|Home | About | Journals | Submit | Contact Us | Français|
Non-fibrillar, water-soluble low-molecular weight assemblies of the amyloid β-protein (Aβ) are believed to play an important role in Alzheimer’s disease (AD). Aqueous extracts of human brain contain Aβ assemblies which migrate on SDS-PAGE and elute from size exclusion as dimers (~8 kDa) and can block long term potentiation and impair memory consolidation in the rat. Such species are detected specifically and sensitively in extracts of Alzheimer brain suggesting that SDS-stable dimers may be the basic building blocks of AD-associated synaptotoxic assemblies. Consequently, understanding the structure and properties of Aβ dimers is of great interest. In the absence of sufficient brain-derived dimer to facilitate biophysical analysis, we generated synthetic dimers designed to mimic the natural species. For this, Aβ(1-40) containing cysteine in place of serine 26 was used to produce disulphide cross-linked dimer, (AβS26C)2. Such dimers had no detectable secondary structure, produced an analytical ultracentrifugation (AUC) profile consistent for an ~8.6 kDa protein, and had no effect on hippocampal long term potentiation (LTP). However, (AβS26C)2 aggregated more rapidly than either AβS26C or wild type monomers and formed parastable β-sheet rich, thioflavin T positive, protofibril-like assemblies. Whereas wild type Aβ aggregated to form typical amyloid fibrils, the protofibril-like structures formed by (AβS26C)2 persisted for prolonged periods and potently inhibited LTP in mouse hippocampus. These data support the idea that Aβ dimers may stabilize the formation of fibril intermediates by a process distinct from that available to Aβ monomer and that higher molecular weight pre-fibrillar assemblies are the proximate mediators of Aβ toxicity.
The amyloid β-protein (Aβ) is believed to play a central role in Alzheimer’s disease (AD) and like several other proteins associated with neurodegeneration, has the ability to self-associate, and can form an array of different assemblies ranging from dimers all the way to aggregates of fibrils (Powers and Powers, 2008). Initially, it was assumed that Aβ toxicity was mediated by fibrils similar to those present in amyloid plaques, but recent data suggest that non-fibrillar, water-soluble assemblies of Aβ may also be important (Klein et al., 2001; Glabe, 2008; Shankar and Walsh, 2009).
Biochemical analysis of brain indicates that the levels of non-fibrillar forms of Aβ correlate well with synaptic loss and presence of dementia (Lue et al., 1999; McLean et al., 1999; Wang et al., 1999; Mc Donald et al., 2010)(Tomic et al., 2009) and that ex vivo such assemblies can impair synaptic form and function (Shankar et al., 2008). Specifically, we have shown that human brain contains Aβ assemblies which migrate on SDS-PAGE and elute from size exclusion as dimers (~8 kDa), block long term potentiation (LTP), inhibit synapse remodeling, and impair memory consolidation (Shankar et al., 2008). Such species are detected specifically in extracts of AD brain suggesting that SDS-stable dimers may be the basic building blocks of AD-associated synaptotoxic Aβ assemblies (Kuo et al., 1996; Roher et al., 1996). The role of low-n oligomers of Aβ in the range of dimer to tetramer is also supported by in vitro studies using peptides bearing design mutations. For instance, substituting glycine for leucine within the GxxxG repeat motif of Aβ indicates that Aβ-mediated neurotoxicity is directly linked to the abundance of mass spectrometry-detected dimers and trimers (Hung et al., 2008). Similarly, peptides containing G33A or G29/33A substitutions form low-n oligomers that fail to block LTP (Harmeier et al., 2009). This latter finding indicates that aggregation size alone is not the sole determinant of synaptotoxicity and that structure is also critical. Consequently, establishing the amyloidogenicity and structure of Aβ dimers in the brain, CSF, and blood of AD patients are of great diagnostic and therapeutic interest.
