Role of α-synuclein penetration into the membrane in the mechanisms of oligomer pore formation

Authors


Dr Eliezer Masliah, Department of Neurosciences, University of California, San Diego, La Jolla, CA 92093-0624, USA
Fax: +858 534 6232
Tel: +858 534 8992
E-mail: emasliah@ucsd.edu

Abstract

Parkinson’s disease (PD) and dementia with Lewy bodies are common disorders of the aging population and characterized by the progressive accumulation of α-synuclein (α-syn) in the central nervous system. Aggregation of α-syn into oligomers with a ring-like appearance has been proposed to play a role in toxicity. However, the molecular mechanisms and the potential sequence of events involved in the formation of pore-like structures are unclear. We utilized computer modeling and cell-based studies to investigate the process of oligomerization of wild-type and A53T mutant α-syn in membranes. The studies suggest that α-syn penetrates the membrane rapidly, changing its conformation from α-helical towards a coiled structure. This penetration facilitates the incorporation of additional α-syn monomers in the complex, and the subsequent displacement of phospholipids and the formation of oligomers in the membrane. This process occurred more rapidly, and with a more favorable energy of interaction, for mutant A53T compared with wild-type α-syn. After 4 ns of simulation of the protein–membrane model, α-syn had penetrated through two-thirds of the membrane. By 9 ns, the penetration of the annular α-syn oligomers can result in the formation of pore-like structures that fully perforate the lipid bilayer. Experimental incubation of recombinant α-syn in synthetic membranes resulted in the formation of similar pore-like complexes. Moreover, mutant (A53T) α-syn had a greater tendency to accumulate in neuronal membrane fractions in cell cultures, resulting in greater neuronal permeability, as demonstrated with the calcein efflux assay. These studies provide a sequential molecular explanation for the process of α-syn oligomerization in the membrane, and support the role of formation of pore-like structures in the pathogenesis of the neurodegenerative process in PD.

Abbreviations
α-syn

α-synuclein

DLB

dementia with Lewy bodies

GFP

green fluorescent protein

MD

molecular dynamics

PD

Parkinson’s disease

POPC

1-palmitoyl-2-oleoyl-phosphatidylcholine

SNARE

soluble N-ethylmaleimide-sensitive factor attachment protein receptors

TU

titer units

Introduction

α-Synuclein (α-syn) is thought to be a natively unfolded synaptic protein [1,2] that, under physiological conditions, plays a role in synaptic vesicle release via interactions with members of the SNARE family [3,4]. Under pathological conditions, progressive accumulation of α-syn, leading to the formation of oligomers, has been proposed to play a critical role in the pathogenesis of Parkinson’s disease (PD), dementia with Lewy bodies (DLB) and multiple system atrophy, jointly denominated synucleinopathies [5–11].

In PD and DLB, which are common neurodegenerative disorders of the aging population, there is progressive accumulation of α-syn in cortical and subcortical regions of the brain, leading to dementia and movement disorders [12–17]. Moreover, familial forms of parkinsonism have been associated with multiplication of the α-syn gene as well as with mutations in the N terminus (E46K, A53T) of α-syn that enhance oligomerization [18–21]. The precise mechanisms through which accumulation of α-syn leads to neurodegeneration in PD are not completely understood; however, some studies support the notion that α-syn oligomers, rather than large polymers, might be responsible [22]. Supporting this possibility, we showed that the most extensive in vivo neurotoxicity was observed in animals with the α-syn variants that form oligomers (i.e. E57K and E35K), whereas the α-syn variants that form fibrils very quickly are less toxic [23]. Alternative studies suggest that protofibrils and fibrils formed during the process of α-syn oligomerization contribute to cytotoxicity. For example, oligomers with a wreath-like shape appear to be consistent with an on-pathway role for the oligomer in the fibrillation process, suggesting that oligomer-driven α-syn fibrillation has a cytotoxic potential via membrane permeabilization [24]. Previous molecular dynamics (MD) studies suggest that α-syn might undergo conformational changes resulting in the formation of propagating dimers that show an increased tendency to aggregate and form ring-like oligomers [25]. In addition, other MD simulation and biophysical studies support the view that binding of α-syn to the membrane is important in this process [26–29]. A similar mechanism has been considered for amyloid β protein [30,31]. Moreover, a number of studies have proposed that alternative splicing, post-translation modifications (e.g. nitration, oxidation, C-terminal cleavage and phosphorylation) and interactions with phospholipids and polyunsaturated fatty acids in the membrane might play a role in the process of α-syn oligomerization [32–36].

