A Conformation Change in the Carboxyl Terminus of Alzheimer's Aβ(1–40) Accompanies the Transition from Dimer to Fibril as Revealed by Fluorescence Quenching Analysis*

Alzheimer's disease is characterized by the presence of insoluble, fibrous deposits composed principally of amyloid β (Aβ) peptide. A number of studies have provided information on the fibril structure and on the factors affecting fiber formation, but the details of the fibril structure are not known. We used fluorescence quenching to investigate the solvent accessibility and surface charge of the soluble Aβ(1–40) dimer and amyloid fibrils. Analogs of Aβ(1–40) containing a single tryptophan were synthesized by substituting residues at positions 4, 10, 34, and 40 with tryptophan. Quenching measurements in the dimeric state indicate that the amino-terminal analogs (AβF4W and AβY10W) are accessible to polar quenchers, and the more carboxyl-terminal analog AβV34W is less accessible. AβV40W, on the other hand, exhibits a low degree of quenching, indicating that this residue is highly shielded from the solvent in the dimeric state. Correcting for the effect of reduced translational and rotational diffusion, fibril formation was associated with a selective increase in solvent exposure of residues 34 and 40, suggesting that a conformation change may take place in the carboxyl-terminal region coincident with the dimer to fibril transition.

Alzheimer's disease is a progressive neurodegenerative disease that is characterized by the abnormal accumulation of ␤-amyloid in senile plaques. Biochemical analysis of the amyloid peptides isolated from Alzheimer's disease brain indicates that amyloid ␤ (A␤) 1   is the principal species associated with senile plaque amyloid (1), whereas A␤(1-40) is more abundant in a soluble form in cerebrospinal fluid. Synthetic A␤(1-40) is relatively soluble, and its aggregation and assembly is a dynamic process, with a number of factors affecting the rate and equilibrium of fibril assembly (2). Key parameters promoting the assembly of amyloid fibril and sedimentable aggregates include high peptide concentration, long incubation times, low pH (pH [5][6], and mechanical agitation (3)(4)(5)(6)(7)(8). The length of the carboxyl terminus is also critical in determining the assembly dynamics. The longer A␤(1-42) isoform aggregates more rapidly at pH 7.4 (4,7). These observations suggest that aggregation of A␤ may be a critical event in pathogenesis.
The amyloid fibril was shown to be a ␤-pleated sheet structure using x-ray diffraction (6). These studies established that the amyloid fibril is made of an orthogonal lattice of ␤-crystallites having unit dimensions of 9.4 Å in a hydrogen bond direction, 7 Å in the polypeptide backbone direction, and 10 Å inter-sheet spacing, arranged in a cylindrical fashion (6). The peptide is organized in a cross ␤ pattern in which the hydrogen bonding direction is parallel to the fiber axis. The A␤  sequence is divided in two regions. Residues 1-28 make up a relatively hydrophilic domain with a high proportion of charged residues (46%). In the amyloid precursor protein, this domain is extracellular. The carboxyl-terminal 28 -40 residues make up a richly hydrophobic domain that is associated with the cell membrane in the amyloid precursor protein (2). Replacement of hydrophobic residues by hydrophilic residues markedly stabilizes A␤ peptides against aggregation (8), whereas replacement of hydrophilic residues by hydrophobic residues alters the morphology of the fibril, suggesting that hydrophilic residues are largely responsible for the specificity of intermolecular interaction within the fibril (8).
Amyloid fibril formation may involve two basic steps, the initial nucleation of aggregates that establishes the amyloid fibril lattice (6), followed by the elongation of the fibril by the sequential addition of subunits (9). Previous studies indicate that A␤(1-40) forms stable dimers in solution (10 -13) and suggest that dimerization is the initial event in amyloid aggregation and that the dimer represents the fundamental building block for further fibril assembly. However, much remains to be elucidated regarding the structure of the amyloid fibril and the mechanism of amyloid dimer-fibril transition. In this work, fluorescence spectroscopy was used to assess solvent exposure and surface charge around discrete sites along the peptidyl backbone of A␤  in both the dimeric and fibrillar states. Specifically, a functional set of single-residue Trp replacement analogs of A␤(1-40) was synthesized, and the capacity of cationic Cs ϩ , anionic I Ϫ , and neutral acrylamide to dynamically quench the emission from the substituted Trps in these analogs was measured (20). Determining which parts of A␤ are accessible to the solvent and how these environments change during polymerization will be important to understand * This work was supported by National Institutes of Health Grant NS31230 and by the Minority Biomedical Researchers Program, National Institutes of Health Grant GM 55246. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ § To whom correspondence should be addressed. Tel.: 949-824-6081; Fax: 949-824-8551; E-mail: cglabe@uci.edu. 1 The how the peptide is organized within the fiber and to identify critical contact sites that are necessary for higher order assembly into fibrils.

