The balancing act of R-loop biology: The good, the bad, and the ugly

Abstract

An R-loop is a three-stranded nucleic acid structure that consists of a DNA:RNA hybrid and a displaced strand of DNA. R-loops occur frequently in genomes and have significant physiological importance. They play vital roles in regulating gene expression, DNA replication, and DNA and histone modifications. Several studies have uncovered that R-loops contribute to fundamental biological processes in various organisms. Paradoxically, although they do play essential positive functions required for important biological processes, they can also contribute to DNA damage and genome instability. Recent evidence suggests that R-loops are involved in a number of human diseases, including neurological disorders, cancer, and autoimmune diseases. This review focuses on the molecular basis for R-loop–mediated gene regulation and genomic instability and briefly discusses methods for identifying R-loops in vivo. It also highlights recent studies indicating the role of R-loops in DNA double-strand break repair with an updated view of much-needed future goals in R-loop biology.

Introduction

R-loops were first discovered through in vitro studies over 40 years ago (1). Researchers in the Davis laboratory demonstrated that a hybrid DNA molecule composed of yeast DNA and bacteriophage λ DNA could form an R-loop with a strand of yeast 26S rRNA complementary to the yeast DNA in the presence of 70% formamide. It was hypothesized that the 70% formamide was necessary because formation of the DNA:RNA hybrid structure is more favorable near the melting temperature of dsDNA. Furthermore, it was shown that R-loops were very stable once formed, with degradation of the RNA strand being necessary for resolution of the R-loop structure. For these reasons, it was thought that R-loops were not biologically relevant. However, R-loops were discovered in vivo 20 years later (2), raising the question of whether these structures serve a biological purpose in living organisms. In vivo R-loop formation was demonstrated using Escherichia coli strains that had a null mutation for topA, the gene that encodes topoisomerase I (TOPO I),² or a temperature-sensitive mutation for gyrB, the gene that encodes DNA gyrase subunit B. It was thought that DNA gyrase subunit B promoted R-loop formation by resolving positive supercoiling within the DNA, whereas TOPO I opposed R-loop formation by resolving negative supercoiling. Overexpression of RNase H, an endoribonuclease that degrades the RNA in the DNA:RNA hybrid portion of an R-loop, rescued growth defects of the topA mutant. In addition, growth defects associated with loss of RNase H were compensated for by a decrease in DNA gyrase subunit B activity. These observations suggested that R-loop accumulation caused the growth defects.

In the years following this initial observation, R-loops have become the focus of many researchers, especially with emerging evidence of R-loops in gene regulation. One of the most well-characterized examples of biologically important R-loop formation is in class-switch recombination (CSR), which allows for antibody class diversification (3–7). At the beginning of CSR process in B cells, R-loops form at the G-rich switch region of the IgH locus during transcription. R-loop formation augments the action of activation-induced cytidine deaminase (AID) by providing this enzyme with a stable ssDNA substrate. AID then deaminates cytosine residues in the ssDNA portion of the R-loop, converting them to uracils, which are then processed to DNA nicks by base excision repair factors (5, 8). Mismatch repair proteins then process the DNA nicks to double-strand breaks (DSBs) to allow DNA end joining and “switching” of the Ig class, a process necessary for producing different types of Igs crucial for the humoral immune response (9, 10). Although, as mentioned later in this review, inefficient repair of DNA breaks during the process of CSR can lead to genomic instability (11), CSR remains a vital physiological process in which R-loops play an important role to allow for antibody class diversification.

R-loop involvement in gene regulation was also seen in plants (12). In Arabidopsis thaliana, antisense transcripts, named COOLAIR, were found to promote transcriptional silencing of FLOWERING LOCUS C (FLC) during cold exposure (13). FLC is a gene that encodes the floral repressor, FLC, whose epigenetic silencing occurs as a result of effective vernalization, a process of prolonged cold exposure that accelerates flowering. R-loop formation and stabilization at the COOLAIR promoter region was found to reduce the expression of these antisense transcripts (12). Unlike in CSR, R-loops in this example seem to be unfavorable structures down-regulating biologically beneficial transcripts.

Moreover, R-loop formation was found to be responsible for transcriptional blocks in rDNA repeats in the absence of topoisomerase I in yeast (14), which provides another example of R-loops being harmful, and their resolution is necessary for normal growth.

In these few examples, it is apparent that R-loops can be both beneficial and detrimental to cellular growth and cellular functions. In this review, we discuss both the benefits of R-loops and the risks they pose to cells, illustrating the tight balancing act that that R-loops must perform in vivo.

