Membrane interactions of a self-assembling model peptide that mimics the self-association, structure and toxicity of Aβ(1-40)

Abstract

β-amyloid peptide (Aβ) is a primary protein component of senile plaques in Alzheimer’s disease (AD) and plays an important, but not fully understood role in neurotoxicity. Model peptides with the demonstrated ability to mimic the structural and toxicity behavior of Aβ could provide a means to evaluate the contributions to toxicity that are common to self–associating peptides from many disease states. In this work, we have studied the peptide-membrane interactions of a model β-sheet peptide, P_11-2 (CH₃CO-Gln-Gln-Arg-Phe-Gln-Trp-Gln-Phe-Glu-Gln-Gln-NH₂), by fluorescence, infrared spectroscopy, and hydrogen-deuterium exchange. Like Aβ(1-40), the peptide is toxic, and conditions which produce intermediate oligomers show higher toxicity against cells than either monomeric forms or higher aggregates of the peptide. Further, P_11-2 also binds to both zwitterionic (POPC) and negatively charged (POPC:POPG) liposomes, acquires a partial β-sheet conformation in presence of lipid, and is protected against deuterium exchange in the presence of lipids. The results show that a simple rationally designed model β-sheet peptide recapitulates many important features of Aβ peptide structure and function, reinforcing the idea that toxicity arises, at least in part, from a common mode of action on membranes that is independent of specific aspects of the amino acid sequence. Further studies of such well-behaved model peptide systems will facilitate the investigation of the general principles that govern the molecular interactions of aggregation-prone disease-associated peptides with cell and/or membrane surfaces.

1. Introduction

Biological molecules, particularly proteins and peptides, have the ability to self-assemble. Self-assembly of proteins and peptides has been the focus of intense research in recent years, due to its importance in understanding a variety of human pathologies such as Alzheimer’s disease (AD), prion disorders, and type 2 diabetes [1]. A characteristic feature of theses diseases is the deposition of a structure known as amyloid. In the case of AD, the β-amyloid peptide (Aβ), produced by proteolytic processing of an amyloid precursor protein, is believed to play a role in neurodegeneration. Originally it was proposed that the fibrillar Aβ amyloid deposits cause the various symptoms of AD; however growing evidence indicates that the smaller, soluble forms of Aβ are the primary neurotoxic species [2-5]. The mechanism of Aβ aggregation has been subject of many studies [6-10]. Increasing evidence indicates that the neuronal cell membrane participates in the mechanism of Aβ aggregation. Aβ has been shown to interact with phospholipid liposomes and the interaction appears to be aggregation dependent [11], with membrane charge and composition being important [12-19].

Recent studies suggest that multiple self-associating peptides and proteins can exhibit toxicity patterns similar to Aβ. The fact that oligomeric intermediates from multiple misfolding diseases increase membrane permeability [20] and exhibit toxicity that can be inhibited by a single anti-oligomer antibody [21] suggests there is at least in part a common mode of toxicity that is independent of specific aspects of sequence. Hence, while it is important to investigate interactions between membranes that are specific to Aβ, it is also critical to unearth the features of peptide-membrane interactions which are common to all such protein misfolding diseases.

Several research groups have developed simple designed model peptides as novel biomolecular materials that self-assemble in a hierarchical manner producing an array of structures including fibers, scaffolds, and lipid-like peptides [22-26]. A set of designed, self-assembling peptides relevant to amyloid disease are those developed by Boden and co-workers [27]. They exploited the biological β-sheet motif to design simple de novo peptides that self-assemble in a hierarchical manner to form a variety of well defined nanostructures ranging from simple monomers to more complex oligomers to supramolecular fibrils and fibers. A thorough knowledge of their physico-chemical behavior in aqueous solution has been achieved and different states of self-assembly can be controlled and modulated by changes in concentration, pH, ionic strength and composition [27,28]. These peptides adopt extended β-strand conformations in solution, and the β-strands self-assemble in one dimension to form elongated tapes as well as higher order aggregates with pure antiparallel β-sheet structure.

