Amino-terminal Deletions Enhance Aggregation of β-Amyloid Peptides in Vitro

β-Amyloid protein, which assembles into pathological aggregates deposited in Alzheimer's disease brain tissue, exhibits N-terminal heterogeneity both in vitro and in vivo. To investigate the effects of this N-terminal heterogeneity on the assembly characteristics and biophysical properties of β-amyloid, we synthesized a series of peptides with progressively shortened N termini (initial residues at positions β1, β4, β8, β12, and β17) and C termini extending to residue β40 or β42. We report that peptides with N-terminal deletions exhibit enhanced peptide aggregation relative to full-length species, as quantitatively assessed by sedimentation analyses. Overall, sedimentation levels were greater for peptides terminating at residue β42 than for those terminating at residue β40. To determine if established biophysical features of the full-length protein were maintained in the truncated peptides, structural and bioactive properties of these peptides were examined and compared. Full-length and truncated peptides exhibiting aggregation showed circular dichroism spectra consistent with predominant β-sheet conformation, fibrillar morphology under transmission electron microscopy, and significant toxicity in cultures of rat hippocampal neurons. These data demonstrate that N-terminal deletions enhance aggregation of β-amyloid into neurotoxic, β-sheet fibrils and suggest that such peptides may initiate and/or nucleate the pathological deposition of β-amyloid.

␤-Amyloid protein, which assembles into pathological aggregates deposited in Alzheimer's disease brain tissue, exhibits N-terminal heterogeneity both in vitro and in vivo. To investigate the effects of this N-terminal heterogeneity on the assembly characteristics and biophysical properties of ␤-amyloid, we synthesized a series of peptides with progressively shortened N termini (initial residues at positions ␤1, ␤4, ␤8, ␤12, and ␤17) and C termini extending to residue ␤40 or ␤42. We report that peptides with N-terminal deletions exhibit enhanced peptide aggregation relative to full-length species, as quantitatively assessed by sedimentation analyses. Overall, sedimentation levels were greater for peptides terminating at residue ␤42 than for those terminating at residue ␤40. To determine if established biophysical features of the full-length protein were maintained in the truncated peptides, structural and bioactive properties of these peptides were examined and compared. Fulllength and truncated peptides exhibiting aggregation showed circular dichroism spectra consistent with predominant ␤-sheet conformation, fibrillar morphology under transmission electron microscopy, and significant toxicity in cultures of rat hippocampal neurons. These data demonstrate that N-terminal deletions enhance aggregation of ␤-amyloid into neurotoxic, ␤-sheet fibrils and suggest that such peptides may initiate and/or nucleate the pathological deposition of ␤-amyloid.
␤-Amyloid (A␤) 1 is a normal, soluble protein 40 -42 amino acid residues in length that, in neuropathological conditions such as Alzheimer's disease (AD), self-assembles into insoluble fibrils, forming characteristic extracellular deposits termed senile plaques (1). Since a variety of histological, molecular genetic, and in vitro and in vivo studies provide evidence consistent with the possibility that A␤ significantly contributes to the initiation and/or progression of neurodegenerative changes in AD (2)(3)(4), investigation of the production, assembly, and bioactivity of A␤ is crucial for the successful understanding of and therapeutic intervention in AD and related disorders.
A␤ is derived from proteolytic processing of its precursor, A␤ precursor protein (A␤PP), by least two distinct and incompletely defined pathways (see Ref. 5 for review). In the secretory pathway, transmembrane A␤PP is cleaved by ␣-secretase between ␤16 and ␤17 (6), thus precluding the formation of full-length A␤ (␤1-40/42) but generating a 3-kDa ␤17-40/42 fragment (7)(8)(9)(10). In the endosomal/lysosomal pathway, A␤PP is degraded into several C-terminal fragments containing the complete A␤ sequence (11,12) that require additional cleavage at the A␤ amino (␤-secretase) and carboxyl (␥-secretase) termini to generate A␤. Although the predominant form of A␤ contains the ␤1-40 sequence, significant N-and C-terminal heterogeneity has been reported recently in studies of both cell culture (7,9,(13)(14)(15) and human fluids and tissues (13, 16 -23), suggesting multiple forms of A␤ that vary in primary structure by a few to several amino acids.
