An improved immunoassay detects Aβ oligomers in human biofluids: their CSF levels rise with tau and phosphotau levels

Abstract

Background: Diffusible Aβ oligomers (oAβ) confer cytotoxicity in Alzheimer's disease. The dynamic complexity of this hydrophobic analyte means few immunoassays exist to quantify oAβ in CSF and plasma.

Methods: We characterized antibody 71A1 to a cyclized dimer of Aβ9-18 for oAβ preference over monomers by surface plasmon resonance. We improved an earlier bead-based immunoassay by using 71A1 streptavidin plates for capture and N-terminal antibody 3D6 for detection. Numerous controls systematically validated accuracy.

Results: 71A1 showed highly selective binding kinetics to Aβ oligomers over monomers. It enriched bioactive oligomers from AD brain that altered neuronal excitatory currents and calcium transients. 71A1/3D6 immunoassay exhibited specificity and reproducibility in human biofluids. CSF oAβ levels correlated positively with CSF tau and phosphorylated-tau-181. APP and PS1 FAD mutations increased oAβ levels in human neuronal media.

Conclusions: CSF oAβ levels rise in concert with rising tau levels. A new plate-based ELISA offers improved consistency, less sample volume, and lower cost, thus better suited to quantify this challenging analyte.

Keywords: Alzheimer’s disease; Amyloid β-protein; Biomarkers; Monitoring; Oligomeric Aβ; Tau.

Fig. 1

Electrophysiological analysis of bioactive 71A1-reactive Aβ species purified from AD brain. 71A1-reactive Aβ species impact excitatory synaptic signaling and calcium dynamics in rat primary neurons. A, B Changes in amplitude and frequency of spontaneous excitatory postsynaptic currents (sEPSCs) in DIV16-19 wt rat hippocampal neurons treated with aCSF (n = 15) or 71A1-affinity purified oAβ for 2 min (n = 11) and 30 min (n = 13). C, D Changes in amplitude and frequency of spontaneous inhibitory postsynaptic currents (sIPSCs) in neurons treated with aCSF (n = 15) or 71A1-purified oAβ for 2 min (n = 13) and 30 min (n = 14). E–H Calcium transient amplitude and frequency in jRGECO1a-expressing rat primary cortical neurons treated with TBS (control) or 71A1-purified Aβ at 2 min (E, n = 483, n = 531; F, n = 15; events across 15 individual cells) (E, F) and 5 min (G, n = 522, n = 625; H, n = 16; events across 16 individual cells) (G, H). * p < 0.05, ** p < 0.01, **** p < 0.001

Fig. 2

New plate-based assay platform and optimized protocol for oAβ detection. A Schematic of the updated plate-based sandwich immunoassay with biotinylated 71A1 capture antibody and alexa647-labelled 3D6 detector antibody. B Step-by-step overview of the assay protocol, detailing incubation conditions and workflow specific to the new plate-based format. C Consistency of calibrator curves: readouts of 71A1/3D6 oAβ assay using two-fold dilution series from 600 pg/mL using synthetic Aβ-derived diffusible ligands (ADDLs) as calibrator across five independent daily runs. Points on the graph represent the relative response units produced at each concentration of ADDLs

Fig. 3

Validation of signal specificity of the plate-based 71A1/3D6 oAβ immunoassay. A Specific detection by the 71A1/3D6 assay of synthetic (S26C)₂ Aβ1-40 disulfide-crosslinked dimers (black) vs. the depolymerized Aβ1-40 monomers post-βME reduction (red). Inset: SDS-PAGE WB of these samples. B Specificity of the 71A1/3D6 assay for synthetic Aβ-derived diffusible ligands (ADDLs, in black) vs. synthetic Aβ1-40 monomers (blue); and synthetic Aβ1-42 monomers (red) before (red squares) and after (red triangles) depolymerization in 7 M GuHCl. C Signal recovery (percentages) of 8 individual human cerebrospinal fluids (CSF) measured at 4x, 8x, or 16 × dilutions by the 71A1/3D6 assay, with results normalized to 4 × dilution. D Signal recovery (percentages) of 8 individual human plasmas measured at 2x, 4x, and 8 × dilutions by the 71A1/3D6 assay, with results normalized to 2 × dilution. E Signal recovery (percentages) in the 71A1/3D6 assay by detecting 500 pg/mL of ADDLs spiked into 6 individual human CSF samples. F 71A1/3D6 oAβ signal stability in three individual human plasmas over four freeze–thaw cycles

Fig. 4

Validation of plate-based 71A1/3D6 oAβ assay consistency over time. Correlation of 71A1/3D6 oAβ readouts from 8 individual human plasmas (A) and 8 individual human CSFs (B) across three independent assay runs, with oAβ concentrations from experiment 1 on the x-axis and oAβ concentrations from experiments 2 and 3 on the y-axis. C Correlation of oAβ measurements of 11 individual human CSFs on the previous bead-based assay [17] versus on the new plate-based assay, demonstrating consistency between the assay platforms. D plate-based 71A1/3D6 oAβ measurement of post IP’ed supernatants from one human (AD) brain soaking extract immunodepleted with 71A1 conjugated to either C1 streptavidin Dynabeads or protein G Dynabeads or Streptavidin microplates

Fig. 5

Longitudinal tracking of pathological Aβ change in aging NL-G-F mice with 71A1. A Plate-based 71A1/3D6 oAβ quantification of NL-G-F mouse brain homogenates at increasing ages (2 to 12 mo; n = 8–10 mice per age) before (red) and after (blue) depolymerization in 7 M GuHCl, confirming signal specificity for aggregated Aβ. WT mouse brain homogenates at 2 and 12 mo (n = 4 mice per age) were measured concurrently (black) as negative controls to highlight the absence of detectable oAβ, as expected (empty triangles (GuHCl +) overlap black triangles (GuHCl-) along x = 0). B Immunofluorescent staining by 71A1 of Aβ deposits from 12 mo WT mice. C-H Immunofluorescent staining by 71A1 of Aβ deposits in NL-G-F mouse hippocampal cryosections from 2 mo old (C) to 12 mo old (H) at 2 mo intervals, confirming an age-dependent rise in Aβ deposition

Fig. 6

oAβ levels correlate significantly with tau and p-tau¹⁸¹ levels but not Aβ42 monomer levels. Human CSFs (n = 95) show highly significant positive correlations between 71A1/3D6 oAβ levels and total tau (t-Tau) levels (A) and phospho-tau (p-Tau¹⁸¹) levels (B) measured on the Alzheimer’s Disease Evaluation (ADEVL) platform at Mayo Clinic Laboratories. See also Supp Fig. 1. C Lack of correlation between 71A1/3D6 oAβ levels and Aβ1-42 monomer levels in the CSFs. D Reproducibility of plate-based 71A1/3D6 oAβ assay across two independent runs of another BWH cohort of patient CSFs (n = 108). Pearson correlation is used for all analyses