In this work, we have characterized the peptide-membrane interactions of the model self-assembling peptide P_11-2 (CH₃CO-Gln-Gln-Arg-Phe-Gln-Trp-Gln-Phe-Glu-Gln-Gln-NH₂) by fluorescence and deuterium exchange techniques. These structural measurements and corresponding toxicity trends have been compared with those obtained for Aβ(1-40). The results show that by employing a simple rationally designed model β-sheet peptide it is possible to recapitulate several features of the Aβ peptide structure and function, reinforcing the idea that there is a common mode of action and toxicity that is independent of specific aspects of the amino acid sequence. Since Aβ peptides are prone to uncontrolled aggregation in solution and therefore much more difficult to handle experimentally than P_11-2, the P_11-2 model peptide promises to be very useful to uncover many of the principles that govern the molecular interactions of conformational diseases in the presence of membrane surfaces. At same time, the controlled self-assembly of P_11-2 on membrane surfaces may open new routes to novel nanostructured materials with a wide range of applications in biomedicine and nanotechnology.

2. Materials and Methods

Peptides, Lipids and Chemical Compounds

P_11-2 (CH₃CO-Gln-Gln-Arg-Phe-Gln-Trp-Gln-Phe-Glu-Gln-Gln-NH₂) was synthesized and its purity was confirmed by analytical high-performance liquid chromatography and mass spectrometry by the Biomolecular Research Facility at the University of Virginia. P_11-2 solutions in 5 mM PBC buffer, pH 7.4 were prepared by dilution of a stock solution of known concentration. Small aliquots of stock solutions were frozen immediately after preparation and stored protected from light at −20 °C. Thawed stock solutions were used on the same day and never refrozen. Native and fluorescent (labeled with AMCA [6-((7-amino-4-methylcoumarin-3-acetyl)amino) hexanoic acid) Aβ(1-40) peptides were purchased from Anaspec, Inc. (San Jose, CA). Aβ samples were prepared as described before [29]. Briefly, Aβ was dissolved in 10 μL 0.1 % trifluoroacetic acid (TFA, Fluka) to a concentration of 10 mg/mL at 25 ± 0.5 °C. After 50 minutes, “fresh samples” were made by diluting the stock to a final concentration of 100 μM in PBS (10 mM NH₂PO₄, 150 mM NaCl, pH 7.4). Lipids were purchased from Avanti Polar Lipids (Alabaster, AL). Salts were generally obtained from Sigma Chemical Co. (St. Louis, MO), with the exception of KI which was purchased from MP Biomedicals, LLC (Solon, OH).

Preparation of Liposomes

Solutions of phospholipid in chloroform at desired ratios of POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) and POPG (1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-rac-1-glycerol [sodium salt]) were evaporated under a N₂ stream and each sample was dried overnight under vacuum. Lipid films were suspended in 5 mM phosphate-borate-citrate (PBC) buffer, pH 7.4 and vortexed to give multilamellar lipid dispersions. To prepare large unilamellar vesicles (LUVs), the dispersion was five times frozen and thawed in liquid nitrogen and in a 37 °C water bath, followed by extrusion (15 times) through polycarbonate membranes with 100 nm pores (Avestin, Ottawa, Canada). To prepare small unilamellar vesicles (SUVs), the dispersion was sonicated in an ice-water bath using a Branson titanium tip ultrasonicator for ~ 1 h at 50 % duty cycle until it was transparent. When only a single lipid was used, the powder was dissolved in buffer directly followed by the steps described above.

Fluorescence Measurements

Fluorescence studies were carried out using a Fluorolog-3 spectrofluorometer (Jobin-Yvon, Edison, NJ). Intrinsic Trp or AMCA fluorescence of the peptides was measured before and after addition of different amounts of phospholipid vesicles to a fixed peptide concentration. Peptide-lipid interactions were further assessed by changes in the accessibility of the peptides to aqueous quenchers (acrylamide and iodide) upon addition of LUVs. Experiments were carried out by addition of aliquots of concentrated acrylamide or potassium iodide stock solution to peptide samples in the absence or presence of LUVs.