Methods to identify R-loops in vivo

To fully understand how these DNA:RNA hybrids function, it is necessary to characterize the sequences in which they form, and under what conditions. There are currently several techniques available to identify and characterize R-loop structures in vivo (Fig. 1). Until recently, most studies have relied on a single antibody to detect and isolate R-loops in vivo through a method termed DNA:RNA immunoprecipitation (DRIP) (15). This method is based on the use of S9.6 antibody, an mAb raised to a DNA:RNA hybrid in 1986, and has been extensively used in research, to specifically immunoprecipitate DNA:RNA hybrids (16). R-loop profiling can then be achieved by sequencing the DNA strand of an R-loop (DRIP-Seq) (15) or sequencing the RNA strand of an R-loop after cDNA synthesis (DRIPc-Seq) (17). Another method termed S1-DRIP-Seq involves the use of S1-nuclease to degrade the displaced ssDNA of an R-loop (18). Sequencing of the immunoprecipitated DNA allows for stranded R-loop profiling. The DRIP technique is currently the standard for mapping and measuring the abundance of R-loops. However, recent studies have demonstrated that there are various issues with the S9.6 antibody (19, 20). There is evidence that it has some sequence bias and off-target binding (19) to structures such as dsRNA (21). Several new techniques have been developed in recent years for more accurate and precise mapping. One such technique is R-ChIP, which uses a catalytically dead mutant of RNase H1 for immunoprecipitation (22). bisDRIP-Seq is another recently developed approach that has enabled mapping of R-loops at a high resolution, by combining the use of the S9.6 antibody with sodium bisulfite treatment, which deaminates unpaired cytosines in the displaced DNA strand (23). R-loops are then mapped using high-throughput DNA sequencing. With these improved techniques, however, we are still unable to identify precise start sites of R-loops in the genome, ask if they are continuous or discontinuous, or understand how prevalent these structures are in single cells.

Figure 1.

Current R-loop–mapping methods. A, S9.6 antibody-based methods. 1) DRIP-Seq: S9.6 antibody is used to specifically immunoprecipitate DNA:RNA hybrids after fragmentation. R-loop profiling can then be achieved by sequencing the DNA strands of an R-loop (15). 2) S1-DRIP-Seq: This method involves a slight modification of the DRIP-Seq. It involves an S1-nuclease treatment step, which preferentially degrades the displaced ssDNA. This is followed by fragmentation and immunoprecipitation (IP). Sequencing of the immunoprecipitated DNA allows for stranded R-loop profiling (18). 3) DRIPc-Seq: This method involves a DNase I treatment step after the immunoprecipitation of the DNA:RNA hybrid, followed by sequencing the RNA strand of an R-loop (17). 4) bisDRIP-Seq: This method combines the use of the S9.6 antibody with sodium bisulfite treatment, which deaminates unpaired cytosines in the displaced DNA strand. This modification enables for a near-nucleotide resolution mapping (23). B, R-ChIP-Seq. This method involves using cells expressing a V5-tagged catalytically dead mutant of RNase H1, which binds DNA:RNA hybrids without degrading the associated RNA. Anti-V5 antibodies are then used for immunoprecipitation, which is followed by sequencing the DNA strand of an R-loop (22). This method has been reported to yield strand-specific R-loop maps.

R-loops can form in cis and in trans

Genome-wide mapping studies for R-loop formation have shown that R-loops tend to form near transcriptionally active genes (15). In particular, they tend to form in promoter regions (17). R-loops commonly form during transcription when a nascent transcript reanneals to the template strand of DNA (Fig. 2) (24). This is the most accepted model of R-loop formation and is known as the “thread back model.” The thread back model is consistent with the crystal structure of RNA polymerase II (RNAPII), which shows that there are two separate channels for template DNA and nascent RNA (25).

Figure 2.

R-loop formation can occur in cis and in trans. A, R-loops can form co-transcriptionally when a nascent RNA transcript anneals to the template strand of DNA behind the extending RNAPII (cis R-loops). B, R-loop can form in trans when a transcript can hybridize to a DNA strand at a distant location. Examples of trans R-loops include but may not be limited to noncoding RNA–based R-loops and guide RNA (gRNA)–based R-loops formed when a guide RNA associated with CRISPR/Cas9 protein hybridizes to its target sequence.

Pausing of RNAPII at transcriptional start sites is also correlated with R-loop formation in human cells (22). R-loops may form at pause sites because of the additional time for the nascent strand to interact with and potentially hybridize with the template DNA. However, this model has not been fully tested.

