to 7B),7B), further suggesting that these changes represent a continuum. Collectively, these data show that α-synuclein over-expression in our system leads to global diminutions in the levels of several presynaptic proteins involved in exo- and endo-cytosis; and that subsets of such over-expressing boutons do not contain any detectable amounts of these proteins.
The above data in cultured neurons and intact mouse brain supports a disease-model where excessive h-α-syn is pathologically altered in a subset of presynaptic boutons, inducing depletion of various endogenous presynaptic proteins at these boutons and subsequent functional deficits resulting from the loss of critical synaptic components. In DLB, the prototypical human disease with α-synuclein-related dementia, presynaptic aggregates of proteinase-K-resistant α-synuclein are also seen in frontal cortical sections (Kramer and Schulz-Schaeffer, 2007), prompting the question whether the mechanistic events professed by our single-cell experiments also operate in human disease as well. If so, one would expect that h-α-syn-containing synapses in these diseased brains would have lesser amounts of other presynaptic proteins. To evaluate this, we immunostained frontal cortical sections from autopsy brains of patients with DLB (or age-matched controls) for h-α-syn and synapsin (the protein showing the greatest diminution in our culture studies), and evaluated synaptic protein colocalization. Many h-α-syn-positive synapses in DLB brains had undetectable synapsin levels as well (fig. 9), suggesting that h-α-syn-induced depletion of presynaptic proteins may play a role in α-synuclein-induced human disease as well. We noticed that the colocalization of different presynaptic proteins in tissue sections (even in control brains) was not complete, unlike the situation in WT cultured neurons, where such colocalization consistently approached 100%. The reasons for this are unclear, but may be related to the complex neuronal architecture in brain slices compared to the relatively simple morphology in cultured neurons, where individual boutons can be resolved precisely by light microscopy.
To investigate evolving pathologic events induced by excessive α-synuclein, we developed a model-system reliably over-expressing modest amounts of fluorescent α-synuclein in large neuronal populations and used quantitative cell-biological tools to evaluate the evolving α-synuclein-induced pathology in live and fixed neurons before the onset of overt synaptic losses. We found that in mature (DIV-21) neurons, the expressed α-synuclein was PK-resistant and also acquired disease-associated post translation modifications. At a functional level, there were profound deficits in neurotransmitter release, providing a basis for synaptotoxicity. Ultrastructural examination revealed a peculiar enlargement of synaptic vesicles in transgenic boutons. This constellation of functional and morphologic changes resembled previous animal-models lacking presynaptic proteins, prompting us to quantitatively evaluate endogenous mouse presynaptic proteins in our h-α-syn over-expressing boutons. Indeed we found that several critical exo- and endo-cytic endogenous presynaptic proteins were absent in transgenic boutons (‘vacant synapses’); along with striking diminutions in the levels of such mouse proteins in the remainder. Importantly, these changes may be disease relevant as diminutions in presynaptic protein levels are also visible in human autopsy brains with α-synuclein-related dementias.
Collectively, the data points to a previously unknown mechanistic cascade of events induced by elevated h-α-syn levels that may underlie the early synaptic dysfunction seen in diseased states. Specifically, increased intracellular α-synuclein leads to attenuated levels of several proteins critical to the physiologic operation of the synaptic vesicle machinery; such attenuations lead to functional impairments at synapses manifested by an inhibition of neurotransmitter release; and this toxic chain of events eventually leads to synaptic dysfunction and dementia. Though we have not looked at the entire gamut of synaptic proteins in h-α-syn over-expressing boutons, we rationally selected a few proteins critically important in exo- and endo-cytosis and surprisingly, they were all depleted/diminished in transgenic boutons.
