Abstract
The amyloid aggregation of α-synuclein (αS), related to Parkinson’s disease, can be catalyzed by lipid membranes. Despite the importance of lipid surfaces, the 3D-structure and orientation of lipid-bound αS is still not known in detail. Here, we report interface-specific vibrational sum-frequency generation (VSFG) experiments that reveal how monomeric αS binds to an anionic lipid interface over a large range of αS-lipid ratios. To interpret the experimental data, we present a frame-selection method ("ViscaSelect”) in which out-of-equilibrium molecular dynamics simulations are used to generate structural hypotheses that are compared to experimental amide-I spectra via excitonic spectral calculations. At low and physiological αS concentrations, we derive flat-lying helical structures as previously reported. However, at elevated and potentially disease-related concentrations, a transition to interface-protruding αS structures occurs. Such an upright conformation promotes lateral interactions between αS monomers and may explain how lipid membranes catalyze the formation of αS amyloids at elevated protein concentrations.
Similar content being viewed by others
Introduction
Amyloid aggregation of misfolded proteins is associated with a large variety of diseases1,2,3. Aggregation of α-synuclein (αS) is related to serious and incurable α-synucleinopathies such as Parkinson’s disease4,5,6,7. αS is a 140-residue protein with an N-terminal region (residues ∼1–60) that contains several positively charged residues and a highly negatively charged C-terminal region (residues ∼96–140). The non-amyloid β-component (NAC) region, residue ∼61–95, forms the hydrophobic core in αS fibrils and is thus implicated in αS aggregation8. Lipid interfaces are thought to play an important role in the aggregation of αS9,10. The relevance of the aggregation catalysis by lipids is substantiated by the presence of lipids in Lewy bodies, the neuronal inclusions that are a pathological hallmark of Parkinson’s disease11,12,13 and by the observation of lipid–protein co-aggregates in vitro12,14,15. Obtaining a molecular understanding of the lipid-αS interaction could be of vital importance for resolving the contribution of αS aggregation to Parkinson’s disease, as anionic lipid membranes are known to catalyze αS amyloid formation14,16. The interaction between αS and lipid membranes has therefore been extensively studied and reviewed9,10,17. Structural studies of αS at membrane interfaces have mainly been based on nuclear magnetic resonance (NMR)18,19,20,21,22,23, electron paramagnetic resonance (EPR)18,21 and neutron reflectivity (NR)24 experiments, with αS bound to lipid vesicles19,20,22, surfactant micelles18,19,21,23, and lipid bilayers19,24.
The density of αS at the membrane surface, depending on the αS concentration in solution relative to the number of lipids, is known to be an important factor for the aggregation16,25,26,27. The relative concentration of αS is often expressed as the lipid–protein ratio (LPR), which denotes the molar ratio between lipid and protein, i.e., a low LPR denotes a high relative αS concentration. In studies with lipid monolayers, the LPR can be calculated using the known concentration of added protein, the volume of the solution, and the area per lipid at the interface. It has been shown that below an LPR of 10, αS clusters form that induce membrane damage due to amyloid formation28. NMR, EPR, and NR studies provide a molecular-level image of αS interaction but have mainly been limited to relatively high LPRs (i.e., relatively low αS concentrations)10,17,23. In the literature, there is only little structural information available at the molecular level for lower LPRs, i.e., higher protein concentrations, where αS aggregates into amyloids catalyzed by the lipid interface (starting at ∼10 μM for supported lipid bilayers28, while the physiological concentration of αS is ∼20 μM29). This lack of knowledge may partly be explained by the difficulty of performing NMR measurements to study protein structure at interfaces in the presence of a large number of proteins in the bulk phase. Since αS has a relatively weak affinity to most lipids16,20,30, forming densely packed protein layers at the lipid interface requires high solution concentrations. Such αS concentrations are of particular interest to study, given the strong correlation that has been observed between elevated αS concentrations and aggregation-related toxicity31.
One particularly suited technique to study surface-bound proteins in the presence of high bulk concentrations is vibrational sum-frequency generation (VSFG)32,33. VSFG signals are, per selection rules, only generated at interfaces. Only surface-interacting proteins contribute to the observed signal, while background signals from proteins in bulk do not contribute. To provide structural insights into αS at low concentrations as applied before in the literature, but also at physiological and elevated concentrations, we have performed VSFG amide-I spectroscopy of αS at an anionic lipid monolayer interface. In combination with molecular dynamics (MD) simulations, we obtain a molecular picture of lipid-associated αS structure when bound to an anionic 1,2-dipalmitoyl-\({sn}\)-glycero-3-phosphoglycerol (DPPG) monolayer interface at the physiological ionic strength and pH. While a cellular membrane has a variety of lipids carrying different charges, we choose a uniform DPPG monolayer as a model system so that we can focus on the effect of varying αS concentration near negatively charged membranes.
In recent years, VSFG has matured into a technique that is frequently employed to study proteins at interfaces32,33. Previously, the combination of VSFG with MD simulations has enabled researchers to interpret protein VSFG data at a molecular level34,35,36,37,38. In the present study, we have extended this approach. MD simulations are used to generate a large number of hypothetical αS conformations in a lipid-bound state, which is subsequently evaluated with spectral calculations for each snapshot or ‘frame’ (i.e., conformation) of the MD trajectories. This frame-selection method (using the “ViscaSelect” algorithm, a part of the Visca vibrational spectroscopy calculation tools, see the Code availability section) thus results in an ensemble of conformations that agrees best with the VSFG experiments, which can be structurally inspected with further analyses.
With this methodology, we derive that αS undergoes a conformational transition from a flat to an upright conformation as a function of increasing local concentration. The VSFG spectra that we recorded at relatively low concentrations (50 nM to 20 μM, equaling LPRs of ∼37 to 0.1) are consistent with the flat-lying α-helical structures that have been observed before18,19,20,21,22,23,24. However, at an elevated concentration of ∼50 μM (LPR = ∼0.04, equaling ∼2.5 times the physiological concentration29), the protein adopts an elongated structure with intermediate helicity, which is bound with its first ∼50 residues to the lipid interface, while the rest of the protein is generally oriented upright (i.e., perpendicular) with respect to the lipid surface. Such structures have not been reported in the literature, where typically 10–10,000 times lower αS concentrations are used18,19,20,21,22,23,24. Interestingly, in a small number of previous studies using intermediate concentrations, a relatively rare subspecies has been observed20,22 that closely resembles the structure we report here for elevated concentrations. A previous EPR/fluorescence study based on three fluorescent labels has already hinted at a structural transition at lipid interfaces25. Here, we provide evidence for an upright orientation at elevated αS concentrations. This binding motif allows aggregation-prone middle regions10,28 of adjacent αS to to come in contact, which could be a key driver of amyloid aggregation of αS, possibly linked to the observed increase of aggregation-related toxicity at elevated αS concentrations31.
