Induction of insoluble α-syn aggregates in primary neurons by α-syn pffs
To determine if exogenous human α-syn pffs can seed recruitment of endogenously expressed mouse α-syn into insoluble LB-like and LN-like fibrillar aggregates, we added α-syn pffs generated from full length recombinant α-syn-hWT to primary hippocampal neurons derived from WT C57BL6 mice after culturing them for 5–6 days in vitro
(DIV). These neurons were examined 2 weeks after the addition of α-syn-hWT pffs, when synapses are mature, and α-syn is normally localized to presynaptic terminals (Murphy et al., 2000
). In PBS treated hippocampal neurons, endogenous mouse α-syn localized to presynaptic puncta as visualized using monoclonal antibody (mAB) Syn202, a pan-synuclein antibody (Giasson et al., 2000
) (, top panels). In contrast, in α-syn-hWT pff-treated neurons, α-syn did not localize to the presynaptic terminal (), but instead formed fibrillar LN-like inclusions. To determine if the α-syn aggregates were detergent-insoluble, PBS and α-syn-hWT pff-treated neurons were extracted with buffer containing 1% Triton X-100 (Tx-100) during fixation. Under such conditions, endogenous α-syn within neuronal processes in PBS treated neurons was soluble in Tx-100, but cells incubated with α-syn-hWT pffs showed Tx-100-insoluble aggregates ().
α-syn-hWT pffs recruit endogenous α-syn to form pathologic, insoluble aggregates
α-Syn recruited into pathologic inclusions undergoes extensive phosphorylation at Ser129 (pSer129), thus antibodies against pSer129 selectively recognize α-syn pathology (Fujiwara et al., 2002
). Furthermore, as this modification is absent in recombinant α-syn pffs (, first lane on left, Luk et al., 2009
), the accumulation of phosphorylated α-syn (p-α-syn) reflects an intracellular modification. PBS treated neurons did not show staining with 81A, a mAB specific for pSer129 (, Waxman and Giasson, 2008). However, neurons treated with α-syn-hWT pffs showed intense 81A immunostaining that was Tx-100 insoluble (). Pff-induced aggregates exhibited morphologies ranging from small puncta to LN-like inclusions of variable lengths within neurites (, , –). Within neuronal perikarya, these α-syn accumulations resembled LBs observed in human PD brains ( inset). The α-syn-hWT pff-induced aggregates also occurred in primary cultures of cortical and midbrain dopaminergic neurons (Supplementary Figure 1A
). Furthermore, neurons generated from other strains of mice as well as rats developed LB- and LN-like inclusions when treated with α-syn-hWT pffs, supporting the hypothesis that induction of α-syn pathology is a general feature of primary rodent neurons (data not shown). P-α-syn-positive aggregates (as detected by 81A) did not form in astrocytes (Supplementary Figure 1B
). Moreover, the appearance of α-syn pathology required the presence of endogenous α-syn since α-syn-hWT pffs did not induce any pathology in primary neurons from α-syn −/− mice (). Furthermore, monomeric α-syn did not induce α-syn inclusions (data not shown), demonstrating that α-syn pffs alone seed the aggregates.
Minimal α-syn domain necessary for aggregate formation
Time dependence of aggregate formation
Effects of aggregate formation on neuronal density and expression of synaptic proteins
Immunoblot analyses were conducted on neuron lysates sequentially extracted with 1% Tx-100, followed by 2% SDS (). In contrast to PBS-treated neurons, those treated with α-syn-hWT pffs for 14 days showed >80% reduction of α-syn in the Tx-100-soluble fraction accompanied by a concomitant appearance of α-syn in the SDS-extractable fraction. Immunoblots of the SDS-extractable fraction also showed insoluble p-α-syn. A mouse specific anti-α-syn antibody did not detect α-syn-hWT pffs (, first lane on left), but detected bands in the neuron lysates similar to those labeled by the C-terminus specific α-syn antibody and mAB 81A. In addition, higher molecular weight species of α-syn were detected in the SDS fraction of all α-syn pffs treated cultures, and likely correspond to oligomeric and/or ubiquitinated α-syn (Li et al., 2005
; Luk et al., 2009
; Sampathu et al., 2003
). Sequential extractions of primary hippocampal neurons from α-syn −/− mice 14 days following addition of α-syn-hWT pffs confirmed the absence of pathological α-syn or any other species of immunoreactive α-syn (Supplementary Figure 1C
). Thus, these data demonstrate that α-syn pffs induced recruitment of soluble endogenous α-syn into insoluble, hyperphosphorylated α-syn aggregates.
