|Home | About | Journals | Submit | Contact Us | Français|
Although the presynaptic protein α-synuclein is a recognized player in neurodegeneration, its precise physiologic function(s) and/or role in human disease remains unclear. An emerging consensus from previous studies in lower-order systems is that α-synuclein interferes with vesicle-trafficking pathways; however putative neuronal correlates are unknown. Here we explore consequences of α-synuclein modulation in cultured mouse hippocampal neurons; coupling α-synuclein over-expression and knockout model-systems with contemporary imaging paradigms. Our studies reveal an unexpected role of α-synuclein in attenuating the mobility of recycling pool (RP) vesicles between presynaptic boutons – called “superpool” trafficking – and also in maintaining the overall size of RPs at synapses. While an excess of α-synuclein led to smaller RPs and inhibited inter-synaptic trafficking, an absence of α-synuclein triggered converse changes with larger RPs and enhanced inter-synaptic trafficking. The data collectively suggest a model where α-synuclein maintains RP homeostasis by modulating inter-synaptic vesicular dynamics, and provide a putative neuronal correlate of α-synuclein-induced impairments in vesicle-trafficking previously reported in lower-order systems.
The small presynaptic protein α-synuclein plays a central role in several neurodegenerative diseases, collectively called “synucleinopathies”. Despite considerable interest, neither the normal function(s) of α-synuclein, nor consequences of its over-expression are fully understood. Previous studies in lower-order systems have consistently shown that α-synuclein interferes with vesicle-trafficking pathways. This is best studied in the yeast, where the evidence is compelling. Expression of α-synuclein in yeast inhibits ER→Golgi trafficking (Outeiro and Lindquist, 2003; Cooper et al., 2006; Gitler et al., 2008; Soper et al., 2008) that can be rescued by genes that facilitate this pathway (Cooper et al., 2006; Soper et al. 2008). Vesicular α-synuclein accumulations – likely secondary to trafficking deficits – are also seen at an ultrastructural level (Gitler et al., 2008; Soper et al., 2008; Pranke et al. 2011), consistent with ER→Golgi trafficking impairments, and unbiased yeast genetic screens implicate trafficking-pathways as well (Willingham et al., 2003). Similar α-synuclein-induced phenotypes have been reported in non-neuronal/PC-12 cell-lines (Thayanidhi et al., 2010), C. elegans (Cooper et al., 2006; Gitler et al., 2008; Kuwahara et al., 2008; van Ham et al., 2008) and drosophila (Cooper et al., 2006), collectively making a rather strong case that α-synuclein inhibits vesicle-trafficking pathways.
However, while deficits in secretory-trafficking pathways are important clues; in neurons, the vast majority of α-synuclein is not normally localized around the ER/Golgi, but is instead concentrated at distal presynapses. Thus we reason that putative neuronal α-synuclein induced vesicle-trafficking deficits are likely to first manifest at/around synapses. Here we employ knockout and over-expressing systems to evaluate effects of α-synuclein modulations on synapses. Our data suggest unexpected roles of α-synuclein in regulating synaptic vesicle trafficking and maintaining RP homeostasis.
See Scott et al., 2010 for basic details. Briefly, cells from postnatal (P0-P1) mice were plated (30,000-60,000 cells/cm2) and cultured to maturity (DIV 17–21) before experimental manipulations. For transient FM4-64 loading, neurons were incubated in stimulation solution (5mM HEPES, 90mM NaCl, 1mM MgCl2, 2mM CaCl2, 10mM glucose, and 47mM KCl) containing 10-15µM FM4-64 and NMDA-antagonists (50uM DL-AP5, 20uM CNQX) for 90 s; a protocol known to label functional RPs (Darcy et al., 2006; Gaffield and Betz, 2006). For field-stimulation experiments, neurons were placed in a chamber (EC-B-18 – Chamlide EC) connected to a SD9-square-pulse stimulator (Grass) and imaged every 3s. Resultant data was photobleach-corrected. Alpha-synuclein null mice were obtained from Jackson Labs. For experiments involving paired WT, +/− and −/− littermates, identity of pups was confirmed by genotyping. All UCSD guidelines were followed. Experiments were repeated at least 2–3 times on separate culture-sets. C57/b6 mouse pups of either sex were used.
All FRAP experiments were performed on a Fluoview-1000 confocal microscope (Olympus). FM4-64-loaded, adjacent en-passant boutons (separated by < 7µm – Darcy et al., 2006) were photobleached to 30 ± 1% (mean ± SEM) of the original fluorescence intensity by scanning with a 515nm laser at 100% power for 5s (2.0µs/pixel). Images were corrected for photobleaching, and FRAP kinetics for experimental groups were statistically analyzed by paired t-test and expressed as mean ± SEM. A p value of < 0.05 was considered significant.