Previous studies suggest that C-terminal heterogeneity of A␤ has profound effects upon the initiation and progression of AD. Specifically, increased length of the hydrophobic C terminus both enhances in vitro aggregation of A␤ (24 -26) and appears to promote early deposition of plaque A␤ in AD brains (27). In addition, an increased relative production of longer C-terminal forms of A␤, demonstrated both in vitro (14) and in AD brain (28), appears to underlie the pathologic action of A␤PP Val-717 mutations linked to early onset familial AD.
The effects of N-terminal heterogeneity of A␤ on its assembly and bioactivity characteristics are not well defined. The significance of this issue is underscored by recent data suggesting that early stage plaques may be composed primarily of A␤ peptides with truncated N termini (23,29). The possibility of enhanced amyloidogenicity of A␤ peptides with truncated N termini is consistent with a previous study that reported decreased in vitro solubilities of synthetic A␤ peptides ␤8 -, ␤9 -, and ␤10 -43 relative to ␤1-, ␤2-, and ␤4 -43 (30).
In order to examine the effects of A␤ N-terminal heterogeneity on both peptide assembly and bioactivity, we have synthesized two series of A␤ peptides with progressive N-terminal deletions (beginning at positions ␤1, ␤4, ␤8, ␤12, and ␤17), one of which terminates at residue ␤40 (␤n-40 series) and the other at ␤42 (␤n-42 series). These A␤ peptides have been examined for their rates and levels of peptide assembly, fibrillar ultrastructures under electron microscope, secondary structures by circular dichroism, and bioactivities with cultured neurons.

MATERIALS AND METHODS
Synthetic Peptides-Peptides were synthesized by solid-phase Fmoc (N-(9-fluorenyl)methoxycarbonyl) amino acid chemistry using a continuous flow semiautomatic instrument, as described previously (25). All peptides were synthesized within a single lot, purified by reverse-phase high performance liquid chromatography, and solubilized at 250 M in sterile double deionized H 2 O (ddH 2 O); the purity of this lot was estimated by electrospray mass spectrometry and amino acid sequencing to be approximately 70% expected product. Due to their solubility characteristics, only trace amounts of purified ␤17-40 and ␤17-42 were recovered in the initial synthesis; thus, these peptides were synthesized a second time, partially purified by ether precipitation, and initially solubilized at 25 mM in 1,1,1,3,3,3-hexafluoro-2-propanol prior to dilution to 250 M in ddH 2 O. It is unlikely that these differences in synthesis lot, purification, and solubilization introduced a significant source of variability between the peptides, since fractions of ␤17-40 recovered from the initial lot behaved in a manner nearly indistinguish-able from that in the second lot. Based upon findings from our previous studies (31,32), we examined the peptides at three time points: (i) immediately following solubilization; (ii) 2 days after solubilization; (iii) 7 days following solubilization.
Sedimentation Assay-The proportion of pelletable peptide was determined using our previously described technique (32). Briefly, peptide stock solutions were diluted to 25 M samples in 20 mM MOPS buffer (pH 7.3) and then ultracentrifuged for 1 h at 1 ϫ 10 5 g. The ratio of protein concentrations, as determined by fluorescamine assay, of the supernatant relative to non-centrifuged peptide was used to calculate the levels of sedimentable peptide. Data represent the mean of triplicate samples statistically compared by analysis of variance.
Electron Microscopy-For electron microscope analysis (32), A␤ peptides were diluted from stock solutions to 25 M in 20 mM MOPS buffer (pH 7.3) and equilibrated for 1 h. Peptide samples were then adsorbed onto carbon-stabilized, Formvar-coated grids, rinsed with ddH 2 O, and stained with 2% (w/v) uranyl acetate. Samples were viewed under a Ziess 10CR transmission electron microscope at ϫ80,000 (80 kV).