Fourier Transform Infrared Spectroscopy

Polarized ATR-FTIR spectra were recorded on a Vector 22 Fourier transform infrared spectrometer (Bruker, Billerica, MA). Planar phospholipid bilayers supported on germanium ATR plates were prepared by Langmuir-Blodgett deposition and fusion of small unilamellar vesicles as described previously [30,31]. The substrate-supported monolayer was 1,2-dimyristoyl-3-sn-phosphatidylcholine (DMPC), and the monolayer exposed to the buffer compartment was POPC. Peptides at 20 μM in 5 mM PBC, pH 7.4 were added to the sample cell, incubated for 5 minutes at room temperature, and excess unbound peptide was washed away with 3 volumes of the same buffer made in D₂O (Cambridge Isotope Laboratories, Inc., Andover, MA, D 99.9 %).

Hydrogen-Deuterium Exchange (HX)-Mass Spectrometric (MS) Analysis

The details of HX-MS labeling and HPLC procedures have been reported previously [29]. Briefly, peptide samples were diluted ten fold into D₂O (Acros Organics, Morris Plains, NJ, D 99.8 %) at room temperature. The labeling solution was 90 % D (molar) and the pH was 7.0. After proper labeling time, dimethylsulfoxide (DMSO) and dichloroacetic acid (DCA, Fluka, 95/5 vol/vol) were used as the dissolution and quench buffers with a final pH close to 3.5.

After quenching, the mixture was injected into a 100 μL sample loop and desalted for ~ 1 min through a C8 peptide trap cartridge (Micro Trap™ 1 mm ID × 8 mm, Michrom Bioresources, Inc., Auburn, CA). Then, the peptide was eluted by a 30%-80% acetonitrile gradient at a flow rate of 50 μL/min over 2 min and sent directly to the mass spectrometer. To minimize back-exchange during the analysis, all the columns, loops, lines and valves were precooled for 1 hr and immersed in ice during all the experiments.

The +1 and +4 charge state peaks were selected for P_11-2 and Aβ analysis respectively. Data were collected in positive ion, zoom scan, and profile mode on a Thermo Finnigan LTQ linear quadrupole ion trap mass spectrometer (San Jose, CA) with a standard ESI source. The ESI voltage was 4.5 kV, capillary temperature 275 °C, sheath gas flow rate 20 units, and tube lens voltage 135 V. The centroid mass for each peak was obtained as described before [29] and the protection percentage was calculated from Eq. 1

$$\mathrm{P}=1-\frac{\mathrm{M}-{\mathrm{M}}_0}{\mathrm{N}}\times 100\%$$ (1)

Here, P is the percent of protection. N is the total number of exchangeable amide backbone protons, 11. M and M₀ refer to the centroid mass for labeled and unlabeled P_11-2 peak, respectively.

Toxicity Assay

Viability of cells after 2 hour exposure to P_11-2 peptides was assessed using a rapid Annexin-7AAD assay recently reported [32]. In brief, SH-SY5Y cells were seeded in 96 well plates at a density of 10⁵ cells/well and allowed to adhere to wells for 24 h in a humidified 5% CO₂/air incubator at 37°C. Culture medium was replaced with 100 μL of P_11-2 peptide in MEM and cells were incubated with P_11-2 peptide for 2 hours. To stain cells, medium was replaced with 100 μL of 1X binding buffer (0.01M HEPES/NaOH, pH 7.4, 0.14M NaCl, 2.5 mM CaCl₂). Thereafter, 5μL of both annexin-phycoerythrin (annexin-PE) (5ng/well) and 7-amino-actinomycin D (7AAD) (0.25μg/well) were added, followed by 10 min incubation in the dark at 25°C. 150 μL of 2X binding buffer was then added. 7AAD binds to DNA, and is therefore a measure of membrane permeability. Annexin-PE binds to phosphatidyl serine, which accumulates on the membrane at early periods during apoptosis. Finally, cells were removed from plates by mechanical scraping, and then analyzed using the BD FACSArray Bioanalyzer flow cytometer (San Diego, CA). Annexin-PE was excited with the 532 nm laser and detected using the yellow (564-606 nm) filter, while 7AAD was detected using the red (653-669 nm) filter. Spillover for the two detectors was calculated using 3 samples one of which was unstained, another stained with only annexin-PE and the third stained only with 7AAD. To ensure staining, positive controls were prepared by incubating cells with a 533 μM H₂O₂ solution in 33% (v/v) MEM in deionized water. A quadrant gate was used to estimate the percentage of the total population that was both annexin and 7AAD negative, taken as the viable population. Relative viability (V_relative) was estimated as indicated below where V_treated is the percentage of viable cells in the treated population, V_peroxide is the percentage of viable cells in the peroxide treated population, and V_untreated is the percentage of cells in the negative control.