In the absence of sufficient brain-derived Aβ dimers, we and others have generated synthetic cross-linked Aβ dimers to mimic the natural species (Shankar et al., 2008; Kok et al., 2009). In our studies Aβ(1-40) containing cysteine in place of serine 26 was used to produce disulphide cross-linked dimers, (AβS26C)2. Given that such dimer preparations share a similar synaptotoxic profile with natural dimers, we undertook experiments to investigate the biophysical and aggregation properties of (AβS26C)2 in the hope that this might shed light on processes occurring in AD brain. Here we report that (AβS26C)2 aggregated rapidly to form protofibril-like assemblies and that freshly isolated (AβS26C)2 did not block LTP whereas (AβS26C)2 solutions that were allowed to form protofibrils did. These data support the idea that Aβ dimers may stabilize the formation of fibril intermediates by a process distinct from that available to Aβ monomer and that such intermediates are potent synaptotoxins.
Wild-type human Aβ1-40, DAEFRHDSGY-EVHHQKLVFFAEDVGSNKGAIIGLMVGGVV, and Aβ1-40 in which serine 26 was substituted with cysteine (AβS26C) were purchased from the Keck Biotechnology Center (Yale University, New Haven, CT). Peptide mass and purity were determined by electrospray ionization/ion trap mass spectrometry and reverse-phase HPLC, respectively. All peptides had the correct mass and were >95% pure.
All chemicals were obtained from Sigma-Aldrich (Saint Louis, MI), and unless indicated otherwise were of the highest purity available. Unbranched dextran standards of molecular masses: 43,800; 21,400; 9890, and 4440 were purchased from Pharmacosmos (Holbaek, Denmark). Water was double-distilled and deionized using a Milli-Q system (Millipore, Cork, Ireland).
Peptide was solubilized at ~0.18 mg/ml in milliQ water, diluted 1:1 with 20 mM ammonium bicarbonate, pH 8.2, to generate a ~20 µM (with respect to monomer) peptide solution and bubbled with oxygen for ~5–10 min. Solutions were incubated at room temperature for 5 days and each day bubbled with oxygen. To facilitate disassembly of aggregates formed during the oxidation reaction, the peptide solution was lyophilized and subsequently incubated in 50 mM Tris-HCl, pH 8.0 containing 5 M Guanidine HCl, for ~4 h prior to size exclusion chromatography (SEC).
The oxidized AβS26C was subjected to SEC on a preparative HiLoad™ 16/60 Superdex™ 75 column (GE Healthcare, Uppsala, Sweden) eluted with 25 mM ammonium acetate, pH 8.5 at a flow rate of 0.6 ml/min using an AKTA FPLC system (GE Healthcare). Samples were centrifuged at room temperature and 16,000 × g for 20 min and 2.0 ml of supernatant injected onto the column. Peptides were detected by absorbance at 280 nm and 1.2 ml fractions collected. To minimize peptide aggregation, fractions were immediately put on ice, and if necessary, diluted with eluent so that the peptide content was ≤0.15 mg/ml. Peptide concentration was determined by absorbance at 275 nm using the molar extinction coefficient for tyrosine (ε275 = 1,400 M−1 cm−1). Fractions that contained highly pure (AβS26C)2 or AβS26C monomer were used immediately for biophysical and/or aggregation studies. Wild type Aβ was solubilized in 0.1% ammonium hydroxide to produce a ~0.2 mg/ml solution and monomer isolated by SEC as described for AβS26C. Conformer purity was confirmed by SDS-PAGE and MALDI-ToF MS (Hu et al., 2008).
Samples from peptide cross-linking reactions and aggregation experiments were electrophoresed on 16% polyacrylamide SDS-tris-tricine gels in the presence or absence of 50 mM β-mercaptoethanol and visualized by silver staining (Shevchenko et al., 1996).