Results
VSFG experiment
We record orientation-sensitive VSFG spectra in the amide-I region (1600–1700 cm−1), as this mode is highly sensitive to the protein structure39. The signals are generated by overlapping an IR and a visible pulse in space and time and collecting the sum-frequency photons that are subsequently diffracted with a spectrograph on a CCD camera. By varying the polarizations of the sum-frequency, visible, and IR beams, one can record amide-I spectra in different VSFG polarization combinations, whose lineshapes and intensity ratios are strongly dependent on the structure and orientation (i.e., the conformation) of the protein33. The VSFG spectra are recorded with four different combinations of the polarizations of the two incoming (visible and IR) and one outcoming (SFG) beams. These polarization combinations can be divided into achiral and chiral combinations, where achiral polarization combinations can give SFG signals for both achiral and chiral structures, and the chiral combinations will only give a signal when the sample is chiral. We use a combination of several chiral and achiral polarization combinations to increase the information content of the experiment, which will later lead to a more unique fit with the structural model. Three of the recorded polarization combinations are achiral (SSP, i.e., S-polarized SFG, S-polarized visible, and P-polarized infrared, and PPP and SPS—see Fig. 1A) and is one chiral (PSP).
In Fig. 1A, the experimental setup is depicted. First, a DPPG lipid monolayer with a surface pressure of 15 mN/m is dropcast at the water surface (see Fig. 1B and Supplementary Fig. 2). Then αS is injected into the phosphate-buffered saline (PBS) subphase, leading to a final concentration of 50 nM, 20 μM, or 50 μM (LPR of ∼37, 0.091, or 0.037; see SI section 1.2 for the LPR calculations). At all these αS concentrations, surface binding, and assembly lead to an initial increase to a physiological surface pressure of approximately 30 mN/m40 within 5 h, after which it remains almost constant over ∼24 h (see Fig. 1B and Supplementary Fig. 2), indicating the formation of a stable layer. The strong binding is also visible in the recorded high signal-to-noise VSFG spectra of αS adsorbed to the DPPG monolayer (Fig. 1D, E and Supplementary Fig. 4), as weak and/or disordered binding will lead to very weak VSFG signals. The spectra show broad features centered around 1640 cm−1, which are typically associated with random-coil structures with a certain degree of order and/or α-helices39. The VSFG spectra are also relatively stable for the first ∼24 h, after which the protein and lipid signals start to decrease (see Supplementary Fig. 3A, B). While the intensity of the signal eventually drops over the ∼50 h incubation time, the normalized spectral lineshapes remain identical (Supplementary Fig. 3C). This signal evolution can be explained by an initial formation of an αS monolayer at the lipid interface, after which amyloidogenic protein-lipid aggregates are formed that detach from the interface, similar to what has previously been observed for αS, e.g., at an LPR of 1028, or (in a seeded fashion) at an LPR of 13041. AFM images taken of Langmuir depositions of the interface on mica after 24 h of incubation (Fig. 1C and Supplementary Fig. 5) confirm the absence of fibrillar aggregates. Because the Langmuir depositions are dried before we record the AFM images, we cannot rule out drying-induced oligomerization artifacts, but we observe 5- to 15-nm high spherical species that resemble previously observed isolated oligomers (e.g., those with a hydrodynamic radius of ∼12 nm42 and those with a TEM-derived diameter of ∼13 nm43), whose numbers appear to increase with incubation time. The absence of any changes in the normalized VSFG spectral lineshapes, while the AFM images show an oligomer-covered interface, could – if their presence is not a drying artifact – indicate that the oligomers are either centrosymmetric particles or that they have a random orientation with respect to the interface33.
Images of regions with defects recorded after 24 h of incubation (Supplementary Fig. 5) reveal the presence of areas where proteins have replaced lipids, which could (like the oligomerization) also be an artifact from the drying procedure, or a result of the high binding affinity of αS to highly curved lipid interfaces30 that are likely present at defects in the lipid membrane. Alternatively, the lipid replacement at long incubation times could indicate lipid co-aggregation, a scenario corroborated by the similar decrease in protein and lipid VSFG intensity as a function of time (see Supplementary Fig. 3 and Supplementary Fig. 5 for the associated AFM-based adhesion map that corroborates the AFM layer assignments) and by previous studies15,44. Interestingly, at αS concentrations identical to the concentration applied in the high-concentration experiment (∼50 μM) in our VSFG study, experiments with supported lipid bilayers deposited on an attenuated total reflection (ATR)–IR crystal under the protein solution did not show a decrease in the protein signal during the aggregation45. Instead, in this inverted (upside-down) geometry compared to our VSFG experiments, spectral lineshapes indicative of amyloids grow on a timescale of ∼10 h. This indicates that changes in the lipid composition (1:1 POPG:POPC vs. pure DPPG) or detachment of aggregates due to gravity might play a role in these experiments. In spite of differences in the ionic strength and exact lipid composition, the timescale of the aggregation in this ATR-IR experiment is comparable to the timescale of which the VSFG signal decreases, which seems to indicate that there is a correlation between αS aggregation and VSFG signal loss. In both the ATR-IR and our VSFG experiment, the lipid signal decreases during the incubation. Although some studies report that αS fibrils associate with anionic lipids such as DLPS and DMPS44, other studies using 1:1 POPC/POPG41 or DMPG46 lipids show very weak or negligible binding, respectively, of aggregated αS to lipid membranes compared to monomers. Therefore, it is likely that the aggregates formed by the protein–lipid co-aggregation have no preference to reside at either the remaining air-water or lipid-water interface, which is consistent with our observations.
The constant amide-I lineshapes of the normalized spectra (Supplementary Fig. 3C), and the AFM images that show the absence of fibrils (probably due to detachment) and that possibly indicate VSFG-invisibility of the oligomers (assuming the oligomers observed in the AFM images are not a drying artifact), make the applied experimental geometry well suited to study the binding mode of monomeric αS to DPPG lipids, undisturbed by aggregating species.
Structural analysis by frame selection
Generation of structural hypotheses with MD simulations
To determine the molecular conformations of αS from the VSFG experiments, we have developed a new methodology in which spectral calculations are used to select frames from MD simulation trajectories. The MD simulations are used as a means to generate a large number of hypothetical conformations for which we calculate VSFG spectra that we compare with the experimental spectra. In order to generate the hypothetical structures, we run MD simulations of one αS molecule in water interacting with a DPPG monolayer, starting from a diverse set of eight starting conformations (see Supplementary Fig. 6) so that we sample a large portion of the αS conformational landscape near a DPPG interface. The starting structures include both kinked α-helical conformations based on the NMR-derived PDB entry 2KKW (SLAS-micelle bound αS), upright standing structures, and control simulations with disordered structures and structures penetrating the monolayer. The so-called horseshoe motif (PDB: 2KKW) is experimentally observed when αS is bound to sodium lauroyl sarcosinate (SLAS) micelles21, and refers to the horseshoe-shaped conformation composed of a kinked α-helix that spans from residue 1 to 91 that is tightly surface-bound, while its randomly coiled C-terminus is unbound. The fact that a similar motif is observed when αS interacts with SDS micelles18 (PDB: 1XQ8 (Human micelle-bound αS)), which indicates that it represents a typical αS folding near lipid-like interfaces, further validates including it in the sampling.
