Accounting for RNA polymerase heterogeneity reveals state switching and two distinct long-lived backtrack states escaping through cleavage

Pausing by bacterial RNA polymerase (RNAp) is vital in the recruitment of regulatory factors, RNA folding, and coupled translation. While backtracking and intra-structural isomerization have been proposed to trigger pausing, our understanding of backtrack-associated pauses and catalytic recovery remains incomplete. Using high-throughput magnetic tweezers, we examined the E. coli RNAp transcription dynamics over a wide range of forces and NTP concentrations. Dwell-time analysis and stochastic modeling identified, in addition to a short-lived elemental pause, two distinct long-lived backtrack pause states differing in recovery rates. We further identified two stochastic sources of transcription heterogeneity: alterations in short-pause frequency that underlie elongation-rate switching, and RNA cleavage deficiency that underpins different long-lived backtrack states. Together with effects of force and Gre factors, we demonstrate that recovery from deep backtracks is governed by intrinsic RNA cleavage rather than diffusional Brownian dynamics. We introduce a consensus mechanistic model that unifies our findings with prior models.