Dementia risk genes exhibited broad systemic expression.

A: Venn diagram visualizing the number of filtered risk genes and their overlap across diseases (AD: red, DLB: blue, and FTD: green). B-D: Pie charts depicting the degree of specificity of the neurodegenerative disease risk genes at tissue, brain region, and cell level (again AD: red, DLB: blue, and FTD: green). The specificity categories from highest to lowest specificity are defined as: tissue/brain region/cell type enriched entails ≥ four-fold higher mRNA level in that tissue/brain region/cell type compared to all other tissue/brain region/cell type; group enriched entails ≥ four-fold higher average mRNA level in a group of 2-5 tissues/brain regions or 2-10 cell types compared to all other; tissue enhanced entails ≥ four-fold higher mRNA level in a particular tissue/brain region/cell type compared to average level in all other; and low tissue/brain region/cell type specificity mRNA is detected but the levels are not elevated in any tissue/brain region/ cell type compared to the other. See 1 for the specific compartments of the enriched genes.

Risk gene fold enrichment across tissue, brain region, and cell gene modules.

A and B show tissue module data, C and D brain region mouel data, and E and F cell-type module data. Panels A, C, and E are dot plots depicting the fold enrichment and p-value per module. Each dot represents one module, x-axis is fold enrichment, and y-axis is -log p-value. Dots colored red (AD), blue (DLB), and green (FTD) represent enrichment of padjusted < 0.1, while yellow points represent enriched modules of significance pnominal < 0.05. Panel B,D, and F are venn diagrams illustrating the names and overlaps of enriched modules between disease groups. The numbers is the sum of disease-specific or shared modules, the module numbers and names are summarized in linked boxes. The color coding is as in the dot plots, yellow boxes are significant by pnominal < 0.05 while red, blue or green boxes are significant by padjusted < 0.1. Striped boxes are for modules significant for more than one disease, the red/yellow striped boxes indicate that the module is of significan ce padjusted < 0.1 in AD, but pnominal < 0.05 in DLB or FTD. This indicates the number of interesting gene modules per disease while visualizing similarities and differences across dementias. For enrichment values based on Monte Carlo simulation see 1, 2, and 3

UpSet plot of druggable gene targets across disease risk genes and risk signatures.

The drugs were categorized according to whether they have indications specific for neurodegneration or lowering lipid levels, they target gene products of the dementia risk genes, target genes part of any module of the immune/liver/cilia/neuronal associated risk signatures, and part of any gene module (not part of a risk signature) enriched by dementia risk genes (called no category). Each set of horizontal bars represents a category of druggable gene products: disease risk genes (AD, DLB, and FTD), specific drug indications (neurodegenerative disease or lipid lowering), any gene part of risk signature modules (neuronal, immune, liver, and cilia), or no assigned category. Horizontal bars indicate the total number of unique drugs per category. Vertical bars represent the size of each intersection, with dots and connecting lines below indicating which categories overlap. Only combinations specific to > 2 drugs are shown in the plot. A total of 1777 drugs were analyzed, of which only 35 have indications for neurodegnerative disease, and 40 a lipid-related indication. 809 drugs target gene products of genes part of the neuronal risk signature, and the largest drug target overlap is between the neuronal and immune risk signatures (n = 64), followed by neuronal and liver associated risk gene signatures (n = 58) and AD risk genes and the liver associated gene signature (n = 51). Beyond the currently indicated treatments for neurodegeneration we identified an extensive repertoire of experimental and approved compounds with targets mapping directly to the resolved risk signatures, suggesting significant repurposing potential for dementias.

The HPA GeneSet Explorer Workflow.

The workflow is divided into I: data retrieval and II: a data analysis component. The data retrieval function searches and retrieves a list of risk genes from the GWAS Catalog using ontology-based terms and filtered by coding status and significance threshold. These genes are then mapped based on expression data for the tissue, brain, and single cell resource of the HPA. In the data analysis component, spatial expression modules are analyzed to identify regions, cell types, and functions enriched for risk genes using statistical tests and assigned to UMAPs of the gene network (see 4, 5, and 6). This approach facilitates spatial interpretation of genetic risk by identifying vulnerable systems and organizing risk genes into discrete functional units. By characterizing the molecular landscape of these disease-associated modules, the investigation pipeline expands beyond individual mutation-based risk genes, providing a broader framework for future analysis of disease-specific vulnerability.

Bar plots of individual risk gene enrichment at the different biological scales.

Bar plots representing the risk genes defined as enriched or group enriched in a tissue/brain region/cell type according to the HPA enrichment scoring system mentioned in 1. The y-axis is the number of genes enriched and the x-axis shows the tissue, brain region, or cell type that is enriched. Top panel: tissue, Middle panel: brain regions, and Bottom panel: cell-types. The three diseases are plotted separately and color coded AD: red, DLB: blue, and FTD: green. These plots visualize the low levels of risk gene specific tissue/brain region/cell type enrichment are for each disease.

Lollipop plots showing the Monte Carlo simulation disease risk gene enrichment of the tissue gene modules per disease.

Enrichment of dementia risk genes in tissue gene modules were analyzed through 106 Monte Carlo simulations. All modules enriched pempirical < 0.05 are highlighted by color, the gray lollipops are not significant or of negative Z-score (AD: Red, DLB: Blue, and FTD: Green). The size of the plot point indicates the p-value (lower pempirical has a larger dot size).

Lollipop plots showing the Monte Carlo simulation disease risk gene enrichment of the brain reigon gene modules per disease.

Enrichment of dementia risk genes in brain region gene modules were analyzed through 106 Monte Carlo simulations. All modules enriched pempirical < 0.05 are highlighted by color, the gray lollipops are not significant or of negative Z-score (AD: Red, DLB: Blue, and FTD: Green). The size of the plot point indicates the p-value (lower pempirical has a larger dot size).

Lollipop plots showing the Monte Carlo simulation disease risk gene enrichment of the cell type gene modules per disease.

Enrichment of dementia risk genes in cell gene modules were analyzed through 106 Monte Carlo simulations. All modules enriched pempirical < 0.05 are highlighted by color, the gray lollipops are not significant or of negative Z-score (AD: Red, DLB: Blue, and FTD: Green). The size of the plot point indicates the p-value (lower pempirical has a larger dot size).

HPA tissue UMAP visualization per disease.

UMAP visualization of the gene modules based on the HPA tissue data, highlighting disease gene enriched modules (pnominal < 0.05). Individual genes are marked by cyan triangle points. The risk genes and enriched modules specific to each disease is depicted in their own UMAP, AD in A; DLB in B; and FTD in C.

HPA brain region UMAP visualization per disease.

UMAP visualization of the gene modules based on the HPA brain data, highlighting disease gene enriched modules (pnominal < 0.05). Individual genes are marked by cyan triangle points. The risk genes and enriched modules specific to each disease is depicted in their own UMAP, AD in A; DLB in B; and FTD in C.

HPA cell type UMAP visualization per disease.

UMAP visualization of the gene modules based on the HPA cell data, highlighting disease gene enriched modules (pnominal < 0.05). Individual genes are marked by cyan triangle points. The risk genes and enriched modules specific to each disease is depicted in their own UMAP, AD in A; DLB in B; and FTD in C.

Heatmaps of cell type expression of the liver- and immune associated risk signature genes.

Heat map of the Alzheimer’s disease risk genes part of the liver (panel A) and immune (panel B) associated signatures gene expression in spatial sequencing data of HPA human non-demented frontal cortex data. Expression across the cell types; astrocytes, microglia, neurons, vasculature, and oligodendrocytes. Not all genes of the risk signatures were found in the spatial data.