Thus, interactions between α-syn and lipids in the neuronal cell membrane has been proposed to be an important step in the process of oligomerization and cytotoxicity [26–28,32]. The oligomers might interfere with the normal function of cellular membranes and result in the formation of pore-like structures [37,38]. It has been proposed that these oligomers might display channel-like activity, resulting in abnormal calcium influx leading to neurodegeneration [25,39–41]. However, the molecular mechanisms and the potential sequence of events involved in the formation of these pore-like structures are unclear. To examine this process, we utilized computer modeling and simulations, as well as electron microscopy and cell-based studies, to investigate the course of oligomerization of wild-type and A53T mutant α-syn in membranes. Our studies support the possibility that penetration of the membrane by α-syn plays an important role.

Results

Simulations of membrane penetration by α-syn

To understand the molecular mechanisms of α-syn oligomerization and pore-like complex formation, MD simulations of α-syn in the membrane were performed. To facilitate the analysis of the simulations in the membrane, the initial studies were performed with a single wild-type and A53T mutant α-syn conformer as an example of how groups of α-syn molecules might behave in the membrane. We initiated the simulations by positioning the original, experimentally defined, α-syn conformer (resolved by NMR; pdb ID 1XQ8) [42] on the surface of a 1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC) membrane (Fig. 1A). This conformation was also used as the basis for the structure of the point mutation model of A53T α-syn. The membranes utilized for the simulations are organized into two interfacial regions containing phospholipid polar head groups and an intrinsic hydrocarbon core. To find the proper orientation of the α-syn conformer on the membrane, the MAPAS membrane-contact prediction program was used [43] (Fig. 1A). Residues of the initial contact position are presented in Fig. 1A. Consistent with previous studies [44], analysis of α-syn transmembrane scores using the dense alignment surface method [45] predicted that the region including residues 64–79 resembles a transmembrane helix (Fig. 1B). This region contains a significant number of hydrophobic residues that play a critical role during the process of membrane penetration. Following the initial contact of the N terminus with the surface of the membrane, the α-syn conformer penetrates the membrane, undergoing changes in secondary structure with portions of the α-helices converting into π-helices that will eventually extend into the β-sheet regions (Fig. 1C,D).

Figure 1.

 Simulated penetration of α-syn into the membrane. (A) Initial position of α-syn before membrane penetration. The highlighted residues of α-syn interact with the membrane upon initial contact. This position was obtained using our program MAPAS. (B) Prediction of the possible transmembrane character of the α-syn primary sequence. Score of the possible transmembrane character of the initial α-syn helix. The existing transmembrane helices typically have a score of > 500. A score of ∼ 400 suggests that this helix has a tendency to be transmembrane and may insert into the membrane. (C) Evolution of α-syn α-helices during membrane penetration. (D) Evolution rate of formation of α-syn π-helices during membrane penetration. wt α-syn, wild type α-syn.

After 2 ns of simulation of the α-syn–membrane complex (both wild-type and A53T α-syn), a new coiled and turn region appeared, interrupting the α-helix around amino acids 70–80 (Fig. 1B). During the first 3 ns of MD simulation, the region to the right of residue 38 (data not shown) transformed initially to a π-helical formation (Figs 1 and 2A,B), then to a mixture of coils, 3–10 helices, and bended regions, continuously evolving towards a β-strand structure (Fig. 2A,B,D,E). Note that such conformations with different contents of α-helical secondary structure have been recently shown in experiments [46]. Compared with wild-type α-syn (Fig. 2A–C), in the A53T mutant α-syn–membrane complex (Fig. 2D–F), the α-helices converted into π-helices more rapidly, and a greater proportion of residues were involved in formation of the π-helix (Fig. 1C,D). During penetration of the membrane, the length of the α-helical region closest to the C terminus decreased over time, evolving towards a β-strand structure (Fig. 2A,B,D,E). This transformation produced an elongated structure that participated directly in the penetration of the molecule through the membrane (Fig. 2).

Figure 2.

 Time course and structural analysis of penetration of α-syn into the membrane. Snapshots of wild-type α-syn at different time-points during membrane penetration with (A) no membrane in the image and (B) membrane present. (C) Close-up of the membrane-penetrating region of wild-type α-syn. (D) Snapshots of A53T mutant α-syn at different time-points during membrane penetration with (A) no membrane in the image and (B) membrane present. (F) Close-up of the membrane-penetrating region of the A53T mutant α-syn. Images (A), (B), (D), (E): 0 ns, cyan; 2 ns, violet; 4 ns, pink; 6 ns, yellow; 8 ns, red; and 9 ns, magenta. For A53T the deepest penetration is achieved at 7 ns and therefore the snapshot at 9 ns is not shown. Arrows point to the A53 residues (CPK, red) and T53 (CPK, yellow). wt α-syn, wild type α-syn.

