A streamlined, resource-efficient immunoprecipitation-mass spectrometry method for quantifying plasma amyloid-β biomarkers in Alzheimer's disease

Abstract

High-performance, resource-efficient methods for plasma amyloid-β (Aβ) quantification in Alzheimer's disease are lacking; existing mass spectrometry-based assays are resource- and time-intensive. We developed a streamlined mass spectrometry method with a single immunoprecipitation step, an optimized buffer system, and ≤75% less antibody requirement. Analytical and clinical performances were compared with an in-house reproduced version of a well-known two-step assay. The streamlined assay showed high dilution linearity (r²>0.99) and precision (< 10% coefficient of variation), low quantification limits (Aβ1-40: 12.5 pg/ml; Aβ1-42: 3.125 pg/ml), and high signal correlation (r²~0.7) with the two-step immunoprecipitation assay. The novel single-step assay showed more efficient recovery of Aβ peptides via fewer immunoprecipitation steps, with significantly higher signal-to-noise ratios, even at plasma sample volumes down to 50 pl. Both assays had equivalent performances in distinguishing non-elevated vs. elevated brain Aβ-PET individuals. The new method enables simplified yet robust evaluation of plasma Aβ biomarkers in Alzheimer's disease.

Keywords: Alzheimer’s disease; amyloid beta; blood biomarkers; immunoprecipitation; mass spectrometry.

Figure 1. Schematic illustration of the PAβ V1.0 vs PAβ V2.0 assays.

The PAβ V1.0 assay protocol (A) entails two rounds of immunoprecipitation. In contrast, the PAβ V2.0 assay protocol (B) includes a simplified sample preparation procedure with only a single round of IP, saving time and resources.

Figure 2. Mass spectra representative of Aβ peptides over multiple reagents and blocking buffers.

(A) MALDI-TOF mass spectra of plasma Aβ peptides replicates utilizing the PAβ V1.0 assay procedure, comparing five different buffers or detergents using single IP procedure; 10% N4PE CSF diluent (PAβ V2.0 assay), 10% SuperBlock, 10μg/ml TruBlock, 0.5% Triton100, and 0.5% Tween20. Representative spectra from each experiment are presented. Interference peaks were consistently observed at 5771.1 m/z and 7746.8 m/z across all assays. Additionally, another interference peak at 6631.0 m/z was consistently noted in all assays except the PAβ V1.0 assay. (B) Upon magnification in the range of 4000–4850 m/z, the theoretical m/z values of peptides were as follows: 4132.6 m/z for Aβ1–38, 4144.7 m/z for Aβ3–40, 4231.8 m/z for Aβ1–39, 4330.9 m/z for Aβ1–40, 4515.1 m/z for Aβ1–42, and 4689.4 m/z for APP669–711. Aβ1–38 IS at 4160.7 m/z, Aβ1–40 IS at 4383.3 m/z, and Aβ1–42 IS at 4569.3 m/z were utilized as internal standards for the normalization of mass spectra. Notably, an interference peak was detected at 4153.4 m/z in samples processed using 10% SuperBlock, 10pg/ml TruBlock, 0.5% Triton100, and 0.5% Tween20.

Figure 3. The spectra of IP-MS assays with S/N comparison.

(A) MALDI–TOF mass spectra of Aβ peptides derived from plasma replicates utilizing the PAβ V1.0 assay, 10% N4PE CSF diluent (PAβ V2.0 assay) and PAβ V1.0 assay with 1IP. Representative spectra from each experiment are presented. Interference peaks were consistently observed at 5771.1 m/z and 7746.8 m/z across all assays. Additionally, another interference peak at 6631.0 m/z was consistently noted in all assay formats except the PAβ V1.0 assay. Interference peaks at 3200 m/z to 3500 m/z and 6432.4 m/z were observed in PAβ Vi.0 assay with 1IP only. Upon magnification to the range of 4000–4850 m/z, the theoretical m/z values of peptides are as follows: 4i32.6 m/z for Aβ1–38, 4i44.7 m/z for Aβ3–40, 4231.8 m/z for Aβ1–39, 4330.9 m/z for Aβ1–40, 45i5.i m/z for Aβ1–42, and 4689.4 m/z for APP669–711. Aβ1–38 IS at 4160.7 m/z, Aβ1–40 IS at 4383.3 m/z, and Aβ1–42 IS at 4569.3 m/z were utilized as internal standards for the normalization of mass spectra. Notably, an interference peak was detected at 4153.4 m/z in samples processed using PAβ V1.0 assay with 1IP but not in the other assays. (B) S/N ratios were compared across three assays in triplicates, with asterisks indicating significant differences (*p < 0.05, **p < 0.0i) as determined by the Wilcoxon Rank Sum test. (C) The averages and standard deviations of the S/N ratios are listed.