Assembly of monomeric and dimeric Aβ into higher-ordered aggregates was investigated in quiescent and agitated reactions (5 replicates for each sample time point) using a thioflavin T (ThT) binding assay adapted for use in microtiter plates (Betts et al., 2008). The concentration of peptide in SEC-isolated (AβS26C)2 and monomer fractions was determined by measuring absorbance at 275 nm and peptides then diluted to 0.15 mg/ml with 25 mM ammonium acetate, pH 8.5. Such solutions were further diluted with 45 mM sodium phosphate, pH 7.4, to produce stock solutions of 0.087 mg/ml in 20 mM sodium phosphate, pH 7.4. These solutions were subsequently serially diluted 2–30-fold into assay buffer (20 mM sodium phosphate, pH 7.4) to final concentrations of 0.0029 – 0.044 mg/ml and 100 µl of each added to wells of a 96-well polystyrene microtiter plate (Thermo Fisher Scientific, Langenselbold, Germany). Five µl of 2 mM ThT in MillQ water was added into appropriate wells, including controls that did not contain peptide. For time-course SEC, electron microscopy (EM), or light scattering analyses, samples were incubated without ThT. Instead, the reaction progress for such samples was determined by monitoring the ThT signal of replicate samples that contained the dye. Quiescent peptide aggregation was initiated by incubating the plate at 37 °C, and agitated samples were shaken at 700 rpm and 37 °C in a VorTemp 56™ incubator/shaker (Labnet International, Inc., Woodbridge, NJ). Each reaction was monitored in real-time by ThT fluorescence (Ex435nm and Em485nm) using a SpectraMax M2 multi-detection microplate reader (Molecular Devices Corp., Sunnyvale, CA) (Betts et al., 2008). Prior studies have shown that the continuous presence of ThT did not affect reaction kinetics (Betts et al., 2008).
Experiments were performed on a Beckman XLI analytical ultracentrifuge fitted with an An50-Ti rotor in quartz cells containing two-sector centerpieces. Samples of freshly SEC-isolated (AβS26C)2 were centrifuged at 50,000 rpm and 20 °C and absorbance data were collected at 230 and 278 nm over 20 h, with scans recorded every 10 minutes. Sedimentation velocity data were analyzed using the c(s) distribution method in the software SEDFIT (v11.8) (Schuck et al., 2002). For the analyses, partial specific volume () for (AβS26C)2 was calculated from the amino acid sequence using the software SEDNTERP (Laue et al., 1992).
The molar masses for (AβS26C)2 aggregates was determined by multi-angle laser light scattering (MALLS). One ml of a 3 day incubated 15 µM (AβS26C)2 was loaded onto a Superdex™ 75 10/300 GL column (GE Healthcare) equilibrated in 20 mM sodium phosphate, pH 7.4, and eluted using a Prominence HPLC instrument (Shimdazu Europa GmbH, Duisburg, Germany) daisy-chained with a Dawn HeleosII MALLS detector and Optilab dRX refractometer (Wyatt Technology Corp., Santa Barbara, CA). The column was maintained at 20 °C (± 0.1 °C) using an oven (Wyatt Technology Corp.) and the delay volumes between instruments and band-broadening were determined by injecting a 1 mg/ml BSA solution through the system. Molar masses were calculated from the intensity of scattered light at 18 different angles, as a function of protein concentration and SEC elution volume. The intrinsic instrumental baseline for each data channel was subtracted and the molar mass across a given 2D slice of the elution profile determined using the ASTRA V HPLC software (Wyatt Technology Corp.). To ensure robust data fitting, the apparent molar mass for the elution peak was determined as a function of the width of fitted windows for the front end, centre, and trailing edge of the peak.
Aβ solutions were placed in a 1 mm path length quartz cuvette (Starna Scientific Ltd., Essex, UK) and spectra obtained at 22 °C using a J-810 JASCO spectropolarimeter (JASCO Corp., Tokyo, Japan). Spectra were generated from three data accumulations between ~195–260 nm with 10 nm/min continuous scanning and a 0.5 nm bandwidth. To minimize artifactual CD signal by aggregate-induced light scattering, samples from aggregation studies were centrifuged at 16,000 × g for 20 min at room temperature. Raw data were manipulated by subtraction of buffer spectra and by binomial smoothing according to the manufacturer’s instructions (JASCO Corp.) and data displayed as molar ellipticity (θ).