Since α-syn is ubiquitinated in LBs and LNs, we studied α-syn aggregates that formed 14 days after addition of α-syn-hWT pffs and showed they were also ubiquitin positive (), and colocalized with p-α-syn. Because the exogenous α-syn-hWT pffs are not ubiquitinated or phosphorylated (Luk et al., 2009
), these post-translational modifications must occur intracellularly as endogenous mouse α-syn accumulates. Thus, these α-syn aggregates share hallmark features of PD-like LNs and LBs allowing us to conclude that misfolded α-syn pffs seed and recruit normal, endogenous α-syn to form pathologic aggregates.
Pffs from the NAC domain of α-syn are sufficient to seed intracellular α-syn aggregates
Previous in vitro
studies have shown that recombinant α-syn protein lacking N- or C-terminal residues, or a synthetic peptide containing only the NAC domain (amino acid residues 61–95), assemble into α-syn amyloid fibrils, and nucleate full length α-syn fibrillization (Giasson et al., 2001
; Han et al., 1995
; Kessler et al., 2003
; Luk et al., 2009
; Murray et al., 2003
; Serpell et al., 2000
). Thus, we asked if human α-syn pffs comprised of α-syn-1-120, α-syn-1-89, α-syn-58-140 or α-syn-NAC could seed formation of LBs and LNs in neurons. We observed that α-syn-1-120, and α-syn-1-89 pffs induced robust accumulation of endogenous p-α-syn aggregates that were Tx-100-insoluble (; data not shown), and they are morphologically indistinguishable from those formed by α-syn-hWT pffs. α-Syn-58-140 pffs also seeded formation of endogenous mouse α-syn aggregates that were hyperphosphorylated (). Moreover, pffs comprised of only the central hydrophobic α-syn-NAC domain also resulted in endogenous mouse α-syn fibrillar LB-like aggregates that were Tx-100-insoluble. Overall, our data demonstrate that α-syn pffs containing only the central, hydrophobic portion of α-syn-hWT are sufficient to seed conversion of endogenous α-syn into pathological aggregates.
Mice typically do not develop LBs except in the case of transgenic lines overexpressing mutant human α-syn. We thus asked if the formation of LB-like aggregates required human α-syn or if they can be seeded by α-syn pffs generated from recombinant mouse WT α-syn (α-syn-mWT) (Touchman et al., 2001
). Immunoblots demonstrated that 14 days treatment of primary neurons with α-syn-mWT pffs induced appearance of p-α-syn in the Tx-100-insoluble fraction (). Immunofluorescence also showed that α-syn-mWT pffs induced formation of p-α-syn aggregates in neurites and somata. Thus, pathological PD-like α-syn aggregates can be induced by α-syn-mWT pffs and does not require the human protein.
Ultrastructure analysis of α-syn aggregates
Examination of the α-syn aggregates using transmission and immuno-EM demonstrated abundant filaments in neurons treated with either α-syn-hWT or α-syn-1-120 pffs () for 14 days, but not PBS-treated neurons (data not shown). Remarkably, inclusions comprised of 14–16 nm thick filaments were seen throughout the cytoplasm, visualized by transmission EM. Two different immuno-EM detection systems, horse radish peroxidase (HRP) and immunogold amplification, were used to demonstrate that fibrils comprised of p-α-syn are found throughout the neuron. P-α-syn-positive fibrils were seen in the soma (), adjacent to the active zone of presynaptic terminals (), the post-synaptic terminal () and throughout processes (). These data establish that the seeding and recruitment of endogenous mouse α-syn into hyperphosphorylated insoluble, filamentous aggregates recapitulate features of LBs and LNs in PD and other human synucleinopathies.