$$V_{relative}=\frac{V_{treated}-V_{peroxide}}{V_{untreated}-V_{peroxide}}\times 100\%$$ (2)

Averages and standard errors of at least five repeated experiments are reported for each condition.

3. Results

We set out first to demonstrate that the model peptide P_11-2 recapitulates some of the biology of Aβ(1-40) and other amyloid forming peptides, namely that small assembly intermediates are more toxic than the final large fibril-like self-assembled structures [2-5,21]. The toxicity of P_11-2 to SY5Y cells was measured at three concentrations where the state of peptide oligomerization is known to differ. At a low concentration of 20 μM, P_11-2 does not self-assemble, and remains monomeric [28] with random coil conformation [53]. 100 μM is just above the minimum concentration (70 μM) at which β structures begin to form and flexible β-sheet ribbons form with width less than 5 nm [28]. As shown by CD, samples at 100 μM have obvious ordered β-sheet structure [53]. At 300 μM the extent of β-structure is even greater [53], but no rigid fibrils were formed [28].

Figure 1 shows results of the toxicity assay in which changes in membrane permeability were detected within 2 hours of the addition of the peptide [32]. The results demonstrate that the intermediate oligomers of P_11-2 are significantly (p<0.01) more toxic than the monomers and large assemblies at low and high concentrations, respectively. Therefore, P_11-2 toxicity correlates well with the state of self-assembly as has been described before for Aβ using the same cell line, assay, and lot of Aβ [32]. It should be noted that the cell viability reported here is the relative value compared to untreated control cells. Thus, the 300 μM peptide sample having a higher viability than the controls resulted in a value over 100%. This suggests that the peptide may coat the cells, making the membrane less permeable, or the phosphatidyl-serine on the surface less available. The involvement of membranes in peptide cytotoxicity is consistent with the fact that the P _11-2 peptide has also been shown to display ion conduction across lipid bilayers [33]. This behavior recapitulates the well-reported ability of Aβ peptides to form ion channels and change conductance in planar lipid bilayers [34-38].

Figure 1.

Effect of the P_11-2 on viability of SY5Y cells. Peptide was incubated with cells for 2 hours prior to viability measurement using annexin-PE and 7 AAD staining, followed by flow cytometry. Viability was normalized relative to untreated control cells. The error bars are the standard error of the mean of replicate samples.

Once we had shown that some aspects of Aβ biology were captured by our model peptide, we sought to utilize P_11-2 to examine features of the peptide-membrane interaction that might be important in biological function that have been difficult to probe (or have led to contradictory results) using Aβ [6,12]. Several biophysical techniques were employed to elucidate the molecular details of the interaction of P_11-2 with lipid liposomes. First, we studied P_11-2-phospholipid interactions by monitoring the changes in tryptophan (Trp) fluorescence upon the addition of large unilamellar vesicles (LUVs). The Trp fluorescence intensity decreased with the addition of either zwitterionic POPC or net negatively charged POPC:POPG (2:1 mol ratio) liposomes to the peptide. Figure 2 shows the corresponding titration curves obtained with 20 μM P_11-2. Although the fluorescence intensity decreased with liposome concentration, the wavelength of the maximum intensity was neither blue nor red shifted, but did not vary significantly beyond the measurement error of ± 1nm (Figure S2, Supplementary Information). Light scattering attenuation from the liposomes was also tested by addition of liposomes to solutions of the fluorophore AMCA and found to be negligible (data not shown) compared to the drops in Figure 2. The binding isotherm data in Figure 2 were fit with a partition equilibrium model [39-41] to obtain peptide-lipid partition constants (K_app, M⁻¹). The partition constants (averaged from 5 independent experiments) were (3.7 ± 0.2) × 10³ M⁻¹ for POPC and (4.2 ± 0.3) × 10³ M⁻¹ for POPC:POPG (2:1). These constants are apparent because no attempt was made to correct for electrostatic double layer effects at the membrane surface.