During the process of membrane penetration, the hydrophilic groups of α-syn (wild-type protein shown; Fig. 3) contact the polar lipid groups on the surface of the membrane bilayer. At the initial stage, Gln62 and Asn65 contact the oxygen atoms (hydrogen-bond acceptors) situated on the surface of the lipid structure (Fig. 3A). In subsequent stages of the process of α-syn penetration (3 and 5 ns), hydrophobic residues, such as Val63 and Val66, begin to play a role, leading to interactions with the hydrophobic parts of the membrane lipids (Fig. 3B,C). During the final stage (7 ns), most of the α-syn N terminus has been transformed to an extended β-strand-like structure (Fig. 3D) that facilitates penetration to the full depth of the membrane. Moreover, hydrophobic interactions of the valine residues with the membrane phospholipids continue to play a role at this stage. In addition, the leading hydrophilic groups (particularly Asn65) of the moving α-syn molecule interact with polar groups on the opposing side of the membrane and with water molecules on the membrane (Fig. 3D).

Figure 3.

 Characterization of the amino acid residues interacting with lipids in the membrane in the membrane-penetrating region of wild-type α-syn. (A) Initial position, (B) 3 ns, (C) 5 ns and (D) 7 ns snapshots of membrane penetration by wild-type α-syn. The stick-like structures represent the residues that interact with the phospholipids and the insets represent an overview of the position of the interaction of α-syn with the membrane.

Further analysis of the course of α-syn penetration was performed utilizing a lateral projection in the membrane for wild-type α-syn (Fig. 4A,C,E) and for the A53T mutant α-syn conformers (Fig. 4B,D,F). These studies showed that the leading penetrating residues in both wild-type and A53T mutant α-syn are residues 59–72, with the first penetrating residue being Asn65 (Fig. 4A,B), as shown in detail in Fig. 3. Both the wild-type and A53T mutant α-syn molecules displayed greater penetration of the N-terminal domain after ∼ 4 ns of MD simulation (Fig. 4C,D), followed by extension of the leading-edge region of the molecule at 7–8 ns (Fig. 4E,F). When compared with the wild-type α-syn, penetration of the A53T mutant α-syn molecule across the membrane was ∼ 20% faster (Fig. 4G). Consistent with the rate of penetration, the levels of van der Waals energy diminished more rapidly for the A53T mutant than for the wild-type α-syn during MD simulations (Fig. 4H).

Figure 4.

 Analysis of membrane penetration of α-syn utilizing lateral projection for the wild-type α-syn and for the A53T mutant α-syn. (A, B) Initial position, (C, D) 4-ns position and (E, F) 7-ns position of wild-type (yellow) and A53T mutant α-syn (red) monomers, respectively. (G) Distance between the front penetrating residue of α-syn to the median of the membrane (red, wild type; blue, A53T mutant). (H) Van der Waals energies of α-syn during membrane penetration (red, wild type; blue, A53T mutant). Arrows point to the A53 residues (CPK, red) and T53 (CPK, yellow). wt α-syn, wild type α-syn.

To further confirm the specificity of the results for the α-syn MD simulation studies in membranes, a similar analysis was performed with β-syn, a member of the synuclein family that lacks the hydrophobic NAC [Non-amyloid β-protein (Aβ) component] domain [47]. Compared with wild-type α-syn (Fig. 4), β-syn did not penetrate the membrane, even after 7 ns of simulation (Fig. 5A). Next, studies were performed with conformers of the α-syn-bearing mutations on the amino-acid residues (E57A, E61A, K58A, Q62A, NS65A, V63A, V66A) predicted by the MD simulations to be critical for the process of membrane penetration (Fig. 3). Similarly, these studies showed that, compared with wild-type α-syn (Fig. 4), the mutated α-syn was unable to penetrate the membrane even after 7 ns of simulation (Fig. 5B).

Figure 5.

 Analysis of β-synuclein and mutated α-syn embedding into the membrane utilizing lateral projection. (A) Initial position of β-synuclein on the membrane and progression after 4 and 7 ns. (B) Docking and MD simulation with mutated (E57A, E61A, K58A, Q62A, N65A, V63A and V66A) α-syn. These amino acid substitutions are predicted to reduce interactions with lipids in the membrane.