Negative contrast EM was performed as described previously (Walsh et al., 1997). Aliquots (10 µl) of peptide sample were applied to carbon-coated Formvar grids (Electron Microscope Sciences, Washington, PA), cross-linked with 0.5% (v/v) glutaraldehyde, stained with 2% (w/v) uranyl acetate solution (Ted Pella, Inc., Redding, CA) and examined using a Tecnai™ G2 Spirit BioTWIN electron microscope (FEI, Hillsboro, OR).
Six to eight week old male C57/BL6 mice were anaesthetised with isoflurane/O2 and decapitated. Brains were rapidly removed and immersed in ice-cold sucrose-based artificial cerebrospinal fluid (ACSF) that contained: 75 mM Sucrose, 87 mM NaCl, 2.5 mM KCl, 25 mM NaHCO3, 25 mM Glucose, 1.25 mM NaH2PO4, 0.5 mM CaCl2, 7 mM MgCl2. Parasagital sections (350 µm) were prepared using a vibratome VT1000S (Leica, Bensheim, Germany) and slices allowed to recover for at least 90 min in normal ACSF (nACSF contained: 119 mM NaCl, 2.5 mM KCl, 1.3 mM MgSO4, 2.5 mM CaCl2, 26.2 mM NaHCO3, 11 mM d-glucose, 1 mM NaH2PO4) in a BSC-PC submerged incubation chamber (Warner Instruments, Hamden, CT). Thereafter, slices were continuously perfused with oxygenated nACSF at a rate of 2–3 ml/min at 30 °C. A stainless steel microelectrode (FHC, Bowdoin, ME) was used to stimulate the hippocampal Schaffer collateral pathway, and extracellular excitatory postsynaptic potentials (fEPSPs) were recorded in the stratum radiatum of the CA1 region using glass microelectrodes (2–4 MΩ) filled with nACSF. Test stimuli were delivered once every 30 s (0.033 Hz) and the stimulus intensity was adjusted to produce a baseline fEPSP of 30–40% of the maximal response. A stable baseline was recorded for at least 20 min before addition of fresh or aggregated (AβS26C)2. LTP was induced by theta-burst stimulation (TBS, 4 pulses delivered at 100 Hz, ten times, with an interburst interval of 200 ms) delivered at baseline intensity. Field potentials were recorded with an Axon CNS Muticlamp 700B amplifier coupled to a Digidata 1440A digitizer (Molecular Devices). Data was recorded using a pClamp10 (Molecular Devices, Sunnyvale, CA) and later analyzed using Clampfit.10 software (Molecular Devices).
In all experiments, fresh or aggregated (AβS26C)2, or buffer was added to the perfusate after a stable baseline period and circulated in the bath for the remainder of the experiment. Immediately prior to use, the 3 day aggregated material was spun at 16,000 × g for 20 min at room temperature and the supernatant added to the perfusate. LTP was induced 30 min after introduction of peptide or vehicle and recorded for at least 60 min post-tetanus.
Data points from aggregation time courses were fitted using SigmaPlot 2000, version 6 (Systat Software, Chicago, IL).