Circular Dichroism-To study secondary structure of A␤ peptides in solution, circular dichroism (CD) spectra were examined using a Jasco J-720 spectropolarimeter, as described previously (32). Briefly, stock solutions of A␤ peptides were diluted to 25 M in 5 mM potassium phosphate (pH 7.3) and then loaded into a 1-cm pathlength quartz cell. Measurements were taken at 0.5-nm steps over a 195-250-nm wavelength range, averaged over eight scans, and subtracted from base-line (buffer only) values. Data are presented as mean residue ellipticity (degrees⅐cm 2 /dmol).
Cell Culture-Primary cultures of hippocampal neurons were established from embryonic day 18 Sprague-Dawley rat pups, as described previously (31). Cultures were maintained in serum-free Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with N2 components. After 2 days in vitro, cultures were exposed to 25 M A␤ peptides; cell survival was determined 48 h later based on counts of viable cells using the ethidium homodimer/calcein AM (Molecular Probes, Eugene, OR) combination of vital dyes (31,33,34). Experiments were performed in triplicate wells and repeated in at least three independent experiments. Data were statistically examined with analysis of variance followed by pairwise comparisons using Scheffé F-test.

RESULTS AND DISCUSSION
Given the N-and C-terminal heterogeneity of A␤ isolated from AD brain tissue (16 -23) as well as the demonstrated profound influence of C-terminal heterogeneity on A␤ assembly in vitro (24 -26) and apparently in vivo (27), the potential influence of N-terminal heterogeneity on A␤ assembly is an intriguing possibility that may significantly contribute to the pathological deposition of A␤. In this study, we sought to determine (i) how the length of the N terminus influences the overall levels and kinetics of A␤ aggregation and (ii) whether aggregates of N-terminal truncated A␤ retain the biophysical properties of full-length A␤.
To investigate the influence of N-terminal deletions on A␤ assembly properties, we solubilized the ␤n-40 and ␤n-42 peptides and quantitatively determined the levels of peptide aggregates at various time points by sedimentation assay. As illustrated in Fig. 1A, the ␤n-40 peptides exhibited initially low levels of sedimentation, which gradually increased over 7 days. This progressive increase in peptide aggregation is consistent with previous observations of the time-dependent nature of A␤ aggregation (25,26,31).
In comparison to the ␤n-40 series, the ␤n-42 peptides exhibited more rapid and extensive assembly as evidenced by high initial sedimentation and early attainment of maintained peak values; maximal levels were reached within 2 days for ␤1-, ␤4 -, and ␤8 -42 and by the initial time point (i.e. within minutes of solubilization) for ␤12and ␤17-42 (Fig. 1B). The significantly greater amounts of sedimentable aggregates and more rapid kinetics observed in the ␤n-42 relative to the ␤n-40 peptides are consistent with the conclusions of previous studies that the length of the C terminus is a critical variable in A␤ assembly (24 -26, 31). More importantly, we observed a general trend in the ␤n-42 series of increasing levels of sedi-mentation with decreasing length of N terminus, with the exception of increased sedimentation levels for ␤8 -42 at the 2and 7-day time points; a similar relationship was also apparent in the ␤n-40 peptides. Note that in both series of peptides, the greatest levels of sedimentation were observed in those peptides beginning at either residue ␤8 or ␤17.
The increased relative aggregation of A␤ peptides with Nterminal deletions shown by the above data suggests that these and/or related peptides may be the initial A␤ species deposited in senile plaques; recent observations in AD brain support this possibility (27,29). Interestingly, A␤ deposited in early stage or diffuse plaques generally lacks the ␤-sheet fibrillar structure and associated degenerating neurites characteristic of classic senile plaques. Thus, we sought to determine whether N-terminal deletions might affect previously defined bioactive and structural properties of full-length A␤, including neurotoxicity, ␤-sheet conformation, and fibrillar structure.