Dynamics of the process of α-syn oligomer embedding into the membrane

We have shown that α-syn molecules might penetrate into the membrane after ∼ 7 ns of unrestrained MD simulation; then we examined whether the conformational changes of the α-syn molecule that take place during the process of membrane penetration might allow the generation of stable annular multimers at each stage of penetration. In order for consistent docking to proceed and annular structures to develop, the α-syn monomers present at each stage of membrane penetration should have attractive energies of intermolecular contact. To obtain the energy-per-molecule value, the total energy was calculated for each complete oligomeric structure, and this value was divided by the number of individual α-syn molecules in the oligomer. This allowed the comparison of oligomers containing different total numbers of α-syn molecules. We found that annular oligomers with an 8-mer array had a highly favorable energy of interaction between neighboring molecules, of around −60 kcal·mol−1. The 9-mer and 11-mer arrays had energies of interaction between ∼ −30 and −50 kcal/mole, respectively. All other possible oligomers had less favorable energy values. In the next step, we evaluated the energies of membrane–oligomer interactions. The structure of the octamer α-syn ring presented in Figs 6 and 7 has significant negative electrostatic energy. Note that this energy is decreased during the possible growth of the multimer from a single molecule to eight molecules within the multimer (Fig. 7D). This supports the possibility of the generation of such annular oligomers at all stages of transmembrane penetration.

Figure 6.

 Model of the penetration of α-syn octomer into the membrane. (A, B, C) Initial position, (D, E, F) snapshot at 4 ns of MS simulation and (G, H, I) snapshot at 9 ns of MS simulation of the α-syn octomer on the membrane in semitransparent membrane view, side view and view from beneath, respectively.

Figure 7.

 Structure of possible octamer constructed from the 9-ns conformers of α-syn. α-Syn octomer viewed from (A) above and (B) the side. The region highlighted in the yellow box is depicted in (C) and shows, in greater detail, the contact residues of neighboring α-syn molecules in the octamer. (D) Electrostatic energy changes during possible growth from the single wild-type α-syn molecule to octamer for the conformers taken at different times of MD simulation of the single α-syn in the membrane. (E) Overlay of the model for the α-syn octomer with that obtained by incubating α-syn with synthetic lipid membrane and analysis by electron microscopy; the merged image is shown in the panel at the far right. wt α-syn, wild type α-syn. El. Energy, Electrostatic energy.

Using the conformers of wild-type α-syn taken at different time-points of MD simulation, the program constructed a set of octamers embedded in the membrane at each MD time step. Figure 6 shows the possible membrane penetration of the α-syn octamer at the initial position (Fig. 6A–C), then at 4 ns (Fig. 6D–F) and 9 ns (Fig. 6G–I) of MD simulation. This reveals that at each stage of simulation, an oligomer can be generated which can penetrate into the membrane. The 9-ns position is shown in Fig. 7A,B as a Corey-Pauling-Koltun (CPK) space-filling model, with intermolecular and multimer membrane interactions characterized in the image inset (Fig. 7C).

Next, we examined the organization of the wild-type α-syn octameric structures and intermolecular interactions within the multimeric complex. This assembly possesses an intrinsic diameter of the opening in the middle of the oligomer of around 35 Å. The external diameter of the 9-ns oligomer is ∼ 130 Å. In this conformation, the C-terminal tail of each α-syn molecule in the octamer, comprising residues 94–140, is extended beyond the oligomer structure and equally distributed in a concentric manner surrounding the interacting N termini of the molecules. The intermolecular interactions of a single α-syn from the octamer with the membrane are described above. We analyzed the interactions between the neighboring molecules of α-syn in the octamer that are involved in organizing the oligomeric structure. We observed that several critical interactions support the close contact of single molecules of α-syn and their binding within the oligomer. Table 1 shows the possible interactions that support the organization of the octamer after 9 ns of MD simulation. Hydrophobic interactions were counted when the distance between the side-chain heavy atoms of the neighboring residues was < 4 Å. Hydrogen bonds were counted when the distance between the corresponding side-chain donor and acceptor atoms of the neighboring residues was < 2.9 Å. π–cation interactions were counted between that center of the plane of the benzene ring of phenylalanine and the positive charge of the lysine residues < 4 Å.

Table 1.   α-Syn amino acid residues involved in intermolecular interactions during oligomerization.
Type of interactionResidue of the ‘left’α-synResidue of the ‘right’α-syn
HydrophobicVal66Val71
Hydrogen bondGln62Thr75
Hydrogen bondBackbone nitrogen of Val55Thr81
HydrophobicVal55Val82
HydrophobicVal55Ala85
Hydrogen bondGln24Thr92
HydrophobicLys23 (hydrophobic region of side chain)Val95
Pi-cationLys23Phe94

These residues participate in the intermolecular interactions depicted in the ‘node of interactions’ that appears at the upper part of the oligomer in the octamer structure (Fig. 7C). The overall organization of the oligomer shows the C-terminal tails of α-syn to be equally distributed in a radial configuration around the center of the structure; this arrangement prevents significant repulsive electrostatic interactions between these negatively charged tails. Next, we evaluated, by electron microscopy, the predictive value of MD simulations by comparing the dimensions of α-syn generated during the MD simulations with membrane-bound recombinant α-syn aggregates analyzed. For this purpose we selected annular oligomers that displayed the best energy of interaction and had an octamer array and overlaid them into the electron micrographs of the α-syn aggregates (Fig. 7E). Annular oligomers obtained by MD simulations or by incubation in synthetic membranes had similar characteristics ranging in diameter from 120 to 150 Å (Fig. 7E).