The oxidized AβS26C peptide produced four major elution peaks (~48, ~55, ~72, and ~84 ml) when chromatographed on a preparative HiLoad 16/60 Superdex 75 (GE Healthcare, Uppsala, Sweden) column (Fig. 1A). Calibration of size exclusion columns using globular protein standards does not allow accurate assessment of Aβ monomer size (Walsh et al., 1997), whereas the use of dextran standards does (Paivio et al., 2004; Walsh et al., 2005). Based on the use of unbranched dextran standards the apparent molecular weight of AβS26C conformers in the two largest peaks (~72 and ~84 ml) were ~10 (dimer) and ~5.7 (monomer) kDa, respectively (Fig. 1A). SDS-PAGE and mass spectroscopy analyses confirmed that these fractions contained (AβS26C)2 and monomeric Aβ (Fig. 1A, B). The minor SDS-PAGE bands in the (AβS26C)2 fractions that migrated >9 kDa were transiently and artificially induced by SDS (Wahlstrom et al., 2008). Reducing agent completely dissociated (AβS26C)2 to monomer, indicating that dimers were indeed stabilized by an intermolecular disulfide bond (Fig. 1B). Furthermore, AUC sedimentation velocity studies demonstrated that the dominant peptide conformer had an apparent molecular weight of ~10+/− 2 kDa (Fig. 1C). However, trace quantities of higher ordered assemblies of ~95 kDa were also observed. Given that the AUC experiment took ~20 h to complete it seems likely that these higher molecular weight species formed during the experiment. In accord with this observation re-chromatographying of the sample incubated for 16 h at room temperature revealed the appearance of species eluting in the void volume (Fig. 1D). These finding suggest that the (AβS26C)2 has a high propensity for aggregation and prompted us to compare the aggregation of (AβS26C)2 versus Aβ monomer.
We have previously demonstrated that (AβS26C)2 are potent neurotoxins (Hu et al., 2008; Shankar et al., 2008). Thus, we investigated whether (AβS26C)2 had a propensity to aggregate and if their toxicity was dependent on aggregation. Freshly isolated (AβS26C)2 was diluted as described in the Materials and Methods and aggregation monitored using thioflavin (ThT) binding. Under quiescent conditions (AβS26C)2 (5–10 µM) rapidly formed ThT positive assemblies without a measurable lag time (Fig. 2A, E), but unlike typical amyloid fibrils the (AβS26C)2 aggregates did not readily sediment when centrifuged (16,000 × g for 20 min) (Fig. 2B). Moreover, CD studies indicated that the (AβS26C)2 aggregates contained significant β-structure (Fig. 2D), whereas freshly isolated (AβS26C)2, and AβS26C and wild type monomers incubated for 1 day had little secondary structure (Fig. 2C, D). As previously documented, the secondary structure of SEC-isolated wild type monomer changed slowly (Walsh et al., 1999) and AβS26C monomer appeared to change even more slowly (Fig. 2D). Dimer aggregation was critically dependent on peptide concentration with no discernible ThT binding evident at (AβS26C)2 concentrations ≤ 2.5 µM in samples incubated at 37 °C for 44 h (Fig. 2E). The progress curves for (AβS26C)2 aggregation fitted to single exponential linear plots (R2= 0.96 ± 0.04) producing an apparent rate constant of 0.085 ± 0.03 h−1 (Fig. 2F).
Having demonstrated that (AβS26C)2 shows an unusually high propensity for aggregation, we investigated the morphology and size distribution of the aggregates formed. SEC analysis of (AβS26C)2 aggregate species in reactions sampled over a 3 day period indicated that these assemblies eluted in the void volume of the Superdex 75 10/300 GL column (Fig. 3A). The SEC-isolated aggregates were not sedimented by centrifugation and had ~4-fold greater ThT signal than the same weight of SEC-isolated (AβS26C)2 or freshly prepared monomers, and ~2-fold lower than for Aβ wild type fibrils (Fig. 3B, C). Consistent with CD analysis of unfractionated (AβS26C)2 aggregates, the SEC-isolated aggregates were rich in β-sheet structure whereas the (AβS26C)2 fraction had no detectable secondary structure (Fig. 3D). SEC-MALLS indicated that the apex of the (AβS26C)2 aggregate peak contained assemblies with a molecular weight distribution of ~1–2 mega-Dalton, with a leading shoulder containing species ≥ 2 mega-Dalton (Fig. 3E). The distribution of species detected are consistent with Aβ assemblies that contained ~100 to 450 (AβS26C)2 molecules. These aggregates had morphologies highly similar to Aβ protofibrils (Harper et al., 1997; Walsh et al., 1997) appearing as short flexible rods with an average width of 5.8 ± 0.2 nm (n= 9) (Fig. 4). The protofibrils grew in length from 59 ± 12 nm (n= 7) after 1 day to 139 ± 35 nm (n= 7) after 3 days incubation. Importantly these assemblies were relatively stable since protofibrils in certain samples persisted even after a one month long incubation at 37 °C. However, the length of protofibrils appeared to increase with time, with the 30 day incubated sample containing protofibrils twice as long, 284 ± 59 nm (n= 7), as the protofibrils detected at 3 days (Fig. 4). The largest protofibrils in the aged samples pelleted after centrifugation at 16,000 × g for 20 min, leaving ~40 % of the ThT positive material in solution which by EM contained shorter protofibrils, 181 ± 31 nm (n= 7)(Fig. 4D, E).
