Assembly of Aâ Amyloid Protofibrils: An in Vitro Model for a Possible Early Event in Alzheimer’s Disease†

Amyloid fibrils comprising primarily the peptides Aâ40 and Aâ42 are a defining feature of the Alzheimer’s disease (AD) brain, and convergent evidence suggests that the process of their formation plays a central role in the AD pathogenic pathway. Elucidation of fibril assembly is critical for the discovery of potential AD diagnostics and therapeutics, since the pathogenic entity is not necessarily the product fibril, but could be a precursor species whose formation is linked to fibrillogenesis in vivo. Atomic force microscopy allowed the identification of an unanticipated intermediate in in vitro fibril formation, the Aâ amyloid protofibril. This manuscript describes studies of the structure of the Aâ40 protofibril and its in vitro assembly and disassembly using atomic force microscopy (AFM). The Aâ40 protofibril has a height of ca. 4.3 ( 0.5 nm and a periodicity of ca. 20 ( 4.7 nm. The rate of its elongation depends on the total concentration of Aâ40, the temperature, and ionic strength of the medium. Aâ42 and Aâ40 protofibrils elongate at a comparable rate. Statistical analysis of AFM data reveals a decrease in the number of protofibrils with time, indicating that coalescence of smaller protofibrils contributes to protofibril elongation. Similar analysis reveals that protofibrils shorten while the number of protofibrils also decrease following dilution, indicating that protofibril disassembly does not proceed by a reverse of the assembly process. These investigations provide systematic data defining factors affecting Aâ fibrillization and, thus, should be valuable in the design of high-throughput assays to identify agents which alter Aâ protofibril assembly. Cortical amyloid plaques, comprising a fibrillar form of the amyloid-â protein (Aâ),1 are the defining pathological feature of the Alzheimer’s disease (AD) brain (1). The observation of amyloid plaques in the postmortem brain becomes more prevalent with age in AD patients and in cognitively normal individuals (2), although the number of plaques in the former population is far greater. In fact, the number of cortical amyloid plaques at autopsy roughly correlates with the severity of AD symptoms (3). In addition, the brains of Down syndrome patients, who invariably develop clinical AD in their thirties or forties, are characterized by abnormal Aâ deposits at an early age (2, 4). Taken together, these pathological observations suggest that amyloid fibril formation is an early and required event in the AD disease process (5). Genetic studies support this notion. Early-onset familial AD has been linked to three genes: the amyloid precursor protein and presenilins 1 and 2. All of these mutations increase the production of the fasternucleating Aâ variant Aâ42 (6), consistent with the acceleration of amyloid fibril formation being critical (5). Two risk factors for development of late-onset AD also influence amyloid formation. ApoE genotype, a determinant of age of onset of sporadic AD (7), influences the level of amyloid deposition, such that the high-risk genotype (apoE4/4) is correlated with the most extensive amyloid deposition (8-10). In addition, head trauma, known to be a significant risk factor for the later development of AD, produces a transient and selective elevation of Aâ42 production (11, 12). Since genetic factors that cause AD or increase susceptibility to AD also promote amyloid formation, inhibition of the formation of amyloid fibrils may have therapeutic benefit. However, if the pathogenic species is a precursor or an alternative to the fibril, then blocking fibril formation at too late a stage could accelerate disease by causing the toxic precursor to accumulate (13). It is therefore critical to elucidate the molecular mechanism of the fibril assembly process and to determine which species is responsible for pathogenesis (14, 15). An effective therapeutic agent targeting the amyloid formation process should act upstream of the pathogenic species to either halt further aggregation or promote the formation of benign oligomers (15). Since amyloid fibrils produced in vitro from synthetic Aâ variants † This work was supported by a grant to C.M.L. and P.T.L. from the National Institute on Aging (PO1 AG14366). S.S.W. was supported by a fellowship from the Natural Sciences and Engineering Council of Canada. * To whom correspondence should be addressed. (C.M.L.) Phone: (617) 496-3169. Fax: (617) 496-5442. E-mail: [email protected]. (P.T.L.) Phone: (617) 525-5260. Fax: (617) 525-5252. E-mail: [email protected]. ‡ Center for Neurologic Diseases. § Department of Chemistry and Chemical Biology. 1 Abbreviations: Aâ, amyloid-â protein; AD, Alzheimer’s disease; ADDL’s, Aâ-derived diffusible ligands; AFM, atomic force microscopy; DMSO, dimethyl sulfoxide; HOPG, highly ordered pyrolytic graphite; IAPP, islet amyloid polypeptide; LTP, long-term potentiation; MES, 2-(N-morpholino)ethanesulfonic acid; QLS, quasielastic light scattering. 8972 Biochemistry 1999, 38, 8972-8980 10.1021/bi9904149 CCC: $18.00 © 1999 American Chemical Society Published on Web 06/18/1999 resemble those extracted from AD brain, in vitro studies can provide relevant information about the in vivo process (1). Amyloid fibril formation is an ordered polymerization characterized by a slow nucleation, followed by rapid growth (16, 17). Atomic force microscopy (AFM) has been utilized to probe the early events in the process (13, 18-21). Using conventional silicon tip technology, this technique is capable of routinely providing ca. 10-20 nanometer resolution threedimensional information about species that are adsorbed to an atomically smooth surface such as mica (19, 20, 22, 23). Recent advances in the assembly and use of carbon nanotube AFM tips can improve resolution by an order of magnitude, into the ca. 1 nm range (20). The observation of individual adsorbed assemblies allows the