Our previous studies have demonstrated that the presence of peptide aggregates in solutions of full-length A␤ accurately predicts significant neurotoxic activity (31,35,36). To determine whether this assembly/toxicity relationship also characterizes A␤ peptides with N-terminal deletions, we examined potential neurotoxic activities of the ␤n-40 and ␤n-42 peptides in cultured hippocampal neurons. Within the ␤n-40 peptide series, neurotoxicity induced by the truncated peptides was greater than that induced by ␤1-40 ( Fig. 2A). Neurotoxicity generally increased within the peptide series between the 0-and 7-day time points, a trend that parallels the sedimentation data.
Like the ␤n-40 peptides, the ␤n-42 peptides also exhibited significant neurotoxicity (Fig. 2B). The greater toxic activities of the ␤n-42 series may reflect in part their higher sedimentation levels. Toxicity of ␤n-42 peptides was generally greatest at day 0 and showed a significant decrease in intensity with decreasing length of the peptide N termini. At the day 7 time point, neurotoxicity was comparable between the different ␤n-42 peptides but generally reduced within individual peptides relative to the initial time point. The decreased neurotoxicities, both across the peptide series at day 0 and within individual peptides over the two time points, may reflect higher order assembly (i.e. aggregation of A␤ oligomers) of these A␤ peptide aggregates resulting in a corresponding decrease in aggregate-cell interactions, in agreement with previous observations (37). Thus, although the property of neurotoxicity for individual A␤ peptides is predicted well by their tendency to form sedimentable peptide aggregates, the relationship between these two factors does not necessarily exhibit a strict quantitative correlation (31).
In addition to neurotoxicity, aggregating A␤ peptides are also characterized by specific structural features, including ␤-sheet secondary structure (24,37) and fibrillar morphology (25, 38 -40); like aggregation measures, these factors have been demonstrated to be associated with A␤ neurotoxicity (32,37,41). The following structural studies focused on the ␤n-42 series, since these peptides not only exhibited significantly higher levels of sedimentation and neurotoxicity than the ␤n-40 series but also have been demonstrated to be the initial A␤ species deposited within AD plaques (27,29).
Examination of CD spectra was used to provide qualitative information regarding the secondary structure of ␤n-42 peptides in solution. Predominant ␤-sheet structure is recognized by a single negative peak near 218 nm, as opposed to the doublet of negative peaks near 208 and 222 nm observed in ␣-helical structure; both structures exhibit a single positive peak near 200 nm (42,43). An absence of ordered structure, random coil, is indicated by a single negative peak near 198 nm. Like ␤1-42, the truncated ␤n-42 peptides exhibit characteristics of predominant ␤-sheet structure (Fig. 3). Although the CD spectra of all the ␤n-42 peptides are similar, ␤8 -42 and ␤17-42 exhibit slightly greater negative peaks than the other ␤n-42 peptides; such enhanced peak values within predominant ␤-sheet solutions suggest higher ␤-sheet content resulting from a reduced proportion of random coil (43).
These CD observations verify that, like full-length ␤1-42, the truncated ␤n-42 peptide assemblies exist in the predominantly ␤-sheet conformations characteristic of amyloid fibrils. In addition, the CD spectra are consistent with the increased amyloidogenicity of A␤ peptides with N-terminal deletions suggested by the sedimentation data. In particular, the ␤8 -42 and ␤17-42 peptides exhibited both the greatest sedimentation levels and the highest apparent ␤-sheet content. The enhanced relative amyloidogenicity of these peptides likely reflects the influences of altered primary structure in promoting and/or stabilizing the ␤-sheet conformation underlying A␤ fibril formation; this conclusion is consistent with recent data demonstrating significant contributions by N-terminal residues to A␤ amyloidogenicity (44). Further studies are required to determine the relevant contributing factors (e.g. increased hydrophobicity, improved interstrand registration).