Effects of α-syn membrane distribution in oligomerization in a neuronal cell model

To investigate the consequences of α-syn oligomerization in a cell-based system, neuronal cultures were infected with lentiviral vectors expressing wild-type or mutant A53T α-syn. By immunoblot analysis, the monomeric forms of wild type and mutant α-syn were found in equal amounts in cytosolic fractions (Fig. 8A,B). In contrast, in the membrane fraction, mutant A53T α-syn monomer and oligomers were more abundant compared with cells expressing the wild-type construct (Fig. 8A–B). Similarly, in native gels, more abundant α-syn aggregates were detected in the membrane fraction of the neuronal cells infected with the LV-mutant A53T α-syn (Fig. 8C). Immunocytochemical analysis and examination by laser scanning confocal microscopy showed that compared with control cells, the lentivirus-expressed wild-type α-syn was distributed as punctae throughout the cytoplasm of the neurons and along the neurites (Fig. 8D–E). In comparison, the A53T mutant α-syn was more diffusely distributed in the cell bodies and tended to concentrate around the periphery of the neurons (Fig. 8F).

Figure 8.

 Characterization of the effects of α-syn oligomers in a neuronal cell line. (A) Immunoblot analysis of α-syn species in the soluble (cytoplasmic) and membrane-bound (membrane) fractions from B103 cells infected with LV-wild type α-syn or LV-A53T α-syn, uninfected cells were used as a control. (B) Analysis of the levels of α-syn monomers and oligomers in the soluble and membrane-bound fractions from B103 cells infected with LV-wild type α-syn or LV-A53T α-syn. (C) Native gel analysis of membrane fractions from control B103 cells and neurons infected with LV-wild type αα-syn or LV-A53T α-syn. (D) Double labeling co-localization studies of α-syn and the neuronal marker MAP2 in control cells, cells infected with LV-wild type α-syn and cells infected with LV-A53T α-syn. Arrows indicates α-syn accumulating in the cell processes and membrane. (E) Calcein and calcium assays respectively in control neuronal cells or cells infected with LV-wild type α-syn or LV-A53T α-syn. (F) Immunoelectron microscopy studies of α-syn localization to the membrane in neuronal cells infected with LV-wild type α-syn. Arrows indicate gold particles near the plasma membrane. Scale bar = 3mm. wt α-syn, wild type α-syn, RFU, relative fluorescence units.

To investigate whether the accumulation of α-syn in the neuronal membranes altered cellular function, cells expressing α-syn or empty vector were analyzed using the calcein permeability assay and by Fluo-4 calcium imaging to measure intracellular calcium levels. With the calcein assay, neuronal cells expressing both wild-type or A53T mutant α-syn displayed increased permeability; however, this effect was more apparent in the cells expressing A53T α-syn (Fig. 8G). Consistent with this increased permeability, neuronal cell cultures expressing either wild-type α-syn or A53T α-syn displayed increased levels of intracellular calcium (Fig. 8H). To further investigate if the α-syn immunoreactive signal distributed around the neurons was localized to the membrane, ultrastructural immunogold analysis was performed. In neuronal cells infected with the LV-control, no gold particles were detected in the membrane. In contrast, neuronal cells infected with LV-wild type (data not shown) and mutant A53T α-syn displayed the presence of gold particles in close proximity to the plasma membrane as well as in the cytosol, in association with other organelles (Fig. 8J).

Discussion

Previous MD simulation and biophysical studies have shown that binding of α-syn to membranes is important in the process of oligomerization [26–29]. In the present study we utilized molecular modeling and MD simulations of the protein during penetration of the membrane, in combination with biochemical and ultrastructural analysis, to show that α-syn might be able to penetrate the cell membrane. The theoretical studies predict that α-syn molecules initially dock to the membrane surface and begin to penetrate the membrane, resulting in changes to their secondary and tertiary structures. During the initial stages of the molecule’s structural changes, compared with the wild-type α-syn, the mutant A53T molecule possessed a greater percentage of amino-acid residues that were involved in α-helical conformations. After transformation of regions within the α-syn molecule from α-helices to π-helices, the N-terminal region of each α-syn molecule began to shift towards a coiled structure and into a β-strand conformation. Later in the simulation times, a greater percentage of amino-acid residues in the mutant α-syn than in the wild-type α-syn were involved in π-helical structures. This was followed by eventual perforation of the membrane by the leading hydrophobic edge formed by two helices and a linker region of each α-syn molecule.

