The broken Alzheimer's disease genome

Abstract

The complex pathobiology of late-onset Alzheimer's disease (AD) poses significant challenges to therapeutic and preventative interventions. Despite these difficulties, genomics and related disciplines are allowing fundamental mechanistic insights to emerge with clarity, particularly with the introduction of high-resolution sequencing technologies. After all, the disrupted processes at the interface between DNA and gene expression, which we call the broken AD genome, offer detailed quantitative evidence unrestrained by preconceived notions about the disease. In addition to highlighting biological pathways beyond the classical pathology hallmarks, these advances have revitalized drug discovery efforts and are driving improvements in clinical tools. We review genetic, epigenomic, and gene expression findings related to AD pathogenesis and explore how their integration enables a better understanding of the multicellular imbalances contributing to this heterogeneous condition. The frontiers opening on the back of these research milestones promise a future of AD care that is both more personalized and predictive.

Graphical abstract

Figure 1

Intersecting paths: The genome and AD Despite the etiology of AD being dominated by a strong genetic component, nonheritable elements also play a role in shaping disease risk. These encompass behavioral factors, such as nutrition, physical activity, and sleep, in addition to environmental determinants, including pollution and socioeconomical status. Together, they affect gene expression across the lifespan. The genome further interacts with the gradual accrual of neuropathological alterations in the brain, comprising both protective and disease-driving programs in a multicellular web of feedback and feedforward responses.

Figure 2

Functional mapping of late-onset AD genetic etiology (A) Pathway enrichment of genetic loci linked to AD based on statistically significant GWAS signals identified by Bellenguez et al., a two-stage case-control study totaling 111,326 clinically diagnosed/”proxy” AD cases and 677,663 control individuals, published in 2022. Their analysis, the largest to date, led to the identification of 75 risk loci, including 42 novel genetic associations. GWAS hits are genetic variants that occur more frequently in individuals with a particular trait or disease. These polymorphisms are most common in non-coding genomic regions and, as a rule of thumb, require additional validations before annotations can be unambiguously assigned to a specific gene. We used Metascape, SynGO, expression profiles, and manual curation to group genes according to their function;^, a small fraction of hits could not be interpreted confidently and were omitted. Unsurprisingly, various genes overlap across distinct functional processes, particularly with regards to APP metabolism, immunity, and the endo-lysosomal pathway, such are the interdependencies between these processes. We note, in addition, that ADAM10 and ADAM17 are α-secretases; their identification suggests that altered non-amyloidogenic APP processing can influence AD pathogenesis. (B) Genomic distribution of lead SNPs associated with increased AD risk. GWAS data were mined from panel compiled by Andrews et al. Functional annotations were performed using gnomAD and GWAS Catalog.

Figure 3

Cellular map of AD genetic risk Shown is an expression profile overview of prioritized genetic loci harboring genome-wide significant signals linked to AD risk. (A) Simplified Manhattan plot derived from the dataset collected by Bellenguez et al. The p values were calculated using a fixed-effect meta-analysis. Brackets indicate different prioritized genes in the same locus. Note also that GWAS signals from independent studies linked to EPHA1 stem from two adjacent loci. (B) Proportional expression of GWAS hits across parenchymal dorsolateral prefrontal cortex cells. Ex. neuron, excitatory neuron; In. neuron, inhibitory neuron; OPC, oligodendrocyte precursor cell. Expression values are weighted on a 0–100 scale; darker colors indicate progressively higher expression values. (C) Differential gene expression analysis centered on the interaction between APOE genotype and other AD GWAS hits. Data points in (B) and (C) were mined from snRNA-seq analyses performed by Blanchard et al. (D) Proportional expression of GWAS hits across major vascular cell types (hippocampus and superior frontal cortex). Asterisks mark the cell type carrying the strongest overall expression for each gene. aSMC, vascular smooth muscle cell; aaSMC, arteriolar smooth muscle cell; T-Pericyte, solute-transport pericyte; M-Pericyte, ECM, extracellular matrix-regulating pericyte; P-Fibroblast, perivascular fibroblast; M-Fibroblast, meningeal fibroblast; P-Macrophage, perivascular macrophage. Vessel isolation and nuclei extraction for sequencing (VINE-seq) data were extracted from Yang et al. (E) Gene expression comparison between AD and control cases. This analysis was circumscribed to the cell type most highly enriched in each gene (i.e., those marked with an asterisk in D). All transcriptional trends shown should be regarded as indicative due to limited sample sizes and, at times, high interindividual variation.

Figure 4

Somatic mosaicism in the AD brain As we age, DNA damage accumulates due to oxidative stress and other factors. Building on top of previous studies, a recent high-resolution survey of genomic integrity in the brain revealed that, compared to control cells, mutational rates are higher in excitatory neurons from AD donors than in normal aging. The mutational signatures in AD neurons are also different, hinting that the underlying processes sparking these changes are unique to AD. In addition to SNVs, which are discussed in detail in the main text, DNA double-strand breaks leading to gene fusions in AD excitatory neurons have also been reported recently. It is speculated that increased oxidative damage related to AD pathology acts as a primary genotoxic trigger. ROS, reactive oxygen species.

Figure 5

The AD transcriptome at scale (A) It has long been known that women are disproportionately affected by AD. Investigations into gene expression patterns associated with the disease have revealed that this sexual dimorphism is evident also at the transcriptional level, with female and male individuals responding differentially to pathology. snRNA-seq analyses have made it possible to pinpoint these differences to specific cell types, with neurons and oligodendrocytes emerging as hotspots of sex-biased responses. (B) Single-nucleus transcriptional profiling has also been instrumental in furthering our understanding of the effects mediated by APOE4 on the human brain. This led researchers to zero in on cholesterol dyshomeostasis in oligodendrocytes as a root cause of myelination-related defects in APOE4 carriers. Specifically, aberrant cholesterol deposition in APOE4 oligodendrocytes leads to endoplasmic reticulum stress pathway activation and ATF6 translocation to the nucleus, a transcription factor known to mediate lipotoxic responses. While the underlying molecular mechanisms remain incompletely defined, the upregulation of cholesterol metabolism genes in APOE4 oligodendrocytes coincides with decreased expression of myelin-associated genes and reduced overall myelination levels. (C and D) High-resolution spatial transcriptomics applied to the TauPS2APP animal model has revealed that Aβ plaques are surrounded in their immediate vicinity by a core shell structure of DAMs. Disease-associated astrocyte-like cells, oligodendrocytes, OPCs, and endothelial cells are found more distally, showing small but significant enrichments within 10–30  μm from plaques. In contrast, hyperphosphorylated tau (p-tau), which is found primarily in hippocampal CA1 excitatory neurons, is associated with a localized enrichment of oligodendrocyte subtypes. Interestingly, while both pathologies trigger oligodendrocyte reactivity, these analyses suggest that different subpopulations are recruited in response to amyloid and tau. The finding that microglia accumulate around Aβ plaques is not new, but methods like STARmap PLUS are opening doors to an unprecedented view of the molecular and cellular features related to AD neuropathology.