An Aβ concatemer with altered aggregation propensities

Biochimica et Biophysica Acta 1804 (2010) 2025–2035

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / b b a p a p

An Aβ concatemer with altered aggregation propensities L. Giehm a,b,1, F. dal Degan c,2, P. Fraser d, S. Klysner e, Daniel E. Otzen a,b,⁎ a Interdisciplinary Nanoscience Centre (iNANO), Center for insoluble Protein Structures (inSPIN), Department of Molecular Biology, University of Aarhus, Gustav Wieds Vej 10C, DK-8000 Aarhus C, Denmark b Department of Life Sciences, Aalborg University, Sohngaardsholmsvej 49, DK-9000 Aalborg, Denmark c Pharmexa, Kogle Allé 6, DK-2970 Hørsholm, Denmark d Centre for Research in Neurodegenerative Diseases and Department of Medical Biophysics, University of Toronto, 6 Queen's Park Crescent West, Toronto, ON, Canada e Nordic Vaccine A/S, Fruebjergvej 3, DK-2100 Copenhagen OE, Denmark

a r t i c l e

i n f o

Article history: Received 12 March 2010 Received in revised form 25 May 2010 Accepted 28 June 2010 Available online 7 July 2010 Keywords: Amyloid Aggregation kinetics SDS Oligomer Membrane permeabilization Aβ peptide

a b s t r a c t We present an analysis of the conformational and aggregative properties of an Aβ concatemer (Con-Alz) of interest for vaccine development against Alzheimer's disease. Con-Alz consists of 3 copies of the 43 residues of the Aβ peptide separated by the P2 and P30 T-cell epitopes from the tetanus toxin. Even in the presence of high concentrations of denaturants or fluorinated alcohols, Con-Alz has a very high propensity to form aggregates which slowly coalesce over time with changes in secondary, tertiary and quaternary structure. Only micellar concentrations of SDS were able to inhibit aggregation. The increase in the ability to bind the fibril-binding dye ThT increases without lag time, which is characteristic of relatively amorphous aggregates. Confirming this, electron microscopy reveals that Con-Alz adopts a morphology resembling truncated protofibrils after prolonged incubation, but it is unable to assemble into classical amyloid fibrils. Despite its high propensity to aggregate, Con-Alz does not show any significant ability to permeabilize vesicles, which for fibrillating proteins is taken to be a key factor in aggregate cytotoxicity and is attributed to oligomers formed at an early stage in the fibrillation process. Physically linking multiple copies of the Aβ-peptide may thus sterically restrict Con-Alz against forming cytotoxic oligomers, forcing it instead to adopt a less wellorganized assembly of intermeshed polypeptide chains. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Alzheimer's disease (AD) is the most prevalent dementia disease affecting more than 5% of all people older than 65. The main hallmarks of AD are extracellular senile plaques (SP) and intracellular neurofibrillary tangles (NFT), both found in brain areas important for memory and cognition. The NFT are composed of the microtubule-associated protein Tau, a protein that stabilizes and promotes formation of axonal microtubules [1]. The major constituent of SP is the 39–43 residue amyloid β peptide (Aβ), an aberrant proteolytic fragment of the membrane protein amyloid precursor protein (APP). Aβ can form βsheet rich structures called amyloid fibrils, which in vitro form via a complex multi-step nucleation polymerization mechanism involving intermediate species. These are called Aβ derived diffusible ligands (ADDLS) [2] or protofibrils [3] and disappear upon fibril formation. ⁎ Corresponding author. Interdisciplinary Nanoscience Centre (iNANO), Center for insoluble Protein Structures (inSPIN), Department of Molecular Biology, University of Aarhus, Gustav Wieds Vej 10C, DK-8000 Aarhus C, Denmark. Tel.: + 45 89 42 50 46; fax. + 45 86 12 31 78. E-mail address: [email protected] (D.E. Otzen). 1 Present address: Copenhagen University, Department of Pharmaceutics and Analytical Chemistry, DK-2100 Copenhagen OE, Denmark. 2 Present address: Zealand Pharma, Smedeland 36, DK-2600 Glostrup, Denmark. 1570-9639/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2010.06.023

During the last few years, there has been a paradigm shift towards these pre-fibrillar species rather than the mature fibrils being the main cause of neurodegeneration [4,5]. Aβ's cytotoxicity is believed to be linked to their ability to form membrane permeable pores via an oligomeric state [6,7]. Current AD treatment only relieves some symptoms for a period of time for a subset of patients, and does not address the underlying pathologic process or substantially slow clinical progression [8]. Several curative strategies are being explored. One is to target the intermediate Aβ aggregates, since antibodies directed against these aggregates can rescue cells from the deleterious effects of protein aggregation [9]. However, while such an antibody reduced plaque load in AD-transgenic mice, it had no functional effect [10]. Another approach is to reduce the concentration of the Aβ in the AD patients using Aβ-specific antibodies (reviewed in [11]). This may be done by administering therapeutic antibodies to the patients (passive immunization) [12] or by vaccination with Aβ [13] or derivatives thereof (active immunization). The Aβ analog K6Aβ1–30[E18E19], designed to remove two T-cell epitopes which have been linked to the microencephalitis observed in a few cases when vaccinating with full-length pre-aggregated Aβ [14], was shown to improve cognition and reduces Aβ burden when used with an adjuvant suitable for humans, without increasing vascular Aβ deposits or microhemorrhages [15]. An 11-mer tandem repeat of Aβ1–6 also led

2026

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

to a predominantly IgG1 response, where the induced antibodies reacted strongly with Aβ plaques without inflammation-related pathology [16]. However, the structural properties of these constructs and their possible link to the extent of the immune response have not been reported. As Aβ is an endogenous peptide, the immunogenicity of an Aβ containing vaccine may be increased by fusing Aβ molecules or fragments thereof to known T-cell epitopes from the tetanus toxin, (P2 and P30), which are known to enhance immune responses [17]. Such a promising vaccine candidate is the protein Con-Alz, which contains three copies of the Aβ(1–43) sequence connected by P2 and P30 (see Fig. 1). The addition of epitopes P2 and P30 in the vaccine serves a double purpose: 1) P2 and P30 are potentially more efficient at activating Helper T cells (CD4+ T cells) than Aβ itself, leading to a potentially higher titer of Aβ-specific antibodies; 2) since P2 and P30 are immunodominant, there should be no response towards the T cell epitopes contained within Aβ upon vaccination with the concatemer, and thus no activation of Aβ-specific Th1 T cells and, in turn, no adverse auto-immune response. Thus, the Con-Alz construct is a potentially safer and more efficient vaccine than Aβ alone. In vivo the precursor protein APP is cleaved in different positions by the γ-secretase, leading to Aβ peptides of 38–43 residues [18]. Our Aβ construct represents the maximum possible length of the Aβ peptide to ensure that in its eventual use as a vaccine, all Aβ residues which may occur in vivo are exposed to the immune system. An obvious question is whether Con-Alz has inherited the fibrillogenic properties of Aβ. It could be expected that the juxtaposition of several highly fibrillogenic sequences would accelerate fibrillation since the rate-limiting nucleation step would become intra- rather than intermolecular. This has been shown to be the case for several other systems [19,20]. On the other hand, when two copies of one of the shortest fibrillating sequences known (KFFE) are linked together by a four-residue intervening sequence, the resulting dodecamer either remains as a monomeric random coil (if the linker is flexible) or forms a stable oligomeric β-hairpin (if the linker is a turn) [21]. An increased aggregation propensity would be a challenge to a reliable and reproducible formulation of the vaccine species due to both the variability in the relative amount of differentially aggregated species and the formation of potentially cytotoxic species. Here we show that even under strongly denaturing conditions Con-Alz has a very high propensity to form aggregates which slowly coalesce over time with changes in secondary, tertiary and quaternary structure. However, this strong aggregative behavior is not linked to pronounced vesicle permeabilizing abilities presumably because the linking together of Aβ concatemers sterically restrict Con-Alz against forming cytotoxic oligomers, forcing it instead to adopt a less well-organized assembly of intermeshed polypeptide chains.

