R-loops as a threat to genome stability
Transcription can result in the formation of R-loops, in which the newly synthesised RNA hybridises with the DNA template strand prior to re-annealing of the transcription bubble to form a stable DNA-RNA hybrid structure (Figure 1A)(reviewed in (Santos-Pereira and Aguilera, 2015)) . R loops have physiological functions in transcription and in accurate chromosome segregation (Boque-Sastre et al., 2015; Kabeche et al., 2018), however, the pathological accumulation of R-loops has been repeatedly shown to increase genetic instability (reviewed in (Brickner et al., 2022; Huertas and Aguilera, 2003; Neil et al., 2018)). Genome-wide profiling shows that R-loops form more readily following RNA polymerase II pausing near transcription starting sites or transcription termination sites (TSS or TTS, respectively) but also form stochastically, particularly with polymerase stalling or in regions of high GC content, and formation is facilitated by negative supercoiling accumulated behind RNA polymerase (Chan et al., 2014; El Hage et al., 2010; Skourti-Stathaki et al., 2014; Skourti-Stathaki et al., 2011; Wahba et al., 2016).
Budding yeast utilises various mechanisms to ameliorate the impact of R-loops on genome stability. Firstly, coating of nascent mRNA with processing and export factors including THO/TREX reduces R-loop formation and replication fork slowing (Garcia-Rubio et al., 2018; Gomez-Gonzalez et al., 2011; Luna et al., 2019; Wellinger et al., 2006), with THO/TREX mutants having unstable genomes (Prado et al., 1997; San Martin-Alonso et al., 2021; Selvam et al., 2014). Another RNA-binding protein, Npl3, which is involved in mRNA splicing and export in yeast, was shown to decrease R-loop dependent genomic instability and replication stress, and physical proximity of genes to nuclear pores which accelerates mRNA export was shown to reduce R-loop accumulation (Gaillard et al., 2018; Garcia-Benitez et al., 2017; Santos-Pereira et al., 2013). Secondly, helicases can unwind R-loops; the transcription termination factor Sen1 has a particular role in preventing R-loop formation during S phase, with Sen1 depletion causing accumulation of both R-loops and the ɣH2A histone marker which indicates the concomitant onset of DNA replication stress (Khurana and Oberdoerffer, 2015; Mischo et al., 2011; San Martin-Alonso et al., 2021). Thirdly, both RNase H1 and RNase H2 can degrade the RNA in R-loops.  Mutants lacking RNase H2 have a much greater impact on R-loop mediated genome instability (O’Connell et al., 2015; Zimmer and Koshland, 2016), but RNase H1 has an ’on-demand’ activity at the critical time in S phase (Lockhart et al., 2019). Fourthly, chromatin modifiers including Rtt109 and FACT counter the formation of R-loops (Canas et al., 2022; Herrera-Moyano et al., 2014).
This multi-layered defence against R-loop formation and persistence is not surprising given the genome instability that results when cells with increased R-loops undergo DNA replication. However, the mechanism by which R-loops cause genome instability remains unclear as replicative helicases can unwind RNA-DNA helices in vitro and replication forks can overcome DNA–RNA hybrids formed in co-directional transcription units (Garcia-Rubio et al., 2018; Hamperl et al., 2017; Shin and Kelman, 2006). Moreover, hyperrecombination phenotypes caused by R-loops in yeast depend on Histone H3S10 phosphorylation suggesting an indirect mechanism involving chromatin (Garcia-Pichardo et al., 2017). One possible explanation is that in head-on encounters the RNA polymerase becomes trapped between the R-loop and the replisome, particularly in the absence of Sen1; this could impair normal mechanisms for removing the RNA polymerase holoenzyme from DNA in front of the replisome and form a potent roadblock (Felipe-Abrio et al., 2015; Zardoni et al., 2021). Alternatively, as noted above ssDNA is more prone to base damage, which would therefore accumulate on the non-template strand in transcriptional R-loops. However, experiments to address this mechanism discovered lesions arising from the replication bubble but there was no dependence on R-loops (Williams et al., 2023). Finally, R-loops can be targeted by the TC-NER or global genome NER pathway, resulting in cleavage and removal of the RNA:DNA duplex followed by gap repair (Crossley et al., 2023; Sollier et al., 2014), which forms a short-lived ssDNA gap that could in theory be processed into a DSB by DNA replication.
Negative supercoiling behind RNA polymerase can also promote the formation of G-quadruplexes - non-B form four-stranded nucleic acid structures made between several guanosines, often stabilised by monovalent cations (Selvam et al., 2014). These structures can form throughout the S. cerevisiae genome and particularly at telomeres though they are not required for telomere function (Esnault et al., 2023; Skourti-Stathaki et al., 2014). Like R-loops, these structures have been implicated in replication fork stalling and leading to genome instability, and are removed by the helicase Pif1 (Lopes et al., 2011; Paeschke et al., 2011; Piazza et al., 2010).
Interruption of replisome progression by R-loops and G-quadruplexes has the potential to be recombinogenic, and may contribute to genetic heterogeneity even in unperturbed cells in a transcription-dependent manner. However, R-loops and G-quadruplexes are common – the fact that these species can be detected in genome wide assays even in wild-type yeast, indicates that at any given site they are present in a significant fraction of cells. If each R-loop even delayed the replisome, let alone caused a time-consuming and mutagenic repair process, this would dramatically slow DNA replication irrespective of mutational burden. In reality, even RNase H1/H2 double mutants in which R loops are highly stabilised are viable and do not show a substantial growth defect (Zhao et al., 2018), suggesting that at most only a tiny fraction of R-loops actually perturb replisome progression.