Figure 4. Analytical performance assessment of the IP-MS assays.

(A) The calibration curves were generated using Aβ1–40 concentrations of 400pg/ml, 200pg/ml, 100pg/ml, 50pg/ml, 25pg/ml, 12.5pg/ml, and 0pg/ml, and Aβ1–42 concentrations of 100pg/ml, 50pg/ml, 25pg/ml, 12.5pg/ml, 6.25pg/ml, 3.125pg/ml, and 0pg/ml, normalized with Aβ1–38 IS. (B) The matrix effect recovery was assessed across three different concentrations, each with three replicates, utilizing Aβ1–38 IS normalization. (C) The IP recovery was measured through the SIMOA assay. (D) The relationship between plasma dilution and normalized intensity of the PAβ V1.0 and PAβ V2.0 assays. Three replicates were performed for each volume. Both Aβ1–40 and Aβ1–42 were normalized by Aβ1–38 IS. (E) The S/N ratios of plasma samples with various volumes were compared between PAβ V1.0 and PAβ V2.0 assays for Aβ1–40 and Aβ1–42.

Figure 5. Calibration curve and matrix effect recovery assessment using Aβ1–40 IS and Aβ1–42 IS normalization.

(A) The calibration curves were generated using Aβ1–40 concentrations of 400pg/ml, 200pg/ml, 100pg/ml, 50pg/ml, 25pg/ml, 12.5pg/ml, and 0pg/ml, and Aβ1–42 concentrations of 100pg/ml, 50pg/ml, 25pg/ml, 12.5pg/ml, 6.25pg/ml, 3.125pg/ml, and 0pg/ml, normalized with Aβ1–40 IS and Aβ1–42 IS. (B) The matrix effect recovery was assessed across three different concentrations, each with three replicates, utilizing Aβ1–40 IS and Aβ1–42 IS normalization.

Figure 6. Clinical performance of plasma Aβ biomarkers.

(A) Box and whisker plot categorizes the ADRC cohort into clinically assessed probable AD and normal control groups, analyzed using the Wilcoxon Rank Sum test, with p-values indicated. N represents the sample size. (B) Box and whisker plot shows the AGUEDA cohort split into Aβ PET positive and PET negative groups, analyzing three assay formats: PAβ V1.0 assay Aβ1–42/Aβ1–40, PAβ V2.0 assay Aβ1–42/Aβ1–40, and PAβ V2.0 assay Aβ1–42/Aβ1–40 normalized with Aβ1–40 IS and Aβ1–42 IS. Differences between groups were evaluated using the Wilcoxon Rank Sum test, with p-values provided. (C) Box and whisker plot dividing the AGUEDA cohort into CL positive, CL transition, and CL negative groups, with differences assessed using the Kruskal-Wallis test and p-values noted.

Figure 7. Correlation tests of PAβ V1.0 and PAβ V2.0 assays.

The correlation between the PAβ V1.0 assay and PAβ V2.0 assay, normalized using Aβ1–38 IS, was illustrated for the ADRC (A) and AGUEDA (B) cohorts. Spearman correlation was employed to evaluate the strength of the correlation between Aβ peptide measurements across the two assays. Additionally, Aβ1–40 and Aβ1–42, normalized using Aβ1–40 IS and Aβ1–42 IS in the PAβ V2.0 assay, were further assessed for correlation with their respective Aβ peptides in the PAβ V1.0 assay, normalized using Aβ1–38 IS.