2. Materials and methods 2.1. Materials Con-Alz (batch: LEP0149 D, suspended in 10 mM TRIS pH 8, purity N 90%) was kindly provided by Lundbeck A/S (Valby, Denmark). Western blots run together with SDS-PAGE showed that the protein formed higher order structures as well as monomers (see Results) but also confirmed that all protein bands were Con-Alz. 1,2-Dioleoyl-sn-glycero3-phosphocholine (DOPC), and 1,2-dioleoyl-sn-glycero-3-phospho-rac1-glycerol-sodium salt (DOPG) were from Avanti Polar Lipids. Aβ(42) was from American Peptide Company Inc. (Evelyn Ave, Sunny Vale, CA). All other chemicals were from Sigma-Aldrich (St. Louis, MO). 2.2. Buffer compositions at different pH values pH 1: 100 mM HCl, pH 2: 10 mM HCl, pH 3: 50 mM Glycine (adjusted with HCL), pHs 4 and 5: 50 mM sodium acetate, pH 6 and pH 7: 10 mM phosphate, pH 10: 50 mM Glycine (adjusted with NaOH), pH 10.8 50 mM Glycine (adjusted with NaOH). PBS: 10 mM sodium phosphate pH 7 150 mM NaCl. All buffers were filtered through a 0.22 μm filter. 2.3. Stock solutions of Con-Alz Con-Alz was dialyzed extensively against water, lyophilized and resuspended to a clear solution in (a) 15 mM SDS 10 mM phosphate pH 7 150 mM NaCl for CD measurements or (b) 5 M GdmSCN pH 10 for all other experiments. These two solvents lead to the smallest aggregate sizes; in addition it is stable against further aggregation over a time period of several days (data not shown). The high far-UV absorption of GdmSCN precludes its use for CD spectroscopy. Con-Alz concentration (~500 μM) was determined using an estimated ε280 of 10.81 mM−1 cm−1. 2.4. Fluorescence and absorption assays The affinity of Con-Alz towards Thioflavin T (ThT) was followed using a SpectraMax Gemini XS fluorescence plate reader (Molecular Devices, Sunnyvale, CA) or a LS55 Luminescence Spectroflourometer (Perkin Elmer, Wellesley, MA). Unless otherwise stated all measurements were conducted in triplicate at 25 °C with a total sample volume of 100 μl and a final concentration of 40 μM ThT and 20 μM Con-Alz 10 mM phosphate 150 mM NaCl pH 7. Contributions from ThT were subtracted. ThT excitation was at 450 nm with emission at 485 nm, Trp excitation was at 295 nm with emission at 330 and 350 nm. ThT was dissolved in 10 mM phosphate buffer pH 7 and the

Fig. 1. Primary structure of Con-Alz. The three white boxes represent three Aβ(43) peptides connected by the P2 epitope and P30 epitope (bold black line). Amino acid sequence of Con-Alz is illustrated below, where bold black letters denote amino acids in the Aβ(43) black letters P2 epitope and italic black letters are the P30 epitope, underlined letters are amino acids incorporated for cloning purposes.

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

2027

concentration was measured by absorbance at 412 nm with ε412 = 36.000 M−1 cm−1 [22]. For fluorescence plate reader experiments, data points were collected every 30 min, and each reading was preceded by 300 s of auto-mixing. To prevent evaporation during long incubations, the plates were covered with Crystal Clear Sealing Tape (Hampton Research, Aliso Viejo, CA).

2.11. Determination of reaction order

2.5. Turbidity measurements

where k is the rate constant. Thus by measuring the initial aggregation rate v at several concentrations of Con-Alz, we obtain n by plotting log (v) versus log[Con-Alz]. The Con-Alz concentration was varied from 5 to 40 μM in 10 mM phosphate pH 7 with 150 mM NaCl and affinity towards ThT was followed using a plate reader.

Measurements were performed at room temperature using a UVIKON 943 Double Beam UV/VIS Spectrophotometer (BIO-TEK KONTRON, Milano, Italy). The absorbance at 320 nm was used as an estimate of aggregated material. Contributions from solvent were subtracted. 2.6. Aggregation of Con-Alz in denaturants A 20 μM Con-Alz was diluted in freshly prepared samples of 0–8 M urea in 10 mM phosphate pH 7 in 150 mM NaCl. Trp emission was measured from 305 to 400 nm (with excitation at 295 nm) using a bandwidth of 5 nm for excitation and emission, scan-speed 200 nm/ min and averaging five spectra. At 72 h the samples were incubated with 40 μM ThT and the emission recorded. 2.7. Circular dichroism Far-UV CD spectra of 20 μM Con-Alz were recorded on a JASCO J810 Spectropolarimeter (Jasco Spectroscopic Co. Ltd, Hachioji City, Japan) using a 1 mm path-length cuvette. Unless otherwise stated the ellipticity was measured in the wavelength range of 250–190 nm with the following settings: resolution 0.2 nm; bandwidth 1.0 nm; sensitivity 50 mdeg; response 2.0 s; speed 50 nm/min. Ten accumulations were averaged to yield the final spectrum from which a spectrum of corresponding buffer/solvent was subtracted. 2.8. Secondary structure estimations Estimation was done using the program Jasco 2nd Structure estimation Program. This program is based on the method of least squares, using a far-UV CD reference set of Yang et al. [23]. The extent of agreement between the calculated CD spectrum and the measured CD spectrum was evaluated using the method of normalized standard deviation (NRMSD) [24]. If Jasco 2nd Structure Estimation Program did not provide an acceptable estimation (NRMSD N 0.2 and defective areas in the residual plot), the neural network based algorithm k2d was used [25]. 2.9. Congo Red assay A 1 ml 20 μM Con-Alz at pH 7 10 mM phosphate 150 mM NaCl was incubated in triplicate at room temperature. At ~0, 24 and 72 h 3 × 100 μl sample was taken and separately incubated with 10 μM Congo Red (CR) for 20 min before absorbance measurements (400– 600 nm). Buffer contributions were subtracted. 2.10. ANS assay The affinity of ANS towards Con-Alz was measured using the fluorescence plate using an ANS excitation wavelength of 350 nm with emission at 480 nm. ANS was dissolved in 1 ml DMSO and diluted to 100 ml with MilliQ water. The concentration was determined spectrophotometrically using a molar extinction coefficient of 4,990 M−1 cm−1 [26]. The final concentration of ANS was 10 μM, 20 μM Con-Alz, pH 7 10 mM phosphate 150 mM NaCl.

The reaction order (n) of aggregation of Con-Alz was determined from the relationship n

ν = k½Con
Alz ⇔ logðνÞ = n log ½Con
Alz + log k

ð1Þ

2.12. Field flow fractionation with multi-angle laser light scattering (FFF-MALLS) FFF-MALLS [27] was performed on an Eclipse F™ (Wyatt Technology, Santa Barbara), light scattering (LS) detector Wyatt miniDawn with three detectors at 41°, 90° and 139° (Wyatt Technology). Pump, degasser and separation system was controlled by Eclipse version 2.2.24. Prior to analysis samples of 0.65 mg/ml were centrifuged for 5 min at 13,000 ×g. After injection of the sample it was focused for 3 min using a balanced flow from each ends of the chamber of 2 ml/min followed by cross flow of 2 ml/min for 1 min, and afterwards a channel flow of 2 ml/min with a gradient of cross flow from 2 to 0 ml/min over a period of 14 min. Measurements were performed at 25° C. Processing of light scattering data was made by Astra software version 4.90.07 obtaining the weight-average molar mass (Mw), z-average radius of gyration (rg)z and Degree of Polydispersity (D.O.P (Mw/Mn where Mn is the number-average molar mass)). Peaks were integrated and compared with the amount of protein added to estimate recovery. 2.13. Vesicle perforation assay 2.13.1. Lipids A 12.5 mg of DOPG and DOPC was dissolved in methanol to a homogenous mixture followed by evaporation overnight by a rotary evaporator (Büchi Rotavapor R-114, Switzerland). The dried lipid films were re-suspended at room temperature in a pH 7 150 mM NaCl buffer containing 40 mM calcein to a final lipid concentration of 12.5 mg/ml. The lipid suspensions were freeze-thawed 10 times using liquid nitrogen and a water bath of 40 °C and extruded 12 times through a 100 nm polycarbonate membrane (Nucleopore filters, Whatman, Madison, UK and Northern Lipids, Vancouver, BC, Canada) to obtain unilamellar vesicles with size of ~ 100 nm. To remove non-encapsulated calcein, the vesicles were passed through a PD10 desalting column (Sephadex G-25M, Amersham Biosciences, GE Healthcare, Hilleroed, DK). A 500 μl lipid suspension was diluted to 2.5 ml and loaded on the gel filtration columns, which had been pre-equilibrated with pH 7 150 mM NaCl. The vesicles were eluted with 3.5 ml pH 7 150 mM NaCl, samples were collected of ~ 0.5 ml obtaining a total volume of 3.5 ml. These fractions were then examined for the presence of vesicles by observing changes in fluorescence upon addition of 0.1% (w/v) Triton-X. A 5 μl of each lipid/calcein aliquot in a total volume of 100 μl (pH 7 150 mM NaCl) was examined. Samples that exhibited a change in calcein fluorescence (excitation at 490 nm and emission at 520 nm) were pooled (fractions 1–5) and adjusted to 1 mg/ml, assuming no lipid loss during desalting. 2.13.2. Perforation of vesicles by Con-Alz Calcein fluorescence data points were collected on a plate reader every 30 min and each reading was preceded by 300 s of auto-mixing. Plates were covered with clear sealing tape. Three different lipid/ protein molar ratios (5:1, 1:1 and 0.2:1) were prepared with a concentration of 20 μM Con-Alz using 100% DOPG, 100% DOPC and