EXPERIMENTAL PROCEDURES
Materials-All peptides were synthesized by fluorenyl-9-ylmethoxycarbonyl chemistry using a continuous flow semiautomatic instrument as described previously (4). A single tryptophan residue was substituted into the sequence of the A␤(1-40) peptide at positions 4 (A␤F4W), 10 (A␤Y10W), 34 (A␤L34W), and 40 (A␤V40W). The peptides were purified by reverse-phase high performance liquid chromatography. The purity and expected structure was verified by electrospray mass spectrometry. Only peptides exhibiting 90.0% or greater purity with less than 5.0% of a single contaminant were used. All other reagents were of the highest analytical grade commercially available.
Aggregation Measurements-Aggregation was determined by using a sedimentation assay as described previously (4,11). A␤ Fibril Formation Assays-Fibril formation was monitored using a thioflavin T (ThT) fluorescence assay (14,15). The peptides (230 M) were incubated for 24 h at room temperature in 50 mM Tris, 0.1 M NaCl, pH 7.4, with continuous stirring. After 24 h, 10-l aliquots of each sample were transferred to a cuvette containing 2 ml of 3 M ThT, pH 7.4. The fluorescence emission was monitored at 482 nm with excitation at 450 nm using a Spex Fluorolog-2 spectrofluorometer. For electron microscopy analysis, the peptides were incubated for 24 h at room temperature with continuous stirring, at a concentration of 230 M in 10 mM MOPS, pH 7.4. A drop of each sample was placed on a carbon coated copper grid, negatively stained with 2% aqueous uranyl acetate, and visualized with a Zeiss 10CR microscope (80 kV).
Fluorescence Measurements-Fluorescence spectra were measured with an Aminco SLM 4800 spectrofluorometer. Tryptophan emission from of the A␤(1-40) Trp analogs was recorded between 310 and 550 nm, with excitation at 295 nm. Fluorescence lifetimes were determined by the time-correlated single photon counting technique using an EEY scientific nanosecond fluorescence spectrofluorometer (La Jolla, CA), equipped with an IBH hydrogen arc lamp (Glasgow, United Kingdom). Emission decays were analyzed with the GLOBALS UNLIMITED computer program (version 1.01; Laboratory of Fluorescence Dynamics, Urbana-Champaign, IL).
Fluorescence Quenching Experiments-Steady-state fluorescence quenching experiments were performed using two different conditions: with the dimeric and fibrillar peptides. For the dimer, aliquots of the stock quenching solutions (5 M) were added into a 0.5 ϫ 0.5-cm cuvette containing 10 M of the peptide in Buffer A (0.1 M NaCl, 20 mM Tris, pH 7.4). For the fibrillar state measurements, the peptides (230 M) were previously incubated for 24 h in Buffer A with continuous stirring. Steady-state quenching experiments were performed with excitation at 295 nm, emission at 350 nm. Corrections were made for dilution. Stock quenching solutions of KI, acrylamide, and CsCl were freshly prepared at 5 M. Quenching data were fit to the Stern-Volmer equation, where F o and F are the fluorescence intensity in the absence and in the presence of quencher [Q], respectively, K SV is the Stern-Volmer quenching constant. The apparent bimolecular quenching rate constant, k q , a measure of solute accessibility, was calculated with the following expression, where ϽϾ is the geometric average fluorescence lifetime obtained from time-resolved measurements (16,17).