We run 16 MD production simulations (two repeats of 150 ns for each of the 8 starting structures, corresponding to a total simulation time of 2.4 μs) where we employ the DES-amber force field, recently developed for both folded and intrinsically disordered proteins (IDPs)47, together with the Slipids force field48,49,50. During the simulations, the αS starting structures partly unfold and reorient, thus generating a large dataset of 48.016 hypothetical αS conformations near a DPPG monolayer. We subsequently use excitonic spectral calculations51,52 to compare all the generated MD conformations with the experimentally observed VSFG spectra (SSP, SPS, PPP, and PSP) recorded during the first stable ∼15 h of incubation.
In previous studies where excitonic Hamiltonian VSFG calculations have been combined with MD simulations, snapshots of the lowest-energy states36, most-densely populated states35, or a representative set of MD snapshots37 have been used as the basis for the calculations. Other strategies have used full MD trajectories34 or particular trajectories out of a set of replicates38. The unfolded nature of medium-sized IDPs like αS requires relatively large MD systems that, in turn, make it prohibitively slow to obtain an equilibrium distribution of conformational states. While recent advances47,53,54 in MD force fields for IDPs have made MD simulations of both folded and disordered states accurate, sampling full MD folding trajectories requires very long simulations or accelerated sampling and is often only computationally possible for small peptides. Therefore, here we have expanded the methodology of combining VSFG with MD by using out-of-equilibrium MD simulations to generate a large set of protein conformations, after which we use spectral calculations (based on an amide-I one-exciton Hamiltonian model33,37,38,52; see “Methods” section for details) to determine which conformations give a match with the experimentally obtained spectra55,56. This removes the requirement of the MD simulation to fully converge as the spectral calculations can be used to select the MD-derived conformations that correctly describe the experiment via spectral comparison. The VSFG-MD frame-selection method (using the ViscaSelect algorithm) is illustrated in Fig. 2, which shows how the structures obtained from the MD trajectories are validated with spectral calculations, using their spectral similarity to the experimental VSFG spectra as the selection criterion.
Evaluating and selecting MD frames by VSFG spectral match
The evaluation of the experimental fitness of the MD conformations through the RSS allows us to select an ensemble of MD frames with the best spectral match. We set the cut-off of the ensembles such that the standard deviation of the spectra within the best-matching ensemble equals the experimental standard deviation. Thus we obtain two ensembles of MD frames, one for each experimental VSFG dataset. In Fig. 3, the spectral and structural ensembles on both sides of the transition are described: on the right for the low-LPR (high αS concentration) case, and on the left for the high-LPR (low αS concentration) case; here only depicted for the 20 μM dataset — the analysis for the 50 nM dataset gives a nearly identical result. On the top row of Fig. 3, the RSS values are given for the two cases; on the middle row, the range and average spectra of the ensembles are depicted, and on the bottom row, associated structures of each of the ensembles are shown. We note that the MD simulations do not explicitly incorporate the different LPR values. The same set of αS conformations is thus used as the collection to select from for both LPRs. In each case, an ensemble of best-matching structures is selected from this collection of frames based on the spectral match between the experimental and calculated VSFG spectra.
From the top row of Fig. 3, one can see that only a small fraction of the frames match the posed RSS criterion. The low- and high LPR ensembles contain 55 and 164 structures, respectively. Interestingly, there are no shared frames between the ensembles.
In the second row of Fig. 3, the transparent bands indicate the range of the spectra within the ensembles, which all show a reasonable overlap with the experimental spectra. Interestingly, their average spectra (solid lines) reproduce the experimental spectra very closely. This indicates that — as expected — the experimental spectrum is probably the result of an ensemble with a certain degree of variation in the tertiary structure and hydrogen bonding that is captured by the selected ensembles, which, upon averaging, leads to a close match with the observed spectra. The match of the weaker polarization combinations is slightly less good than the match with the SSP spectra, which can be understood from the fact that the selection — based on the total RSS of all polarization combinations — results in a higher sensitivity to stronger polarization combinations. The lack of any PSP intensity, which would originate from chiral amide-I normal modes, is reproduced well. The transparent band, which indicates the bandwidth, shows that some structures within the ensemble have a small non-zero chiral (PSP) response nonetheless, so including the PSP intensity in the frame selection criteria appears to help with filtering out frames with potentially larger chiral responses.
In the last row of Fig. 3, some typical structures are depicted (see Supplementary Fig. 8 for the 10 best matching structures) that are found within the ensembles. Visual inspection of the top-scoring conformations reveals a clear transition from a flat-lying αS to an upright orientation when going from a low αS concentration (high LPR) to a high αS concentration (low LPR). The absence of a chiral spectral response is consistent with these structures, as they are solely composed of random coils and α-helices, which are both known to be chiral-VSFG silent secondary structures in the amide-I region57.
The interfacial structure of αS at low and high concentrations
Now that the best-matching frames have been selected from the MD trajectories, one can obtain a quantitative structural image of the high- and low-LPR ensembles and determine to which extent there are commonalities and differences between the two ensembles. In Fig. 4, both the helicity (given as the fraction of the residues along the protein sequence that are adopting a helical structure within the ensembles) and the distance to the closest DPPG molecule are given for each protein residue. The plots reveal that both the low- and high-LPR ensembles contain several broken helices in the residue 1–95 segment, while the low-LPR ensemble contains conformations that are, on average, further away from the DPPG surface compared to the high-LPR ensemble conformations — in particular in the residue 50–120 segment. The small standard error of the mean, especially on the distance to the nearest DPPG molecule, indicates that the frame-selection method yields ensembles that are characterized by a structural similarity (see Supplementary Fig. 9 for the distances of each of the structures in the ensembles). This allows one to obtain a good understanding of what structural features remain the same and which change when the protein concentration is varied.
While the proteins within the high- and low-LPR ensembles adopt a very different orientation, surprisingly, the regions that adopt a helical structure are similar in the two ensembles. This could indicate a cooperative folding behavior along the protein chain, where — even in the protruding helical structure adopted at high protein concentrations — N-terminal lipid binding and subsequent local helix formation results in secondary structure changes further down the chain. This is supported in the αS sequence (Uniprot: P37840) by several imperfect KTKEGV repeats in the N-terminus that promote helix formation.
Summarizing, Fig. 4 supports what was observed in the visual inspection of the best-matching structures for the two LPRs (Fig. 3 and Supplementary Fig. 8): (1) for all proteins in each of the two ensembles, the best-matching structures all have an intermediate degree of helicity (see also Supplementary Fig. 10 for the secondary structure per protein segment), divided over several broken helices, and a tight lipid binding by the first ∼50 N-terminal residues, and (2) a clear distinction between the two ensembles is the orientation of the remainder of the protein (especially of the NAC region), where an upright orientation is an important prerequisite for a low-RSS fit to the low-LPR measurements, while a flat orientation is a prerequisite for a good spectral match with the high-LPR measurements.
Discussion
Here, we will first compare our derived structural ensembles with previously published models. The derived high-LPR (low αS concentration) ensemble shows remarkable similarities to reported literature structures obtained at similar LPRs, while the low-LPR (high αS concentration) ensemble, derived for a concentration that is less accessible with most experimental techniques, deviates significantly.