2028

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

20:80% DOPG/DOPC vesicles. Negative controls were vesicles lacking Con-Alz and positive controls were vesicles with added Aβ(42) and with 0.1% (w/v) Triton X, all in triplicates. Experiments were carried out in 10 mM phosphate buffer pH 7 150 mM NaCl. The degree of released calcein was estimated using the following equation:

RC% =

FðtÞ−Fð0%Þ Fð100%Þ−Fð0%Þ

ð2Þ

where RC% is the fluorescence of released calcein in percent, F(0%) is a negative control and F(100%) is the maximum fluorescence corresponding to the release of calcein after addition of 0.1% (w/v) Triton X. F(0%), F(100%) and F(t) are the average of triplicates. 2.13.3. Dynamic light scattering (DLS) The DLS measurements were made using a Dyna Pro MSTC Dynamic Light Scattering instrument (Protein Solutions Inc., Charlottesville, VA). Three repetitions at each sample were carried out. The samples were filtered through a 0.22 μm filter to remove any dust particles and injected into a 50 μl quartz cuvette (Hellma, Plainview, NY, USA). The sample was illuminated by a miniature 808.4 nm laser. Forty pulses were accumulated at 25 °C. Data was analyzed with the program Dynamic (version 5.24.02). 2.13.4. Transmission electron microscopy analysis Purified and lyophilized Con-Alz was resuspended at 1.0 mg/ml in 10 mM Tris, pH 8.0. Immediately prior to analysis, aliquots were diluted into PBS to a concentration of 0.25 mg/ml and incubated at room temperature. Morphology of the aggregates was examined by transmission electron microscopy over a period of 10 days. For negative staining, carbon-coated polyvinyl butyryl grids (Pioloform (R), SPI Supplies, West Chester PA), were floated on the protein solutions, blotted and air-dried. The samples were stained with 1% (w/v) phosphotungstic acid (pH 7,0) and examined using a Hitachi H7000 electron microscope with an accelerating voltage of 75 kV. 3. Results 3.1. Con-Alz undergoes slow conformational changes over time without a lag phase to form structures binding ThT, ANS and Congo Red to a smaller extent than Aβ As an initial experiment to test the fibrillogenic properties of ConAlz relative to its mother peptide Aβ(42), we incubated Con-Alz in the presence of ThT which displays a 10–1000 fold increase in emitted fluorescence when bound to amyloid [28]. Time profiles of 60 μM Aβ(42) and 20 μM Con-Alz (corresponding to 60 μM Aβ-monomers) in 10 mM phosphate pH 7 and 150 mM NaCl showed important differences (Fig. 2A). Aβ's aggregation gave rise to a sigmoidal kinetic curve, with a lag phase of 5 h typical of the nucleation-dependent polymerization process that underlies amyloid formation. In contrast, Con-Alz showed an increase in fluorescence without lag-phase which levelled out after 30 h at 40% of the Aβ level. In contrast to Aβ(42), Con-Alz had affinity towards ThT from the very start of the experiment and this affinity only increases around 50% during the experiment. Con-Alz's lack of a lag-phase could arise for several reasons: (a) nucleation within the sample preparation time (~15 min) and subsequent fibrillation, (b) fibrillation from the monomeric state without a lag phase (as seen for e.g. acylphosphatase [29]), and (c) formation of an amorphously aggregated state during purification, or within the sample preparation time, which subsequently rearranges to a state which has higher ThT affinity but is not necessarily fibrillated.

Fig. 2. (A) Change in ThT fluorescence as a function of time when incubated with 20 μM Con-Alz and 60 μM Aβ(42) at pH 7 10 mM phosphate 150 mM NaCl. (B) Absorption spectra of Congo Red in the absence and presence of 20 μM Con-Alz after incubation of 0, 24 and 72 h in 10 mM phosphate pH 7 150 mM NaCl. (C) Change in ANS fluorescence as a function of time when incubated with 20 μM Con-Alz and 60 μM Aβ(42) at pH 7 10 mM phosphate 150 mM NaCl.

In order to distinguish between these possibilities, we need to characterize the species accumulating during incubation. ThT does not discriminate absolutely between fibrils and more amorphous aggregates, as shown for e.g. α- SN [30] and human growth hormone (D.E.

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

O., unpublished observations). For further insight, we therefore incubated Con-Alz with Congo Red (CR) a compound which binds amyloid fibrils with a shift in its absorption spectrum [31]. Spectra of Con-Alz incubated in 10 mM phosphate 150 mM NaCl pH 7 for several days and subsequently mixed with CR, revealed a red shift in λmax of ~ 10 nm within the first 24 h which did not change further over the next 72 h (Fig. 2B). Likewise, we saw an increase in absorbance at 24 h which remained constant up to 72 h. The magnitude of increase in absorbance and the red shift for Con-Alz is significantly less than those published for Aβ, where a red shift of 40–60 nm is observed [31]. Thus, aggregated Con-Alz is most likely less ordered than Aβ fibrils. Further insight was provided by incubating the proteins with the dye ANS which binds with particular affinity for contiguous hydrophobic patches with an accompanying increase in fluorescence [32]. Con-Alz' ANS time-profile showed an initial increase similar to the ThT-profile; however after 40 h Con-Alz's fluorescence decreased to a level slightly below the starting level suggesting additional structural changes (Fig. 2C). Con-Alz had a much higher starting fluorescence value than Aβ and remained more fluorescent throughout, implying that more hydrophobic area was exposed in Con-Alz than in Aβ. This could arise from the formation of less structured fibrils or aggregates. The P2 sequence is quite hydrophilic and P30 is only slightly more hydrophobic. Their average Kyte-Doolittle hydropathy values [33] are −0.293 and 0.065, respectively, which is significantly less than the values for Aβ(43) (0.271) and the C-terminal 26 residues of Aβ (1.164). Therefore they are unlikely to give rise to ANS binding by themselves. Thus the ANS data suggest that Con-Alz is irregularly structured from the very start of our measurements.

3.2. Con-Alz shows a broad size distribution from the onset which decreases in polydispersity over time due to accretion To determine the size distribution of Con-Alz at the start of our experiment in 10 mM phosphate 150 mM NaCl pH 7 and its evolution over time, we initially used dynamic light scattering. The sample is polydisperse right from the onset. Monomeric Con-Alz is estimated to have a hydrodynamic radius Rh of ~ 2–4 nm [34]. Over 72 h there is a clear shift towards larger particles, with the smallest particles at 0 h having an Rh of 4.9± 0.9 nm and at 72 h 91.5 ± 3.2 nm, while the size of the largest aggregates increased from 101.6 ± 6.2 nm at 0 h to 520 ± 21.5 nm at 72 h (Table 1). This agrees with the increase in turbidity (see below, including Fig. 5B insert). The samples older than 72 h were turbid, preventing detection of smaller species. For further resolution we turned to field flow-fractionation (FFF), a matrix-free device which separates particles according to size coupled on-line with a multi-angle laser light scattering (MALLS) system and UV- and refractive index detectors [35]. The FFF-MALLS analysis in 10 mM phosphate 150 mM NaCl pH 7 (Fig. 3) shows a clear increase in the distribution of molecular weights between 0 and 72 h, with no particles below 106 g/mol observed after 72 h. The degree of polydispersity decreases from 10.5 ± 3.9 at 0 h to 2.76± 2.1 at 72 h as the smaller particles accrete to form larger structures.

