reflect an enrichment of AD GWAS associations for genes that are expressed in myeloid cells rather than specifically among PU.1 target genes. To attempt to address this issue, we performed stratified LD score regression of AD heritability partitioned by functional annotations obtained from SPI1 (marking the PU.1 cistrome) and POLR2AphosphoS5 (marking actively transcribed genes) ChIP-Seq experiments, performed in duplicate, using a human myeloid cell line (HL60) by the ENCODE Consortium37. We observed a significant enrichment for SPI1 (PU.1) (34.58 fold enrichment, P=1.31×10−3 in first replicate; 58.12 fold enrichment, P=4.95×10−3 in second replicate) much stronger than that for POLR2AphosphoS5 (15.78 fold enrichment, P=1.71×10−2 in first replicate; 16.34 fold enrichment, P=1.25×10−1 in second replicate), consistent with our hypothesis (Supplementary Table 12).
