Citations

AbstractBackgroundLate Onset Alzheimer’s Disease (LOAD) is the most common neurodegenerative disorder. Carriers of an ɛ4 allele of the apolipoprotein E gene (APOE) have significantly increased risk of developing LOAD. LOAD is also strongly sex biased. Besides the well‐established role of sex hormones, the impact of genetic sex combined with APOE genotypes has not been studied. Sex chromosome dosage (SCD) is the main genetic difference between females (XX) and males (XY). Despite dosage compensation by X chromosome inactivation (XCI), some X‐linked genes remain expressed from the inactive X (Xi) and are female biased in expression. Many of these genes are related to neurological and immune functions and could contribute to neuroinflammation in AD.MethodTo understand the genetic interactions between autosomal risk (APOE) and sex‐bias in AD, we are generating isogenic sets of hiPSCs with various sex chromosome complements and APOE genotypes. We have derived XXY/XY, XY/X, and XX/X isogenic pairs by reprogramming cells from naturally mosaic individuals or by selective removal of the Xi in vitro. These lines uniquely position us to differentiate the effects of X chromosome dosage and/or presence of a Y chromosome, while minimizing genetic variability and environmental or hormonal confounders. Currently, we are also constructing APOE ɛ3 and ɛ4 alleles in our paired hiPSC lines with differing sex chromosome content using gene editing.ResultStarting from an XXY/XY isogenic pair, we have generated an hiPSC series with six different genotypes of APOE ɛ3/ɛ3, ɛ3/ɛ4 or ɛ4/ɛ4 alleles with either the XXY or XY genotype. Preliminary results in differentiated neural precursor cells show differential expression of sex‐linked genes in relation to SCD. We also observed genome‐wide changes in gene expression and DNA methylation. Subsets of autosomal sex‐biased genes were differentially affected either by X chromosome dosage, presence of a Y, or both.ConclusionThe new hiPSC lines are being differentiated to AD‐relevant cell types, including neurons, microglia, and cortical organoids, for transcriptomic and functional analyses to elucidate how APOE alleles and genetic sex interact to modulate risk in LOAD pathology.

AbstractBackgroundLewy body dementia (LBD) is the second most prevalent dementia in the United States after Alzheimer’s disease (AD). Recent studies have implicated rare mutations in two lipid transport genes, ABCA1 and ATP8B4, in Alzheimer’s disease. Substantial co‐pathology and shared risk factors indicate an intersectional genetic architecture between LBD and AD; therefore, we investigated the association of rare mutations in ABCA1 and ATP8B4 with LBD. Additionally, we examined rare mutations in 25 genes involved in ABCA1‐related lipid‐specific pathway with risk of developing LBD.MethodWe annotated rare missense and loss‐of‐function (LOF) mutations in ABCA1 and ATP8B4 genes, using Ensembl Variant Effect Predictor, in previously published whole‐genome sequence data from 2,591 LBD cases and 4,027 healthy controls of European ancestry. To test the association with LBD risk, we applied the optimized sequence kernel association test (SKAT‐O), followed by single‐variant association test (Fisher’s exact test), and multiple testing (Bonferroni correction). Additionally, we evaluated rare coding variants of 25 genes in ABCA1‐related positive regulation of the cholesterol efflux pathway (GO: 0010875) by performing single‐gene burden and gene‐set burden analyses.ResultsWe found a significant enrichment of rare ABCA1 missense mutations in LBD (SKAT‐O p = 2.68E‐02). However, no significant associations were found between LBD and ABCA1 LOF, and ATP8B4 LOF or missense mutations. Interestingly, the single‐variant analysis identified two ABCA1 missense variants, c.5398A>C; p.N1800H (Fisher’s test: p‐value = 1.16E‐04) and c.3544G>A; p.A1182T (p‐value = 4.29E‐02), with the p.N1800H substitution surviving Bonferroni multiple testing. The p.N1800H variant is reported to be pathogenic, and it was observed in twelve LBD patients and one control, with an odds ratio of 18.48 (95%CI = 2.4‐142.8). This mutation is reported to be pathogenic, in our LBD cohort, six out of twelve ABCA1 p.N1800H carriers also carry the APOE ε4 risk allele. However, neither a single‐gene burden nor a gene‐set burden analysis in 25 genes in the positive regulation of the cholesterol efflux pathway revealed any significant associations with LBD.ConclusionOur findings reveal significant enrichment of ABCA1 missense mutations in LBD emphasizing the genomic complexity surrounding LBD and AD.