Table 1 Estimated Rh of 20 μM Con-Alz in 10 mM phosphate pH 7 and 150 mM NaCl at 0, 24 and 72 h based on dynamic light scattering (DLS). Each Rh is the mean of a peak or sub-peak. 0h

Peak Peak Peak Peak

1 2 3 4

24 h

72 h

Rh (nm)

Intensity (%)

Rh (nm)

Intensity (%)

Rh (nm)

Intensity (%)

4.9 ± 0.9 21 ± 3.1 101.1 ± 6.2 –

12 47 41 –

18.2 ± 5.1 130.1 ± 9.8 330 ± 14.5 480.7 ± 21.5

8 15 26 51

91.5 ± 3.2 131.5 ± 11.7 334 ± 34.8 520 ± 21.5

6 18 17 59

2029

Fig. 3. The cumulative mass distribution of 0.65 mg/ml Con-Alz at 0 and 72 h determined by Field-Flow Fractionation at 25 °C in 150 mM NaCl and 10 mM sodium phosphate pH 7.

3.3. The Con-Alz aggregates are non-fibrillar but show elements of ordered structure which form in a first-order reaction before precipitating Full-length human Aβ peptide (residues 40 or 42) assembles into classical amyloid fibrils following incubation from 24 to 72 h depending on concentration and temperature. These structures are long unbranched fibers which are ~ 80 Å in diameter and have an underlying filament substructure (Fig. 4A). In contrast, the Con-Alz peptide was unable to form typical amyloid fibrils after prolonged incubation. After 4 days of incubation, largely amorphous aggregates were observed which were accompanied by small particulate structures (Fig. 4B). These appeared in some cases to be semi-fibrous structures approximately 5–7 Å in diameter. These resembled fragments of protofibrillar aggregates which are observed during the early transition of Aβ from monomeric to amyloid fiber. Following extended incubation (10 days and greater), Con-Alz did undergo a slow conversion to protofibril-like aggregates although very short and fragmented (Fig. 4C). The electron microscopy observations indicate that the Con-Alz construct readily forms aggregates similar to those observed in the early stages of amyloidogenesis. However, these are unable to convert into amyloid fibrils which are normally seen for the Aβ peptide. It is possible that the semi ordered structure formed by Con-Alz could arise from kinetic partitioning in which relatively unspecific contacts between mostly unstructured protein molecules form more quickly than those which require prior formation of fibrillogenic folds at the monomer level [36]. Lowering the protein concentration might favor the more ordered pathway. However, ThT fluorescence experiments at 5–40 μM Con-Alz in 10 mM phosphate 150 mM NaCl pH 7 all show essentially the same time profile (Fig. 5A insert). A plot of the log of the initial velocity of fibrillation versus log[Con-Alz] reveals a slope of 1.1 ± 0.1 indicating a first order reaction (Fig. 5A). However this does not necessarily mean that the aggregation or fibrillation process does not include higher order reactions describing formation of dimer/tetramer/oligomer etc., but rather that the overall rate limiting step in the process monitored by ThT does not involve a change in molecularity. This contrasts with Aβ, where formation of higher-order species is rate-limiting [37]. Instead, it is possible that the initially formed aggregate simply undergoes an internal rearrangement. We thus conclude that of the 3 scenarios suggested to account for the lack of a lag phase, only the scenario involving preformed aggregates which rearrange over time to form semi-ordered structures is compatible with our data.

2030

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

Fig. 4. Negative-stain transmission electron microscopy images of Aβ(42) and Con-Alz. (A) 0.25 mg/ml Aβ(42) displays standard fibrillar morphology after 48 h incubation. (B) Con-Alz peptide at 4 days of incubation with amorphous and particulate aggregates. (C) Con-Alz after 10 days incubation shows an increased density of aggregates with a more protofibril-like morphology. The scale bar is 250 nm.

The CD spectrum of fresh Con-Alz in 10 mM phosphate pH 7 and 150 mM NaCl has a global minimum at 218 nm, strongly indicating βsheet structure (Fig. 5B). Over 8 days the global minima of the CD spectra only shifted slightly, but a decrease in signal intensity was observed (Fig. 5B). The signal from Con-Alz decreased between 48 and 72 h, reaching a plateau at 120 h. Visible aggregates appeared at 96 h for

Con-Alz in parallel with the decrease in residue ellipticity at 218 nm and an increase in absorbance measured at 320 nm (Fig. 5B insert). This suggests that the decrease in CD-signal intensity after 48 h arises from precipitation of Con-Alz rather than intramolecular changes.

3.4. The affinity of ThT towards Con-Alz is not prevented at elevated pH The fibrillation of Aβ is inhibited above pH 9 [28,38] and significantly retarded at pH 1 [39]. However, over the entire pH range 1–10, the time profile of Con-Alz aggregation shows an increase in emitted fluorescence of ThT (Fig. 6 insert) although the emitted fluorescence at pH 10 is significantly lower than at pH 1 and 7, related to the reversible isomerization of ThT at alkaline pH [40]. Clearly some kind of aggregated structure is present in all samples independent of the pH. Due to the pH-titration of CR below pH 5, we did not use this dye to investigate the structure of these Con-Alz aggregates. Far UV CD spectra showed that all samples had global minima at 218 nm, strongly indicating β-sheet structure (Fig. 6), although the CD-signal intensity of the samples at pHs 1 and 10 was decreased compared to pH 7. This may be due to an increase in random coil and turn but structure estimation programmes failed to successfully deconvolute these spectra (data not shown). According to DLS, the sizes of the species found at pHs 7 and 10 were in the same order of magnitude, whereas the sizes of the species found at pH 1 were smaller, indicating that acidic pH affects the oligomerization process to a higher extent than basic pH (Table 2).

Fig. 5. (A) Initial velocity of the aggregation process of Con-Alz as function of protein concentration. Insert: time profiles of ThT fluorescence in 5–40 μM Con-Alz. (B) CD spectra of 20 μM Con-Alz in 10 mM phosphate 150 mM NaCl pH 7 at time 0 and 192 h. Insert: change in ellipticity and turbidity at 320 nm over time.

Fig. 6. Far-UV CD spectra of 40 μM Con-Alz in pHs 1, 7 and 10 at 25 °C. Insert: change in ThT fluorescence as a function of time when incubated with Con-Alz at pHs 1, 7 and 10.

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035 Table 2 Estimated Rh of 40 μM Con-Alz at pHs 1, 7 and 10 based on DLS. Each Rh is the mean of a peak or sub-peak. pH 1 Rh (nm)

pH 7 Intensity (%)

Peak1 3.4 ± 0.9 8 Peak2 18 ± 3.0 51 Peak3 91.5 ± 6.2 41 Peak4 – –

pH 10

Rh (nm)

Intensity (%)

Rh (nm)

Intensity (%)

6.9 ± 1.5 49.6 ± 8.9 131.5 ± 11.7 334 ± 34.8

11 13 11 65

5.8 ± 1 28.2 ± 5.5 112 ± 28 –

41 13 46 –

3.5. Urea leads to conformational changes, but does not abolish Con-Alz aggregates The strong propensity of Con-Alz to aggregate prompted us to investigate whether different solvents could monomerize Con-Alz and thus provide a method to investigate early precursors of the ConAlz-aggregates using far UV CD and Trp fluorescence. We first investigated the chemical denaturant urea at pH 7 10 mM phosphate 150 mM NaCl. As shown in Fig. 7A, an 8 nm red shift in the Trp emission peak was observed for Con-Alz dissolved in urea, suggesting that the single Trp found in the P30 loop is buried in the absence or at