To investigate the formation and stability of protofibrils over more tractable time scales we studied aggregation of (AβS26C)2, AβS26C and wild type monomers under conditions where the samples were vigorously agitated. In contrast to quiescent conditions, agitation initiated aggregation of all peptides within a couple of hours (compare Fig. 2A and SFigs. 1A – C). The concentration of (AβS26C)2 required to facilitate aggregation was reduced almost 4-fold when samples were agitated, with fully formed ThT positive assemblies evident within ~4 h at (AβS26C)2 concentration as low as 1.25 µM (SFig. 1A). In contrast, without shaking no ThT positive aggregates were formed in (AβS26C)2 solution <5 µM (Figs. 2E). The progress curves for the agitated (AβS26C)2 reactions were sigmoidal with short lag phases, tlag <1 h. The absence of a measurable lag phase for the quiescent (AβS26C)2 reactions is somewhat puzzling but presumably reflects the significantly larger peptide concentrations used in the quiescent reactions. The maximum ThT signal amplitude for (AβS26C)2 aggregation was proportionate to the amount of peptide used (SFig. 1A). Wild type monomer was considerably less amyloidogenic than (AβS26C)2 since they only formed ThT positive aggregates at concentration ≥ 5 µM, whereas (AβS26C)2 aggregated even at 1.25 µM (SFig. 1A, B). Consistent with quiescent experiments, 20 µM (AβS26C)2 formed aggregates in agitated samples without a lag (SFig. 1C), were not pelleted by centrifugation (SFigs. 1 and 2) and had protofibril-like morphology. The average lengths and widths of protofibrils formed in agitated samples (78 ± 22 nm, n= 8 and 6.0 ± 1.0 nm, n= 8, respectively) were highly similar to those of protofibrils formed under quiescent conditions (compare Fig. 4C and SFig. 2B). In contrast, the products for the AβS26C and wild type monomer reactions readily pelleted and had typical amyloid fibril morphology when examined by EM. Interlaced networks of fibrils with diameters ranging from 7 to 11 nm and up to several microns in length were detected in incubates of both S26C and wild type monomers (SFig. 2).
We have previously shown that AβS26C dimers can potently block LTP both in vivo (Hu et al., 2008) and in vitro (Shankar et al., 2008). In those studies (AβS26C)2 was isolated by SEC, frozen at −80 °C and shipped on dry ice to our collaborators, where the samples were stored frozen, then thawed and used for electrophysiology experiments. However, given the findings presented above regarding the high propensity for dimer to form aggregates we set out to determine if the previously documented plasticity-impairing activity of S26C derived from authentic dimers or aggregates of (AβS26C)2. First we used analytical ultracentrifugation and SEC to assess aggregation in samples that had been frozen and thawed versus samples that had been freshly isolated and immediately used for AUC. When freshly isolated (AβS26C)2 was immediately re-chromatographed it produced a single peak, whereas when the same solution was frozen and stored at −80 °C for 1 week and then re-chromatographed a small second high molecular weight peak was detected (SFig. 3). Next we performed experiments in which (AβS26C)2 was isolated and used for electrophysiology experiments within 4 h (between 1–4 h) of collection (Figs. 5C and D). Hippocampal slices perfused with vehicle produced robust LTP measuring 160 ± 8 %, at 1 h post-TBS (n=7), while slices perfused with 50 nM of freshly isolated (AβS26C)2 produced a highly similar potentiation (155 ± 8 %, n=7). The concentration of (AβS26C)2 used here is significantly higher than the minimal concentration of (AβS26C)2 previously shown to impair hippocampal LTP (Shankar et al., 2008) and strongly suggests that the previously documented impairment of plasticity was not mediated by authentic (AβS26C)2. Given the demonstration that (AβS26C)2 can aggregate when frozen (SFig. 3), we set out to determine if deliberately aggregated (AβS26C)2 could block LTP. When the same (AβS26C)2 solution that had been tested when fresh and failed to block LTP was aggregated for 3 days it completely inhibited LTP (103 ± 10 %, n=7) (Fig. 5C and D). Moreover, EM examination of the 3 day aggregated material demonstrated the presence of abundant protofibrils (Fig 5B) indistinguishable to those observed in prior aggregation experiments (e.g. Fig. 4). Thus together these results indicate that S26C dimers rapidly assemble into relatively stable protofibrils that can block LTP.