These algorithms automated the imaging process, sampled large areas, and significantly scaled up sample size. Regularly spaced 6 × 6 (or 8 × 8) field grids (696 × 520 pixels/grid) were distributed evenly over each coverslip (fig. 1). Z-series images were acquired from each grid (36/64 z-series stacks for each coverslip) and de-convolved (Huygens). An auto-focusing algorithm facilitated rapid search for an optimal focal plane for each microscopic field while minimizing latent exposure of the field to incident light before acquisition. All image sampling and focusing algorithms were implemented as scripts for use within Metamorph.
Image acquisition, deconvolution and colocalization-algorithms were similar to our previously published protocols (Scott et al. 2010). For retrospective correlation experiments, FM4-64+ve boutons were imaged live and cultures were fixed/stained for mouse presynaptic proteins. Images were retrospectively aligned, and FM4-64/synaptic-protein intensities were quantified. Fidelity of the datasets was cross-checked with extensive manual sampling. For bouton-correlation experiments, some paired intensity measurements were square root transformed, and expressed as a correlation coefficient (Pearson’s). Contour-plots were calculated using the 2-D kernel density estimation function supplied with the MCLUST package for R. Experimental groups were statistically analyzed using a nonparametric t-test and expressed as mean ± SEM; a p value of < 0.05 was considered significant.
Algorithms were developed in Matlab/Metamorph. (1) Movies (120 sec) of FM4-64-loaded boutons/intervening axons were acquired (2 Hz), corrected for photobleaching and background subtracted. (2) Flux was calculated as absolute change in intensity/pixel for each successive time-point by subtracting consecutive frames in a movie. This is represented as [I(tn) - I(tn - I(tn-1)] where I is the absolute fluorescence intensity of each pixel, and t represents a given time-point. Resulting image-stacks were noise-corrected (Dussault and Hoess, 2004). (3) 8–12 pixel polylines were placed in axonal regions between FM4-64 positive boutons, and 2-D noise-corrected kymographs were generated. Pixel-intensities in kymographs was averaged and represented graphically as “average FM4-64-flux”.
Three types of cultured hippocampal neurons were used in this study – neurons expressing wild-type (WT) α-synuclein, neurons lacking α-synuclein (from α-synuclein null mice), and neurons over-expressing α-synuclein. Briefly, dissociated hippocampal neurons were obtained from mice over-expressing human α-synuclein tagged to GFP (α-syn:GFP) – or lacking α-synuclein altogether (α-syn −/−), and all experiments were performed in neurons cultured for 17–21 days in-vitro (DIV). WT/+/− littermates were used as paired controls. There is ~ two-fold higher level of α-synuclein in DIV-17–21 α-syn:GFP neurons, with no detectable synaptic losses (Scott et al. 2010 - also see Unni et al., 2010). This model-system offers several advantages. The α-synuclein-positive boutons can be visualized directly (by their native GFP-fluorescence), and the system is amenable to live-cell experiments. Moreover, large numbers of boutons can be analyzed (several thousand per experiment – see below) offering significant methodological advantages and statistical power over strategies using transient transfections. Upgrades to our previous protocols now allow us to evaluate considerably larger numbers of boutons in a relatively unbiased manner (fig. 1A and methods), minimizing the impact of intrinsic local variations in synaptic quantitation.
Working with α-synuclein over-expressing model-systems, we and others previously reported reduced presynaptic uptake of the styryl dye FM4-64 – a marker of functional RPs (Nemani et al. 2010; Scott et al. 2010). The extent of these diminutions, as well as the heterogeneity amongst individual synapses is shown in figure 1C, D. We used our bouton-sampling algorithms to compare RPs in neurons from α-synuclein null mice (Abeliovich et al., 2000) with WT mice – processing cultures in parallel. Surprisingly, average FM4-64 intensities were significantly greater in α-synuclein −/− boutons (fig. 1E). We also repeated these experiments in neurons from littermate-paired mice containing or lacking endogenous α-synuclein (WT/+/− and −/− respectively). Average intensities of both FM-dye uptake (fig. 1F, left) and endogenous presynaptic protein levels were higher in −/− neurons (fig. 1F, middle and right); unlike over-expressing neurons (see below).
Previously, we reported reduced presynaptic protein levels in α-synuclein over-expressing boutons (Scott et al. 2010). Using A53T-mutant α-synuclein mice, Lim and colleagues also found similar diminutions in vesicular presynaptic proteins in-vivo, but with relative preservation of t-SNARE components (Lim et al. 2011). Revisiting the issue, we found that while levels of several membranous presynaptic proteins are diminished in α-synuclein over-expressing boutons, levels of the t-SNARE protein SNAP-25 were actually higher than littermate-controls (fig. 2A, B – also see Burre et al., 2010; Garcia-Reitböck et al., 2010). Notably, we also repeated these experiments in neurons from an identical mouse model that lacks the GFP-tag (PDGF promoter-human-α-synuclein mice) with similar results (not shown) – indicating that diminutions are independent of the GFP-tag. Hence α-synuclein-induced diminutions in presynaptic protein levels are likely a bona-fide observation.