AbstractBackgroundAlzheimer’s Disease (AD) presents complex molecular heterogeneity, influenced by a variety of factors including heterogeneous phenotypic, genetic, and neuropathologic presentations. Regulation of gene expression mechanisms is a primary interest of investigations aiming to uncover the underlying disease mechanisms and progression.MethodWe generated bulk RNA‐sequencing in prefrontal cortex from 565 AD brain samples (non‐Hispanic Whites, n = 399; Hispanics, n = 113; African American, n = 12) across six U.S. brain banks, and conducted differential gene expression and enrichment analyses. We sought to identify cross‐ancestry and ancestry‐specific differentially expressed genes (DEG) and pathways across Braak stages, adjusting for sex, age at death, and RNA quality metrics. We validated our findings using the Religious Orders Study/Memory and Aging Project study (ROS/MAP, n = 1,095). Lastly, we validated top DEG using publically‐available human single‐nucleus RNA sequencing (snRNAseq) data.ResultAD‐known genes VGF (LFC = ‐0.661, padj = 3.78) and ADAMTS2 (padj = 1.21) were consistently differentially expressed across statistical models, ethnic groups, and replicated in ROS/MAP (Figure 1). Genes from the heat shock protein (HSP) family, e.g. HSPB7 (padj = 3.78), were the top DEG, also replicated in ROS/MAP. Ethnic‐stratified analyses prioritized TNFSF14 and SPOCD1 as top DEG in Hispanic samples. Gene set enrichment analysis highlighted several significantly pathways, including “TYROBP causal network in microglia” (WP3945; padj = 1.68) and “Alzheimer Disease” (WP5124; padj = 4.24). snRNAseq validated several DEG, including VGF downregulated in neurons (padj = 1.1).ConclusionTo our knowledge, this is the largest diverse transcriptome study for AD in post‐mortem tissue. We identified perturbated genes and pathways resulting in cross‐ethnic and ethnic‐specific findings, ultimately highlighting the importance of diversity in AD investigations.

AbstractBackgroundWe recently developed a novel tau‐PET summary measure THETA, capturing regional heterogeneity and identifying tau status, using ground truth visual assessments from a large single‐center cross‐sectional dataset and validated on independent cohorts [1, 2]. In this study, we aimed to evaluate the performance of THETA on longitudinal and histopathology data.MethodWe included longitudinal tau‐PET ([18F]flortaucipir) data from 696 Mayo Clinic Study of Aging (MCSA) and ADRC participants, with histopathology in n = 90. Fig. 1 shows the model that uses regional standard uptake value ratios (SUVR) and a target of binary class of tau positivity for prediction of THETA. This model was applied to predict tau status on each followup scan. Slopes of the meta‐ROIs’ tau SUVRs and THETA were evaluated by diagnostic group and as a function of baseline amyloid SUVR in discordant CU participants (where meta‐ROI and visual assessments did not match at baseline) and in incident‐THETA+ CU participants (whose THETA moved from below to above 1 over serial tau‐PET scans). We evaluated tau measurements as a function of Braak stages for neurofibrillary tangles.ResultLongitudinal plots of meta‐ROIs for diagnostic groups were very similar, but the expanded range of THETA based on visual positivity prediction clearly identified tau positive or tau‐negative scans. In discordant‐CU (n = 97) and incident‐THETA+ CU participants (n = 14 all amyloid positive at follow‐up but only 65% at baseline), the relationship between baseline amyloid and rate of tau increase was stronger for THETA than temporal meta‐ROI (Fig. 2). Separation between baseline and follow‐up was greater for THETA (t‐statistics = 90) compared to temporal meta‐ROI (t‐statistics = 20) (p<0.01) in the incident‐THETA+ CU participants. THETA showed a slightly stronger association with Braak stage than meta‐ROIs (rho = 0.87 vs. ≤0.83, p<0.05), with better separation of clinical diagnoses (Fig. 3).ConclusionTHETA remained clearly negative or positive in MCI and AD, providing consistent information on underlying etiology of impairment at both baseline and follow‐up. Although binary in its construction, THETA both provided separation of values based on tau status and its change correlated with baseline amyloid burden especially in discordant‐CU where tau deposition is not in typical meta‐ROIs. Further work is needed to confirm if THETA captures early tau changes.