Time and concentration dependence of α-syn aggregate formation
To determine the temporal sequence of α-syn aggregate formation, α-syn-hWT pffs were added to the neurons at DIV5. P-α-syn immunostaining was not detectable until 4 days later when small aggregates began to appear, exclusively in the neurites, albeit at low levels (, upper series). By 7 days after α-syn-hWT pffs addition, there was an increase in α-syn pathology with some cell bodies showing α-syn accumulations. By 10 days post α-syn-hWT pff addition, the overall p-α-syn immunostaining was more intense, and p-α-syn aggregates in the neurites appeared both punctate and fibrillar resembling LNs that were longer than the aggregates observed 4 or 7 days after α-syn-hWT pffs addition.
The sequence of events revealed by immunofluorescence was confirmed by biochemical experiments of sequentially extracted neurons (). Four days after α-syn-hWT pffs addition, the majority of α-syn was found in the Tx-100-soluble fraction and showed levels similar to PBS-treated neurons. In PBS-treated control neurons, there was an increase in α-syn levels by DIV10 as demonstrated previously (Murphy et al., 2000
). In contrast, 7–10 days following α-syn-hWT pff treatment, soluble levels of α-syn were reduced, accompanied by a concomitant increase of α-syn into the Tx-100-insoluble fraction. Thus, these data indicate that α-syn-hWT pff-induced recruitment of mouse α-syn into the insoluble fraction with a lag phase of a few days followed by a progressive increase in insoluble p-α-syn.
Since levels of α-syn, and its concentration at the presynaptic terminals increase as primary neurons mature, (Murphy et al., 2000
; , day 4 PBS vs. day 10 PBS), we asked if adding pffs to mature neurons would enhance the rate of aggregation. When α-syn-hWT pffs were added to DIV 10 neurons, aggregates were visible in neurites 2 days later (, lower series), in contrast to 4 days required after addition of pffs to DIV5 neurons. By 4 days following α-syn-hWT pff treatment of DIV10 neurons, small punctate aggregates were detected throughout the neurites and some somata also showed accumulations, again unlike 4 days after adding pffs to DIV5 neurons in which α-syn pathology was exclusively in neurites. Seven days after α-syn-hWT pff treatment of DIV10 neurons, the pathology was extensive, similar to 10 days α-syn-hWT pff treatment of DIV5 neurons (). Thus, α-syn aggregates develop faster in mature neurons, consistent with in vitro
studies demonstrating that the rate of fibril formation positively correlates with α-syn concentrations (Wood et al., 1999
We next examined if the amount of α-syn pathology correlated with the amount of fibrils added. We found progressive decreases in the amount of somatic and neuritic pathology correlated with 10-fold serial dilutions of α-syn-hWT pffs added (in ng/mL: 100, 10, 1, 0.1; Supplementary Figure 2
). Thus, the rate and extent of pathology depends on the amount of α-syn pffs, and that small quantities of α-syn pffs are sufficient to seed α-syn aggregate formation, consistent with in vitro
studies showing that the rate of seeded assembly depends on the initial concentrations of α-syn pffs (Wood et al., 1999
Initial formation of α-syn pathology within axons
Because α-syn normally localizes to the presynaptic terminal and since α-syn puncta initially appeared in neurites, we hypothesized that α-syn-hWT pffs recruited presynaptic α-syn into insoluble aggregates that then propagate from the axons to the cell bodies. To demonstrate that the pathology initiated in axons, we conducted double labeling immunofluorescence studies using a mAB specific for mouse tau (T49, an axonal marker) and 81A. P-α-syn aggregates colocalized predominately with tau 4 days after pff addition (; upper panel), but not with the dendritic marker, microtubule associated protein 2 (MAP2) (, upper panel), indicating that α-syn accumulations were initiated in axons. However, by 14 days, when more accumulations appeared in the somata, the α-syn aggregates were seen in axons (, lower panel), in cell bodies, and proximal dendrites where they colocalized with MAP2 (, lower panel). Thus, α-syn is recruited away from the presynaptic terminal with subsequent spread via axons to other parts of the polarized neuron.