Figure 2.

The relative fluorescence of P_11-2 at 20 μM in 5 mM PBC, pH 7.4 is shown as function of lipid concentration. Binding of P_11-2 to liposomes composed of POPC (●) and POPC/POPG (2:1) (■). The curve represents a fit of a partition binding model [41].

The accessibility of the water-soluble fluorescence quencher acrylamide to the P_11-2 tryptophan residue yielded information about the mode of peptide insertion in the model membranes (Figure 3A). Fluorescence intensity was measured as a function of quencher concentration in the outer aqueous phase of phospholipid vesicles [42]. The order of the fluorescence quenching was (solution) > (POPC) ≅ (POPC/POPG) showing that P_11-2 is buried to approximately the same extent in both model membranes. Due to the incidence of static quenching at higher acrylamide concentrations [43], the Stern-Volmer constants (K_SV) were calculated from the plots between 0 and 0.1 M acrylamide. The measured fluorescence values were corrected for absorption of acrylamide at the absorption wavelength of Trp as we have previously [56]. K_SV values for P_11-2 were 44, 31 and 29 M in the absence of lipid, presence of POPC, and presence of POPC/POPG liposomes, respectively. The same general trends were observed with another soluble quencher of Trp fluorescence, iodide (Figure 3B) with and without POPG liposomes. The K_SV values dropped from 10.6 M⁻¹ to 6.5 M⁻¹ in presence of the POPG liposomes. The ratio of KI quenching K_SV values in absence and presence of the liposomes (1.6) is in good agreement with the ratio obtained using acrylamide (1.4; Figure 3A) with POPC liposomes. The depletion of iodide in the vicinity of the membrane surface means that measured K_SV values should be considered upper limits.

Figure 3.

Stern-Volmer plot for the quenching of P_11-2 fluorescence by acrylamide (A) and KI (B) in aqueous buffer (■), and bound to liposomes composed of POPC (●) and POPC/POPG (2:1) in panel A or POPG in panel B (▲). The curve represents a fit of the data points using the method previously described [42].

Because Aβ lacks a Trp reporter residue, similar experiments were performed using Aβ labeled with the fluorophore, AMCA. EM showed that concentrated, 100 uM stock solutions of AMCA-Aβ were mostly devoid of any large aggregates, although smaller amorphous structures (10-15 nm) were observed (See Supplementary Information). A similar study with Aβ(1–40) labeled with a 7-diethylamino-coumarin-3-carbonyl (DAC) group at the N-terminus (DAC-Aβ) reported little change in aggregation behavior [55].

Experiments with freshly dissolved AMCA-labeled Aβ(1-40) at 10 μM exhibited a very similar behavior to P_11-2. The AMCA fluorescence decreased upon Aβ binding to negatively charged liposomes and the liposomes offered protection against the quencher iodide when compared with the quenching of the peptide in the absence of vesicles (Fig. 4). K_SV values for Aβ were 74 and 32 M⁻¹ in absence and presence of POPG liposomes, respectively. Also supporting the solvent exposed nature of fresh AMCA-Aβ, experiments with aged showed lower K_SV values than freshly prepared Aβ (See Supplementary Information).

Figure 4.

Stern-Volmer plot for the quenching of Aβ(1-40) fluorescence by KI in aqueous buffer solution (■) and in presence of liposomes composed of POPG (●). The curve represents a fit of the data points as in Figure 3.