AbstractBackgroundAlthough the rate of Alzheimer’s disease (AD) in African‐ancestry (AA) Americans is higher than that of persons from European‐ancestry (EA) populations, AA participants have been underrepresented in AD neuropathological studies.MethodUtilizing the AD Research Centers (ADRC) infrastructure, we obtained AA donor pre‐frontal cortex (PFC) tissue from brain repositories of 12 ADRC and generated bulk RNA sequencing (RNA‐seq) data for 179 samples that met QC and inclusion criteria. Previously generated PFC RNAseq data were obtained for 28 additional AA donors from the Columbia University ADRC. Differential gene expression was evaluated among 125 donors with a neuropathological diagnosis of AD (NIA‐Reagan intermediate or high likelihood) and 82 neuropathologically confirmed controls using regression models including covariates for age at death, sex, cell‐type frequencies, and RNA integrity number (RIN) calculated with Limma. FDR‐corrected p‐values (padj) were calculated to control for the 33,611 genes examined.ResultA total of 482 genes surpassed the multiple‐testing threshold. The most significant, ADAMTS2 (p = 2.96 × 10‐8, padj = 0.001), showed increased expression in AD cases (see Table/Figure). We note that ADAMTS2 was differentially expressed in a prior EA study of neuropathologically confirmed AD cases and controls (Panitch et al. Molecular Psychiatry 2021). Additionally, a recent analysis of cognitive resilience in EA neuropathological AD cases identified a strong association with ADAMTS2 (See Li et al. AAIC2024 abstract). Of the differentially expressed genes observed in the Panitch et al. EA study, 385 (35%) were nominally significant,65 (5.8%) were corrected significant, and most (89%) of these genes had the same effect direction in the AA cohort. Some of the observed associations appear to be AA specific (e.g., EFR3B, IRS4, and CA12; see Table). Additionally, we found nominally significant (p<0.05) associations with expression of APOE (Log2 fold change [L2FC = ‐0.20, p = 0.012) and several other established AD genes including SORL1 (L2FC = ‐0.10, p = 0.014) and IGF1R(L2FC = 0.11, p = 0.0013).ConclusionThis largest‐ever (to our knowledge) gene expression study of AD in postmortem brain tissue from AA donors implicates many more genes as having a role in AD in this population than previously identified in genome‐wide association studies and provides insight into trans‐ancestry differences risk for AD.

AbstractBackgroundBiomarker cutpoints have received significant importance in AD research. Reliable worsening method based on serial biomarker data has been proposed (Jack et al., 2017 Alz & Dem). Advanced machine learning models (ML) and explainable AI provide an opportunity to enhance existing techniques. We aimed to estimate biomarker cutpoints on cross‐sectional data using ML by identifying important transitions in the biomarker values.MethodWe first trained an ML model with the biomarker of interest as an input including other available biomarkers, demographics, and diagnosis, and predicted global cognition z‐score. We then obtained SHAP values (Shapley Additive exPlanations), fit a curve between the SHAP values and biomarker values, and calculated the first derivatives (slope) of the curve. Then, a CUSUM (Cumulative Sum) analysis was used to identify SHAP change points and cross‐referenced to values where the slopes were steepest near the zero‐crossing location of the curve. A zero SHAP is an important indicator of a feature’s progressive importance for model prediction because a feature’s influence on prediction significantly shifts when its SHAP values go from positive to negative or vice versa. However, relying on zero‐crossing alone may be insufficient when the relationship is non‐linear. Consequently, we calculated the second derivatives to identify inflection points (peaks and troughs) along the curve. Lastly, we compiled all the potential cutpoints and performed post‐processing as shown in Fig. 1.ResultThe algorithm identified cutpoints at 1.44 SUVR for all and at 1.42 SUVR for CU participants (Fig. 2). At age 70, both cutpoints estimated the percentage of abnormals at less than 40% (Fig. 3).ConclusionTPE amyloid‐PET SUVR cutpoints for CUs equal the reported RW values (1.42 SUVR). A strength of the TPE is its application on cross‐sectional data making it versatile for various biomarkers across different cohorts.