The helicity in both the high- and low-LPR ensemble is comparable to the starting, horseshoe-shaped conformation, which has been observed with NMR for αS adsorbed to sodium dodecanol sulfate (SDS)18 and NMR/EPR for αS adsorbed to SLAS21 experiments at low protein concentrations. However, the close lipid association observed in the latter study, performed at a lipid-to-protein ratio of ∼30 and higher21,22, is only seen for the high-LPR (∼0.1 and ∼37) VSFG-derived ensembles. Other EPR and NMR studies at lower relative αS concentrations also showed tightly lipid-bound, largely α-helical structures near phospholipid vesicles, rod-like SDS micelles, or lipid bicelles with LPRs in the range of 50–200019,20,23, albeit in these studies a straight α-helix was observed, similar to the structure observed with NR at LPRs ranging between 100 and 30024. Interestingly, two NMR studies that also investigated relatively high αS concentrations (LPRs of around 120 and 3022, respectively), closer to the here-employed low LPR value of ∼0.04, found a small second αS population that is less lipid-associated in co-existence with the closely lipid-associated species. This less lipid-associated αS conformation is likely the same binding motif we derive from our low-LPR experiments.
Obviously, the exact αS concentration where the protein conformation undergoes the transition from flat to upright will depend on many factors, like surface curvature, temperature, lipid composition, lipid phase, and salt concentration. This will lead to different transition concentrations, for example, for experimental systems utilizing SUVs (e.g., studies refs. 20,22) vs. flat Langmuir monolayers (this study), for studies using different lipid mixtures (although interestingly, there is no difference in the low αS-concentration binding mode to pure POPG vesicles and to vesicles composed of a 5:3:2 of DOPE:DOPS:DOPC lipid mixture22), and when applying lower temperatures (e.g., −19 or +4 °C22 or 10–20 °C20) vs. the 23 °C applied in the current study. It is nevertheless insightful to compare the results of various studies, e.g., the various reports on the high-LPR (low concentration) structure of αS show great similarities while providing new information from the various techniques. It is, thus, for example, interesting to note that the authors of ref. 22 observe that even at relatively low αS concentrations (i.e., high LPRs), the contacts between the NAC region and the lipids have a transient nature, while also in the high-LPR ensemble, we see a few structures that indeed show an upright NAC orientation (see Supplementary Fig. 8), but other than that, the frame-selection method in its here-presented form will not provide information on the dynamics of the conformational states. The previous and current studies are thus complementary to each other, and our results bring new and potentially important findings to the table, especially in providing new structural detail of the concentration-dependent structural transition25, which might bear relevance through a potentially disease-related elevation of the αS concentration31 that can thus lead to membrane-adsorbed αS structures in which the aggregation-prone NAC regions are in close proximity.
To structurally compare our results directly with the previously-proposed models, we calculate NMR chemical shifts (CSs) and VSFG spectra based on previously-reported structures. For this, we analyze the frames from the present simulation within the best-matching ensembles, defined such that their standard deviation is similar to the experimental one (see previous sections). NMR CSs are then calculated from each MD frame using the SPARTA+58 artificial neural network, and the average CS of each MD-VSFG ensemble was uploaded to the neighbor-corrected structure propensity calculator59 (ncSPC) webserver60 to obtain a secondary structure propensity score (Fig. 5). An ncSPC score of 1 indicates a fully formed helix, -1 indicates β-structure, and 0 is a coil/loop disordered structure. We do the same for the CSs obtained in the SLAS micelle-associated αS NMR experiment (BRMB 16302)21, which we also used as one of the starting structures (see Supplementary Fig. 6), and, as references, for disordered αS in solution as determined with NMR (BRMB 25227)61 and with a 73 μs solution MD trajectory kindly provided by Robustelli et al.53. The ncSPC propensity scores yield a very similar picture for the high- and low-LPR ensembles as the DSSP helicity values depicted in Fig. 4A, which demonstrates the consistency between the two methods to quantify the secondary structure. The experimental data for both low- and high LPR experiments is again most consistent with a sequence of broken α-helices, which follows the secondary-structure trend of αS adsorbed to SLAS micelles surprisingly well. Such a ∼50/50 α-helical/random-coil conformation has also been observed with other techniques, like UV-CD at a high LPR of 75025 and, at a similarly low LPR value (∼0.033; see SI for LPR determination) as the one we employ in the low-LPR experiment45, with ATR-IR.
Because the VSFG signal is not only sensitive to secondary structure but also to tertiary structure and protein orientation, we further investigate these aspects in relation to literature models by calculating the VSFG spectra of the straight19,20,24,62 and kinked18,21, mainly α-helical structures and comparing them with the experimental high- and low-LPR datasets for all possible protein orientations (see Supplementary Fig. 11). Interestingly, for the high-LPR VSFG spectra, we find a close spectral match for the flat-lying orientations that have been observed before for the two protein structures, which corroborates the results obtained with the frame-selection method (Figs. 3 and 4). In contrast, for the low-LPR spectra, the RSS values are generally significantly larger for both the straight and kinked helices, especially for the protein orientations reported in the literature, which have RSS values at least 10 times as large as the lowest RSS values that we find for the MD frames. The large RSS values obtained for these literature structures also substantiate the result from the frame-selection method at the low LPR of 0.04, as the derived N-terminally bound and otherwise protruding helical structure is inconsistent with the two structural models that are available in the literature, which have been obtained for higher (∼120–∼200019) LPR values18,19,20,21,24,62.
Neuronal cell membranes contain significant fractions of neutral lipids like cholesterol, a majority of zwitterionic lipids like PC and PE, and a minority of anionic lipids like PI, PA, and PS63,64. Reducing anionic membrane charge from the 100% anionic lipid contents used in this study is known to lead to decreased interaction with αS as has been shown with, e.g., UV-CD spectroscopy65, fluorescence correlation spectroscopy30, and voltage-dependent anion channel nanopore measurements66, indicating that αS-lipid membrane interaction is driven by negatively charged lipid species. But as shown by Fusco et al.22, the low αS concentration binding mode of the rigid parts of the protein is actually similar to vesicles composed of pure POPG and to vesicles composed of a 5:3:2 of DOPE:DOPS:DOPC lipid mixture, which shows that the anionic vs. mixed lipid composition does not strongly affect the αS conformation. The conformation of αS bound to zwitterionic lipids will probably mainly differing from the orientation derived in the present study by the aspect that the negatively-charged C-terminus will not be pointing away from the lipid interface so much66 due to charge repulsion that occurs with negatively charged lipids like DPPG.
Now, we will focus on the implications for health and disease. The strong experimental VSFG signals indicate that at low- (∼50 nM), approximately physiological (∼20 μM29), and elevated (∼50 μM) protein concentrations, αS forms densely packed monolayers at lipid interfaces. Recently, αS has been shown to adsorb to lipids in a strongly cooperative manner67, which enhances the formation of such high-density αS layers. As previously suggested22, the upright αS conformation — here derived at elevated αS concentrations — may be relevant both in a healthy context where it can form a bridge between lipid vesicles or between vesicles and membranes in order to maintain the stasis of neuronal vesicles. It has also been shown that such a protruding form of αS plays a role in the clustering of synaptic vesicles68. In the elevated-concentration conformation, the C-terminus of αS is maximally available for synaptobrevin-2 binding69,70, thereby promoting SNARE-complex assembly, which facilitates membrane fusion. The derived high-concentration structure can also be consistent with the relatively-high concentration EPR observations that reveal that the hydrophobic side of the αS molecules is shielded from the solvent62, only not interpreted such that the αS molecules lie flat on the lipid membrane, but instead form “bundles” at the interface20. In such a conformation, the hydrophobic sides of the amphipatic helices could come together, potentially interacting with extracted lipid tails. Furthermore, as also previously suggested20,71, it could be a particularly relevant structure in the development of synucleopathies as well, given the fact that this upright structure results in a high local concentration where the amyloidogenic NAC regions of the lipid-bound αS molecules line up, which can explain the catalytic effect that lipids have on αS aggregation16. We speculate that the fact that the eventually formed fibril structure is not sensitive to the LPR72, combined with the fact that the solely N-terminally bound αS structures reported in this study have been observed in co-existence with the flat-lying αS structure20,22, could indicate that the extended structure purely observed here at elevated αS concentrations, is the intermediate structure that leads to lipid-catalyzed fibril formation at all LPRs.