Enhanced endocytosis of α-syn pffs increases the extent of pathology
To determine if α-syn-hWT pffs can gain access to the cytoplasm to seed recruitment of endogenous α-syn, we performed two-stage immunofluorescence using antibodies that recognize only human α-syn pffs. Live neurons were labeled at 4°C with mAB Syn204 followed by fixation, permeabilization, incubation with the antibody, LB509 (Giasson et al., 2000
),. Thus, mAB Syn204 labeled only extracellular hWT pffs whereas LB509 recognized both extracellular and intracellular hWT pffs. Many α-syn-hWT pffs remained outside the neuron and were double-labeled with both mAB Syn204 and LB509 (yellow in the merged image, ). However, significant amounts of small puncta labeled exclusively with LB509 (green, arrowheads highlight examples in the merged image), suggesting that α-syn-hWT pffs gain entry inside the neuron, as demonstrated previously for both α-syn and tau amyloid fibrils (Luk et al., 2009
; Guo and Lee, 2011
). Furthermore, double labeling immunofluorescence in fixed, permeabilized neurons with mAB 81A and mAB Syn204 showed p-α-syn accumulating near seeds of α-syn-hWT pffs (). A 3D view constructed from serial confocal images demonstrated colocalization between α-syn-hWT pffs (Syn204) and p-α-syn (81A) in the XY, XZ and YZ planes (), further confirming that intracellular pffs seed recruitment of endogenous α-syn. Since p-α-syn is exclusively intracellular, our data indicate that pffs enter the cytoplasm where they initiate accumulation of pathologic p-α-syn.
α-syn-hWT pffs are internalized into neurons
To begin assessing the mechanism by which pffs gain entry to the cytoplasm, we treated neurons with α-syn-hWT pffs in the presence of wheat germ agglutinin (WGA) which binds N
-acetylglucosamine (GlcNAC) and sialic acids at the cell surface and induces adsorptive-mediated endocytosis (Banks et al., 1998
; Broadwell et al., 1988
; Gonatas and Avrameas, 1973
). To determine the effects of WGA on formation of α-syn aggregates, neurons were treated at DIV5 and fixed for immunofluorescence 4 days later. When incubated with α-syn-hWT pffs alone, few p-α-syn puncta were visible in a subset of neurites (). Co-incubation of pffs with WGA dose-dependently increased the extent of p-α-syn pathology. In addition to small puncta, longer, continuous p-α-syn filaments were visible, and α-syn pathology was present in the cell body, particularly with 5 μg/mL of WGA treatment. Furthermore, the addition of 0.1 M GlcNAc, a competitive inhibitor of WGA, reduced the effects of WGA on α-syn pff induced aggregate formation. Immunoblots of sequentially extracted neurons confirm that WGA-mediated endocytosis enhances formation of pathologic α-syn. Four days after treatment with α-syn-hWT pffs alone, the majority of α-syn remained in the Tx-100 extractable fraction, whereas co-incubation of α-syn-hWT pffs with 5 μg/mL of WGA increased the amount of Tx-100 insoluble α-syn. Taken together, our findings indicate that α-syn pffs gain access to the neuronal cytoplasm by adsorptive endocytosis.