AbstractBackgroundTau aggregation is the major cause of several neurodegenerative tauopathies. Tau interaction with other proteins affects the formation of tau aggregates with seeding activity but less is known about its effects on tau‐seed properties. Our previous study revealed that Bassoon (BSN), a presynaptic protein, interacts with tau‐seed, exacerbating its toxicity in vivo. Bsndownregulation reduced tau spreading and overall pathology. Intriguingly, a parallel study associated missense mutations in BSN with tau aggregation in patients, prompting an investigation into the influence of genetic mutations in BSN on tau pathology for potential therapeutic insights.MethodWe generated a knock‐in mouse model (BSNKI) harboring the disease‐associated p.Pro3866Ala mutation in endogenous Bsn. Cognitive and motor abilities were assessed in aged heterozygous and homozygous BSNKI mice, followed by analyses of BSN and tau patterns, gliosis, and gene expression changes in their brains. Additionally, we validated our findings in a human BSN mutation carrier.ResultAt 10 months, BSNKI mice displayed motor impairments on the rotarod, and grip strength assays compared to WT mice. Their brains displayed somatic BSN and pathological tau accumulation, with the gene expression changes indicating alterations in microglia activation, protein degradation pathways, complement activation, and synapse pruning. We also observed an accumulation of pathological tau at the synapses and synapse engulfment by microglia. Histopathological analyses revealed robust microglia activation and co‐deposition of proteasomal subunits with BSN. The human BSN mutation carrier displayed inclusions of BSN and tau pathology similar to observations in the BSNKI model.ConclusionOur BSNKI mouse model, reflecting a disease‐associated BSN mutation, revealed motor impairments and pathological tau and BSN deposits, mirroring observations in BSN mutation carriers. Notably, BSN seems to play a dual role, promoting tau aggregation and sequestering protein degradation molecules, leading to tau and protein accumulation at the soma and synapse, triggering microgliosis and neuroinflammation. These findings propose BSN as a promising therapeutic target for tauopathies, underscoring the need for further exploration to elucidate underlying mechanisms and therapeutic implications.

AbstractBackgroundPosttranslational modifications of tau occurs throughout disease progression in Alzheimer’s disease (AD) and non‐AD tauopathies. Phosphorylation of tau (p‐tau) in the proline‐rich region is a common target of immunohistochemical and fluid biomarker evaluation. P‐tau217 has emerged as a highly accurate fluid biomarker in AD, however elevated levels are not observed in non‐AD tauopathies.MethodA neuropathologic overview of tau lesions immunopositive for phosphorylation at threonine 217 (pT217) in AD and non‐AD tauopathies will be provided in a didactic format with an emphasis on disease spectrum. The relevant biomarker literature with neuropathologic validation will be highlighted.ResultImmunohistochemical evaluation of pT217 in AD demonstrates immunostaining in pretangles and mature tangles, but rarely observed in ghost tangles. Neuritic pathology in AD with pT217 immunopositivity include both neuropil threads and dystrophic neurites that cluster in neuritic plaques. Lesions in non‐AD tauopathies are also immunopositive for pT217, as exampled by immunolabelling of argyrophilic grains and tufted astrocytes. Digital pathology measures of pT217 and amyloid‐β will be compared with plasma p‐tau217 levels to contextualize the relationship. Additionally, the relationship between neurotransmitter hub vulnerability will be explored to consider upstream factors that may influence soluble tau release into plasma.ConclusionNeuropathologic insights discussed through visual and quantitative comparison of insoluble phospho‐tau accumulation and soluble levels of p‐tau217 may provide neurobiologic insights into the intersection between amyloid‐β and tau in the human brain.