Monitoring peptide aggregation in solution and in the presence of liposomes is important to understand aspects of their molecular interactions, in particular with regard to pore formation. The use of hydrogen-deuterium exchange coupled with electrospray ionization mass spectrometry has proven to be a versatile technique to study different aggregation states of Aβ peptides [44]. HX/MS has been used to study structural aspects and the kinetics of Aβ peptide aggregation [45,46]. In order to further test whether P_11-2 can be a good model for Aβ aggregation, we examined the protection from deuterium exchange of P_11-2 at 20 μM where it is not likely to self-associate and at higher concentrations where it likely forms oligomers. Figure 5 shows that monomeric P_11-2 gains almost as much deuterium as fully labeled P_11-2 (Fig. 5B), while 250 μM peptide (Fig. 5C) that adopts an oligomeric state protects ~5 of its exchangeable amide hydrogens. The protection percentages for 20 μM and 250 μM samples were 3% and 47%, respectively. This result demonstrates that P_11-2 possesses a core that is resistant to deuterium exchange, recapitulating another important feature of the structural behavior of Aβ [29].

Figure 5.

Hydrogen-deuterium exchange of P_11-2 in solution: (A) 20 μM P_11-2; solid line, unlabeled control; dashed line, fully labeled control. (B) P_11-2 labeled at 20 μM for 10 s. (C) P_11-2 at 250 μM labeled for 10 s.

HX-MS experiments were also carried out for P_11-2 at low concentration (20 μM) and compared with fresh samples of Aβ(1-40). Figure 6 shows the relative abundance of unlabeled P_11-2 increases as a function of increasing lipid concentration. Similar results were obtained with fresh preparations of the Aβ peptide in absence (Figure 7B) and presence (Figure 7C) of the same negatively charged liposomes. In the absence of liposomes (Figure 7B) a fully solvent exposed species is accompanied by a protected species that we have observed previously and shown by SEC under identical conditions to be a mixture of monomer and dimer [29, 46]. Murphy and colleagues have also reported equilibrium mixtures of monomer and dimer at similar peptide concentrations [54]. While the quantitative distributions differ, the addition of lipid model membranes increased the fractions of both the Aβ and P_11-2 peptides that remained resistant to deuterium exchange. This indicates that P_11-2 models the behavior of Aβ quite well with regard to self-association at membrane surfaces.

Figure 6.

Representative mass spectra of P_11-2 at 20 μM bound to POPG liposomes at different lipid concentrations: (A) 0.005 mM; (B) 0.513 mM; (C) 1.19 mM. The corresponding mass spectrum in the absence of POPG is shown in Figure 5B.

Figure 7.

Representative mass spectra of freshly prepared Aβ(1-40) at 100 μM in the (B) absence and (C) presence of 1.19 mM POPG liposomes and incubated for 30 min. Deuterium labeling time was 10 s. In Panel A, mass spectra for unlabeled Aβ(1-40) (solid line) a fully labeled Aβ(1-40) (dashed line) in solution are shown for comparison. All spectra are normalized to their maximum signal intensity.

P_11-2 is known to form ion pores in lipid bilayers [33], which suggests that the peptide undergoes a conformational change on lipid binding. In order to more directly explore the conformation of P_11-2 bound to liposomes, we examined the peptide bound to supported bilayers by ATR-FTIR spectroscopy. Representative polarized FTIR spectra of 20 μM P_11-2 bound to POPC bilayers are shown in Figure 8A. The amide I’ band at 1600-1700 cm⁻¹ is sensitive to the conformation of the peptide. Bands at 1625-1640 cm⁻¹ are related to peptides adopting a β-sheet conformation, and bands at 1640-1650 cm⁻¹ are related to deuterated irregular (random coil) secondary structures [30]. The higher intensity band at around 1730 cm⁻¹ arises from the lipid ester carbonyl stretch vibration. The intensities of component bands were estimated from Gaussian curve fitting and indicated a β-sheet fraction of 10-20% and 80-90% irregular structures. For P_11-2 in the absence of liposomes, ATR-FTIR is too insensitive to assess secondary structure; however CD measurements showed no evidence of any β structure (Figure 8B). In addition, by using the online CD spectra deconvolutor (SOMCD http://geneura.ugr.es/cgi-bin/somcd/index.cgi), random coil was determined to be the major secondary structure, 96%, with β-sheet only 2% and turns and helix making up the remaining 2%. Negligible β structure is also consistent with previous solution measurements in the same concentration range [53]. It is challenging to be quantitative comparing small amounts of β structure by two techniques, but the results are consistent with an increase in β-sheet in the presence of liposomes.