What is the basis for these diminutions? Since presynaptic proteins associate with synaptic vesicles (either integrally or peripherally), we asked whether such diminutions correlated with reduced RPs. WT/transgenic boutons were loaded with FM4-64, imaged, then fixed and stained the with mouse VAMP-2/synapsin-1 antibodies; retrospectively comparing VAMP-2/synapsin-1 fluorescence with corresponding FM4-64-intensities (see methods). Bouton-crops (fig. 2C) highlight correlated diminutions in FM4-64 and endogenous presynaptic proteins; quantitatively depicted in figure 2D. Moreover, corresponding diminutions in synapsin-1/VAMP-2 levels were seen in the same α-syn:GFP over-expressing boutons (fig. 2F-left, but not with synapsin-1/SNAP-25, fig. 2F-right – see above). Though a diminished RP-level is likely responsible for the ‘reduced presynaptic-protein’ phenotype, effects of other processes (e.g. reduced axonal transport) cannot be excluded and will require more focused experiments.
Collectively, the above data suggest that α-synuclein may be involved in maintaining RPs at steady-state levels. RP-levels are maintained at equilibrium by replenishment from local and imported sources – reviewed in Denker and Rizzoli, 2010 – the former by endocytic retrieval of recycling vesicles within individual boutons post-exocytosis, and the latter via refilling of RPs via trafficking vesicles. Specifically, recent studies have shown that RP vesicles are continuously exchanged within adjacent boutons in resting neurons – a phenomenon dubbed “superpool” trafficking (Krueger et al., 2003; Darcy et al., 2006; Staras et al., 2010; Herzog et al., 2011; Ratnayaka et al., 2012).
Given the known effects of α-synuclein on vesicle-trafficking in lower-order systems, we asked whether superpool-trafficking was altered in neurons over-expressing/lacking α-synuclein. Towards this we first used an “FM4-64-FRAP assay” that reports the exchange of RP-vesicles between boutons (Staras et al. 2010; Darcy et al., 2006). Boutons were loaded with FM-dyes to label RPs, fluorescence within a single bouton is photobleached, and FM-dye recovery within the bleached bouton was monitored. As recovery is contingent upon movement of unbleached fluorescent molecules from adjacent boutons into the bleached bouton (Lippincott-Schwartz et al., 2003), the extent of FM-dye recovery reports the overall mobility of the superpool (fig. 3A). In FM4-64-FRAP experiments with boutons containing α-syn:GFP, we found a significant negative correlation between the maximum extent of FM4-64 recovery and the intensity of α-syn:GFP within the same bouton (fig. 3B, top), suggesting that α-synuclein may have an inhibitory effect on inter-synaptic vesicle-trafficking. Furthermore, kinetics of FM4-64 recovery were significantly slower in high α-syn:GFP over-expressers – quantitatively defined as simply the upper 50th percentile of the average GFP intensity-range across all boutons (fig. 3B, bottom).
Intriguingly, similar FM4-64-FRAP experiments in α-synuclein −/− neurons revealed converse changes, where the extent of FM4-64 recovery was higher in neurons from α-synuclein −/− mice, compared to their WT/+/− littermates (fig. 3C). Remarkably, when we compare all our FM4-64-FRAP data from WT, +/− and −/− DIV 17–21 neurons, there appears to be a dose-dependent influence of α-synuclein on FM4-64 recovery (fig. 3D). Specifically, the extent of FM-dye recovery is lowest in boutons containing the highest levels of α-synuclein (Up-50 h-α-syn in fig. 3D); whereas the most rapid FM-dye recovery occurs in the α-synuclein −/− boutons, suggesting that α-synuclein may have a role in ‘fine-tuning’ this trafficking process.