2031

low concentration of denaturant, whereas in high denaturant concentration it is solvent-exposed. Despite the superficial resemblance of Fig. 7A to conventional unfolding of globular proteins, the Con-Alz system is not in equilibrium, since measurements were done at time zero and change over time. The Con-Alz consists of an ensemble of different sizes of species with different amounts of secondary structures, depending on the oligomerization state. The secondary structure remained constant from 0 to 6 M urea, followed by a steep decrease in CD-signal intensity from 6 to 8 M and a plateau from 8 to 10 M urea. λmax increased slightly from 0 to 2.5 M, followed by a plateau from 2.5 to ~ 5 M and increasing significantly from 5 to 8 M followed by a plateau at 8–10 M. We made similar observations in terms of changes in secondary structure and λmax when we used GdmCl instead of urea (data not shown). The slight increase in λmax from 0 to 2.5 M could be a consequence of urea altering the oligomerization without affecting the secondary structure elements, since the far UV CD was approximately constant from 0 to 6 M. Thus the species formed in 2.5–5 M urea could be monomers/ dimers/small oligomers still able to fold in essentially the same way as the species in 0 M urea. Clearly the size distribution is an important factor in discussing Con-Alz behavior in denaturants. We therefore analyzed the size distribution of Con-Alz in 0, 4 and 8 M urea. Size exclusion chromatography showed that there was a dynamic equilibrium between species of different sizes, since individual peaks which were collected and re-injected led to a multitude of peaks including some of larger molecular weights than the original fraction (data not shown). However, there were also peaks corresponding to apparent molecular weights well below monomeric Con-Alz (data not shown). This probably arose due to interactions between Con-Alz and the chromatographic material. For more reliable data, we therefore turned to Field Flow Fractionation, where the sample flows through a channel lined with inert material. At all 3 concentrations, we see a broad size distribution (Fig. 7B). Interestingly, 4 M urea produced the smallest species while 0 and 8 M urea had more or less the same cumulative mass distribution. Increasing denaturant in FFF increased the recovery of injected protein, leading to full recovery in 8 M urea (Table 3). Samples in the range 5–20 kDa, (corresponding to a Con-Alz monomer), made up 0% of the total mass in 0 and 8 M urea but 1.4% in 4 M urea. In 8 M urea, the smallest species found were in the range 200–400 kDa corresponding to 10–22-mers of Con-Alz and they constituted 2.0% of the entire mass. In 4 M urea only 30% of the total mass was in the 103–105 kDa range; the corresponding figures in 8 and 0 M urea are 74% and 70%. This tendency is likewise reflected in the values of the mean square moments radii, which are 163.0 and 169.6 nm in 0 and 8 M urea respectively, but only 56.9 nm in 4 M urea. Thus, these data indicate that 4 M urea prevents oligomerization better than 8 M urea. In addition the polydispersity was lower in 4 M urea than in 0 and 8 M urea (Table 3). We also tested the affinity of ThT towards Con-Alz as a function of denaturant concentration (Fig. 7A insert), and found that ThT emission intensity decreased by a factor of only ca. 2 at high concentrations of urea. This indicates—in line with size determination experiments—that Con-Alz retains structure even at high denaturant concentration.

Table 3 The weight average molar mass Mw, z-average radius of gyration (rg)z and Degree Of Polydispersity (D.O.P) and recovery percentage of Con-Alz in 0, 4 and 8 M urea determined using FFF. Errors on Mw and rz are ~ 10%. Fig. 7. (A) The residue ellipticity at 218 nm and the emitted fluorescence of Trp as a function of urea. Insert: affinity of ThT towards Con-Alz in 0–8 M urea, illustrated by the fluorescence emission intensity of ThT at 72 h for each denaturant concentration. (B) Cumulative mass distribution of 0.65 mg/ml Con-Alz in 0–8 M urea measured by Field-Flow Fractionation at 25 °C.

[Urea] (M)

Mw (g/mol)

(rg)z (nm)

D.O.P. (Mw/Mn)

Recovery (%)

0M 4M 8M

1.21e + 07 1.02e + 06 6.84e + 06

163.0 56.9 169.6

10.51 ± 3.9 3.91 ± 1.79 5.05 ± 2.99

45.0 69% 99%

2032

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

3.6. Con-Alz undergoes structural changes in TFE but retains aggregative properties In view of the strong propensity of Con-Alz to form large molecular species even in strongly denaturing conditions, we turned to other solvents which might disaggregate Con-Alz. The fluorinated alcohol HFIP is generally used as a reliable defibrillating solvent (cfr. [41]), but we were unable to dissolve Con-Alz in this solvent. When incubated in 10–100% TFE, a solvent which normally is a strong inducer of α-helical structure, Con-Alz showed increasing ellipticity but no overall deviation from the typical β-sheet spectrum with a minimum around 218 nm (Fig. 8A insert). It was not possible to deconvolute spectra at TFE concentrations below 50% TFE to obtain estimates of secondary structure contents. Between 50% and 100% TFE, the α-helix content remained around 22–24% and coil content around 18%, while β-content slightly decreased from 60% to 52% with a parallel rise in turn content. The estimated α-helix content of Con-Alz is low compared to Aβ, where 60% TFE induces 80–90% α-helicity [42]. Over time, structural changes occurred for Con-Alz in the presence of 50% TFE at about the same rate as in the absence of TFE (Fig. 8A). Although the starting CD-signal intensity was higher in the presence of TFE, Con-Alz ended with the same final ellipticity (Fig. 8A) and same ability to bind ThT (data not shown) irrespective of TFE content, indicating that the structural changes induced by TFE do not prevent Con-Alz's aggregation.

3.7. Micellar SDS prevents Con-Alz aggregation Having failed to curb Con-Alz aggregation with fluorinated alcohols, we tried anionic detergents. At 100 mM, which is well above the critical micelle concentration (around 0.5–5 mM, depending on buffer conditions), SDS can solubilize mM concentrations of Aβ [43,44]. In contrast, sub-cmc concentrations increase Aβ fibrillation rates [45,46] and stabilize β2-microglobulin fibrils [46]. We carried out experiments with Con-Alz in SDS in 10 mM sodium phosphate pH 7 only; we omitted NaCl in order to raise the cmc from 0.67 mM (10 mM sodium phosphate, 100 mM NaCl) to 5.2 mM in 10 mM sodium phosphate (measured using an ANS assay [47]). Since we were able to dissolve 500 μM Con-Alz completely in 20 mM SDS and could dilute it no more than 25 times, the 10 mM phosphate buffer allowed us to straddle the cmc for our experiments. The far-UV CD spectra of Con-Alz in different SDS concentrations retained the same overall appearance (Fig. 8B insert), although there was a slightly increase in CD-signal intensity with increasing SDS concentrations, similar to results with TFE. However, at concentrations above cmc, no visual changes or decreases in CD-signal intensities were observed over the same period (144 h) where significant changes in TFE had been seen. In contrast, at the sub-cmc concentration of 2 mM SDS, we saw a time dependent CD-signal intensity decrease (Fig. 8B) which also led to turbidity increases (Fig. 8B) and visual particles after 144 h. In addition, there was no detectable binding of ThT by any Con-Alz samples incubated at 10 and 67 mM SDS, but clear binding at 0.6 and 2 mM (data not shown). In contrast to the powerful solubilizing effect of SDS, the non-ionic detergent dodecyl maltoside had no effect on secondary structure and did not affect aggregation behavior (data not shown). Incubation with the positively charged detergent LTAC led to immediate precipitation, possibly due to the overall negative charge of Con-Alz at pH 7. 3.8. Ability of Con-Alz to perforate vesicles

Fig. 8. (A) The residue ellipticity at 218 nm as a function of time at 0% TFE (10 mM phosphate pH 7 150 mM NaCl) (open circles) and 50% (v/v) TFE (filled circles). Insert: Far UV CD-spectra of 20 μM Con-Alz. The black arrow denotes increase in TFE, where the lowest CD-signal correspond to 0% TFE and the highest CD-signal to ~ 100% TFE. (B) Residue ellipiticity at 218 nm and turbidity at 320 nm for Con-Alz in 2 mM SDS as a function of time. Insert: Far UV CD-spectra of 20 μM Con-Alz in 0.6, 2, 10 and 67 mM SDS. Black arrow indicates increase in SDS-concentration.

In view of the interest in using Con-Alz as a vaccine, it is relevant to examine whether this protein has inherited the cytotoxic properties of their mother-peptide Aβ. We have tried to study this using the release of calcein, a fluorescence probe, entrapped in vesicles at selfquenching concentrations (40 mM). If Con-Alz perforates the vesicles at pH 7 10 mM phosphate 150 mM NaCl, leakage of calcein will occur, causing a dilution of calcein in the sample volume and giving rise to an increase in fluorescence [48]. We used lipids of different net change, namely 100% DOPC (zwitterionic), 100%DOPG (anionic), 80:20% DOPC/DOPG, and 95:5% DOPC/SA (cationic). Three different protein/lipid molar ratios (0.2:1, 1:1 and 5:1) were investigated. The two first ratios did not result in any perforation of the above described vesicles, nor did the lipids affect Con-Alz's affinity for ThT (data not shown). Only an unphysiologically high protein/lipid molar ratio of 5:1 led to a perforation of vesicles containing 80:20% DOPG/DOPC and 100% DOPG by Con-Alz (Fig. 9A). However the ThT binding pattern in these vesicles did not differ significantly from that in the absence of lipid, indicating that the lipid does not stimulate aggregation (Fig. 9B). Note also that the calcein release was extremely slow, occurring over a time-scale of many hours in contrast to the almost instantaneous (second–minute) release of calcein by bona fide oligomers [49]. For Con-Alz, the release of calcein included a lag phase, suggesting that Con-Alz needed to adopt a certain conformational structure or oligomeric state, before it could perforate the negatively charged vesicles. This lag phase was followed by an increase that led to a plateau for 20%DOPG, and for 100%DOPG the calcein release continued over 70 h. Preincubation of Con-Alz in 10 mM phosphate pH 7 150 mM NaCl for 24–72 h prior to mixing with calcein resulted in a significant decrease in perforation ability (Fig. 9B insert). During this pre-incubation period, the size distribution of Con-Alz species shifts towards higher order aggregates which clearly have a reduced ability

