Fig 3.

Identification of open chromatin regions in tonsillar B cells and prediction of genes linked to differentially accessible EBNA3A enhancer loci. (A) Schematic of ATAC-seq experiment indicating that data were generated from donor-matched CD19⁺ B cells sorted for naïve, memory, GC DZ, GC LZ, and plasma cells (PC). (B) MA plots (log₂ fold change vs. mean average) showing differentially accessible regions between fractions of interest from tonsils. Red points indicate regions with a false discovery rate (FDR) <0.05. (C) Upset plot of genome-wide overlapping accessible chromatin regions among tonsillar B-cell subsets and LCLs. (D) Bullseye plots depicting the fraction of intervals with differential accessibility among tonsillar subsets, their co-incidence with EBNA3A and enhancer ChIP-seq signatures in LCLs, and cis-linked gene predictions. Logical gates were applied to filter ATAC peaks present in one subset but not in others (e.g., LZ ! DZ) and to intersect these differentially accessible loci against ChIP-seq peaks for EBNA3A and histone enhancer patterns (e.g., ∩ EBNA3A and/or ∩ enh). Regions matching given logical gating criteria were filtered to exclude any intervals shorter than 100 bases. Genes with predicted cis-regulatory linkages to intervals matching each set of gates were identified using GREAT (44). (E) Distribution of DZ-unique EBNA3A- and enhancer-associated gene-linked intervals relative to gene TSSs. This gate recipe [(DZ ! LZ ! N ! M ! P) ∩ EBNA3A ∩ enh] yielded no significant linked genes. (F) Distribution of LZ-unique EBNA3A- and enhancer-associated gene-linked intervals relative to gene TSSs. This gate recipe [(LZ ! LZ ! N ! M ! P) ∩ EBNA3A ∩ enh] yielded multiple statistically significant linked genes (binomial gene test FDR Q <0.05), including BCL2A1 (BFL-1).