A sensitive aβ oligomer assay discriminates Alzheimer's and aged control cerebrospinal fluid

Abstract

A hallmark of Alzheimer's disease (AD) brain is the amyloid β (Aβ) plaque, which is comprised of Aβ peptides. Multiple lines of evidence suggest that Aβ oligomers are more toxic than other peptide forms. We sought to develop a robust assay to quantify oligomers from CSF. Antibody 19.3 was compared in one-site and competitive ELISAs for oligomer binding specificity. A two-site ELISA for oligomers was developed using 19.3 coupled to a sensitive, bead-based fluorescent platform able to detect single photons of emitted light. The two-site ELISA was >2500× selective for Aβ oligomers over Aβ monomers with a limit of detection ∼ 0.09 pg/ml in human CSF. The lower limit of reliable quantification of the assay was 0.18 pg/ml and the antibody pairs recognized Aβ multimers comprised of either synthetic standards, or endogenous oligomers isolated from confirmed human AD and healthy control brain. Using the assay, a significant 3- to 5-fold increase in Aβ oligomers in human AD CSF compared with comparably aged controls was demonstrated. The increase was seen in three separate human cohorts, totaling 63 AD and 54 controls. CSF oligomers ranged between 0.1 and 10 pg/ml. Aβ oligomer levels did not strongly associate with age or gender, but had an inverse correlation with MMSE score. The C statistic for the Aβ oligomer ROC curve was 0.86, with 80% sensitivity and 88% specificity to detect AD, suggesting reasonable discriminatory power for the AD state and the potential for utility as a diagnostic marker.

Keywords: Abeta oligomer; Alzheimer.

Figure 1.

The 19.3 antibody selectively binds to oligomeric Aβ. The differential binding of the 19.3 antibody to Aβ monomer and oligomers was quantified in (A) a one-site ELISA and in (B) a competitive binding ELISA. In the one-site ELISA, 19.3 showed only a modestly different EC₅₀ for monomer (0.64 pg/ml, ■) and Aβ oligomers (0.25 pg/ml, ▴). The maximal binding was greater for Aβ oligomers. In the competitive binding ELISA, 39 ng/ml of Aβ oligomers was sufficient to displace 50% binding to coated Aβ40 monomers (■), whereas 24.8 μg/ml Aβ monomer was needed to displace 50% of the Aβ oligomer (▴) binding. RLU, relative luminescence units.

Figure 2.

19.3/82E1 ELISA shows differential sensitivity toward Aβ oligomers compared with Aβ monomer and a larger dynamic range compared with the homotypic 6E10/6E10 ELISA. A, Plates coated with 19.3 were exposed to increasing concentrations of Aβ oligomers (▴) or Aβ42 monomer (♦). Binding was detected with the 82E1 antibody. Cross-reactivity with the monomer preparations was limited, with 1000-fold difference in sensitivity. B, C, Plates coated with 6E10 were exposed to increasing concentrations of Aβ oligomers (▴) or Aβ42 monomer (♦) with binding detected using 12F4 Aβ42-specific antibody (B) or 6E10 (C) antibodies. Equivalent signals were observed in the Aβ42-specific ELISA assay, while 350-fold lower oligomer sensitivity was seen in the 6E10/6E10 homotypic ELISA compared with 19.3/82E1. RLU, relative luminescence units.

Figure 3.

The Aβ oligomer preparation is enriched for HMW Aβ species. A, Aβ oligomers were fractionated by SEC-MALLS and monitored by absorbance at 280 nm. A primary broad HMW peak was detected at minutes 16.5 through 20.5. MALLS analysis indicated that the peak was a polydisperse mixture of oligomers ranging in size from 310 to 7700 kDa, with a mean molecular weight of ∼1117 kDa. A minor LMW peak was detected at minute 28–32. Five additional peaks, corresponding to potentially LMW species, were detected at minutes 34 through 45. B, SEC profile of F12 media alone is shown. F12 media is used in the preparation of Aβ oligomers. The peaks at minutes 34 through 45 are present in F12, indicating these are likely not Aβ related and that the peak at 28–32 min corresponds to the smallest Aβ species. C, HFIP-treated Aβ42 monomer was separated by SEC and had a prominent peak at minute 28–32, suggesting this is an accurate size for monomers. MALLS analysis confirmed Aβ monomeric peak molecular weight at 4.8 kDa (actual weight is 4.514 kDa). Globular protein size standards were run on the same column under the same conditions and their peak elutions are indicated at the top of each chromatogram: (1) thyroglobulin (670 kDa), (2) gamma globulin (158 kDa), (3) ovalbumin (44 kDa), (4) myoglobin (17 kDa), and (5) vitamin B12 (1.35 kDa). The standards serve as calibration controls and estimation of the LMW peaks rather than to accurately predict HMW fraction sizes.