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

2033

why Con-Alz does not form amyloid fibrils in the same manner as Aβ could be due to the fact that the amorphous aggregation pathway is kinetically favored, or that Con-Alz is hindered from forming amyloid fibrils due to its primary structure. 4.2. Proposed model for aggregation of Con-Alz

Fig. 9. (A) Perforation of 100% DOPG and 80/20% DOPC/DOPG vesicles measured by release of entrapped calcein for 20 μM Con-Alz. (B) Emission of ThT measured in the presence of 100% DOPG, 80/20% DOPC/DOPG and buffer alone (pH 7, 150 mM NaCl, 10 mM phosphate). Protein/lipid molar ratio 0.2:1 and each curve is an average of three curves. Insert: release of calcein by Con-Alz from 100% DOPG vesicles, using 0.2:1 lipid/ protein molar ratio Each data-point is an average of three samples, with its standard deviation. Blank was subtracted. The release was measured over a period of 72 h, where the x-value of each data-point corresponds to the time the protein was incubated before vesicles were added, and the y-value corresponds to the release of calcein after 72 h.

Work by the groups of Tycko and Morimoto have led to a model for Aβ fibrils [50–52], in which each Aβ molecule forms two β-strands; the β-strand containing the hydrophobic C-terminal is in proximity to another β-strand likewise containing the hydrophobic β-strand, leading the cross β-unit with two Aβ molecules which stack on another cross-β-unit in an infinite propagating arrangement. This arrangement induces favorable hydrophobic interactions between residues in the two Aβ molecules. Such an arrangement between the individual Aβ molecules would be very difficult to construct in ConAlz due to linking of three Aβ, unless the bridging sequences P2 and P30 allowed three Aβ sequences to stack upon each other. This would allow them to form one half of the equivalent to three cross-β-units on top of each other, but the high intra-molecular cooperativity needed to form such a structure makes it a very unlikely arrangement. Definite conclusions on the nature of the Con-Alz structure will require detailed information, which could be provided by e.g. solid state NMR. However, such studies typically require highly organized samples [50]. The lack of regular higher order structures revealed by TEM makes this a very challenging project. The aggregation process of Con-Alz is a first order reaction. How could this form the basis for aggregation? We speculate that at least one Aβ sequence (or part thereof) in a Con-Alz molecule interacts specifically with another Aβ sequence in another Con-Alz molecule, forming the parallel β-sheet. Similar interactions involving the remaining parts of the molecule would lead to a criss-crossed mesh of joined molecules rather than a simple one-dimensional fibril. The rate-limiting step in this Con-Alz aggregation model could then be the individual Con-Alz molecule adopting a conformation that allows a part of the entire molecule to form parallel β-sheet, which in turn would give a rate limiting step corresponding to a first order dependence of Con-Alz monomers towards aggregation. However, due to the complexity of the fibrillation process as well as the effect of linking Aβ and incorporating P2 and P30, it cannot be ruled out that other structural arrangements that favor fibrillation could be induced. 4.3. The secondary structure of Con-Alz is difficult to modulate

to perforate vesicles. The residual activity may be due to the medium sized aggregates still present (albeit at reduced concentration) after 24–72 h. 4. Discussion 4.1. Does Con-Alz form amyloid fibrils? Several observations suggest that Con-Alz retains a partial ability to form amyloid structure. Firstly, Con-Alz showed affinity towards both CR and ThT, secondly it exhibited β-sheet structure in aqueous solution and thirdly it is capable of permeabilize vesicles. Nevertheless the high affinity towards ANS compared to Aβ and reduced spectral shifts with ThT and CR point at a relatively reduced level of ordered structure compared to fibrils formed by monomeric Aβ. In support of this, electron microscopy (EM) studies revealed that ConAlz could not form classical amyloid fibrils, even after prolonged incubation, but exhibited a protofibril-like morphology, normally observed as an early stage of the fibrillation process of amyloidogenic peptides. On the basis of this, it would be reasonable to assign the detected amyloid-like tendencies of Con-Alz to aggregates containing some proto-fibril like structures, and not to amyloid fibrils. The reason

The transmembrane part of Aβ as well as the residues comprising the hydrophobic cluster (residue 17–20) forms helices in TFE, preventing fibril formation in this solvent. However, far-UV CD spectra of Con-Alz in different concentrations of TFE, did not change the spectra from having a minimum at 218 nm, to a classical α-helix spectrum with local minimum at 208 and 222 nm, but rather just increased its signal at 218 nm, suggesting increase of β-sheet. This increase in β-sheet structure does not lead to bona fide fibrillation since the binding of ThT in 10–100%TFE was unaltered. Con-Alz shows a very extreme tendency to aggregate, and exists as high molecular weight species already from the onset. Using FFF the estimated aggregates have about ~ 70% of the measured mass corresponding to 50–5000-mers of Con-Alz in 10 mM phosphate pH 7 and150 mM NaCl, the remaining 30% consists of species corresponding to 5–55-mers of Con-Alz. This oligomerization process is not prevented at elevated pH, high denaturant concentration or a combination of both. At high urea concentrations, the hydrophobic interactions between individual Con-Alz molecules should be less favorable due to the ability of urea to interact with hydrophobic areas in protein. However, since aggregates with around 24% β-sheet structure survive

2034

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035

in 8 M urea, inter- and intra-molecular interactions obviously remain more favorable. Micellar concentrations of SDS are the only conditions that prevent binding of ThT and precipitation of Con-Alz. This effect must mainly be attributed to the negative charge of the micelles, since neither DM nor LTAC had the same tendencies. This could indicate an electrostatic interaction between positively charged clusters of amino acids in ConAlz and the surface of the SDS micelles. However, interactions between the transmembrane part of the Aβ in Con-Alz and the interior of the hydrophobic SDS micelles probably also occur. Con-Alz exhibited mainly β-sheet in SDS, but had a weak shoulder at 208 nm, indicating some α-helical structure. 4.4. Electrostatic interactions with anionic lipid surface is needed for Con-Alz to perforate vesicles Con-Alz could perforate vesicles but only under physiologically unrealistic conditions, and this lytic effect confined itself to negatively charged vesicles (100% DOPG and 80:20% DOPC/DOPG). The degree of vesicle perforation was dependent on the aggregation state of ConAlz, with a correlation between longer incubation time and decreased ability to perforate vesicles, suggesting that the species formed within 24 h shared the highest cytotoxicity. Since no perforation is observed for neutral or positively charged vesicles, electrostatic interactions must be the dominant factor determining the perforation even in the presence of 150 mM NaCl, which would increase the dielectric constant and lower the strength of ionic interactions. Given that Con-Alz has a theoretical pI of 5.67, and is therefore overall negatively charged at pH 7, there must be positively charged patches in its amino acid sequence that can interact with the negatively charged vesicles. Two likely candidates are the sequences E3FRHD7 and V12HHQKL17 (assuming that the pKa of His increases beyond the usual value of 6–7 in the presence of the negatively charged vesicles), although we note that the positive patch in E3FRHD7 is flanked by negative charges. In general, interactions with vesicles are believed to enhance the permeabilizing properties of the monomeric protein, as well as catalyze the conversion of monomeric species into toxic amyloid protofibrils [53] [54] [55]. The catalytic effect is believed to rely on local increase of the concentration of proteins that exhibit a favorable conformation for assembly of amyloid fibrils [55]. The Aβ has high affinity towards negatively charged vesicles, which accelerate the conversion of monomeric Aβ into amyloid fibrils [56], suggesting that Con-Alz has inherited these properties. However, the ThT binding pattern towards Con-Alz remained unaltered in the presence of negatively charged vesicles compared to buffer alone. Nevertheless it cannot be ruled out that the molecular structure of the aggregates formed by Con-Alz in the presence of negative vesicles is altered, without altering the binding pattern of ThT. Such alterations could involve uptake of lipids from the membrane incorporating them into the aggregates [54], which also would lead to a lytic effect [53]. Importantly, mammalian cell membranes mostly contain zwitterionic lipids with b20% negatively charged lipids. Thus, in view of the fact that Con-Alz has no perforating activity towards zwitterionic lipids and only perforate 100% negatively charged lipids at unrealistically high protein lipid concentration, the risk of Con-Alz perforating mammalian cells in vivo must be considered very low. We therefore conclude that from a biophysical perspective, the Con-Alz construct represents a promising strategy as a potential vaccine against Alzheimer's disease on several grounds: Firstly, conditions (micellar concentrations of SDS, here ≥10 mM) can be found which keep the concatemer in a stably aggregated state, indicating that formulations mimicking these circumstances can be identified. Secondly, even when transferred to conditions that stimulate aggregation (such as will occur when the formulated sample is injected), the ensuing aggregates do not show vesicle-permeabilizing properties. According to the pore-forming theory of Aβ cytotoxicity [57], this argues that