Interestingly, and in support of our observations that the polar lipid heads play an important role in α-syn–membrane interactions, previous studies have shown that α-syn binds preferentially to membranes composed of lipids with small polar head groups (e.g. phosphatidic acids), where inherent membrane instabilities may be present and may contribute to the binding of α-syn with the membrane [48]. Similarly, a recent study demonstrated that binding of α-syn to negatively charged lipid membranes is much stronger than the interactions of α-syn with neutral lipid bilayers, and this interaction is less sensitive to other membrane alterations [49]. Intrinsic surface disruptions present in negatively charged membranes could play a role in promoting penetration of the membrane by α-syn, as bilayer charge and accessibility of the hydrocarbon core modulates oligomer–membrane interactions [48]. Consistent with these observations, the present study revealed that interactions between hydrophobic valine residues of α-syn and the membrane hydrocarbon core play a pivotal role in the process of membrane penetration by α-syn molecules. The MD simulations propose that penetration of the membrane by α-syn facilitates oligomerization and embedding that could eventually result in the formation of pore-like structures. In support of this possibility, we found similar ring-like structures when comparing the α-syn model with oligomers obtained by incubating α-syn with a synthetic lipid membrane. The characteristics and dimensions (120 and 150 Å) of our α-syn model and oligomers are very similar to those reported by Lashuel et al. [50] and Quist et al. [39].

In the present study we performed the MD simulations with full-length α-syn in a neutral POPC membrane and showed a fast speed of penetration. Previous MD studies have used segments of α-syn, including residues 1–41 [51], residues 35–43 [26] and residues 1–99 [27]. In the latter study it was shown that the 1-99 fragment penetrates into the membrane on the length around 5 Å. For this study, a negatively charged membrane comprising DOPS was utilized. The N-terminal region of α-syn has an overall positive charge (and the C-terminal region is negatively charged). This suggests that the truncated N-terminal region, as a result of its positive charge, was probably attracted by the negative phospholipid heads and penetrated slightly beyond their level. The position of the truncated α-syn immediately below the charged heads of the lipids can be explained by the sharp decrease of dielectric constant when moving from water to the area inside the lipid layer. The constant changes from around 80 to around 5 would cause the sharp (around 16-fold) increase of electrostatic energy of lipid–α-syn interactions.

While the present study proposes that the process of membrane penetration was energetically favorable for both wild-type and mutant A53T forms of α-syn, we observed that immersion of α-syn in the membrane occurs more rapidly for the mutant A53T monomer than for the wild-type molecule and with a more favorable van der Waals energy of interaction. This observation provides a potential molecular mechanism that may explain the previous findings that neural cells expressing mutant α-syn (A30P and A53T) have higher plasma membrane ion permeability than control cells [40]. In the present study we demonstrated, using MD simulations, that the mutant A53T form of α-syn possesses a greater proportion of residues involved in a π-helix formation, and, over time, the length of the α-helical region closest to the C terminus decreased, evolving towards an elongated, β-strand structure that facilitates the penetration of the molecule into the membrane. Enhanced membrane penetration and distinct changes in the secondary and tertiary structures of mutant A53T α-syn may also contribute to altered aggregation kinetics compared with the wild-type molecule. In support of this possibility, previous studies have revealed that mutant A53T α-syn displays accelerated aggregation compared with wild-type α-syn [52–54], which may be related to an increase in the β-strand character of the molecule [43].

In summary, these results suggest a model where penetration of membranes by α-syn gives rise to the formation of annular pore-like oligomer structures with the ability to increase cell permeability and calcium influx. As an example, we focused on α-syn octamers because they display favorable negative energies of interaction (around −60 kcal·mol−1); however, other multimeric arrays are also possible, in particular 9-mers and 11-mers. The MD simulations suggest three stages in this process (Fig. 9). In the first, hydrophilic residues around amino acids 59–72 of α-syn interact with the polar heads on the surface of the membrane (Fig. 9A). Next, as α-syn could penetrate into the membrane, hydrophobic valine residues interact with the nonpolar tails of the lipids (Fig. 9B). Finally, the hydrophilic amino-acid residues – particularly Asn65 at the leading edge of the molecule – could participate in hydrogen-bonding interactions with the polar heads of the lipids on the opposing side of the membrane bilayer and allow penetration of the molecule through the membrane (Fig. 9C). Understanding the precise mechanisms and the specific amino-acid residues involved in the interactions of α-syn with the membrane and with other α-syn monomers will aid in the development of molecular structure-based therapeutic approaches for disorders of α-syn aggregation.