Figure 4.

Comparison of the SEC fractionated synthetic Aβ oligomers by ELISA shows 19.3/82E1 ELISA detects HMW Aβ species. Aβ oligomer preps were fractionated by SEC and the fractions tested in a 6E10/12F4, 6E10/6E10, or 19.3/82E1 ELISA. A, The 6E10/12F4 Aβ42 ELISA shows binding to LMW (fractions 28–32) and HMW (fractions 16.5–20.5) Aβ species. B, The 6E10/6E10 ELISA recognized HMW Aβ (fractions 17–21) with no reactivity to LMW Aβ. C, The 19.3/82E1 ELISA also shows selectivity toward HMW Aβ, but recognizes a broader range and higher concentration of Aβ species than 6E10/6E10.

Figure 5.

SEC fractions of human cortex aqueous supernatants revealed more HMW Aβ oligomers in AD compared with non-AD cortex, while similar total Aβ levels were found. A, 19.3/82E1 ELISA assay on SEC fractions of 0.07 mg of AD (left) and non-AD (right) cortex shows ∼3-fold elevation in the mean concentration of HMW fractions in AD compared with non-AD samples. B, Similar levels of LMW fractions were observed between AD and non-AD PBS extracts when examined using the 6E10/4G8 ELISA of total Aβ signal. Globular protein size standards were run as noted in Figure 3 legend.

Figure 6.

Use of Erenna flow-based detection provides increased sensitivity while preserving selectivity with Aβ monomers. A, The MP-based 19.3/82E1 assay shows dilution linearity from 0.04 to 0.72 pg/ml. Variability in the lower part of the curve constrains the LLORQ to 0.18 pg/ml and higher. B, Aβ monomers show limited cross-reactivity with the 19.3/82E1 assay, 2500-fold lower signal compared with oligomers.

Figure 7.

Aβ oligomer levels in the CSF discriminate between AD patients and controls and correlate with MMSE values. A, Aβ oligomer levels were significantly higher in the AD CSF samples than in aged control samples from Cohort 1 (geometric mean with 95% CI, p < 0.008). Dotted lines in A and B indicate the lower limit of reliable quantification of the assay. B, Aβ oligomers in Cohort 2 are significantly elevated in the AD CSF compared with age-matched healthy controls (geometric mean with 95% CI; p < 0.0001). C, Aβ oligomer concentrations in Cohort 2 were used to generate a ROC curve to assess the predictive validity of the assay. The area under the curve of the ROC curve is 0.86. D, There is a significant correlation between MMSE and Aβ oligomers when all samples from Cohorts 1 and 2 are analyzed either by pooling healthy controls and AD samples (r = 0.67, p < 0.0001) or within AD samples alone (r = 0.28, p < 0.05). There is no correlation for the control samples. AD Cohort 1 (●), Control Cohort 1 (○), AD Cohort 2 (■), Control Cohort 2 (□), Total AD only correlation (- - -), total control correlation (···), Combined correlation (——).

Figure 8.

Aβ40 monomer concentration is unchanged between the AD and aged controls in Cohort 1 and 2 (A and C, respectively) while Aβ42 values are reduced by twofold in the AD samples (B, Cohort 1, p < 0.002; D, Cohort 2, p < 0.001; geometric mean and 95% CI). A Cohort 2 subset was analyzed because of limited CSF volumes available. The correlation between Aβ40 and oligomers was significant for the AD CSF (r = 0.49, p = 0.03; ■), but not aged control CSF (E; □). Correlation between Aβ42 and oligomers was not significant for either group (F; AD, ♦; control, ◊).

Figure 9.

ROCs using either Aβ42 (A) or oligomer (B) values from AD or control CSF. The oligomer values provided improved sensitivity compared with Aβ42 to detect disease.

Figure 10.

AD CSF samples contain significantly higher oligomers and lower Aβ42 monomers compared with other diagnostic groups. A, Aβ oligomer levels were significantly higher (p < 0.01) and (B) Aβ42 levels were significantly lower (p < 0.01) in AD CSF samples when compared with aged control or ALS, PD, or Schizophrenia samples from Cohort 3. C, Aβ40 levels were unchanged among all tested CSF groups of Cohort 3.

Figure 11.

AD CSF samples across both genders contained more Aβ oligomers than control. No differences were detected when males were compared with females, in AD, controls, or pooled (one-way ANOVA, Bonferroni correction).