The FANTOM5 project has generated a CAGE-based transcription start site (TSS) atlas across a broad panel of primary cells, tissues, and cell lines covering the vast majority of human cell types6. Within that dataset, well-studied enhancers often have CAGE peaks delineating nucleosome-deficient regions (NDRs) (Supplementary Fig. 1). To determine whether this is a general enhancer feature, FANTOM5 CAGE (Supplementary Table 1) was superimposed on active (H3K27ac-marked) enhancers defined by HeLa-S3 ENCODE ChIP-seq data7. CAGE tags showed a bimodal distribution flanking the central P300 peak, with divergent transcription from the enhancer (Fig. 1a). Similar patterns were observed in other cell lines (Supplementary Fig. 2a). Enhancer-associated reverse and forward strand transcription initiation events were, on average, separated by 180 bp and corresponded to nucleosome boundaries (Supplementary Figs 3 and 4). As a class, active HeLa-S3 enhancers had 231-fold more CAGE tags than polycomb-repressed enhancers, suggesting that transcription is a marker for active usage. Indeed, ENCODE-predicted enhancers7 with significant reporter activity8 had greater CAGE expression levels than those lacking reporter activity (P<4e-22, Mann-Whitney U test). A lenient threshold on enhancer expression increased the validation rate of ENCODE enhancers from 27% to 57% (Supplementary Fig. 5).