Figure 9.

 Schematic hypothesis of α-syn annular oligomer penetration into the membrane. (A) Hydrophilic residues of α-syn interact with polar heads on the surface of the membrane. (B) As α-syn begins to penetrate into the membrane, the hydrophobic residues interact with the nonpolar tails of the lipids. (C) Finally, the hydrophilic residues participate in hydrogen-bonding interactions with polar heads of lipids on the opposing side of the membrane bilayer, allowing the molecule through the membrane.

Materials and methods

Molecular dynamics simulations and modeling of the α-syn–membrane system

Modeling and simulations were based on the previously reported NMR structure of α-syn (PDB index 1xq8 [42]). MD simulations of the α-syn molecule-membrane system were conducted as previously described [25,55] using periodic boundary conditions at constant pressure (1 atm) and temperature (310 K) initially with the α-syn molecule in the water box, in which the shortest distance between the protein molecule and the box walls was 30 Å. For the α-syn–membrane model, the NMR structure of α-syn was located on the surface of the membrane according to the predicted membrane-contacting surface (the prediction was performed using the mapas program) [43]. This program was applied to the initial α-syn conformer, and the top-scoring predicted membrane-attractive surface was used to dock the protein molecule to the membrane. For modeling α-syn–membrane interactions, a rectangular POPC polar lipid bilayer membrane with fixed edges and dimensions of 100 × 180 Å was used.

The namd md program (version 2.5 [56]) was used with the CHARMM27 force-field parameters [57] to simulate the behavior of α-syn molecules in water under normal conditions and the interaction of the POPC membrane with the α-syn aggregates. The temperature was maintained at 310 K by means of Langevin dynamics using a collision frequency of 1/ps. A fully flexible cell at constant pressure (1 atm) was employed using the Nosé-Hoover Langevin Piston algorithm [58,59], as used in the namd software package. Initial coordinates were taken from a previously equilibrated 500-ps system. The van der Waals interactions were switched smoothly to zero over the region 10 Å, and electrostatic interactions were included via the smooth particle-mesh Ewald summation [60–62].

The simulation was performed in four steps. The systems were minimized for 10 000 iterations, heated in 0.1° increments and equilibrated for 10 ps, and then the MD simulation was conducted. Data for analysis were taken between 50 ps and 10 ns of the simulation.

Cell culture of neuronal cells expressing wild-type or mutant A53T α-syn

B103 rat neuroblastoma cells were grown in 10-cm dishes at 50% confluence in Dulbecco’s modified Eagle’s medium (DMEM) (Mediatech, Manassas, VA, USA) containing 10% fetal bovine serum at 37 °C, in a 5% CO2 atmosphere. Cells were infected with lentiviruses expressing α-syn (wild-type or mutant A53T) or GFP (each at 1.0 × 107 TU), prepared as previously described by transient transfection in 293T cells [63–65]. After incubation for 48 h, the efficiency of transduction of lenti–GFP was more than 90%. Cells were then plated onto 12-mm coverslips, or into 6-well or 96-well plates, and then processed for subsequent biochemical and functional assays.

Subcellular fractionation and immunoblot analysis of α-syn distribution in lentivirus-infected neuronal cells

To analyze the distribution and levels of α-syn in B103 cells expressing α-syn, briefly, as previously described [65], lentivirus-infected cells grown on six-well plates were homogenized and fractioned into cytosolic and membrane fractions by ultracentrifugation at 627 000 g for 1 h in a TL-100 rotor (Beckman-Coulter). Approximately 20 μg of protein was loaded into each lane of 4–12% Bis-Tris gels with Mops/SDS buffer and blotted onto poly(vinylidene difluoride) (PVDF) membranes. Blots were incubated with antibodies against α-syn (1 : 1000 dilution; Chemicon, Temecula, CA, USA or BD Transduction Laboratories, San Jose, CA, USA), followed by incubation with secondary antibodies tagged with horseradish peroxidase (1 : 5000 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Immunoreactive signals were visualized by enhanced chemiluminescence and analyzed using a Versadoc XL imaging machine using the QuantityOne program (BioRad, Hercules, CA, USA). Analysis of actin levels was used as a loading control.

