Analysis of gene expression in the postmortem brain of neurotypical Black Americans reveals contributions of genetic ancestry.

Study design for the examination of the genetic and nongenetic contributions to genetic ancestry-associated gene expression differences.

Extensive ancestry-associated expression changes across brain regions.a, Circos plot showing global ancestry DEGs across the caudate nucleus (red), dentate gyrus (blue), DLPFC (green) and hippocampus (purple). b, GSEA of differential expression analysis across brain regions, highlighting terms associated with increased African or European ancestry proportions based on normalized enrichment score (NES) direction of effect. c. UpSet plot showing large overlap between brain regions. Green is shared across the four brain regions; blue is shared across three brain regions; orange is shared between two brain regions; and black is unique to a specific brain region. The single asterisk indicates significant pairwise enrichment (two-sided Fisher’s exact test; P = 2.0 × 10−135 (caudate nucleus versus dentate gyrus), 4.9 × 10−324 (caudate nucleus versus DLPFC), 2.8 × 10−288 (caudate nucleus versus hippocampus), 1.8 × 10−166 (dentate gyrus versus DLPFC), 9.8 × 10−169 (dentate gyrus versus hippocampus) and approximately 0 (DLPFC versus hippocampus) or significant overlap between all four brain regions (Monte Carlo simulation). d, Heatmaps of the proportion of global ancestry DEG sharing with concordant direction (top, sign match) and within a factor 0.5 effect size (bottom) e, Metaplot showing examples of brain region-specific ancestry effects.

Ancestry-associated genes and canonical transcripts are evolutionarily less constrained.a, Significant depletion of ancestry DEGs for evolutionarily constrained genes (canonical transcripts) across brain regions. Significant depletion and enrichments (two-sided Fisher’s exact test, FDR-corrected P, −log10-transformed) are annotated within the tiles. Odds ratios (ORs) were log2-transformed to highlight depletion (blue) and enrichment (red). b, A similar trend of depletion of ancestry DETs (all, canonical and noncanonical) for evolutionarily constrained transcripts across brain regions. ORs were log2-transformed to highlight depletion (blue) and enrichment (red). c, The mean of ancestry-associated differential expression (that is, genes and transcripts) LFSR as a function of loss-of-function observed/expected upper bound fraction (LOEUF). The decile shows a significant negative correlation for genes (left; caudate nucleus (n = 122), dentate gyrus (n = 47), DLPFC (n = 123) and hippocampus (n = 133): two-sided Pearson correlation, r = −0.20, −0.20, −0.21 and −0.21; P = 3.0 × 10−122, 7.6 × 10−113, 8.6 × 10−126 and 1.2 × 10−122) and transcripts (right; caudate nucleus (n = 122), dentate gyrus (n = 47), DLPFC (n = 123) and hippocampus (n = 133): two-sided Pearson correlation, r = −0.05, −0.05, −0.04 and −0.04; P = 8.6 × 10−13, 1.7 × 10–11, 9.0 × 10−11 and 3.2 × 10−10). The error bars correspond to the 95% confidence intervals.

Genetic contribution of genetic ancestry differences in expression across the brain.a, UpSet plot showing large overlap of eGenes between brain regions. b, Heatmap of the proportion of global ancestry DEG sharing with concordant direction (sign match). c, Significant enrichment of ancestry-associated DEGs for eGenes (unique gene associated with an eQTL) across brain regions separated by the direction of effect (increase in AA or EA proportion). d, Density plot showing a significant increase in absolute AFDs (one-sided Mann–Whitney U-test, P < 0.05) for global ancestry-associated DEGs (red) compared with non-DEGs (blue) across brain regions. A dashed line marks the mean absolute AFD. Absolute AFD was calculated as the average absolute AFD across a gene using a significant eQTL (LFSR < 0.05). e, Correlation (two-sided Spearman) of elastic net predicted (y axis) versus observed (x axis) ancestry-associated differences in expression among ancestry-associated DEGs with an eQTL across brain regions. A fitted trend line is shown in blue as the mean value ± s.d. The s.d. is shaded in light gray.

DNA methylation-based contributions to global ancestry-associated differential expression.a, Circos plot showing local ancestry-associated DMRs across the caudate nucleus (red), DLPFC (blue) and hippocampus (green). Methylation status is annotated in red for hypermethylation and blue for hypomethylation. b, Gene term enrichment (hypergeometric and FDR-corrected) of DMRs across brain regions. c, Histograms showing the distribution of ΔPST associated with the impact of unknown environmental factors as captured by residualized VMR (corrected according to local ancestry, age, sex and unknown biological factors captured by principal component analysis (PCA)) for nearby global ancestry-associated DEGs. A dashed line marks the mean ΔPST. A solid line shows the density overlay.

Global ancestry-associated DEGs stratified according to coding or noncoding DEGs show general enrichment for heritability of several neurological and immune-related traits, but depletion for brain-related behavioral traits.Heatmap for ancestry-associated DEGs that show enrichment (red) or depletion (blue) for heritability of brain-related and immune-related traits from S-LDSC analysis. Significant enrichment for heritability traits disappears when limited to noncoding DEGs. Numbers within the tiles are the levels of enrichment (>1) or depletion (<1) that are significant after multiple testing correction (FDR < 0.05). Left, Results for all DEGs in each brain region. Middle and right, Results for DEGs increased with AA or EA proportions for each brain region, respectively.