Finally, we provide some summarizing remarks. To study the interaction of αS with lipid interfaces as a function of concentration, we have developed an effective approach for extracting structural information from experimental VSFG spectra using a combination of non-equilibrium MD simulations and spectral calculations. We apply it to investigate αS binding to lipid surfaces, from low concentrations — which have been studied previously using other methods — to physiological and elevated protein concentrations. The method has allowed us to study the structure and orientation of αS with a high structural resolution.
Using the ViscaSelect algorithm to implement the frame-selection method, we reveal a structural transition as a function of concentration from a low αS conformation in which it adopts a flat binding pose with its helices parallel to the lipid surface to a more upright orientation of the proteins as the αS concentration increases (see Fig. 6). The low-concentration (high-LPR) structure is consistent with the derived structures at similarly high LPRs, which benchmarks the here-presented method. The degree of helicity observed in the best-matching ensemble is also comparable to what has previously been observed.
In the upright conformation, derived here for elevated αS concentrations, the first ∼50 N-terminal residues anchor the protein to the lipid membrane, similar to what we derive for the relatively lower-αS-concentration conformations. However, the upright orientation of the remainder of the protein (especially of the aggregation-probe middle NAC region) provides an opportunity for αS to to engage in more extensive intermolecular contacts and may pave the way for subsequent aggregation. This molecular mechanism can explain both the concentration-dependent aggregation catalysis by lipids16 and the aggregation-related toxicity observed at elevated αS concentrations31,73.
Methods
αS expression and purification
Purified αS was expressed and purified as follows, and as also described in74. The plasmid vector pET11-d was used to express αS in Escherichia coli BL21 (DE3). The cells were then pelleted by 20 min of centrifugation at 2657 × g, at 4 °C. The pellet of 1 L of culture was resuspended in 100 mL of osmotic shock buffer (composed of 30 mM Tris-HCl, 40% sucrose, and 2 mM EDTA at pH 7.2) and incubated for 10 min. Subsequently, the suspension was centrifuged for 30 min at 9000 × g and 20 °C. The resulting pellet was then resuspended in 90 mL of ice-cold deionized water, and 40 μL of saturated MgCl2 was added. Subsequently, the pellet was incubated on ice for 3 min, and the supernatant, containing the periplasmic preparation, was collected by 20 min centrifugation at 9000 × g and 4 °C. The periplasmic preparation was subjected to acid precipitation with drop-wise addition of 1 M HCl to a final pH level of 3.5 and then incubated for 5 min. The supernatant was collected by 20 min of centrifugation at 9000 × g and 4 °C. The pH of the supernatant was immediately adjusted to pH 7.5 with the drop-wise addition of 1 M NaOH. The solution was filtered (0.45 μm) and loaded on a Q-Sepharose column (HiTrap Q HP) pre-equilibrated with 20 mM Tris-HCl pH 7.5. The column was washed with three column volumes of 0.1 M NaCl in buffer followed by elution of αS with a NaCl gradient from 0.1 to 0.5 M. SDS-PAGE analysis was used to identify fractions with αS and to ensure protein purity. Finally, the αS was dialyzed exhaustively against deionized water, lyophilized, and stored at −20 °C.
Prior to use, the lyophilized powder was dissolved in PBS (0.01 M phosphate buffer, 0.0027 M KCl, and 0.137 M NaCl, pD 7.2, Sigma Aldrich) in D2O (99.9%D, Eurisotop) to avoid interference from H2O bending modes. The αS solution was prepared either by filtration with 100 kDa Nanosep filters (Pall Corporation, USA) or without this filtration step because we observed that this did not affect any of the kinetic or spectral features of the measurements (see Supplementary Fig. 13).
Lipid monolayer preparation and protein injection
A lipid monolayer of DPPG (1,2-dipalmitoyl-sn-glycero-3-phospho-(1'-rac-glycerol) (sodium salt), Avanti Polar Lipids) is assembled at the air–water interface within a 2 mL trough at room temperature. The DPPG is first dissolved in chloroform and then dropcast at the air–water interface to an initial surface pressure of ∼15 mN/m (see Supplementary Fig. 2), corresponding to a liquid-condensed (LC) phase75. After equilibrium, VSFG spectra are recorded on the pure lipid monolayer (see Supplementary Fig. 3). After injection of the protein solution, the surface pressure increased to ∼30 mN/m (see Supplementary Fig. 2).
The motivation for performing these measurements with a DPPG monolayer at ∼30 mN/m are as follows: (1) we have to use DPPx lipids instead of unsaturated lipids (like DOPx and POPx, which are more abundant in biological membranes76) because the C=C modes are close to the amide-I band77 that we want to use to determine the protein conformation, and mainly because these kinds of lipids are very unstable and can oxidize quickly when exposed to air. Because of the long measurement times required to follow the αS aggregation (tens of hours), the oxidative changes in the model surface would render the experiments less reliable and reproducible. (2) We aim for an area per lipid that is comparable to the value in an average idealized mammalian plasma membrane (50 Å2/molecule) and obtained this by measuring at the surface pressure to 30 mN/m78, which is also considered to be the physiological surface pressure40. (3) Many other biomimetic monolayer studies have also been performed around this surface pressure (e.g., refs. 40,79,80), so for comparability with respect to such studies, the choice for 30 mN/m is ideal. (4) Anionic DPPG lipids bind relatively strongly to αS30, and in aging mice brains, the levels of various anionic lipids are higher than in young mice64, so there might be biomedical relevance of such a relatively negative lipid interface as well. (5) Finally, the binding mode of αS — at least at relatively low αS concentrations — appears to be similar to vesicles composed of pure POPG as it is to vesicles composed of a 5:3:2 of DOPE:DOPS:DOPC lipid mixture22, which indicates that the anionic component of the lipid mixture dominates the protein–lipid interaction.
In the 50 nM, 20 μM, and 50 μM experiments, the protein was dissolved at 250 nM, 100, and 250 μM concentration, respectively, in 400 μL of PBS-D2O buffer before injection into a total subphase of 2 mL below the lipid monolayer, while stirring the subphase with a magnetic stirring bar. The concentration before protein injection of each experiment was measured by UV–vis spectroscopy using the absorbance at 280 nm (NanoDrop 2000c, Thermo Scientific).