Aggregation and Fibril Forming Properties of A␤(1-40) Tryp-
tophan Analogs-To validate the utility of the quenching experiments employing A␤ substitution analogs, the aggregation properties of the Trp substitution analogs were characterized. A␤(1-40) was chosen for these studies because its dimeric state is stable over the time frame of the experiments and the transition to the fibrillar state can be accomplished by stirring the peptide for 24 h under the same physiological conditions (10 -13).
The oligomeric assembly state of the amyloid tryptophan analogs was characterized by gel filtration chromatography. The peptides eluted at the same position as wild type A␤(1-40) (Fig. 1). The elution position corresponded to an apparent molecular mass of 9 kDa, determined by the elution behavior of a series of calibration standards as reported previously (8, 10 -13) (Fig. 1, inset). The other tryptophan analogs used in this study also eluted as a dimer (data not shown).
Aggregation was determined using a sedimentation assay as described under "Experimental Procedures." The aggregation properties of the tryptophan analogs and wild type A␤(1-40) are shown in Fig. 2, under several conditions that are known to modulate the assembly state of A␤. At pH 7.4, in the presence or absence of zinc ions, and at pH 5.0, the sedimentation behavior of all of the tryptophan analogs used is comparable to wild type A␤ . Several other substitution analogs were synthesized, but they exhibited significantly altered aggregation properties and were not studied further (data not shown).
Thioflavin T binding assays were also used to follow the progress of amyloid fibril formation for the tryptophan analogs. In comparison to the wild type A␤(1-40), all the tryptophan analogs were able to bind thioflavin T as judged by the high fluorescence, but A␤Y10W showed a significantly lower fluorescence (Fig. 3). To further verify the fibril forming properties of the tryptophan analogs, their structures were examined by electron microscopy. Electron microscopy indicated that the tryptophan analogs A␤Y10W and A␤L34W form fibrils that are indistinguishable from wild type amyloid fibrils (Fig. 4). The other two tryptophan analogs were also able to form morphologically normal fibrils (data not shown). The fact that A␤Y10W displays a low yield of fluorescence in the ThT assay but forms morphologically normal fibrils suggests that tyrosine 10 may play a critical role in ThT binding.
Steady-state Fluorescence Quenching Studies-The assembly of a polypeptide chain such as A␤(1-40) to form a relatively stable dimer or a fibril is frequently associated with the shielding of certain amino acid residues from the external aqueous environment. Other residues lie on the surface, where they are exposed to the polar solvent. We employed a strategy often used in studying the solution structure of proteins or peptides to map those residues that are exposed and those that are buried in the dimeric and fibrillar A␤ (1-40). The exposure of the single tryptophan moiety in different peptide assembly states was measured by using neutral, anionic, and cationic quench-ers for each of the A␤(1-40) analogs. Stern-Volmer plots of fluorescence quenching of A␤(1-40) tryptophan analogs are shown in Fig. 5 for dimeric and fibrillar peptides with three different quenchers. The lines in Fig. 5 were obtained by fitting the data to the Stern-Volmer equation. Linear Stern-Volmer plots (linear correlation coefficients Ͼ0.98) indicated that the quenching was dynamic and that no detectable static quenching occurred. Table I summarizes the collisional quenching parameters for the four tryptophan analogs in both dimeric and fibrillar states with iodide, cesium, and acrylamide quenchers: Stern-Volmer quenching constant (K SV ), the bimolecular quenching rate constant (k q ), and the fluorescence lifetime ().
Acrylamide is a polar, uncharged dynamic quenching agent that in an aqueous phase interacts with all surfaces, independent of charge. Consequently, for proteins and peptides, the biomolecular quenching rate constant (k q ) for acrylamide quenching of tryptophan emission is a measure solvent exposure of tryptophan residues. Merril et al. (21) reported that acrylamide quenching of NATA in aqueous solution was highly efficient (k q ϭ 8 -8.5 M Ϫ1 ns -1 ). With acrylamide, the degree of Trp solvent exposure to solvent is defined by Merrill et al. (21) as follows: exposed to the solvent, 1.5 Ͻ k q Ͻ 5; moderately exposed, 0.6 Ͻ k q Ͻ 1.5; moderately buried, 0.2 Ͻ k q Ͻ 0.6; and buried, k q Ͻ 0.2.