What is the specific defect induced by excessive α-synuclein at the synapse? The available evidence strongly suggests that inhibition of neurotransmitter release is the overall pathologic mechanism induced by excessive α-synuclein, as shown here and in a recent study (Nemani et al., 2010). However the specific synaptic defect(s) induced by excessive α-synuclein is less obvious. Previous unbiased studies of genetic modifiers of α-synuclein toxicity clearly implicate multiple vesicular trafficking pathways in the pathogenesis of α-synuclein toxicity (Gitler et al., 2008; Soper et al., 2008), and a similar approach in C. elegans α-synuclein models implicated genes involved in the endocytosis, in addition to other vesicular trafficking pathways (Kuwahara et al., 2008; van Ham et al., 2008). Other groups have implicated SNARE proteins in α-synuclein pathogenesis (Chandra et al., 2005). Recently, Larsen et al. suggested that excessive α-synuclein may impair vesicular release machinery by inhibiting “priming”- the step immediately preceding vesicular docking to presynaptic membranes (Larsen et al., 2006). Most recently, Nemani et al. rigorously demonstrated that excessive α-synuclein inhibits neurotransmission, reduces the recycling pool of vesicles, and impairs the activity-dependent mobility of synaptic vesicles; suggesting that excessive α-synuclein induces a specific defect in the synaptic vesicle recycling pathway by preventing the re-clustering of vesicles after endocytosis (Nemani et al., 2010).
The focus of our studies was to elucidate the evolution of pathologic events resulting from α-synuclein over-expression and culminating in synaptic dysfunction. Though we are agnostic about the precise defect(s) induced by α-synuclein at synapses, we favor the view that the pathologic effects are pleiotropic and not restricted to a specific and exclusive impairment of a singular step in the synaptic machinery. First, as previously noted, unbiased screens of α-synuclein toxicity modifiers suggest involvement of multiple vesicular trafficking pathways including those involved in exo- and endo-cytosis. Second, our FM data shows that ≈25% of the transgenic boutons fail to uptake any dye (fig. 5B). Though a diminution in the recycling pool as proposed by Nemani et al. should technically reduce FM-levels within boutons as well, a complete absence of FM loading may reflect accompanying endocytic defects. Third, abnormally enlarged synaptic vesicles, as seen in our EM studies have been repeatedly observed in animal-models deficient in proteins associated with endocytosis (Zhang et al., 1998; Nonet et al., 1999; Deak et al., 2004; Poskanzer et al., 2006; Ferguson et al., 2007), suggesting that such morphologic changes may be due to (at least in part) abnormalities in endocytosis. Finally, we show that α-synuclein over-expressing synapses also have clear diminutions of amphiphysin- a protein involved in endocytosis, further implicating the involvement of multiple exo- and endo-cytic pathways in α-synuclein pathogenesis.
In contrast to the prevalent viewpoint, our data suggest that the eventual pathology in α-synuclein over-expressing synapses is likely to be a complex phenotype induced by the absence of an assortment of presynaptic proteins, including SNARE and vesicle-associated proteins involved in recycling and mobilization of synaptic vesicles (VAMP-2 and synapsin-1), active-zone proteins that act as scaffolds at synapses and are thought to play a role in exo- and endo-cytosis (piccolo), and proteins critical for the endocytosis (amphiphysin). The absence/diminution of these assorted proteins would likely induce defects in recycling, priming as well as endocytosis, perhaps reconciling evidence from existing studies. We note that Nemani et al. also showed changes in some protein levels from brains of α-synuclein transgenic mice, but it is difficult to compare their biochemical data from brain homogenates to our examination of single-boutons that unequivocally contain α-synuclein and can be analyzed with precision.
The enlarged vesicles seen in our EM studies also resemble the h-α-syn-positive vesicular clusters that we and others reported in the yeast model-system over-expressing h-α-syn:GFP (Gitler et al., 2008; Soper et al., 2008). Though the yeast data were somewhat puzzling at the time, in retrospect, perhaps the vesicular clusters represent failed cycles of vesicular fusion/fission and/or endocytosis in yeast, in the presence of excessive h-α-syn. Regarding neurons, though the exact reason for the presence of enlarged vesicles is unclear, given the existing EM evidence from endocytic mutants (cited above), it seems plausible that this synaptic phenotype is simply due to the lack of proteins involved in endocytosis. Alternatively/additionally, α-synuclein binding to synaptic vesicles may directly cause vesicular fusion abnormalities as well (Kamp and Beyer, 2006).