VSFG spectroscopy
The SFG setup has been previously described in ref. 81. In short, a narrowband (FWHM ∼15 cm−1) visible beam was temporally and spatially overlapped with a broadband IR beam at the sample surface, and the SFG signal was focused into a spectrograph and detected by an EMCCD camera. VSFG spectra were recorded between 1600 and 1800 cm−1 in SSP (S-SF, S-visible, P-IR), PPP, SPS, and PSP polarization combinations. All spectra were background subtracted and normalized using a reference spectrum obtained from gold. The various polarization combinations were recorded in a sequential manner, frequently going back to the SSP polarization combination in order to monitor the development of the overall signal intensity (see Supplementary Fig. 3).
MD simulations
Eight systems of αS in different conformations and in contact with a DPPG monolayer were created (see Supplementary Fig. 6) using the CHARMM-GUI webserver82. MD simulations were performed using GROMACS83 v2021.4. We use the DES-amber force field47 for protein, water, and ions and the Slipid forcefield48,49,50 for lipids. Slipids are compatible with the AMBER99SB-ILDN force field branch that DES-amber is built upon. After minimization and 10 ns equilibration of the DPPG monolayer, production runs of 150 ns were simulated for each of the eight systems, with two repeat simulations in each. Frames were saved every 50 ps, totaling 48.016 individual αS conformations. See Supplementary Information for more details.
Spectral calculations
The spectral calculations are based on an excitonic Hamiltonian approach developed and first described in ref. 52 and in detail in the Supplementary Information. We performed the calculations for a total of 48.016 different frames, with varying protein structures and orientations in each frame. Because we did not find an improvement in the spectral match when we averaged over multiple frames (see Supplementary Table 2), all presented spectral calculations are performed by directly obtaining the eigenmodes from each frame by evaluating the excitonic Hamiltonian model. The residual sum-of-squares (RSS) difference between the experimental and calculated spectra was used to find the frames that describe the experimental response best.
Statistics and reproducibility
All VSFG, AFM, and surface-pressure measurements have been performed twice. The derived structural ensembles are robust over trajectories starting from different starting structures (given the similarity of, e.g., the 10 best-matching structures that come from various trajectories with different starting structures, with generally similar helicity-RSS trends depicted in Supplementary Fig. 12), the assumed Lorentzian normal mode width (up to a full-width-at-half-max of 12 cm−1, see Supplementary Table 3), and over reproduced experimental datasets (Supplementary Fig. 13), in that the qualitative conclusion remains the same irrespective of the parameters. We also explored whether averaging multiple frames would result in a better match by using an average over 10 and 100 successive frames (see Supplementary Table 4), but as the spectral match from single structures are already very good, it turned out to be impossible to achieve lower RSS values, so for the sake of simplicity we have focused on single structure results. Experimental modifications, like the removal of the DPPG monolayer from the air–water interface, result in markedly different best-matching structural ensembles (see Supplementary Fig. 14).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The data generated in this study are provided in the Source Data file. The data used in this study are also available in the Zenodo database under accession code 7916728 [10.5281/zenodo.7916728]. Finally, the PDB structures can be found in the protein database, see in particular 2KKW (SLAS-micelle bound αS) and 1XQ8 (Human micelle-bound αS). Source data are provided in this paper.
Code availability
The code used to calculate VSFG spectra and the ViscaSelect algorithm used to perform the frame-selection method is available from the authors upon reasonable request and on the Visca GitHub. We are currently working on making all code available in a GUI-based user-friendly fashion, which will be available for download from the same GitHub.
References
Chiti, F. & Dobson, C. M. Protein misfolding, functional amyloid, and human disease. Annu. Rev. Biochem. 75, 333–366 (2006).
Eisenberg, D. & Jucker, M. The amyloid state of proteins in human diseases. Cell 148, 1188–1203 (2012).
Knowles, T. P., Vendruscolo, M. & Dobson, C. M. The amyloid state and its association with protein misfolding diseases. Nat. Rev. Mol. Cell Biol. 15, 384–396 (2014).
Breydo, L., Wu, J. W. & Uversky, V. N. α-Synuclein misfolding and Parkinson’s disease. Biochim. Biophys. Acta 1822, 261–285 (2012).
Wong, Y. C. & Krainc, D. α-synuclein toxicity in neurodegeneration: mechanism and therapeutic strategies. Nat. Med. 23, 1–13 (2017).
Singh, S. K., Dutta, A. & Modi, G. α-Synuclein aggregation modulation: an emerging approach for the treatment of Parkinson’s disease. Fut. Med. Chem. 9, 1039–1053 (2017).
Alafuzoff, I. & Hartikainen, P. Alpha-synucleinopathies. Handb. Clin. Neurol. 145, 339–353 (2018).
Miraglia, F., Ricci, A., Rota, L. & Colla, E. Subcellular localization of alpha-synuclein aggregates and their interaction with membranes. Neural Regen. Res. 13, 1136 (2018).
Ruipérez, V., Darios, F. & Davletov, B. Alpha-synuclein, lipids and Parkinson’s disease. Prog. Lipid Res. 49, 420–428 (2010).
Iyer, A. & Claessens, M. M. Disruptive membrane interactions of alpha-synuclein aggregates. Biochim. Biophys. Acta 1867, 468–482 (2019).
den Hartog Jager, W. A. Sphingomyelin in Lewy inclusion bodies in Parkinson’s disease. Arch. Neurol. 21, 615–619 (1969).
Gai, W. et al. In situ and in vitro study of colocalization and segregation of α-synuclein, ubiquitin, and lipids in Lewy bodies. Exp. Neurol. 166, 324–333 (2000).
Shahmoradian, S. H. et al. Lewy pathology in Parkinson’s disease consists of crowded organelles and lipid membranes. Nat. Neurosci. 22, 1099–1109 (2019).
Grey, M., Linse, S., Nilsson, H., Brundin, P. & Sparr, E. Membrane interaction of α-synuclein in different aggregation states. J. Parkinson’s Dis. 1, 359–371 (2011).
Hellstrand, E., Nowacka, A., Topgaard, D., Linse, S. & Sparr, E. Membrane lipid co-aggregation with α-synuclein fibrils. PloS ONE 8, e77235 (2013).
Galvagnion, C. et al. Lipid vesicles trigger α-synuclein aggregation by stimulating primary nucleation. Nat. Chem. Biol. 11, 229–234 (2015).
Galvagnion, C. The role of lipids interacting with α-synuclein in the pathogenesis of Parkinson’s disease. J. Parkinson’s Dis. 7, 433–450 (2017).
Ulmer, T. S., Bax, A., Cole, N. B. & Nussbaum, R. L. Structure and dynamics of micelle-bound human α-synuclein. J. Biol. Chem. 280, 9595–9603 (2005).
Georgieva, E. R., Ramlall, T. F., Borbat, P. P., Freed, J. H. & Eliezer, D. Membrane-bound α-synuclein forms an extended helix: long-distance pulsed ESR measurements using vesicles, bicelles, and rodlike micelles. J. Am. Chem. Soc. 130, 12856–12857 (2008).
Bodner, C. R., Dobson, C. M. & Bax, A. Multiple tight phospholipid-binding modes of α-synuclein revealed by solution NMR spectroscopy. J. Mol. Biol. 390, 775–790 (2009).