Intracellular propagation of pathologic α-syn
To determine if direct addition of α-syn pffs to either neurites or somata leads to propagation of pathologic α-syn aggregates throughout the neuron, we utilized microfluidic culture devices which isolate the neuronal processes from the cell bodies via a series of interconnected microgrooves (Taylor et al., 2005
). C-terminally myc-tagged α-syn-1-120 pffs added to the neuritic chamber () resulted in p-α-syn-positive aggregates within axons and cell bodies (). Aggregates were morphologically identical to those seen in primary neurons directly exposed to pffs, and they were also insoluble in Tx-100 (). Anti-myc immunostaining suggested that pffs did not enter into the somal compartment () or microgrooves. Thus, these data indicate that pathological p-α-syn can form within isolated neurites, and is propagated retrogradely to the cell bodies.
Intracellular propagation of pathologic α-syn aggregates
We also exposed neuronal somata that were isolated from neurites in the microfluidic devices to α-syn-1-120-myc pffs and assessed the extent of α-syn pathology in the processes (). As expected, neurons treated with α-syn-1-120-myc pffs formed somatic p-α-syn pathology (). P-α-syn aggregates were also detected in axons that extended through the microgrooves into the neurite chamber, as revealed by co-labeling with tau (). Again, α-syn aggregates throughout the axon were Tx-100-insoluble, and immunofluorescence using the anti-myc antibody demonstrated that α-syn-1-120-myc pffs were confined to the somatic compartment (). Thus, we conclude that pathologic p-α-syn aggregates also propagate in the anterograde direction.
Formation of aggregates leads to neuron loss and diminished levels of select synaptic proteins
α-syn resides predominantly at the presynaptic terminal and previous reports indicate that it acts as a co-chaperone, in concert with another chaperone, cysteine-string protein α (CSPα), to maintain SNARE complex formation by binding to VAMP2/synaptobrevin 2 (Burre et al., 2010
; Chandra et al., 2005
; Greten-Harrison et al., 2010
). Thus, we examined the consequences of recruitment of α-syn into insoluble aggregates on synaptic protein distribution and expression. In PBS-treated control neurons, α-syn colocalized with VAMP2 at the presynaptic terminal. Addition of α-syn-hWT pffs led to a depletion of α-syn from the presynaptic terminal such that it showed minimal colocalization with presynaptic VAMP2 (). To further investigate the molecular consequences of recruitment of endogenous α-syn into insoluble aggregates, we examined additional synaptic proteins that could be impacted by the pathological sequestration of α-syn into aggregates and away from the presynaptic terminal. Although β-synuclein (β-syn), another member of the same family of neuronal proteins as α-syn, but lacking the NAC domain, colocalized with α-syn at presynaptic terminals in control neurons (Murphy et al., 2000
), α-syn-hWT pff addition did not change the presynaptic localization of β-syn (Supplementary Figure 3
). Furthermore, Tx-100 extraction showed that, unlike pathological α-syn, which localized to detergent insoluble aggregates, β-syn remained soluble (Supplementary Figure 3
). Immunoblot analyses showed that endogenous β-syn was Tx-100 soluble 14 days after adding α-syn pffs () and protein levels in pff treated neurons were not statistically significantly different from PBS treated neurons. Thus, like LBs in PD brains, the aggregates that developed in primary neurons are comprised of insoluble α-syn, but not β-syn (Spillantini et al., 1998
). Importantly, this is consistent with the selective recruitment of α-syn by pffs as opposed to the indiscriminate disruption of adjacent presynaptic components.
Nonetheless, we were able to detect statistically significant reductions in a subpopulation of synaptic proteins two weeks after the addition of α-syn-hWT pffs, including the synaptic vesicle-associated SNARE proteins, Snap25 and VAMP2, as well as soluble proteins that participate in SNARE complex assembly or the exo-endocytic synaptic vesicle cycle such as CSPα, and synapsin II (). Levels of other synaptic proteins showed slight, but not statistically significant reductions. Changes were not observed in GAPDH, the plasma membrane-associated SNARE protein, syntaxin 1, or the transmembrane synaptic protein synaptophysin.