Burgeoning evidence suggests that non-fibrillar water-soluble forms of Aβ are the principal mediators of neurotoxicity in AD, but, as yet the precise conformation and assembly form(s) of Aβ responsible remain unidentified (Klein et al., 2001; Hardy and Selkoe, 2002; Klein et al., 2004). Recent work has indicated that SDS-stable Aβ dimers present in the water-soluble phase of human brain are strongly associated with AD-type dementia (Mc Donald et al., 2010) and possess disease-relevant toxic activity (Shankar et al., 2008). However, due to the lack of highly pure brain-derived Aβ it has not been possible to study the conformation and aggregation kinetics of this material. Thus we have chosen to study synthetic dimers which mimic the toxic activity of the natural SDS-stable dimer (Hu et al., 2008; Shankar et al., 2008). Unlike prior work with covalently-linked Aβ which employed crude mixtures containing both cross-linked and uncross-linked species (Siegel et al., 2007; Hartley et al., 2008; Moore et al., 2009), and which lacked definition regarding aggregation state and the sites of cross-linking, we used highly pure Aβ dimer with a defined linkage site. Specifically, dimers were formed by the substitution of serine 26 with cysteine and subsequent disulphide bond formation and isolated free of other assemblies by SEC.
Like the wild type monomer, (AβS26C)2 was devoid of discernible secondary structure and exhibited little or no binding to thioflavin T. De novo fibrillogenesis from Aβ monomer is believed to involve a series of conformational alterations that include the formation of thermodynamically unstable amyloidogenic intermediates, self-association and stabilization, and finally, protofibril association into mature fibrils (Harper and Lansbury, 1997; O'Nuallain and Wetzel, 2002; Glabe, 2008; Roychaudhuri et al., 2009). In contrast, (AβS26C)2 aggregation proceeded rapidly with the formation of stable protofibril-like assemblies which persisted for relatively long periods. In agreement with prior reports, protofibrils appeared as a continuum of structures which by EM ranged in size from imperfect spheres of 5 nm diameter to flexible rods up to 200 nm long and approximately 5–6 nm wide (Harper et al., 1997; Walsh et al., 1997; Harper et al., 1999; Johansson et al., 2006) and included structures sometimes referred to as Aβ-derived diffusible ligands (ADDLs) (Lambert et al., 1998; Hepler et al., 2006). This increase in the propensity to form protofibrils coupled with a reduced tendency to form mature fibrils suggests that there may be some subtle microstructural differences between the monomer and (AβS26C)2 which act to change the rate constant for protofibril formation, and/or reduce the rate constant for conversion of protofibrils into fibrils (Fig. 6). Conformational changes in the mid-region of Aβ are thought to be required for the transition of protofibrils into fibrils (Williams et al., 2005; Kheterpal et al., 2006) and cross-linking at residue 26 may act to suppress this transition.