Con-Alz is unlikely to be cytotoxic in vivo. Finally, given that our data indicate that the Con-Alz construct has a high tendency to aggregate but does not form proper fibrils, we would predict that the Con-Alz immunization strategy would also be more beneficial than conventional immunization with unmodified Aβ, as the antibodies should be targeted to oligomers rather than random monomers or mature fibrils found in plaques. Current efforts are therefore devoted to the translation of these studies into a biologically efficacious vaccine preparation. Acknowledgements We thank EMBO for generously funding this work via an EMBO Young Investigator Grant to D.E.O, Martin Rask for his excellent assistance with light scattering and Dr. Lars Østergaard Pedersen for constructive and stimulating discussions. Con-Alz was provided by Lundbeck A/S. Funding to P.F. was provided by the Canadian Institutes of Health Research, Ontario Mental Health Foundation and the Alzheimer Society of Ontario. References [1] M. Higuchia, J.Q. Trojanowski, V.M.Y. Lee, Pathobiological features in neurodegenerative diseases: an overview, Int. Congr. Ser. 1260 (2004) 69–75. [2] M.P. Lambert, A.K. Barlow, B.A. Chromy, C. Edwards, R. Freed, M. Liosatos, T.E. Morgan, I. Rozovsky, B. Trommer, K.L. Viola, P. Wals, C. Zhang, C.E. FInch, G.A. Krafft, W.L. Klein, Diffusible, nonfibrillar ligands derived from Aβ1–42 are potent central nervous system neurotoxins, Proc. Nat. Acad. Sci. U.S.A. 95 (1998) 6448–6453. [3] D.M. Walsh, A. Lomakin, G.B. Benedek, M.M. Condron, D.B. Teplow, Amyloid βprotein fibrillogenesis. Detection of a protofibrillar intermediate, J. Biol. Chem. 272 (1997) 22364–22372. [4] M.D. Kirkitadze, G. Bitan, D.B. Teplow, Paradigm shifts in Alzheimer's disease and other neurodegenerative disorders: the emerging role of oligomeric assemblies, J. Neurosci. Res. 69 (2002) 567–577. [5] A. Lord, H. Englund, L. Söderberg, S. Tucker, F. Clausen, L. Hillered, M. Gordon, D. Morgan, L. Lannfelt, F.E. Pettersson, L.N.G. Nilsson, Amyloid-β protofibril levels correlate with spatial learning in Arctic Alzheimer's disease transgenic mice, FEBS J. 276 (2009) 995–1006. [6] B.A. Chromy, R.J. Nowak, M.P. Lambert, K.L. Viola, L. Chang, P.T. Velasco, B.W. Jones, S.J. Fernandez, P.H. Lacor, P. Horowitz, C.E. Finch, G.A. Krafft, W.L. Klein, Selfassembly of Aβ(1–42) into globular neurotoxins, Biochemistry 42 (2003) 12749–12760. [7] M.P. Lambert, K.L. Viola, B.A. Chromy, L. Chang, T.E. Morgan, J. Yu, D.L. Venton, G.A. Krafft, C.E. Finch, W.L. Klein, Vaccination with soluble Aβ oligomers generates toxicity-neutralizing antibodies, J. Neurochem. 79 (2001) 595–605. [8] D.J. Selkoe, Alzheimer disease: mechanistic understanding predicts novel therapies, Ann. Intern. Med. 140 (2004) 627–638. [9] R. Kayed, E. Head, J.L. Thompson, T.M. McIntire, S.C. Milton, C.W. Cotman, C.G. Glabe, Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis, Science 300 (2003) 486–489. [10] A. Lord, A. Gumucio, H. Englund, D. Sehlin, V.S. Sundquist, L. Söderberg, C. Möller, P. Gellerfors, L. Lannfelt, F.E. Pettersson, L.N.G. Nilsson, An amyloid-β protofibrilselective antibody prevents amyloid formation in a mouse model of Alzheimer's disease, Neurobiol. Dis. 36 (36) (2009) 425–434. [11] T. Wisniewski, U. Konietzko, Amyloid-β immunisation for Alzheimer's disease, Lancet Neurol. 7 (2008) 805–811. [12] N.R.S. Relkin, P.B. Adamiak, T. Burgut, C. Monthe, R.W. Lent, S. Younkin, L. Younkin, R. Schiff, M.E. Weksler, 18-Month study of intravenous immunoglobulin for treatment of mild Alzheimer disease, Neurobiol. Aging 30 (2009) 1728–1736. [13] C. Holmes, D. Boche, D. Wilkinson, G. Yadegarfar, V. Hopkins, A. Bayer, R.W. Jones, R. Bullock, S. Love, J.W. Neal, E. Zotova, J.A. Nicoll, Long-term effects of Aβ42 immunisation in Alzheimer's disease: follow-up of a randomised, placebocontrolled phase I trial, Lancet 372 (2008) 216–233. [14] S. Gilman, M. Koller, R.S. Black, L. Jenkins, S.G. Griffith, N.C. Fox, L. Eisner, L. Kirby, M.B. Rovira, F. Forette, J.M. Orgogozo, AN1792(QS-21)-201 Study Team, Clinical effects of Abeta immunization (AN1792) in patients with AD in an interrupted trial, Neurology 64 (2005) 1553–1562. [15] A.A. Asuni, A. Boutajangout, H. Scholtzova, E. Knudsen, Y.S. Li, D. Quartermain, B. Frangione, T. Wisniewski, E.M. Sigurdsson, Vaccination of Alzheimer's model mice with Aβ derivative in alum adjuvant reduces Aβ burden without microhemorrhages, Eur. J. Neurosci. 24 (2006) 2530–2542. [16] H.D. Kim, A. Maxwell, F.K. Kong, D.C. Tang, K.I. Fukuchi, Induction of antiinflammatory immune response by an adenovirus vector encoding 11 tandem repeats of Aβ1–6: toward safer and effective vaccines against Alzheimer's disease, Biochem. Biophys. Res. Commun. 336 (2005) 84–92. [17] D.J. Fryauff, E. Mouzin, L.W. Church, S. Ratiwayanto, H. Hadiputranto, M.A. Sutamihardja, H. Widjaja, G. Corradin, B. Subianto, S.L. Hoffman, Lymphocyte response to tetanus toxin T-cell epitopes: effects of tetanus vaccination and concurrent malaria prophylaxis, Vaccine 17 (1999) 59–63.