Since loss of synaptic proteins may correlate with neurodegeneration, we asked if the accumulation of α-syn aggregates leads to neuron loss. NeuN-positive neurons were counted in cultures treated with PBS or α-syn-hWT pffs 4, 7 or 14 days after α-syn pff addition. While there was a slight, but not statistically significant decrease in number of neurons 7 days following α-syn-hWT pff treatment, by 14 days post pff treatment, there was a significant 40% decrease in neurons relative to PBS controls (). Cell death did not occur in α-syn-hWT pff treated neurons derived from α-syn −/− mice, demonstrating that intracellular aggregates, rather than the mere addition of exogenous pffs, caused neuron death. Finally, using ethidium homodimer to detect dead cells and Hoechst 33342 to detect total cells, we demonstrated a ~68% increase in cell death 14 days after pff-treatment (56.6%) versus PBS-treated (33.6%) neurons.
Disruption in network activity matches the progression of α-syn pathology
The decreased levels of synaptic proteins suggest impairment in neural network activity following accumulation of α-syn inclusions. Calcium imaging of hippocampal neurons loaded with the calcium-sensitive fluorescent dye, Fluo-4 AM, was performed to investigate the effect of α-syn aggregates on the activity patterns of the in vitro neural network established by these cultured neurons. The spontaneous activity of neurons treated with PBS was characterized by flickering events, intermixed with network-wide bursts when nearly all the neurons were simultaneously firing as reflected by a high synchronization index (). In contrast, neurons treated with α-syn-hWT pffs showed a significant decrease in synchronized activity as early as 4 days after treatment. At this time point, low levels of α-syn aggregates were visualized exclusively in axons by immunofluorescence microscopy, and no pathological α-syn was detected biochemically (). Yet, this was sufficient to impair coordinated network activity. This reduction in synchronized activity persisted at 7, 10 and 14-days after α-syn-hWT pff treatment (). In contrast, α-syn-hWT pff-treated neurons from α-syn −/− mice showed no impairments in the synchronization index, indicating that these effects are selective for neurons harboring α-syn aggregates and do not result from exogenously added pffs.
Effect of aggregate formation on neural network activity
We next determined whether the progressive recruitment of α-syn into pathologic aggregates correlated with changes in the excitatory tone of the network. First, synchronous oscillations were forced using the GABA(A) antagonist, bicuculline, to abolish inhibitory input, followed by increasing doses of the AMPA receptor antagonist, NBQX, until synchronous oscillations stopped (). The final concentration of NBQX required to impair activity within the excitatory network determined the excitatory tone. No significant changes in excitatory tone was detected in cultures 4 or 7 days after α-syn-hWT pff treatment but by 10 and 14 days after treatment, when increasing accumulation of neuritic and perikaryal pathology was observed, there were significant reductions in excitatory tone (), reflecting compromised synaptic activity. Again, neurons from α-syn −/− mice did not show impairments in excitatory tone at 10 and 14 days post-pff treatment, confirming that the effects result from the accumulation of endogenous α-syn aggregates.
Since spatiotemporal patterns of activity are shaped by the underlying connectivity architecture and the relative balance of excitation and inhibition, we used network activity patterns to determine the functional connectivity in PBS and α-syn-hWT pff treated neurons. As neurons matured in vitro
, the number of functional connections increased and eventually plateaued (). The timeframe for this correlated well with neurite sprouting and synapse stabilization based on previous studies of developing connections in vitro
(Soriano et al., 2008
However, in α-syn-hWT pff treated WT, but not α-syn −/− neurons, the maturation of functional connections never reached the level achieved in PBS treated cultures, as a significant reduction was observed 10 days after α-syn-hWT pff treatment (). This functional connectivity was severely compromised 14 days post treatment and the network consisted of just a few sparse connections at this time point (). In summary, the formation of insoluble aggregates of endogenous α-syn results in early disruption in coordinated network activity. Later, as more α-syn inclusions develop and propagate throughout the neuron, excitatory tone is decreased and functional connectivity is greatly reduced.