But how does this translate to Aβ in the human brain? For instance, to make dimers is it necessary to start from monomeric Aβ, and if it is, how then is it possible that monomers can contribute to two different pathways? First, there is evidence that dimerized APP may undergo amyloidogenic processing (Munter et al., 2007) thus providing a route for the direct production of discrete Aβ dimer subunits. Once formed such dimers may associate with other dimers in a pathway distinct from the aqueous-phase assembly of individual monomers. Second, other factors that preferentially stabilize dimer formation from monomers would be predicted to enable assembly by a pathway similar to that we have observed for (AβS26C)2, whereas conditions that do not favour dimer formation would lead to lower levels of protofibril. It is noteworthy that protofibril formation and stability are modulated by factors that may also facilitate dimer formation and stabilization. Specifically, covalent cross-linking of Aβ by either 4-hydroxynonenal (HNE) or transglutaminase (TGase) or the presence of certain small molecules accelerate formation of protofibrils while inhibiting fibril formation (Siegel et al., 2007; Hartley et al., 2008; Moore et al., 2009; Williams et al., 2005). Although, it has not yet been demonstrated that transglutaminase catalyzed isopeptide ε-(c-glutamyl)lysine bond formation leads exclusively to formation of cross-linked dimer, the data available suggest that intermolecular bond formation between lysine 16 and glutamine 15 is the only likely site of linkage (Ikura et al., 1993). Thus the enhancement in the rate of protofibril formation and persistence observed in prior cross-linking studies is probably attributable to the presence of dimer which aggregates in a manner distinct from monomer. The rate of protofibril formation and the time period over which protofibrils persist are strongly influenced by Aβ primary sequence (Lashuel et al., 2003; Paivio et al., 2004). For example, the population of protofibrils is greater for Aβ1-42 and Aβ1-40E22G than for wild type Aβ1-40 and it is interesting to speculate that this may reflect an increased tendency of Aβ1-42 and Aβ1-40E22G to form dimers(Walsh et al., 1997; Nilsberth et al., 2001).
Moreover, since SDS-stable Aβ dimers appear specific for AD it seems plausible that the presence of dimers and abundance of protofibrils are linked. That is, dimers may exert toxicity as a consequence of their ability to form relatively stable protofibrils (Fig. 6).
In prior studies we have demonstrated that SDS-stable Aβ dimers from human brain and CSF can impair disease-relevant measures of learning and memory (Klyubin et al., 2008; Shankar et al., 2008) and that synthetic (AβS26C)2 exhibited a similar activity. Here we have now shown that freshly prepared (AβS26C)2 does not inhibit LTP as such, but rather aggregated dimers in the form protofibrils mediate this activity. As discussed above this finding is consistent with the previously documented protofibril-mediated inhibition of LTP (Hartley et al., 2008) and with recent observations that Aβ protofibril levels correlate with spatial learning impairment of AD transgenic mice (Lord et al., 2009). Taken together our findings suggest that the activity previously ascribed to SDS-stable Aβ dimers, may not reside in dimers per se, but rather in the higher order assemblies they rapidly form. Of course other factors which increase the kinetic stability of protofibrils by a dimer-independent mechanism could also lead to an increase in synaptotoxicity, but given the link between SDS-stable dimers and AD, a dimer-dependent mechanism seems most likely. Similarly, the pathogencity of assemblies other than protofibrils cannot be discounted and merits further investigation. Moving forward it will also be important to investigate the assembly and toxicity of heterodimers of Aβ1-40S26C and Aβ1-42S26C and of homodimers of Aβ1-42 and to assess the toxicity of protofibril sub-populations. The latter is particularly important since one might expect that different structures may have different activity. Nonetheless, therapeutic targeting of Aβ dimers remains attractive since such a strategy should prevent formation of a substantial population of pre-fibrillar toxic assemblies.
We thank Ms. Gwen Manning and Dr. Elia Giuliano for assistance with mass spectrometery. This work was supported by funding to DMW from Science Foundation Ireland (Grant 08/1N.1/B2033), the Health Research Board (Grant RP/2008/30) and the National Institutes of Health (grant IRO1AGO27443).