Human Subjects
We included 69 patients with MCI and another 69 age-matched cognitively normal (CN) individuals from the Epidemiology of Mild Cognitive Impairment study in Taiwan (EMCIT) study (19). All human subjects were recruited from Far Eastern Memorial Hospital. The participants reside in the communities of neighboring towns of the hospital in New Taipei City. Individuals with active cancer, recent hospitalization, or current infection were excluded. The study was approved by the FEMH’s institutional ethical committee (FEMH 105147-F), and written informed consent was acquired from all participants.
An additional 120 samples (41 MCI, 79 CN) were recruited from the Taiwan Precision Medicine Initiative of Cognitive Impairment and dementia (TPMIC) study. The TPMIC study is an ongoing cohort initiated in 2021, with the ultimate enrollment of more than one thousand elderly Taiwanese to build a precision medicine platform by integrating multiple modalities, including MRI and PET neuroimaging, genetics, and immunoassay, to identify and integrate novel biomarkers linking progression from MCI to dementia. The participants were recruited from FEMH (FEMH 110065-F) and Cardinal Tien Hospital (CTH-110-2-1-014). Both institutions’ ethical committees approved the TPMIC study.
For both cohorts, the diagnosis of MCI was confirmed by an expert panel comprised of a neurologist, a psychiatrist, and a clinical psychologist, based on NIA-AA criteria (3).
Sample processing and PBMCs isolation
For all samples, the blood was sampled when cognitive function tests were performed. On the day of blood sampling, peripheral blood mononuclear cells (PBMCs) were immediately isolated by Ficoll-Paque PLUS gradient centrifugation following the manufacturer’s protocol (GE Healthcare). PBMCs were frozen in liquid nitrogen for long-term preservation.
Full-length (1-42) and peptide pool of amyloid b protein
The full-length amyloid b protein (Ab1-42, A21G, Flemish variant) (20) amino acid sequence is DAEFRHDSGYEVHHQKLVFFGEDVGSNKGAIIGLMVGGVVIA. The full-length wild-type peptide sequence (both from JPT Peptide Technologies) is DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA. Using protein-spanning mixtures of overlapping peptides is highly efficient for stimulating T-lymphocytes for diagnostic applications. We additionally created the overlapping 15mer peptides derived from the full-length amyloid b protein A21G variant.
Peptide pool composed of 10 sequences created based on 15mer amino acid length with sequential three amino acid shifts of the A21G full-length amyloid b protein (Ab1-42) was produced by Kelowna Internal Scientific Inc. The sequences are as the following: DAEFRHDSGYEVHHQ (Ab1-15), FRHDSGYEVHHQKLV (Ab4-18), DSGYEVHHQKLVFFG (Ab7-21), YEVHHQKLVFFGEDV (Ab10-24), HHQKLVFFGEDVGSN (Ab13-27), KLVFFGEDVGSNKGA (Ab16-30), FFGEDVGSNKGAIIG (Ab19-33), EDVGSNKGAIIGLMV (Ab22-36), GSNKGAIIGLMVGGV (Ab25-39) and KGAIIGLMVGGVVIA (Ab28-42). For tests on EMCIT samples, the AbA21G peptide pool and A21G full-length peptide were used for T cell stimulation. For tests on TPMIC cohort samples, Ab1-42 full-length wild-type peptide was also used. The chemicals PMA/ionomycin were used as positive controls.
T cell stimulation and quantification of amyloid-specific T cell response
T cell stimulation and polyfunctionality analyses were performed as described in our previous studies (21, 22). The full-length amyloid b protein, as well as the amyloid peptide pool, were used for PBMC cell stimulation (1mg/ml per peptide) along with costimulation (anti-CD28/CD49d), anti-CD107a, Golgistop (monensin) and Golgiplug (brefeldin A) for six hours at 37ºC. Afterward, cells were stained with anti-CD3, anti-CD8, anti-CD4, and Live/Dead® cell viability assay (Invitrogen) for 20 minutes before fixation with Cytofix/Cytoperm buffer (BD Biosciences). Cells were fixed, washed, and stained with anti-CD40L, anti-IL-2, anti-TNFα, and anti-IFNγ. Results were acquired using a multicolor flow cytometer (Beckman Coulter Cytoflex) at the Far Eastern Memorial Hospital Core Lab. Flow cytometry results were analyzed using FlowJo (Tree Star). After gated on live CD3+ cells, CD4+ and CD8+ cells were analyzed separately for cytokine expression in response to each stimulation (Supplementary Table 1). Frequency of reactive cells expressing each effector function among CD4+ or CD8+ T cells was determined. A combinatorial gating strategy based on the gates of each effector function was further applied to determine co-expression statistics using the FlowJo Boolean gate platform. The total amyloid-specific T cell response frequency was derived by summing the percentages of amyloid-reactive cells expressing at least one effector function among total CD8+ or CD4+ T cells (% CD8+ or % CD4+) (21).
Phospho-Tau 181 Measurement
Phospho-Tau 181 (pTau181) was measured at Veritas laboratory (Taipei, Taiwan). The Simoa Human pTau181 Advantage V2 assay was used according to the manufacturer’s protocol using the Simoa HD-X analyzer.
Statistical Analyses
Comparing T cell responses between groups was investigated using the Student’s t-test. The area under the receiver operating characteristic curve (ROC) was used to examine the discriminative performance of these effector functions as potential biomarkers. Areas under the curve (AUROCs) and their confidence intervals (CI) were calculated based on 2000 bootstrap samples using the non-parametric method (rocreg command in STATA). Due to age difference between MCI and cognitively normal adults in the TPMIC sample, age-adjusted AUROCs were additionally calculated by regressing continuous age in the model. The equality of AUROCs between p-tau 181 and T-cell effector functions was tested using roccomp command in STATA. All statistical analyses were conducted in STATA 17 (College Station, TX: StataCorp LLC).