A common 19 bp APOE enhancer deletion is protective against Alzheimer’s disease in African Americans

Figures

• 

  Fig. 1. A 19 bp deletion is common in African APOE-ε4 Alleles. a Alignment of phased long-read alleles. b gnomAD minor allele frequency of the 19 bp SNP. c Relative frequency of the deletion in 1000 Genomes Project populations ordered by relative abundance. * = African Ancestry in Southwest US. d Number of alleles with the deletion in APOE-ε4 homozygous individuals from the 1000 Genomes Project populations. e UCSC genome browser track showing the location of the deletion allele relative to APOE and transcription factor binding sites on the GRCh37 build. f Presence of the deletion and strong (red) and modest (orange) predicted SPI1 binding sites along with sequence conservation in the mouse genome.

• 

  Fig. 2. The rs148353395 19 bp deletion SNP is more abundant in controls versus AD cases in APOE-ε4 homozygous individuals of African ancestry. a Genotype distribution (top) and odds ratio (bottom) from African American ADSP individuals. b Stratification of cases and controls by local African ancestry at APOE with the percent allele distribution (top) and genotype distribution (Bottom). c, d Forest plot of odds ratios and 95% confidence intervals for the rs148353395 19 bp deletion, rs7412 the APOE-ε2 defining SNP, and rs429358, the APOE-ε4 defining SNP to APOE-ε3 homozygous individuals, adjusting for effects of the other variants in the model, sex, age, sequencing center, sequencing platform, and cohort in c, samples with local African ancestry at APOE (n = 7,248) and d all ADSP Samples (n = 38,994). e Kaplan-Meier survival analysis of age of onset for different genotype combinations for APOE-ε3, APOE-ε4, and the rs148353395 19 bp deletion for individuals of African local ancestry (n = 2,002) at APOE. *p < 0.05 by log rank test versus APOE-ε4/ε4 individuals without the 19 bp deletion.

• 

  Fig. 3. Phenotypes related to APOE-associated SNPs in the NIH All of Us dataset. a–c Phenome-wide association study (PheWAS) Manhattan plot for 1816 traits grouped into related categories on the x-axis compared with significance on the y-axis for a, rs429358 (the APOE-ε4 defining SNP), b rs7412 (the APOE-ε2 defining SNP) and c rs148353395 (the 19 bp deletion). d Forest plot of the PheWAS odds ratio, standard error, and whether Bonferroni significance was reached for the conditions which passed Bonferroni significance for at least the APOE-ε4 or APOE-ε2 defining SNP and had at least 50 cases for rs7412 (green), rs429358 (blue), and rs148353395 (yellow). Also shown are the odds ratio for rs148353395 after adjusting for rs7412 and rs429358 (in purple) and for rs429358 after adjusting for rs148353395 (in red). Significance is denoted by a filled square for conditions which passed Bonferroni correction (p < 2.75 × 10-5), nominal significance (p < 0.05) with a filled triangle, and an unfilled circle for conditions which did not reach significance. The n for each trait and variant can be found in Supplementary Data 1-5. Delirium dementia*: full phenotype name is: “Delerium dementia and amnestic…

• 

  Fig. 4. The region 1 kb distal to the APOE 3′UTR acts as a transcriptional repressor and Aβ and SPI1 treatment impacts APOE-associated transcript expression in HMC3 cells. a Schematic of the luciferase experimental design and constructs. The APOE 3′UTR (APOE 3′UTR only) plus ~400 bp of downstream DNA containing the 19 bp deletion region and SPI1 binding site were cloned into a psiCheck2.2 dual luciferase reporter construct. b Renilla:Firefly luciferase expression data in HMC3 cells after delivery of APOE SPI1 WT sequence (p = 0.0016 relative to the APOE 3′UTR only) or delivery of the 19 bp deletion [p = 0.8252; one-way ANOVA, n = 12 samples/group (3 biological replicates, 4 technical replicates), ± SD]. c–e HMC3 gene expression following SPI1 overexpression. c Effect of SPI1 overexpression on c APOE expression (p = 0.3831, F = 1.789, unpaired two-tailed t-test, n = 12 control and 12 SPI1 biological replicates derived from the average of 4 technical replicates/sample, ± SEM), d APOC1 expression (p = <0.0001, F = 2.987, unpaired two-tailed t-test, n = 11 control and 12 SPI1 biological replicates derived from the average of 4 technical replicates/sample, ± SEM), and e lncRNA ENSG00000280087…

• 

  Fig. 5. PheWAS analysis of intergenic variants between APOE and APOC1 from NIH All of Us. a Pairwise linkage disequilibrium (LD) plot of Pearson correlation (r2) between SNPs present in the intergenic region between APOE and APOC1 with a minor allele frequency of >1% and >0.2 r2 in samples of African ancestry (n = 32,886 individuals). b LD map for the same SNPs in samples of European ancestry (n = 104,107 individuals). c Manhattan PheWAS multi-ancestry plot for rs438811. d Corresponding forest plot of odds ratios and standard error for rs438811 of the top 20 most significant phenotypes. e Manhattan plot for rs72654473. f Corresponding forest plot of odds ratios and standard error for rs72654473 of the top 20 most significant All of Us phenotypes. All values and ns available in Supplemental Tables 19 and 21.