AbstractBackgroundGenetic variations have emerged as crucial players in the etiology of Alzheimer’s disease (AD), and they serve for a better understanding of the disease mechanisms; yet the specific roles of these genetic variants remain uncertain. Animal models with reminiscent disease pathology could uncover previously uncharacterized roles of these genes. Therefore, we generated zebrafish models for AD variants to analyze the in depth molecular and biological functions of these variants.MethodUsing CRISPR/Cas9, we generated a knockout model for abca7, orthologous to human ABCA7. We performed single cell transcriptomics and analyzed the altered genes and molecular pathways in zebrafish. We leveraged data from multiethnic AD cohorts at Mayo Clinic and Columbia University, to perform genetic association studies, co‐expression analyses, in silico interaction mapping, family based variant segregation analyses and epigenetic association studies, and the functional and histological studies in zebrafish.ResultThe abca7± zebrafish reduced astroglial proliferation, synaptic integrity, and microglial response after Aβ42 toxicity. We found that the abca7 loss‐of‐function (LOF) reduced neuropeptide Y (npy) expression as well as Brain‐derived neurotrophic factor (bdnf) and Nerve growth factor receptor (ngfr). Human brain analysis showed reduced NPY in AD, regulatory interaction between NPY and BDNF, genetic variants in NPY associated with AD, and segregation of variants in ABCA7, BDNF and NGFR in families. ABCA7 variants altered the epigenetic codes in NPY, BDNF, and NGFR promoter regions. Human results paralleled with zebrafish findings to indicate an evolutionarily conserved disease mechanism through ABCA7‐NPY signalling axis. NPY administration to zebrafish rescued the phenotypes in abca7 knockout, suggesting a true biological relevance.ConclusionOur results demonstrate a previously unknown link between ABCA7 and NPY in regulation of synaptic integrity and neurogenesis in AD. We propose that ABCA7‐dependent NPY is a resilience factor in vertebrate brains, and this reserve mechanism is impaired in AD.

AbstractBackgroundAlzheimer’s disease (AD) is neuropathologically characterized by amyloid‐β (Aβ) plaques and tau neurofibrillary tangles often quantified by Thal phase and Braak stage, respectively. Aβ also frequently deposits in the cerebrovasculature with severity categorized by a cerebral amyloid angiopathy (CAA) score. These and related measures often show high variability within AD suggesting distinct underlying mechanisms. We hypothesize that, within the AD brain, neuropathology and levels of core AD‐related proteins are influenced by variations in DNA methylation (DNAm). To test this, we performed epigenome‐wide association studies (EWAS) using DNAm measures from the temporal cortex (TCX) and cerebellum (CER) with AD‐related neuropathologic measures (Braak, Thal, CAA) and brain biochemical levels of five proteins (apoE, Aβ40, Aβ42, tau, p‐tau).MethodsDNAm from neuropathologically‐confirmed AD cases was measured by reduced representation bisulfite sequencing (RRBS) from 471 TCX samples, 200 of which also had CER RRBS. TCX levels of five AD‐related proteins from three tissue fractions (buffer‐, detergent‐, and in‐soluble) were measured previously by ELISA (Liu 2020). CpG methylation (CpGm) was binned by a 15‐state chromatin model (Kundaje 2015) and averaged into CpGm clusters (rCpGm). rCpGms were tested for association with each AD endophenotype and expression levels of nearby genes through multi‐variable linear regression. Replication of significant rCpGms was performed in two independent datasets with AD‐related endophenotypes (Shireby 2022, De Jager 2014).ResultsOur innovative binning method demonstrated biologically relevant CpGm patterns. We found epigenome‐wide significant associations primarily with tau‐related endophenotypes including 93 that had significant and concordant associations in the replication datasets. These rCpGms also significantly associated with expression of nearby genes showing enrichment in oligodendrocyte marker genes including those related to myelination. In vitro validation of tau effects on oligodendrocyte gene expression through DNAm is ongoing.ConclusionsAlthough all endophenotypes tested are core to AD pathophysiology, our results suggest each has a distinct epigenetic architecture underlying their variability in the AD brain. In particular, we found evidence of DNAm variability associating with brain oligodendrocyte gene expression and TCX tau levels. By discovering AD brain endophenotype‐specific DNAm changes, we can identify core components of complex mechanisms revealing important biological insights into AD pathophysiology.