Rao, J. N., Jao, C. C., Hegde, B. G., Langen, R. & Ulmer, T. S. A combinatorial NMR and EPR approach for evaluating the structural ensemble of partially folded proteins. J. Am. Chem. Soc. 132, 8657–8668 (2010).
Fusco, G. et al. Direct observation of the three regions in α-synuclein that determine its membrane-bound behaviour. Nat. Commun. 5, 1–8 (2014).
Antonschmidt, L. et al. Insights into the molecular mechanism of amyloid filament formation: Segmental folding of α-synuclein on lipid membranes. Sci. Adv. 7, eabg2174 (2021).
Hellstrand, E. et al. Adsorption of α-synuclein to supported lipid bilayers: positioning and role of electrostatics. ACS Chem. Neurosci. 4, 1339–1351 (2013).
Shvadchak, V. V., Yushchenko, D. A., Pievo, R. & Jovin, T. M. The mode of α-synuclein binding to membranes depends on lipid composition and lipid to protein ratio. FEBS Lett. 585, 3513–3519 (2011).
Galvagnion, C. et al. Chemical properties of lipids strongly affect the kinetics of the membrane-induced aggregation of α-synuclein. Proc. Natl Acad. Sci. 113, 7065–7070 (2016).
Hoover, B. M. et al. Membrane remodeling and stimulation of aggregation following α-synuclein adsorption to phosphotidylserine vesicles. J. Phys. Chem. B 125, 1582–1594 (2021).
Iyer, A., Petersen, N. O., Claessens, M. M. & Subramaniam, V. Amyloids of alpha-synuclein affect the structure and dynamics of supported lipid bilayers. Biophys. J. 106, 2585–2594 (2014).
Wilhelm, B. G. et al. Composition of isolated synaptic boutons reveals the amounts of vesicle trafficking proteins. Science 344, 1023–1028 (2014).
Middleton, E. R. & Rhoades, E. Effects of curvature and composition on α-synuclein binding to lipid vesicles. Biophys. J. 99, 2279–2288 (2010).
Olivares, D., Huang, X., Branden, L., Greig, N. H. & Rogers, J. T. Physiological and pathological role of alpha-synuclein in Parkinson’s disease through iron mediated oxidative stress; the role of a putative iron-responsive element. Int. J. Mol. Sci. 10, 1226–1260 (2009).
Yan, E. C. Y., Wang, Z. & Fu, L. Proteins at interfaces probed by chiral vibrational sum frequency generation spectroscopy. J. Phys. Chem. B 119, 2769–2785 (2015).
Hosseinpour, S. et al. Structure and dynamics of interfacial peptides and proteins from vibrational sum-frequency generation spectroscopy. Chem. Rev. 120, 3420–3465 (2020).
Carr, J. K., Wang, L., Roy, S. & Skinner, J. L. Theoretical sum frequency generation spectroscopy of peptides. J. Phys. Chem. B 119, 8969–8983 (2015).
Bellucci, L. et al. The interaction with gold suppresses fiber-like conformations of the amyloid β (16–22) peptide. Nanoscale 8, 8737–8748 (2016).
Harrison, E. T., Weidner, T., Castner, D. G. & Interlandi, G. Predicting the orientation of protein G B1 on hydrophobic surfaces using Monte Carlo simulations. Biointerphases 12, 02D401 (2017).
Lu, H. et al. Peptide-controlled assembly of macroscopic calcium oxalate nanosheets. J. Phys. Chem. Lett. 10, 2170–2174 (2019).
Alamdari, S. et al. Orientation and conformation of proteins at the air–water interface determined from integrative molecular dynamics simulations and sum frequency generation spectroscopy. Langmuir 36, 11855–11865 (2020).
Barth, A. Infrared spectroscopy of proteins. Biochim. Biophys. Acta 1767, 1073–1101 (2007).
Ameziane-Le Hir, S. et al. Cholesterol favors the anchorage of human dystrophin repeats 16 to 21 in membrane at physiological surface pressure. Biochim. Biophys. Acta 1838, 1266–1273 (2014).
Chaudhary, H., Subramaniam, V. & Claessens, M. M. Direct visualization of model membrane remodeling by α-synuclein fibrillization. ChemPhysChem 18, 1620 (2017).
Lorenzen, N., Lemminger, L., Pedersen, J. N., Nielsen, S. B. & Otzen, D. E. The N-terminus of α-synuclein is essential for both monomeric and oligomeric interactions with membranes. FEBS Lett. 588, 497–502 (2014).
Paslawski, W. et al. High stability and cooperative unfolding of α-synuclein oligomers. Biochemistry 53, 6252–6263 (2014).
Galvagnion, C. et al. Lipid dynamics and phase transition within α-synuclein amyloid fibrils. J. Phys. Chem. Lett. 10, 7872–7877 (2019).
Fallah, M. A. Simultaneous IR-spectroscopic observation of α-synuclein, lipids, and solvent reveals an alternative membrane-induced oligomerization pathway. ChemBioChem 18, 2312–2316 (2017).
Ramakrishnan, M., Jensen, P. H. & Marsh, D. Association of α-synuclein and mutants with lipid membranes: spin-label ESR and polarized IR. Biochemistry 45, 3386–3395 (2006).
Piana, S., Robustelli, P., Tan, D., Chen, S. & Shaw, D. E. Development of a force field for the simulation of single-chain proteins and protein–protein complexes. J. Chem. Theory Comput. 16, 2494–2507 (2020).
Jambeck, J. P. & Lyubartsev, A. P. Derivation and systematic validation of a refined all-atom force field for phosphatidylcholine lipids. J. Phys. Chem. B 116, 3164–3179 (2012).
Jambeck, J. P. & Lyubartsev, A. P. Another piece of the membrane puzzle: extending slipids further. J. Chem. Theory Comput. 9, 774–784 (2013).
Grote, F. & Lyubartsev, A. P. Optimization of slipids force field parameters describing headgroups of phospholipids. J. Phys. Chem. B 124, 8784–8793 (2020).
Hamm, P. & Zanni, M. Concepts and Methods of 2D Infrared Spectroscopy. (Cambridge University Press, 2011).
Roeters, S. J. et al. Determining in situ protein conformation and orientation from the amide-I sum-frequency generation spectrum: theory and experiment. J. Phys. Chem. A 117, 6311–6322 (2013).
Robustelli, P., Piana, S. & Shaw, D. E. Developing a molecular dynamics force field for both folded and disordered protein states. Proc. Natl Acad. Sci. USA 115, E4758–E4766 (2018).
Piana, S., Donchev, A. G., Robustelli, P. & Shaw, D. E. Water dispersion interactions strongly influence simulated structural properties of disordered protein states. J. Phys. Chem. B 119, 5113–5123 (2015).
Snow, C. D., Zagrovic, B. & Pande, V. S. The Trp cage: folding kinetics and unfolded state topology via molecular dynamics simulations. J. Am. Chem. Soc. 124, 14548–14549 (2002).
Duan, L. et al. Accelerated molecular dynamics simulation for helical proteins folding in explicit water. Front. Chem. 7, 540 (2019).
Fu, L. et al. Characterization of parallel β-sheets at interfaces by chiral sum frequency generation spectroscopy. J. Phys. Chem. Lett. 6, 1310–1315 (2015).