Double-labeling immunocytochemical analysis of patterns of α-syn immunoreactivity in B103 cells infected with α-syn lentiviruses

For double-labeling immunocytochemical analyses, cells were infected with lentiviruses, as described above, and cultured on coverslips. The coverslips were then fixed in 4% paraformaldehyde for 20 min and blocked overnight at 4 °C in 10% fetal bovine serum and 5% BSA. The cells on coverslips were then incubated overnight at 4 °C with antibody against α-syn diluted in PBS-T (Triton X) (1 : 2500 dilution; Chemicon or BD Transduction Laboratories). The following day, the antibody was detected using the Tyramide Signal Amplification-Direct (Red) system (NEN Life Sciences, Boston, MA, USA). The cells on coverslips were then double-labeled with an antibody against microtubule-associated protein-2 (MAP2, a marker of dendritic integrity; 1 : 250 dilution) and detected using a fluorescein isothiocyanate (FITC)-conjugated secondary antibody (1 : 75 dilution; Vector Laboratories, Burlingame, CA, USA). The coverslips were air dried overnight, then mounted under glass coverslips with ProLong Gold antifade reagent with DAPI (diamidino-2-phenylindole) (Invitrogen, Carlsbad, CA, USA) and imaged using a Zeiss 63X (N.A. 1.4) objective on an Axiovert 35 microscope (Zeiss, Jena, Germany) with an attached MRC1024 laser scanning confocal microscope system (BioRad) [66]. All sections were processed simultaneously under the same conditions and the experiments were performed twice to assess reproducibility.

Electron microscopy and immunogold labeling

Briefly, as previously described [55], recombinant α-syn was incubated in synthetic lipidic membranes and then aggregates were analyzed by electron microscopy 24 h later. The synthetic lipid monolayer was generated as described by Ford et al. [67]. For studies of α-syn distribution, cells were fixed in 0.1 m cacodylate buffer (pH 7.4) containing 0.25% glutaraldehyde and 3% paraformaldehyde and then pre-embedded with epoxy resin, sectioned with an ultra microtome (Reichert Ultracut E, Leica Microsystems, Buffalo Grove, IL, USA) at 60-nm thickness and collected in nickel grids for immunogold labeling. The grids were treated with sodium periodate saturated in water and then incubated overnight with anti-synuclein Ig (1 : 100 dilution). The next day, the grids were incubated for 2 h at room temperature with the secondary antibody anti-rabbit IgG/10-nm gold particles (AURION Immunogold reagents). The immunostained grids were poststained using saturated uranyl acetate and analyzed using a Zeiss EM10 electron microscope, and electron micrographs were obtained at a magnification of 50 000.

Dye efflux (calcein) assay

Assessment of membrane permeabilization was carried out as previously described [48] using calcein dye as an indicator (Invitrogen). Briefly, lentivirus-infected cells were plated at a density of 25 000 cells per well on Costar 96-well black plates with a flat clear bottom (Corning, Corning, NY, USA). After an additional 24 h of incubation, the complete medium was replaced with 200 μL of phenol red-free medium containing 1 μm calcein dye (Invitrogen). Cells were kept in the incubator at 37 °C for 1 h before measuring fluorescence on a DTX 880 Multimode Detector (Beckman Coulter, Fullerton, CA, USA) with an excitation/emission filter at 470–495/515–575 nm. As a positive control of dye efflux, detergent was added to control wells to lyse the cells.

Calcium mobilization FLUO-4 assay

Assessment of calcium influx was carried out, as previously described [55], using a modified protocol of the FLIPR 4 calcium assay (Molecular Devices, Sunnyvale, CA, USA). Briefly, B103 cells were infected with lentiviral constructs. Cells were infected at a multiplicity of infection of 30, and 2 days after infection, the cells were plated at a density of 30 000 cells per well on Costar 96-well black plates with a flat clear bottom (Corning). After an additional 24 h of incubation, the medium was replaced with 100 μL of HBSS (Hanks buffered salt solution) buffer, and 100 μL of calcium dye (FLUO-4 NW; Invitrogen) was added to each well. Cells were kept in the incubator at 37 °C for 30 min, followed by incubation in the dark at room temperature for an additional 30 min before measuring fluorescence on a DTX 880 Multimode Detector (Beckman Coulter) with an excitation/emission filter at 470–495/515–575 nm. As a positive control of calcium influx, 0.6 μg of ionomycin (Sigma, St Louis, MO, USA) was added to control wells.

Statistical analysis

The data are expressed as mean values ± SEM. Statistical analysis was performed using one-way analysis of variance (ANOVA) followed by post hoc Dunnett’s or Tukey–Kramer tests (Prism Graph Pad Software, San Diego, CA, USA). Differences were considered significant at P < 0.05.

Acknowledgements

This work was supported by NIH grants AG18440 AG11385, AG022074, NSO44233, and HL066012 and by a DOE INCITE grant. The authors are also grateful to DOE for computational support on its BlueGene computers at the Argonne National Laboratory.

Ancillary