The mechanism of R-loop formation in mammalian immunoglobulin class switch recombination

R-loops form at immunoglobulin (Ig) heavy chain locus in germinal center B-cells undergoing class switch recombination (CSR). These are triple-stranded structures formed upon transcription of the G-rich switch sequences where the G-rich transcript hybridizes with the C-rich DNA template strand, leaving the nontemplate DNA strand in a single-stranded conformation. These structures aid in targeting of the CSR initiating activation-induced cytidine deaminase to the participating switch regions to initiate CSR. CSR involves exchanging IgM heavy chain domain gene (Cmu) with another heavy chain (CH) gene located downstream of Cmu to generate IgG, IgA or IgE antibodies. This occurs by a recombination event between the long switch regions located upstream of the CH genes in the Ig locus. To understand the mechanism of R-loop formation and study their biochemical and structural properties, we have performed experiments with a minimal R-loop forming sequence as our model.;We find that R-loops form by a threadback mechanism where the single-stranded RNA coming out of the RNA polymerase invades the upstream edge of transcription bubble to establish a hybrid with the template strand DNA. We observe that R-loop formation efficiency decreases with a decreasing number of G-rich switch repeats and G-clustering is the primary sequence feature that helps in R-loop formation. A sequence with a high G-density but no clustering cannot efficiently form R-loops. We find that R-loop formation on the template strand DNA is not dependent upon G-quartet formation on the nontemplate strand DNA. R-loops can initiate at short G-clustered regions by formation of short thermodynamically stable RNA:DNA [R-loop initiations zones (RIZ)]. Immediately downstream of the RIZ, R-loop elongation zones (REZ) are also G-rich, and they can harbor R-loop structures by providing a longer G-rich region. Whereas a G-clustered RIZ can initiate R-loop formation efficiently, G-rich but unclustered regions cannot effectively serve as RIZs. REZs, on the other hand, have to be G-rich but not necessarily G-clustered. Our experiments also indicate that RIZ sequences can help in R-loop formation at distant REZ regions, and DNA conformation is an important determinant of R-loop formation.