To confirm the activity of newly-identified candidate enhancers, we randomly selected 46 strong, 41 moderate and 36 low activity enhancers (as defined by CAGE tag frequency) and examined their activity using enhancer reporter assays compared to randomly selected untranscribed loci with regulatory potential in HeLa-S3 cells: 15 DHSs10, 26 ENCODE-predicted ‘strong enhancers’7 and 20 enhancers defined as in Figure 1A (Supplementary Tables 2 and 3). While 67.4-73.9% of the CAGE-defined enhancers showed significant reporter activity, only 20-33.3% of the untranscribed candidate regulatory regions were active (Fig. 1c, and Supplementary Fig. 9a). The same trend was observed in HepG2 cells (Supplementary Fig. 10a, b). Corresponding promoter-less constructs showed that the enhancer transcription read-through is negligible (Supplementary Fig. 9b, c). Large fractions of CAGE-defined enhancers overlapped predicted ENCODE ‘strong enhancers’ or ‘TSS’ states (25% and 62%, respectively, for HeLa-S3), but there was no substantial difference in validation rates between these classes (Supplementary Fig. 10c, d). In summary, active CAGE-defined enhancers were much more likely to be validated in functional assays than untranscribed candidate enhancers defined by histone modifications or DHSs.