Figure 8.

Secondary structure of P11-2 in presence and absence of liposomes. (A) Polarized ATR-FTIR spectra of P_11-2 at 20 μM bound to planar lipid bilayers composed of POPC with the polarizers oriented at 0 degrees from the plane of the ATR plate and membrane. (B) CD spectra of P_11-2 at 20 μM without liposomes.

4. Discussion

In this work we employed several biophysical techniques to examine the interactions of the rationally designed peptide P_11-2 with lipid model membranes. These were compared with similar studies of the Aβ(1-40) peptide, and the results showed that P _11-2 recapitulates several structural and functional features exhibited by Aβ(1-40). Binding studies employing fluorescent spectroscopy showed that the peptide P_11-2 binds to both zwitterionic and negatively charged membrane interfaces with a slight preference for the latter. The similar binding of P_11-2 to both zwitterionic and anionic liposomes indicates that hydrophobic interactions dominate the binding. However, the slight preference for negatively charged liposomes shows that electrostatic interactions also contribute in a minor fashion to this process. The present results showing the binding of P _11-2 with zwitterionic and negatively charged lipid agrees with several previous studies that showed that Aβ peptides bind both types of model membranes [12-18]. The similar binding behavior of the P_11-2 and Aβ(1-40) is consistent with their equivalent average hydrophilicity (−0.1) by the scale of Hopp and Woods [57] and their relatively low overall net charge (0 and −3, respectively).

The insertion of a Trp residue into a more hydrophobic environment is usually characterized by a fluorescence blue shift and by an increase in the fluorescence quantum yield. In certain cases such as with penetratin or TAT-PTD peptide [41,47], Trp fluorescence decreased while still exhibiting a blue shift. Here, though the fluorescence intensity decreased with liposome concentration, the wavelength of the maximum intensity was neither blue nor red shifted, but did not vary significantly beyond the measurement error of ± 1nm (Figure S2, Supplementary Information). This unusual intensity decrease could be due to conformational changes and/or self-association upon binding to liposomes. In particular, in-register self-association expected for P_11-2 [25] could lead to self quenching of Trp residues. Such in-register structures are well established for Aβ fibrils [48]. Special interactions with phospholipid headgroups and/or π-cation interactions between the Trp indole moiety and arginine in the peptide might also account for a reduction in Trp fluorescence [41,49]. Fluorescence quenching by acrylamide and iodide of the lipid-free and lipid-bound peptides showed that the presence of liposomes offers protection against quenching for Trps of P_11-2 and AMCA attached to Aβ. The protection might come from the insertion of the probes into the liposome surface, but also may come from the accelerated aggregation induced self-quenching.

It is useful to compare the K_SV values to those of other membrane disrupting peptides, such as mellitin. Because the K_SV values regressed here were not strictly corrected for inner filter effects, it is perhaps most appropriate to compare the ratio of K_SV values in the absence vs. presence of the liposome, R. This ratio might reflect the strength of the peptide-lipid binding interaction. For P_11-2, the values of R for the quenching in presence of POPC, POPC/POPG and POPG are 1.4, 1.5 and 1.6. For AMCA-Aβ, the R in presence of POPG is 2.3. In comparison for mellitin, the R in presence and absence of egg PC is 5.8 [55]. Thus, above results suggest that though Aβ and P_11-2 do interact with the lipids, their interactions are weaker than those of mellitin.