Shen, Y. & Bax, A. SPARTA+: a modest improvement in empirical NMR chemical shift prediction by means of an artificial neural network. J. Biomol. NMR 48, 13–22 (2010).
Tamiola, K. & Mulder, F. A. Using NMR chemical shifts to calculate the propensity for structural order and disorder in proteins. Biochem. Soc. Trans. 40, 1014–1020 (2012).
Neighbor-corrected structure propensity calculator (ncSPC) web-server, accessed February. https://st-protein02.chem.au.dk/ncSPC/ (2021).
Porcari, R. et al. The H50Q mutation induces a 10-fold decrease in the solubility of α-synuclein. J. Biol. Chem. 290, 2395–2404 (2015).
Jao, C. C., Hegde, B. G., Chen, J., Haworth, I. S. & Langen, R. Structure of membrane-bound α-synuclein from site-directed spin labeling and computational refinement. Proc. Natl Acad. Sci. USA 105, 19666–19671 (2008).
Wenk, L. L. M. 9 Neuronal membrane lipids–their role in the synaptic vesicle cycle. Handb. Neurochem. Mol. Neurobiol. 14, 223 (2009).
Rappley, I. et al. Lipidomic profiling in mouse brain reveals differences between ages and genders, with smaller changes associated with α-synuclein genotype. J. Neurochem. 111, 15–25 (2009).
Kjaer, L., Giehm, L., Heimburg, T. & Otzen, D. The influence of vesicle size and composition on α-synuclein structure and stability. Biophys. J. 96, 2857–2870 (2009).
Jacobs, D. et al. Probing membrane association of α-synuclein domains with VDAC nanopore reveals unexpected binding pattern. Sci. Rep. 9, 1–14 (2019).
Makasewicz, K. et al. Cooperativity of α-synuclein binding to lipid membranes. ACS Chem. Neurosci. 12, 2099–2109 (2021).
Fusco, G. et al. Structural basis of synaptic vesicle assembly promoted by α-synuclein. Nat. Commun. 7, 1–12 (2016).
Burré, J. α-Synuclein promotes SNARE-complex assembly in vivo and in vitro. Science 329, 1663–1667 (2010).
Burré, J., Sharma, M. & Südhof, T. C. Systematic mutagenesis of α-synuclein reveals distinct sequence requirements for physiological and pathological activities. J. Neurosci. 32, 15227–15242 (2012).
Dikiy, I. & Eliezer, D. Folding and misfolding of alpha-synuclein on membranes. Biochim. Biophys. Acta 1818, 1013–1018 (2012).
Dubackic, M. On the cluster formation of α-synuclein fibrils. Front. Mol. Biosci. 8, 768004 (2021).
Souza, J. M., Giasson, B. I., Chen, Q., Lee, V. M.-Y. & Ischiropoulos, H. Dityrosine cross-linking promotes formation of stable α-synuclein polymers: implication of nitrative and oxidative stress in the pathogenesis of neurodegenerative synucleinopathies. J. Biol. Chem. 275, 18344–18349 (2000).
Mohammad-Beigi, H. et al. Strong interactions with polyethylenimine-coated human serum albumin nanoparticles (PEI-HSA NPs) alter α-synuclein conformation and aggregation kinetics. Nanoscale 7, 19627–19640 (2015).
Rodrigues, J. C. & Caseli, L. Incorporation of bacitracin in Langmuir films of phospholipids at the air-water interface. Thin Solid Films 622, 95–103 (2017).
Symons, J. L. et al. Lipidomic atlas of mammalian cell membranes reveals hierarchical variation induced by culture conditions, subcellular membranes, and cell lineages. Soft Matter 17, 288–297 (2021).
Withey, P., Shen, L. & Graham, W. Fourier transform far infrared spectroscopy of a C4 bending mode. J. Chem. Phys. 95, 820–823 (1991).
Ingólfsson, H. I. et al. Lipid organization of the plasma membrane. J. Am. Chem. Soc. 136, 14554–14559 (2014).
Li, B. et al. Sum frequency generation of interfacial lipid monolayers shows polarization dependence on experimental geometries. Langmuir 32, 7086–7095 (2016).
Maskarinec, S. A., Hannig, J., Lee, R. C. & Lee, K. Y. C. Direct observation of poloxamer 188 insertion into lipid monolayers. Biophys. J. 82, 1453–1459 (2002).
Golbek, T. W., Schmüser, L., Rasmussen, M. H., Poulsen, T. B. & Weidner, T. Lasalocid acid antibiotic at a membrane surface probed by sum frequency generation spectroscopy. Langmuir 36, 3184–3192 (2020).
Jo, S., Kim, T., Iyer, V. G. & Im, W. CHARMM-GUI: a web-based graphical user interface for CHARMM. J. Comput. Chem. 29, 1859–1865 (2008).
Berendsen, H. J., van der Spoel, D. & van Drunen, R. GROMACS: a message-passing parallel molecular dynamics implementation. Comput. Phys. Commun. 91, 43–56 (1995).
Kabsch, W. & Sander, C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22, 2577–2637 (1983).
Acknowledgements
We thank the Langen group (University of Southern California, USA) for kindly providing the coordinates of the αS fragment (residues 9–89) bound to 7:3 POPC/POPS small unilamellar vesicles used in the calculations presented in Supplementary Fig. 11. We thank Aditya Iyer and Ad Bax for their valuable discussions. S.J.R. acknowledges the Lundbeck Foundation for funding through the post-doctoral award R303-2018-349. K.B.P. and D.E.O. are supported by the Lundbeck Foundation (Grant R276-2018-671). K.B.P. and B.S. acknowledge the Novo Nordisk Foundation (Grant NNF18OC0032608) for computational resources. T.W.G. would like to thank the Lundbeck Foundation for post-doc grant R322-2019-2461. This article is part of a project that has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Grant agreement no. 819039 F-BioIce). S.J.R., K.S., T.W.G., M.B., and T.W. acknowledge support from the Novo Nordisk Foundation (Facility Grant NanoScat, no. NNF18OC0032628).
Author information
Authors and Affiliations
Contributions
S.J.R. and T.W. conceived the study, S.J.R., K.S., K.B.P., and T.W. developed the analysis methodology, S.J.R. and K.S. developed and programmed the spectral calculation scripts, K.S. performed the spectral calculations, K.B.P. performed and analyzed the MD simulations, S.J.R., K.S., K.B.P., and Y.W. made the figures, S.J.R., T.W.G., and M.B. performed the VSFG experiments, Y.Z. and Y.W. performed the AFM experiments, M.D. designed and supervised the AFM experiments, D.E.O. and J.N. provided the αS, advised during the project, helped embedding the work in the literature, and co-composed an important part of the paper, B.S. designed and supervised the MD simulations, and all co-authors wrote the paper.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Shayon Bhattacharya, Gabriele Kaminski Schierle, Poul B. Petersen, and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Roeters, S.J., Strunge, K., Pedersen, K.B. et al. Elevated concentrations cause upright alpha-synuclein conformation at lipid interfaces. Nat Commun 14, 5731 (2023). https://doi.org/10.1038/s41467-023-39843-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-023-39843-1