In order to compare further the amyloidogenic nature of the truncated ␤n-42 peptides relative to full-length ␤1-42, we examined their aggregate structures under electron microscopy. Previous studies have shown that several synthetic A␤ peptides form 5-10-nm diameter fibrils in vitro (25, 38 -40, 41), which are morphologically similar to A␤ fibrils from AD plaques (38,45). We observed that all truncated ␤n-42 peptides exhibited negatively stained fibrils with morphological features similar to ␤1-42 and consistent with previous observations of A␤ fibrils (Fig. 4). Interestingly, although ␤17-42 fibrils usually exhibited morphology comparable with fibrils of the other ␤n-42 peptides, ␤17-42 fibrils in some samples appeared to be relatively shorter in length and narrower in diameter (data not shown). Similar discrete populations of relatively wide and narrow A␤ fibrils have been observed in plaque core preparations (45). In addition, ␤17-42 fibrils were orga-  (Fig. 1A). Note the enhanced toxicity of ␤n-40 deleted peptides relative to ␤1-40. B, ␤n-42 peptides cause the greatest cell loss at day 0, a time point associated with maximal or near maximal sedimentation values (Fig. 1B). At day 7, all ␤n-42 peptides induce comparable but diminished levels of toxicity. * denotes significant (p Ͻ 0.05) differences in cell viability relative to untreated controls; ** denotes significant (p Ͻ 0.05) differences relative to ␤1-42 at the day 0 time point. nized into exceptionally large, dense meshworks (Fig. 4), consistent with observations of ␤17-40 fibrils recently reported by Nä slund et al. (46). However, these authors also reported (46) that ␤17-40 aggregates lack the positive thioflavine staining characteristic of all amyloids. In contrast, using our thioflavine assay (32), we have observed that aggregates formed by all the ␤n-40 and ␤n-42 peptides exhibit positive thioflavine staining. 2 In this paper, we have provided experimental evidence demonstrating a significant influence of N-terminal length on A␤ peptide assembly. Specifically, A␤ peptides with N-terminal deletions exhibit enhanced levels of aggregation in comparison with full-length A␤ peptides, as quantitatively assessed by sedimentation assay. Despite the variable absence of N-terminal residues, the truncated peptides retain the neurotoxicity and ␤-sheet, fibrillar structure associated with aggregated fulllength A␤. The structure/activity relationship suggested by these data may impact current concepts regarding the pathogenic potential of A␤ peptide fragments. The prevailing doctrine within the literature has been that ␣-secretase cleavage, which cleaves A␤PP between A␤ residues ␤16 and ␤17 (6), is the preferred proteolytic pathway in terms of minimizing disease potential since it generates the theoretically benign ␤17-40/42 as opposed to the more pathologic ␤1-40/42. In contrast to this position, our data predict that ␤17-40/42 is actually more likely than ␤1-40/42 to assemble into deposits in vivo and that the shortened peptide retains significant neurotoxic activity. Consistent with the idea of enhanced amyloidogenicity, recent data by Gowing et al. (29) suggest that ␤17-42 is a primary component of early stage, diffuse plaques. We suggest that ␤17-42 and other prevalent ␤n-42 peptides may initiate and/or accelerate plaque formation, perhaps by acting as nucleating centers that seed the subsequent deposition of relatively less amyloidogenic but apparently more abundant fulllength A␤; similar seeding events have been described previously for A␤ peptides in vitro (26).
In summary, the present findings predict that N-terminal heterogeneity of A␤ peptides, demonstrated to occur both in vitro (7,9,13,15) and in AD brain (16,18,(21)(22)(23), may accelerate A␤ deposition into plaques. Thus, proteolytic events contributing to the cleavage of A␤PP within the N-terminal domain of A␤ (e.g. activities of ␣and ␤-secretases) may be of considerable significance in the pathogenesis of AD and related disorders.