Molecular engineering of a secreted, highly homogeneous, and neurotoxic aβ dimer

Abstract

Aβ oligomers play a key role in the pathophysiology of Alzheimer's disease. Research into structure-function relationships of Aβ oligomers has been hampered by the lack of large amounts of homogeneous and stable material. Using computational chemistry, we designed conservative cysteine substitutions in Aβ aiming at accelerating and stabilizing assembly of Aβ dimers by an intermolecular disulfide bond without changing its folding. Molecular dynamics simulations suggested that mutants AβS8C and AβM35C exhibited structural properties similar to those of Aβ wildtype dimers. Full length, mutant APP was stably expressed in transfected cell lines to study assembly of Aβ oligomers in the physiological, secretory pathway and to avoid artifacts resulting from simultaneous in vitro oxidation and aggregation. Biochemical and neurophysiological analysis of supernatants indicated that AβS8C generated an exclusive, homogeneous, and neurotoxic dimer, whereas AβM35C assembled into dimers, tetramers, and higher oligomers. Thus, molecular engineering enabled generation of bioactive, homogeneous, and correctly processed Aβ dimers in vivo.

Keywords: Alzheimer’s disease; Aβ oligomers; computational chemistry; dimers; molecular engineering; neurotoxicity.

Figure 1

Schematic presentation of the Aβ processing and aggregation pathway depicting the sites of Aβ mutations investigated. (a) APP dimer showing that S26 and M35 are located in the transmembrane helix close to the G29−G33 dimerization motif (shown as gray rectangle). S8 is located in the extracellular part of APP. (b) After processing of APP, Aβ undergoes an α-helix to β-sheet conversion and predominantly associates via parallel β-sheets (schematically shown as orange bars). The dimers are generally conformational heterogeneous (c), and the elongation competent U-shaped conformation (left) is in equilibrium with alternative folds (middle, right), which are rather elongation-incompetent and are the most likely candidates for toxic Aβ dimers.

Figure 2

Molecular structure of the most populated Aβ conformation for wildtype (a), S8C (b), S26C (c), and M35C (d) that was present in the respective molecular dynamics simulations. The two chains of the Aβ dimer are shown as red and blue tubes. Residues L17 (green), F19 (orange), and A21 (yellow) of the central hydrophobic core are shown as a space-filled presentation. To emphasize the different degree of solvent accessibility of these core residues, the molecular surface of all remaining residues is shown as a translucent surface. See Supporting Information Table 1 for a detailed analysis of the solvent accessibility.

Figure 3

(a) Western blot of immunoprecipitated Aβ monomers and SDS-stable oligomers derived from SNs of permanently transfected CHO cells. Transfected CHO cells secreting the following Aβ species were used: control (mock), Aβ wildtype, AβS8C, AβS26C, or AβM35C, as indicated on top. Equal expression of APP in the cell lysates used is shown in the top panel. Only monomeric Aβ was produced by cells overexpressing APP-wt or APP(AβS26C). Predominantly dimeric Aβ, but no oligomers of higher order, was present in SNs of APP(AβS8C) cells. A heterogeneous signal pattern in the range of Aβ dimers and tetramers and putative 12−14 mers were detected in APP(AβM35C) transfected cells. Aβ was absent using empty CHO cells (mock). Detection antibody: 4G8. Additional immunoreactivity in the range of 50 and 25 kDa belongs to cross-reactivity of the secondary antibody with heavy and light chain of mAB IC16 used for immunoprecipitation. For comparison, immunoprecipitated Aβ monomers and oligomers derived from conditioned medium of 7PA2 cells (permanently secreting Aβ) are shown on the left. (b) Disulfide stabilization of Aβ dimers leads to accelerated formation of nativelike Aβ dimers. Western blot showing that a significant fraction of Aβ dimers and tetramers immunoprecipitated from AβS8C or M35C SNs remains stable after incubation in 2% β-mercaptoethanol (β-MeOH). Thus, nativelike SDS-resistant Aβ dimers are present. (c) Western blot of size exclusion chromatography (SEC) fractionated SNs from CHO cells secreting either Aβ from 7PA2 cells (top), AβS8C (middle), or AβM35C (bottom). Concentrated SNs were separated on a S75-column. Lyophilized fractions were analyzed by tricine SDS-PAGE and Western blotting using the 4G8 monoclonal antibody. With 7PA2 SN (top) a typical ladder of Aβ trimers (fractions 22−26), dimers (fr. 26−32), and monomers (fr.40−46) was seen. AβS8C (below) was only present as dimer (fr. 26−34) and monomer (fr. 40−46). In addition to dimers (fr. 26−32) and monomers (fr. 38−46), Aβ-M35C (bottom) showed signals in the range of tetrameric Aβ over a broad range between fractions 6 and 24. Higher molecular weight signals in fractions 40−46 are due to unspecific oxidations of the free cysteine in monomeric AβS8C or AβM35C with contaminants present in these fractions during the concentrating lyophilization process.

Figure 4

(a) Western blot of cell lysates derived from wt or mutant APP-overexpressing CHO cells. Western blot developed with an antibody directed against the C-terminus of APP (CT15). In contrast to CHO-mock and CHO cells expressing APP-wt, dimeric APP was present in lysates derived from cells expressing APP(AβS8C), APP(AβS26C), and APP(AβM35C) (top panel, SDS-PAGE gel run without β-MeOH). In lysates from cells treated with the γ-secretase inhibitor LY411575 (γ-i), dimeric APP-CTF was only found in cells expressing APP(AβS26C) (weak) and APP(AβM35C) constructs. Dimeric APP products were no longer visible after incubation with 2% β-MeOH, clearly demonstrating the participation of a disulfide bridge in APP dimer formation. (b) AβS8C dimers decrease miniature excitatory postsynaptic current (mEPSC) frequency and amplitude in cultured cortical neurons. mEPSC frequency and amplitude were significantly decreased in mouse primary cortical neurons incubated for 4 days with SN containing Aβ dimers (S8C; n = 40 cells) when compared with neurons incubated with CHO mock (n = 39 cells) or with the SN that was produced by CHO cells transfected with APP wt (n = 38 cells) that did not contain significant amounts of Aβ dimers (see Figure 3a). *p < 0.025 by Student’s test after Bonferroni correction. Quantitative data represent mean ± SEM. (c) Purified Aβ from CHO cells expressing APP(AβS8C) and APP(AβM35C) reduced neurite outgrowth in PC12 cells. Aβ was immunoprecipitated from 5 mL supernatants of wt or APP overexpressing CHO cells. Eluted Aβ was applied to differentiating PC12 cells. After 5 days, cells with neurites longer than three cell soma were quantified in 10 randomly chosen fields per well with 80−120 cells in each field. Data represent the mean ± SEM for the percentage of cell with long neurites compared to CHO mock calculated for 6 experiments. *p < 0.05 by ANOVA; significant decrease of neurite length after incubation with Aβ derived from CHO cells expressing APP(AβS8C) and APP(AβM35C) but not wt Aβ which did not contain significant amounts of Aβ dimers or other oligomers (see Figure 3a).