Though the trafficking of FM4-64 vesicles between boutons is obvious in videos, the mobile particles are heterogeneous and vary greatly in intensities, making manual tracking/quantification difficult – also see Darcy et al., 2006 – prompting us to develop an “FM-flux” assay to directly monitor the superpool. Transgenic neurons were loaded with FM4-64, both α-syn:GFP and FM4-64 fluorescence was simultaneously imaged, and successive frames in a movie were subtracted to eliminate stationary fluorescence (including synaptic fluorescence that was largely stationary over these short time-scales). These methods robustly highlighted the flux of FM4-64-particles in inter-boutonic axons (fig. 4A). Significant reductions of FM4-64-flux were seen in α-synuclein over-expressing neurons (fig. 4B), arguing that α-synuclein induces a specific trafficking-deficit. Furthermore, such inhibitions resulted in stalling of FM4-64 particles in axons, leading to an increase in cumulative inter-boutonic axonal FM4-64-fluorescence from “un-subtracted” movies (fig. 4C). We also saw instances where α-syn:GFP and FM4-64 particles were co-transported in inter-boutonic axons (48/144 particles, ~33% were positive for both) suggesting physical associations of α-synuclein with trafficking synaptic vesicles. Finally, interruptions of vesicle-trafficking using the microtubule-depolymerizing drug nocodazole also reduced RP-levels (fig. 4D), arguing that an inhibition of trafficking/transport could result in diminished RPs.
We also monitored stimulus-dependent FM4-dye exocytosis in WT and α-syn:GFP +ve boutons. Increased α-synuclein-levels led to reductions in exocytosis, and the data suggest a dose-dependent effect of α-synuclein on release-kinetics (fig. 4E–G).
Exploring the cell biology of α-synuclein in over-expressing and knockout neurons, we found unexpected roles of α-synuclein in modulating the size and dynamics of synaptic vesicle pools. Specifically, α-synuclein inhibits the inter-synaptic vesicle-trafficking and manipulating α-synuclein levels influences the overall size of recycling pools (RPs) at synapses, as well as their release properties. Collectively, the data suggest that α-synuclein maintains vesicle-pools within boutons at equilibrium by modulating/regulating vesicle-motility. This view does not necessarily exclude other possibilities – for instance effects of α-synuclein on local endocytosis and/or membrane-curvatures (Pranke et al., 2011) – that may also be a part of (or lead to) the spectrum of disrupted vesicle-trafficking. However, our data cannot conclusively link the α-synuclein-induced superpool deficits to changes in RP-size, and independent effects of α-synuclein on these two variables cannot be ruled out.
Normal synaptic transmission depends upon a pool of vesicles that recycle with the presynaptic plasma-membrane and release their content (neurotransmitters) into the synaptic cleft – reviewed in Denker and Rizzoli, 2010. Thus the RP is a key player in neurotransmitter release, and release-probability (pr) at synapses is known to correlate with RP-size (Murthy et al., 1997). Our FM-dye uptake experiments in mice over-expressing/lacking α-synuclein (fig. 1) suggest that α-synuclein may have a role in regulating the overall size of RPs; and our FM-dye release experiments (fig. 4E–G) suggest α-synuclein-dependent diminutions in release-kinetics. The notion that α-synuclein is a negative regulator of exocytosis is also supported by recent studies showing exaggerated synaptic responses in αβγ-synuclein-knockout mice (Anwar et al., 2011; Greten-Harrison Becket et al., 2010) – though this view has been challenged (Burre et al., 2011).
Neuronal correlates of α-synuclein-induced vesicle-trafficking deficits found in lower-order systems have been mysterious. Using FRAP as well as direct flux-imaging assays, our data show that excessive α-synuclein inhibits the trafficking of synaptic vesicles between boutons – the “superpool”. The superpool is a large repository of synaptic vesicles (including recycling vesicles) that spans several boutons and is continuously exchanged amongst them in the resting state (Staras et al., 2010; Darcy et al., 2006; Herzog et al., 2011). Estimates suggest that this pool may be >100 fold larger in size than the RP at an individual bouton, and superpool-vesicles are recruited into boutons, where they appear to undergo exocytosis (Staras et al., 2010; Ratnayaka et al., 2011) – suggesting a dynamic repository of vesicles that can be recruited into individual synapses. Emerging evidence extends the superpool to recycling as well as resting vesicle-pools, and also implicates another peripheral presynaptic protein (synapsin) in regulating superpool-dynamics (Orenbuch et al., 2012). Given the enormous distances that may exist between the soma and a presynaptic bouton, the existence of such a process is also conceptually appealing. At this time, the precise molecular details underlying α-synuclein-mediated attenuation of superpool-trafficking are unclear. Further studies are also needed to unequivocally establish a mechanistic relationship between superpool-trafficking and RP-size.
We thank Yong Tang and Utpal Das for help with genotyping, Edward Rockenstein and Eliezer Masliah for α-synuclein:GFP transgenic breeders, Steve Edland for statistical consultation, and Christina Sigurdson for helpful comments on the manuscript (all at UCSD). This work was supported by grants from the Larry Hillblom Foundation, the Alzheimer’s Association (NIRG-08-90769), the American Federation for Aging Research (AFAR), and NIH (P50AG005131 and R01NS075233) to S.R. We also acknowledge the UCSD Neuroscience Microscopy Shared Facility Grant (P30 NS047101) for confocal FRAP imaging.