With acrylamide as a quenching agent, the k q values for the dimeric A␤F4W, A␤Y10W, and A␤L34W (3.5, 3.3, and 2.4, respectively) indicated significant solvent exposure but only moderate exposure for A␤V40W (k q ϭ 0.6 M Ϫ1 ns Ϫ1 ). In good agreement with the A␤(1-42) dimeric model (22) in which only two amino acids (Val-36 and Val-40) contribute to the hydrophobic core, the rest of the side chains are exposed to the aqueous environment. These same analogs were also analyzed in the fibrillar state. Formation of fibrils should be associated with about a 50% reduction in k q , as the fluorophore in the much larger fibril will have dramatically lower translational and rotational diffusion rates (31). As expected if just the diffusion rates decrease without any change in the solvent exposure at positions 4 and 10, the k q values decreased about 50% upon fibril formation for acrylamide quenching of A␤F4W and A␤Y10W (Table I and Fig. 6). In contrast, the k q values increased 29 and 450% for A␤L34W and A␤V40W, respectively, indicating an increase in the solvent exposure for these analogs, especially at position 40.
The surface charge surrounding the tryptophan residues in the A␤(1-40) tryptophan analogs was assessed by employing ionic quenchers. The capacity of anionic iodide to quench tryptophan emission is affected by both solvent exposure and the surface charge adjacent to the indole moiety of the tryptophan (16,18). In the absence of surface charge effects, k q values for iodide quenching of tryptophan greater than 1.5 are indicative of high solvent exposure, whereas k q values less than 1.5 are indicative of partially exposed tryptophans (20). The relative susceptibility of the substituted tryptophans in the dimeric peptides to iodide quenching is similar to that observed with acrylamide and suggests that surface charge has little effect on the iodide quenching (Table I and Fig. 6).
The formation of fibrils was associated with differential changes in the surface charge adjacent to the substituted tryptophans. Correcting for the effect of reduced translational and rotational diffusion (31), fibrillar A␤Y4W should have been associated with a k q value of ϳ0.8 nM Ϫ1 s Ϫ1 , whereas the observed value was significantly lower, 0.3 nM Ϫ1 s Ϫ1 , suggesting a small fibrillar formation-induced increase in the negative surface charge near position 4. Similarly, if only translational and rotational diffusion rates changed upon fibril formation, then the k q values for the A␤Y10W, A␤L34W, and A␤L40W should have been 0.8, 0.5, and 0.2 nM Ϫ1 s Ϫ1 , respectively. However, the k q values were observed to be significantly higher, 3.8, 2.3, and 0.4 nM Ϫ1 s Ϫ1 , respectively (Table I and Fig. 6), suggesting an increase in the positive surface charge near positions 10, 34, and 40 following fibril formation.
Although the k q values for cesium quenching of the dimeric analogs are 80 -86% less than the respective k q values for acrylamide quenching, their rank order is the essentially same as that of acrylamide (A␤Y4W Ϸ A␤Y10W Ͼ A␤L34W Ͼ Ͼ A␤L40W) (Table I). This suggests that the surface charge about each substituted tryptophan has little effect on cesium quenching of the dimeric analogs and that cesium is intrinsically a less efficient tryptophan quencher than acrylamide.