Studies in α-synuclein null mice suggest a physiologic role of α-synuclein in regulating the release of neurotransmitters (Abeliovich et al., 2000), and also in the maintenance of the SNARE machinery itself (Chandra et al., 2004; Chandra et al., 2005). Thus it is conceivable that in diseased states, an imbalance of α-synuclein levels at synapses interferes with its physiologic role of releasing neurotransmitters, eventually leading the exquisite synaptic machinery to go awry. Such mechanisms can be imagined as being analogous to the role of amyloid-beta (Aβ) in Alzheimer’s disease, where emerging evidence implicates a concentration-dependent role for Aβ oligomers in both normal synaptic functioning as well as synaptic damage (Giuffrida et al., 2009; Parodi et al., 2009).
Though our studies are focused on h-α-syn-induced synaptotoxicity, our proposed mechanistic cascade leading to “vacant synapses” may have more general implications for neurodegeneration as well. In a recent report, prolonged treatment of hippocampal neurons with oligomeric Aβ diminished spontaneous synaptic responses in cultured hippocampal neurons, along with a diminution of various presynaptic proteins (Parodi et al., 2009), somewhat similar to our results. Presynaptic protein depletion and synaptic dysfunction was also reported in a model of amyotrophic lateral sclerosis (ALS) associated with mutant superoxide dismutase as well (SOD1) (Wang et al., 2009), suggesting further commonalities between apparently diverse neurodegenerative diseases.
How does excessive α-synuclein lead to a diminution of other presynaptic proteins? One possibility is that excessive α-synuclein may inhibit the axonal transport and/or presynaptic targeting of other proteins. The continuum of protein loss in transgenic boutons as seen in our studies, ranging from a total absence in some (fig. 7), to ≈20–50% diminutions in the remainder (fig. 8) supports this idea. In human DLB brains, wide-spread varicosities containing phospho-α-synuclein are commonly seen (Saito et al., 2003), and may represent axonal transport defects. In this scenario; as α-synuclein itself is conveyed in slow axonal transport with other proteins (Roy et al., 2007, 2008), misfolding/aggregation of the protein may alter its mobility, and also the movements of other co-transported cargoes as they navigate along elongated axons. Indeed we see the greatest diminutions in synapsin, the only proteins examined that moves in slow axonal transport. Recent studies showing that mutant α-synuclein (A53T) diminished levels of various motor proteins in neurons (Chung et al., 2009), and that axonal transport abnormalities are also seen in squid axons treated with 1-methyl-4-phenylpyridinium (MPP+)- a model of experimentally-induced Parkinsonism (Morfini et al., 2007) also implicate axonal transport defects. Alternatively, excessive amounts of mis-folded α-synuclein may aggregate at synapses (Kramer and Schulz-Schaeffer, 2007), physically preventing the targeting of other presynaptic proteins; or alter the biophysical properties/turnover of synaptic vesicles/proteins. Another possibility is that excessive α-synuclein may alter the structural properties of synapses over time, preventing the robust association of synaptic proteins to these sites. Indeed previous studies suggest that actin or other synaptic scaffolding proteins may be involved in α-synuclein pathogenesis (Sousa et al., 2009); Ihara et al., 2007). Future studies will focus on distinguishing between these possibilities.
We thank Margarita Trejo and Anthony Adame for technical assistance with electron microscopy and histology. This work was supported by grants to SR from the Hillblom foundation, the American Parkinson’s Disease Association, the Alzheimer’s Association (NIRG-08-90769), the March of Dimes foundation (Basil O’Connor award), the National Institute for Aging (2P50AG005131-P2), a generous gift from Darlene and Donald Shiley to the UCSD Alzheimer’s Center; and a NIH grant to IT (NS060799).
Author contributions:DS and SR designed and performed the imaging experiments, developed quantitative algorithms, analyzed the data and wrote the manuscript. Electrophysiological studies/analyses were done by IT; YT contributed to immunofluorescence experiments. EM and AS contributed to the ultrastructural studies, human tissue histology and biochemical analyses.