The binding of Aβ peptides to lipid model membranes is a first step towards their established ability to form ion channels and change the conductance and dielectric properties in planar lipid bilayers [34-38, 50-51]. In addition to the binding of P_11-2 to lipid bilayers and liposomes that has been demonstrated in this work, P_11-2 has been previously shown to also form pores and to display ion channel-like activity in planar lipid bilayers [33]. Our HX-MS studies show that both P_11-2 and Aβ(1-40) self-associate into larger aggregates under certain conditions and that these aggregates are protected from the deuterium-hydrogen exchange in lipid environments, probably by membrane-mediated self-association. It is also interesting to notice that there is overlap in degree of HX protection between the aggregates of P_11-2 in solution and the peptide contacted with liposomes. The HX-MS signal intensity for the former ranges from 1592.5 to 1605 m/z (Figure 5C) and the latter from 1592.5 to 1600 (Figure 6C). This overlap in protection might support the idea that the two aggregates may be related in structure. However, the wider, asymmetric labeling distribution exhibited by aggregates in solution suggests the presence of a wider distribution of structures, probably including those with similar solvent accessibility as the more protected population produced by the liposomes. These results are in agreement with and further generalize the fluorescence quenching results to include the labile protons in the P_11-2 and Aβ structures. Therefore, P_11-2 appears to recapitulate lipid binding, self-assembly, and pore formation of Aβ peptides in lipid model membranes.

P_11-2 at 20 μM in aqueous solution is below the critical association concentration of 45 μM and therefore occurs in a monomeric random coil conformation at this concentration. However, when P_11-2 binds to lipid bilayers, our ATR-FTIR spectroscopic studies show that some β-structure is induced. This indicates that the lipid bilayer promotes a conformational change and possibly self-association into larger β-sheet aggregates. Again, similar results were obtained with Aβ(1-40) [13,18,52]. Our HX-MS observations also show an increase in the amount of the protected, presumably self-associated, state in the presence of lipid for both P_11-2 and Aβ(1-40). This lipid-induced β-sheet formation is likely driven by hydrogen-bond formation between neighboring peptide strands in membrane interfaces that, if unsatisfied, would energetically be very unfavorable [53]. Forming higher oligomer β-sheets in membranes may be a prerequisite for P_11-2 to form ion-conducting pores in lipid bilayers. The membrane-induced oligomeric β-structures of P_11-2 may also be responsible for cell toxicity that we determined in this work to be depended on the state of self-assembly, very similar to what has been previously observed with Aβ.

In conclusion, the present work represents the first systematic study of the interaction of the P_11-2 model peptide with lipid model membranes. The results reveal that this model peptide recapitulates several key features of the Aβ amyloid peptide. Why might one want to study such a model peptide rather than Aβ itself? The main reason is an attempt to isolate general from sequence-specific mechanisms that have been implicated in Aβ toxicity. With regard to membranes alone, Aβ has been reported to bind and increase self-association on membranes [12,16,18], to form ion channels [34-37], and to increase membrane conductance [38, 50-51]. Oxidatively damaged lipids have also been purported to stimulate formation of fibrils [52]. While the neurotoxicity of Aβ(1-40) probably involves multiple mechanisms, it is important to know which of these are sequence-specific and which are general. Dissecting the nature and dominance of these diverse mechanisms can be difficult with competing effects and the complex behavior of Aβ. The relatively well-behaved model peptide P_11-2 and its validation in the present work should facilitate future investigations of some aspects of the toxicity of amyloid peptides, namely those that are general and less dependent on peptide sequence. In particular, well defined changes in sequence which alter oligomer size via the rules suggested by solution phase studies of these rationally designed peptides [27,33] could enable investigation of oligomer size on membrane permeability and neurotoxicity. Alternatively, changes in sequence that modulate membrane binding affinity would facilitate the effect of such changes on membrane-induced oligomerization, permeability and toxicity. Such studies will help us better understand the intrinsic ability of proteins and peptides to self-assemble at membrane surfaces and how such self-assembly may lead to pathological situations in amyloid diseases.

Abbreviations

AD

Alzheimer’s disease

Aβ

β-amyloid peptide

POPC

1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

POPG

1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-rac-1-glycerol [sodium salt]

DMPC

1, 2-dimyristoyl-3-sn-phosphatidylcholine

LUV

large unilamellar vesicle

SUV

small unilamellar vesicle

AMCA

[6-((7-amino-4-methylcoumarin-3-acetyl)amino) hexanoic acid)

TFA

trifluoro acetic acid

PBC

phosphate-borate-citrate

ATR-FTIR

attenuated total reflection – Fourier transform infrared spectroscopy

HX-MS

hydrogen-deuterium exchange mass spectrometry

DMSO

dimethylsulfoxide

DCA

dichloroacetic acid