construction of mechanistic models that stimulate experiments using complementary techniques (18, 24). Our preliminary AFM studies revealed the existence of a discrete, dynamic intermediate in Aâ fibril formation, which we designated the amyloid protofibril (19). Oligomeric Aâ species that are probably identical to those observed by AFM have also been observed by analytical ultracentrifugation (25), gel filtration, and electron microscopy (24). The conversion of protofibrils to fibrils is sudden and can be seeded by preformed Aâ fibrils but not by preformed Aâ protofibrils (18). Furthermore, Aâ40 solutions preincubated to form longer protofibrils (up to ca. 200 nm long) convert to fibrils after addition of fibrillar seeds approximately twice as rapidly as freshly prepared solutions containing shorter protofibrils (ca. 50 nm or less) (18). This finding suggests that the process of Aâ protofibril elongation and assembly may be a critical one in the disease process. Recently, protofibril-like species have also been observed during fibril formation by other disease-associated proteins including the islet amyloid polypeptide (IAPP) (26), transthyretin (27), R-synuclein (28), and acyl phosphatase (which is amyloidogenic but not associated with disease) (29). We report herein detailed studies of Aâ protofibril assembly and disassembly, as well as preliminary studies of these species using carbon nanotube AFM tips. MATERIALS AND METHODS Synthetic Aâ Peptides. Aâ40 and Aâ42 were purchased from Quality Controlled Biochemicals Inc. (Hopkinton, MA). Seed-free Aâ40/Aâ42 stock solutions in dimethyl sulfoxide (DMSO) were prepared at concentrations of 12-15 mg/mL. The solution was sonicated (5-10 min in a Branson 1200 water bath sonicator) and filtered through a 0.2 μm nylon microspin filter to remove any undissolved seed. Final peptide concentrations of the DMSO stock solutions, as determined by quantitative amino acid analysis, were typically 2-2.5 mM. Initiation of in Vitro aggregation of Aâ40/42 was accomplished by adding an aliquot of a concentrated DMSO stock of Aâ40 to aqueous buffer (10 mM phosphate, 137 mM NaCl, and 27 mM KCl, pH 7.4, unless otherwise specified) followed by immediate vortexing to mix thoroughly. For experiments investigating the concentration dependence of protofibril elongation, [DMSO] in the aggregation buffer was kept constant (e10 vol %). After initial mixing, the solutions were incubated at room temperature without further agitation except for the removal of aliquots for AFM analysis. Solutions used to investigate the effect of glycerol were prepared without presolubilization in DMSO. Instead, Aâ40 was solubilized at alkaline pH using NaOH solutions before adding buffer to neutralize the solution and generate an aqueous stock solution. Briefly, the peptide was taken up with enough NaOH to titrate the solution to pH 8-9 (with 10 mM NaOH) at which point the solution cleared. After sonicating for 2 min, concentrated buffer was added to create an aqueous Aâ40 stock at pH 7.4 (10 mM phosphate, 100 mM NaCl) which was stored at -20 °C until use. AFM Specimen Preparation. Aliquots of 3-5 μL were removed from peptide incubations and placed on freshly cleaved mica. After incubating (30 s to 2 min), the remaining suspension was removed by rinsing (twice with 50 μL of water) to remove salt and loosely bound peptide. Excess water was removed with a gentle stream of difluoroethane (Dust-Off Plus, Falcon Safety Products Inc.), and the samples were stored in a covered container to protect them from contamination until they were imaged (within 1-2 h). Optimization of Protofibril Adsorption Density. For accurate length measurements, specimens with dispersed protofibrils were produced by diluting the samples at 1525 μM between 30 s and 2 min prior to specimen preparation. The amount of dilution and the incubation time were kept constant within each series of measurements (see below). Atomic Force Microscopy. All images were obtained under ambient conditions with a Nanoscope IIIa Multimode scanning probe workstation (Digital Instruments, Santa Barbara, CA) operating in TappingMode using etched silicon NanoProbes (probe model FESP, Digital Instruments). Scanning parameters varied with individual tips and samples, but typical ranges were as follows: initial root-mean-square amplitude, 1.6 V; setpoint, 1.1-1.4 V; tapping frequency, 70-90 kHz; scan rate, 1.5-2 Hz. Consecutive scans were monitored until distortion due to creep or shifts in the slow scan direction were negligible before collecting scans at sizes of 1 μm with the maximum 512 × 512 pixel resolution (occasionally 2 μm scan sizes were used when protofibril adsorption density was too low to find 50 protofibrils in a 1 μm square region). Preparation and Application of Carbon Nanotube AFM Tips. Carbon nanotube tips were fabricated as reported earlier (20). Counting Adsorbed Species and Measurement of AVerage Protofibril Length and Longest Protofibril. Protofibrils were measured in fields containing 50-100 assemblies by summing the lengths of short line segments (in top view mode in the Nanoscope software) to approximate the curvature of protofibrils. Globular aggregates measuring <20 nm and showing no evidence of elongation were counted in each region and included in the calculation of average protofibril length as described below. Only protofibrils that could clearly be distinguished (including ∼4 nm high globular aggregates, which were assumed to be roughly spherical, with diameter ) length ) 4.3 nm) were included in the average. Before calculating the average, 8.5 nm was subtracted from the length of the elongated features to compensate for the contribution of tip-related broadening artifacts [calculated by subtracting the approximate protofibril height (4.3 nm) from the average protofibril width measured at half-height (12.8 ( 2.5 nm)] (20). The adjusted lengths were summed In Vitro Assembly of Aâ Amyloid Protofibrils Biochemistry, Vol. 38, No. 28, 1999 8973