Abstract
R-loops are nucleic acid hybrids which form when an RNA invades duplex DNA to pair with its template sequence. Although they are implicated in a growing number of gene regulatory processes, their mechanistic origins remain unclear. We here report real-time observations of cotranscriptional R-loop formation at single-molecule resolution and propose a mechanism for their formation. We show that the bacterial Mfd protein can simultaneously interact with both elongating RNA polymerase and upstream DNA, tethering the two together and partitioning the DNA into distinct supercoiled domains. A highly negatively supercoiled domain forms in between Mfd and RNA polymerase, and compensatory positive supercoiling appears in front of the RNA polymerase and behind Mfd. The nascent RNA invades the negatively supercoiled domain and forms a stable R-loop that can drive mutagenesis. This mechanism theoretically enables any protein that simultaneously binds an actively translocating RNA polymerase and upstream DNA to stimulate R-loop formation.