Because of the similarity with PROMPTs, we reasoned that capped enhancer RNAs might be rapidly degraded by the exosome. Indeed, siRNA-mediated depletion of the hMTR4 (SKIV2L2) co-factor of the exosome complex resulted in a median 3.14-fold increase of capped enhancer-RNA abundance (Fig. 2e, and Supplementary Fig. 14a, b), but only a negligible increase at mRNA TSSs. This increasing trend is similar to that of PROMPT regions upstream of TSSs, although the increase of enhancer RNAs was significantly higher (P<4.6e-67, Mann-Whitney U test, Fig. 2e, and Supplementary Fig. 14b, c). Thus, the bidirectional transcriptional activity observed at enhancers is also present at promoters, as suggested previously17, but in promoters only the antisense RNA is degraded. Furthermore, the CAGE expression of enhancers in control and hMTR4-depleted cells was proportional (Supplementary Fig. 14d), suggesting that virtually all identified enhancers produce exosome-sensitive RNAs. The number of detectable bidirectional CAGE peaks increased 1.7-fold upon hMTR4 depletion and novel enhancer candidates had on average similar, but weaker, chromatin modification signals compared to control HeLa cells (Supplementary Fig. 14e).