Fibril formation was associated with differential effects on the capacity of cesium to quench the emission from the substituted tryptophan analogs. Assuming that there were no changes in the surface charge upon fibril formation and that the only changes in quenching are due to reduced translational and rotational diffusion (31), the cesium k q values should have been about 0.25, 0.26, 0.18, and 0.07 nM Ϫ1 s Ϫ1 for A␤Y4W, A␤Y10W, A␤L34W, and A␤L40W, respectively. However, the observed k q values for the A␤Y4W, A␤Y10W, and A␤L34W analogs were 2.3-3.4 times higher and 2.5 times lower for the A␤L40W analog that what would be expected if there were changes in surface charge around the substituted tryptophans    tryptophan analogs Fluorescence emission intensities were recorded from each analog at varying concentrations of acrylamide, KI, or CsCl. The Stern-Volmer dynamic quenching constants (K SV ) were calculated from the slopes of the data plots in Fig. 6, and bimolecular collisional (k q ) quenching constants were obtained using Equations 1 and 2. Also shown are the values of the maximum emission wavelengths ( cm ) and fluorescence lifetimes () of dimeric and fibrillar A␤(1-40) tryptophan analogs. em Acrylamide KI CsCl upon fibril formation (Table I and Fig. 6). This suggests that fibril formation is associated with an increased negative charge around positions 4, 10, and 34 and an increase in positive charge around position 40. These results are only partially consistent with the iodide quenching results. Whereas both the iodide and cesium results indicated increase negative surface charge around position 4 and increased negative charge around position 40, the iodide suggested increased positive and the cesium increased negative surface charge around positions 10 and 34. Collectively, the cesium and iodide quenching results would indicate that fibril formation was associated with a homogeneous increases in negative surface charge around position 4 and positive surface charge around position 40 but inhomogeneous increase in both positive and negative charge around positions 10 and 34. DISCUSSION The primary objective of the present work was to map the solvent-accessible surface of A␤  in the dimeric and fibrillar states by using fluorescent quenching analysis. The aromatic amino acids that are present at positions 4 and 10 (Phe and Tyr) of A␤(1-40) were substituted with a tryptophan. Hydrophobic carboxyl-terminal residues 34 and 40 (Met and Val) were also substituted with tryptophan. Substitution of the other aromatic residues at positions 19 and 20 significantly inhibited the aggregation of the peptides and were not studied further. 2 This general strategy has been used to study the structure and function of proteins and peptides from which the Trp aromatic amino acid is absent (21,(23)(24)(25)(26). This approach was recently used (27) to study ligand dependent changes in the accessibility of P-glycoprotein induced by different antitumor agents by using quenching analysis. They found that upon addition of the substrate, the enzyme adopts a different tertiary structure, resulting in a significantly increased solvent accessibility.
The single tryptophan analogs examined in this work can be classified into two groups on the basis of their sensitivity to collisional quenchers: exposed and minimally exposed. Tryptophan residues in dimeric A␤F4W, A␤Y10W, and A␤L34W are exposed and display high bimolecular quenching rate constants with acrylamide and iodide (Fig. 6). Comparison of the degree of quenching of these residues with iodide and cesium indicates that in the dimeric state, residues 4 and 10 are surrounded by positively and negatively charged amino acids. This is in good agreement with the published model of the A␤ dimer (22), in which the negatively charged residues (Asp-1, Glu-3, Asp-7, Glu-11, and Asp-23) are on the surface of the molecule, and the basic residues (His-13, His-14, and Lys-16) represent a positively charged domain on both outer surfaces of the participating ␤ sheets. This positively charged cluster appears to participate in the activation of microglia (28), which are involved in the inflammatory response to A␤ observed in Alzheimer's disease. However, the tryptophan residue in dimeric A␤V40W seems to be relatively inaccessible to the solvent. The carboxylterminal residues 28 -40 constitute a richly hydrophobic domain. Accompanying the transition to the fibrillar state, A␤ tryptophan analogs at positions 4 and 10 are less exposed, consistent with this region participating in assembly as a contact site. Residues 10 and 34 are surrounded by both positive and negative charges, position 4 by primarily negative charges, and position 40 by positive charges. The tryptophan residue at position 40 is highly exposed to the solvent in the fibrillar state but is significantly less exposed in the dimeric state (Fig. 6). The measurements described in this paper clearly suggest that this change in exposure is indicative of a conformational change that occurs in this region accompanying the dimer to fibril transition, which may be an important step in amyloid assembly. The existence of such a conformational change could be due to the involvement of distinct peptide domains in the association of amyloid dimers, which eventually lead to higher order aggregates and subsequent fibril formation.
This observation of a conformation change is consistent with previous studies of the temperature dependence of fibril formation, which suggested such a conformational change on the basis of the unusually high activation energy for the transition (29). Although these studies were conducted in 0.1 M HCl, we also find evidence of a conformational change under more physiological conditions. Our data further indicate that this conformation change takes place at the carboxyl terminus. This conformation change may confound attempts to elucidate the structure of the A␤ peptide in amyloid fibrils, because most of the molecular models have assumed that no change in conformation takes place after initial dimerization (22,30).