L. Giehm et al. / Biochimica et Biophysica Acta 1804 (2010) 2025–2035 [18] V.W. Chow, M.P. Mattson, P.C. Wong, M. Gleichmann, An overview of APP processing enzymes and products, NeuroMol. Med. 12 (2010) 1–12. [19] R. Bader, R. Bamford, J. Zurdo, B. Luisi, C.M. Dobson, Probing the mechanism of amyloidogenesis through a tandem repeat of the PI3-SH3 domain suggests a generic model for protein aggregation and fibril formation, J. Mol. Biol. 356 (2006) 189–208. [20] C.F. Wright, S.A. Teichmann, J. Clarke, C.M. Dobson, The importance of sequence diversity in the aggregation and evolution of proteins, Nature 438 (2005) 881. [21] W. Hosia, N. Bark, E. Liepinsh, A. Tjernberg, B. Persson, D. Hallén, J. Thuberg, J. Johansson, L. Tjernberg, Folding into a beta-hairpin can prevent amyloid fibril formation, Biochemistry 43 (2004) 4655–4661. [22] J.L. Johnson, B. Cusack, M.P. Davies, A. Fauq, T.L. Rosenberry, Unmasking tandem site interaction in human acetylcholinesterase. Substrate activation with a cationic acetanilide substrate, Biochemistry 42 (2003) 5438–5452. [23] J.T. Yang, C.S. Wu, H.M. Martinez, Calculation of protein conformation from circular dichroism, Methods Enzymol. 130 (1986) 208–269. [24] S. Brahms, J. Brahms, Determination of protein secondary structure in solution by vacuum ultraviolet circular dichroism, J. Mol. Biol. 138 (1980) 149–178. [25] M.A. Andrade, P. Chacón, J.J. Merelo, F. Morán, Evaluation of secondary structure of proteins from UV circular dichroism using an unsupervised learning neural network, Prot. Eng. 6 (1993) 383–390. [26] W.R. Kirk, E. Kurian, F.G. Prendergast, Characterization of the sources of protein– ligand affinity: ANS-binding to intestinal fatty acid binding protein, Biophys. J. 70 (1996) 69–83. [27] J.C. Giddings, A new separation concept based on a coupling of concentration and flow nonuniformities, Sep. Sci. 1 (1966) 123–125. [28] H.I. Levine, Thioflavine T interaction with synthetic Alzheimer's disease β-amyloid peptides: detection of amyloid aggregation in solution, Prot. Sci. 2 (1993) 404–410. [29] F. Chiti, N. Taddei, F. Baroni, C. Capanni, M. Stefani, G. Ramponi, C.M. Dobson, Kinetic partitioning of protein folding and aggregation, Nat. Struct. Biol. 9 (2002) 137–143. [30] W. Hoyer, T. Antony, D. Cherny, G. Heim, T.M. Jovin, V. Subramaniam, Dependence of α-synuclein aggregate morphology on solution conditions, J. Mol. Biol. 322 (2002) 383–393. [31] W.E. Klunk, R.F. Jacob, R.P. Mason, Quantifying amyloid-peptide (A-) aggregation using the Congo Red-A0- (CR-A-) spectrophotometric method, Anal. Biochem. 266 (1999) 66–70. [32] G.V. Semisotnov, N.A. Rodinova, O.I. Razgulyaev, V.N. Uversky, A.F. Gripas, R.I. Gilmanshin, Study of the “molten globule” intermediate state in protein folding by a hydrophobic fluorescent probe, Biopolymers 31 (1991) 119–128. [33] J. Kyte, R.F. Doolittle, A simple method for displaying the hydropathic character of a protein, J. Mol. Biol. 157 (1982) 105–132. [34] D.K. Wilkins, S.B. Grimshaw, V. Receveur, C.M. Dobson, J.A. Jones, L.J. Smith, Hydrodynamic radii of native and denatured proteins measured by pulse field gradient NMR techniques, Biochemistry 38 (1999) 16424–16431. [35] J.C. Giddings, Field-flow fractionation: analysis of macromolecular, colloidal, and particulate materials, Science 260 (1993) 1456–1465. [36] S. Frokjaer, D.E. Otzen, Protein drug stability—a formulation challenge, Nat. Rev. Drug Delivery 4 (2005) 298–306. [37] M.M. Pallitto, R.M. Murphy, A mathematical model of the kinetics of β-amyloid fibril growth from the denatured state, Biophys. J. 81 (2001) 1805–1822. [38] D. Burdick, B. Soreghan, M. Kwon, J. Kosmoski, M. Knauer, A. Henschen, J. Yates, C. Cotman, C. Glabe, Assembly and aggregation properties of synthetic Alzheimer's A4/β amyloid peptide analogs, J. Biol. Chem. 267 (1992) 546–554. [39] J.D. Harper, S.S. Wong, C.M. Lieber, P.T.J. Lansbury, Assembly of Aβ amyloid protofibrils: an in vitro model for a possible early event in Alzheimer's disease, Biochemistry 38 (1999) 8972–8980.

2035

[40] V. Foderà, M. Groenning, V. Vetri, F. Librizzi, S. Spagnolo, C. Cornett, L. Olsen, M. Van de Weert, M. Leone, Thioflavin T hydroxylation at basic pH and its effect on amyloid fibril detection, J. Phys. Chem. B 112 (2008) 15174–15181. [41] R. Srinivasan, E.M. Jones, K. Liu, J. Ghiso, R.E. Marchant, M.G. Zagorski, pHdependent amyloid and protofibril formation by the ABri peptide of Familial British Dementia, J. Mol. Biol. 333 (2003) 1003–1023. [42] Y. Fezoui, D.B. Teplow, Kinetic studies of amyloid β-protein fibril assembly. Differential effects of alpha-helix stabilization. J. Biol. Chem. 277 (2002) 36948–36954. [43] M. Coles, W. Bicknell, A.A. Watson, D.P. Fairlie, D.J. Craik, Solution structure of amyloid β-peptide(1–40) in a water-micelle environment. Is the membranespanning domain where we think it is? Biochemistry 37 (1998) 11064–11077. [44] A.M. D'Ursi, M.R. Armenante, R. Guerrini, S. Salvadori, G. Sorrentino, D. Picone, Solution structure of amyloid β-peptide (25–35) in different media, J. Med. Chem. 47 (2004) 4231–4238. [45] T.A. Pertinhez, M. Bouchard, R.A. Smith, C.M. Dobson, L.J. Smith, Stimulation and inhibition of fibril formation by a peptide in the presence of different concentrations of SDS, FEBS Lett. 529 (2002) 193–197. [46] S. Yamamoto, K. Hasegawa, I. Yamaguchi, S. Tsutsumi, J. Kardos, Y. Goto, F. Gejyo, H. Naiki, Low concentrations of sodium dodecyl sulfate induces the extension of β2-microglobulin-related amyloid fibrils at neutral pH, Biochemistry 43 (2004) 11075–11082. [47] J.E. Mogensen, P. Sehgal, D.E. Otzen, Activation, inhibition and destabilization of Thermomyces lanuginosus lipase by detergents, Biochemistry 44 (2005) 1719–1730. [48] J.N. Weinstein, S. Yoshikami, P. Henkart, R. Blumenthal, W.A. Hagins, Liposome– cell interaction: transfer and intracellular release of a trapped fluorescent marker, Science 195 (1977) 489–492. [49] M.J. Volles, S.J. Lee, J.C. Rochet, M.D. Shtilerman, T.T. Ding, J.C. Kessler, P.T. Lansbury, Vesicle permeabilization by protofibrillar α-synuclein: implications for the pathogenesis and treatment of Parkinson's Disease, Biochemistry 40 (2001) 7812–7819. [50] R. Tycko, Insights into the amyloid folding problem from solid-state NMR, Biochemistry 42 (2003) 3151–3159. [51] A. Morimoto, K. Irie, K. Murakami, Y. Masuda, H. Ohigashi, M. Nagao, H. Fukuda, T. Shimizu, T. Shirasawa, Analysis of the secondary structure of β-amyloid (Abeta42) fibrils by systematic proline replacement, J. Biol. Chem. 279 (2004) 52781–52788. [52] A.K. Paravastu, R.D. Leapman, W.M. Yau, R. Tycko, Molecular structural basis for polymorphism in Alzheimer's β-amyloid fibrils, Proc. Nat. Acad. Sci. U.S.A. 105 (2008) 18349–18354. [53] E. Sparr, M.F. Engel, D.V. Sakharov, M. Sprong, J. Jacobs, B. de Kruijff, J.W. Hoppener, J.A. Killian, Islet amyloid polypeptide-induced membrane leakage involves uptake of lipids by forming amyloid fibers, FEBS Lett. 577 (2004) 117–120. [54] H. Zhao, E.K.J. Tuominen, P.K.J. Kinnunen, Formation of amyloid fibers triggered by phosphatidylserine-containing membranes, Biochemistry 43 (2004) 10302–10307. [55] J.D. Knight, A.D. Miranker, Phospholipid catalysis of diabetic amyloid assembly, J. Mol. Biol. 341 (2004) 1175–1187. [56] M. Bokvist, F. Lindstrom, A. Watts, G. Grobner, Two types of Alzheimer's βamyloid (1–40) peptide membrane interactions: aggregation preventing transmembrane anchoring versus accelerated surface fibril formation, J. Mol. Biol. 335 (2004) 1039–1049. [57] H.A. Lashuel, P.T. Lansbury, Are amyloid diseases caused by protein aggregates that mimic bacterial pore-forming toxins? Q. Rev. Biophys. (2006) 1–35.