Neurogenetic contributions to amyloid beta and tau spreading in the human cortex

Abstract

Tau and amyloid beta (Aβ) proteins accumulate along neuronal circuits in Alzheimer's disease. Unraveling the genetic background for the regional vulnerability of these proteinopathies can help in understanding the mechanisms of pathology progression. To that end, we developed a novel graph theory approach and used it to investigate the intersection of longitudinal Aβ and tau positron emission tomography imaging of healthy adult individuals and the genetic transcriptome of the Allen Human Brain Atlas. We identified distinctive pathways for tau and Aβ accumulation, of which the tau pathways correlated with cognitive levels. We found that tau propagation and Aβ propagation patterns were associated with a common genetic profile related to lipid metabolism, in which APOE played a central role, whereas the tau-specific genetic profile was classified as 'axon related' and the Aβ profile as 'dendrite related'. This study reveals distinct genetic profiles that may confer vulnerability to tau and Aβ in vivo propagation in the human brain.

Conflict of interest statement

Competing Financial Interests Statement: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Figures

Figure 1.. Propagation Routes from Prominent Areas of Tau and Amyloid Deposits.

Cortical connectivity maps of a priori selected areas, namely inferior temporal for tau (left) and posterior cingulate cortex (PCC) for Aβ (right), from cross-sectional (a and c, N=69; HABS healthy adults) and longitudinal (b and d, N=19; HABS healthy adults) data. Scatterplots of spatial similarity between cross-sectionally predicted and longitudinally observed network connectivity profiles (green dots). X- and Y-axis show the z-scores of the connectivity values. Network diagrams illustrate theoretical examples of seed-based connectivity in cross-sectional (no arrows) and longitudinal (arrows) samples.

Figure 2.. Hubs of Pathology Propagation of Tau and Amyloid.

Cortical maps of longitudinal weighted-degree connectivity (directional relationships) of tau (a; Out-degree maps in top-left; In-degree maps in middle-left, N=19; HABS healthy adults) or Aβ pathways (b; Out-degree maps in top-right; In-degree maps in middle-right, N=19; HABS healthy adults). Network diagrams illustrate high and low theoretical examples of out-degree and in-degree. Overlap maps with an arbitrary threshold of z-score≥3SD across the cortical mantle of tau and Aβ are presented at bottom left and right.

Figure 3.. Tau and Amyloid Propagation-Based Staging.

Characterization of propagation pathways of tau and Aβ deposits in individuals of the cross-sectional data (healthy HABS study participants (SP) with complete neuropsychological assessment, N=64) and Alzheimer’s disease (AD/MCI, N=19) subjects; polar and violin plots; a and b). Mean is measure of center in violin plots (dark line; b). Smooth histograms of SP (light gray dots and histograms) and AD/MCI (dark gray dots and histograms) data in violin plots are obtained by using the cumulative histogram, a smoothening spline and the analytical derivative. Characterization of the relationship between propagation pathways of tau and Aβ and cognitive levels in SP (c and d). X-axis shows the targeted regions of interest in b, and cognitive scores in d. Y-axis shows Propagation-based Staging scores in b and d. MMSE=Mini-Mental State Examination. Two-tailed unpaired t-test corrected by multiple comparisons was used in b and Pearson’s correlation (and related p-value from one-tailed t-test non-corrected by multiple comparisons) was used in c and d.

Figure 4.. Brain Co-localization of In Vivo Propagation Patterns and Allen Gene Expression Data.

Matrix of similarity scores (hot scale) between in vivo propagation patterns of tau and Aβ (columns) and Allen gene expression levels of 21 previously related AD genes (a; star symbols show highest similarity scores). Scatterplots between in vivo propagation patterns of tau and Aβ and MAPT expression (top; red dots; b) and CLU expression (top; blue dots; d) genes, as well as their cortical distributions (bottom; hot color scale; b and d). Null hypothesis distributions of similarity scores between in vivo propagation patterns of tau (c) and Aβ (e) and the entire protein-coding transcriptome. Similarity scores were converted to z-scores, and corresponding one-tailed p-values were obtained for MAPT and CLU z-scores. Red areas show similarity scores above 2σ. Scatterplots are shown between in vivo regional intensity of tau or Aβ and MAPT (top; red dots; f) and CLU (bottom; blue dots; f) genes. Note that scatterplot in f shows SUVR and DVR values that represent linearly transformed values from the original maps, including down-sampling, conversion to 68-regions Desikan space, and whole sample averaging of the data. Pearson’s correlation was used in a to f to determine linear similarity scores between cortical maps of 68-regions (dots in b, d and f).

Figure 5.. Interactome and Gene Ontology Analyses of Imaging-Genetic Profiles.

Interactome network showing significant Gene Ontology (GO) overrepresented functionality (FWE corr p<0.05) and node centrality of common (a), or only tau (b), or only Aβ (c), genetic sets obtained from the neuroimaging-genetics similarity approach (Figure 4). Blue color scale in the interactome networks (left) shows types of genetic interactions. Labels in interactome networks (left) and in centrality bar graphs (right) show genes associated to the main GO overrepresented domains (middle). Top 25 most central genes of the interactome networks are shown in green in centrality bar graphs (right). Of note, Supplementary Tables 7, 9 and 11 show the GO functional labels and values, and Supplementary Tables 8, 10 and 12 show the list of genes and node centrality values for each interactome network.

Figure 6.. Propagation-Based Staging of Tau and Aβ and APOE Genotype in Individuals.

Diagram of neuroimaging-genetic relationships between in vivo PET phenotypes, cortical gene expression levels, and interactome profiles, aimed to identify biomarkers conferring vulnerability for tau and Aβ accumulation. Scatterplots show a linear relationship between tau and Aβ individual staging only in APOE ε4+ individuals (N=68; HABS healthy adults; 21 APOE ε4+ and 47 APOE ε4−, b). X- and Y-axis show propagation-based staging scores for tau and Aβ in b. Pearson’s correlation (and related p-value from one-tailed t-test non-corrected by multiple comparisons) was used in b.