Head-on collision of the transcription and replication machinery can also lead to R-loop formation. In Bacillus subtilis, R-loop accumulation leads to replisome stalling and mutagenesis (26). Similar results have also been shown in human cells (27). Note that it takes longer than one cell cycle to transcribe human genes that are greater than 800 kb in length (27); thus, collisions between replication and transcription machinery at these genes are inevitable.

R-loops have also been shown to form in trans, using RNA molecules transcribed at a distant genomic locus (Fig. 2). This was first demonstrated in a study using Saccharomyces cerevisiae whereby transcripts produced from the yeast genome were shown to form R-loops on a yeast artificial chromosome (28). R-loop formation depends on Rad51, a protein involved in double-strand break repair via strand exchange. However, the precise mechanism is not fully understood.

Factors promoting R-loop formation and stabilization

There are various factors (including DNA sequences, DNA topology, and proteins) that can increase the propensity of R-loop formation (Fig. 3). R-loop formation is more efficient when a G-rich transcript is produced from a C-rich template (29). In particular, in vitro assays have shown that G clusters—short stretches of G nucleotides—on the RNA transcript favor R-loop formation. Once R-loop formation is initiated, G clustering is less important for extension of the R-loop structure as long as the density of G nucleotides in the transcript is high—sometimes as high as 40–50% in common R-loop–forming regions in many organisms (29). It was concluded that the increased stability of a DNA:RNA hybrid between a G-rich strand of RNA and a C-rich strand of DNA is responsible for the likelihood of R-loop formation in these regions.

Figure 3.

Schematic representation of factors promoting R-loop formation and stabilization. Negative supercoiling promotes R-loop formation (24, 34–38). Topoisomerases inhibit R-loop formation by resolving negative supercoiling (37, 60). The presence of a nick in the nontemplate strand downstream of a promoter makes DNA reannealing less efficient, which favors nucleation of an R-loop (29, 35). DNA secondary structures (e.g. G-quadruplexes that form on the displaced strand of DNA) can also stabilize the R-loop structures (30). G-clusters on the RNA transcript favor R-loop formation (29). Defects in anti-R-loop factors like RNase H and RNA-DNA helicases also lead to increased R-loop formation (18, 46).

In addition, DNA secondary structures that form on the displaced strand of DNA can also further stabilize the R-loop structures. G-quadruplex (G4) structures are secondary structures that are composed of quartets of guanine bases held together by Hoogsteen base pairing and base stacking (30). DNA G4 structures can form on the displaced strand of DNA in R-loops formed across immunoglobulin loci during CSR (31), thereby stabilizing the R-loop by disfavoring reannealing of the two DNA strands. DNA-binding proteins can also stabilize R-loops by binding the displaced strand of DNA (12, 32).

Recently, researchers have developed a mathematical model to study ideal circumstances for R-loop formation (33). This model suggests that R-loops can absorb negative supercoiling, thereby relieving topological strain on the overall DNA molecule and stabilizing the R-loop structure. Negative supercoiling can also promote R-loop formation (24, 34–38). DNA:RNA hybrids are thermodynamically more stable than dsDNA (39), which may be due to their conformation as an intermediate between the A-form of a dsRNA and the B-form of a dsDNA (40, 41).

Moreover, the presence of nicks downstream of the promoter in the nontemplate strand favors nucleation of an R-loop structure across a given gene (29, 35). During transcription, RNA polymerase transiently unwinds duplex DNA, which reanneals efficiently behind elongating RNA polymerase. The presence of a nick in the nontemplate strand of DNA makes this reannealing less efficient, thus increasing the likelihood of the template strand to hybridize to the nascent RNA (35).

Mechanisms preventing R-loop formation

In addition to factors promoting R-loop formation, there are a variety of processes that inhibit formation of these unique structures, underscoring the need to tightly regulate R-loop abundance in vivo. For example, cells encode helicases and nucleases that remove R-loops once they form. Helicases that resolve DNA:RNA hybrids in vitro include senataxin in humans (42–44), Sen1 in budding yeast (45, 46), which is an RNA-DNA helicase. In S. cerevisiae, Sen1 is a helicase protein that forms part of the NNS complex (Nrd1, Nab3, and Sen1), which is involved in transcription termination (47). S. cerevisiae sen1-1 mutants harboring a temperature-sensitive mutation in the helicase domain exhibit increased levels of R-loop accumulation compared with WT cells, thereby suggesting that Sen1 is involved in resolving R-loops (46). Similar findings have been shown for senataxin, as knockdown of senataxin using RNAi leads to increased R-loop formation in human cell lines (43, 44). Senataxin promotes binding of XRN2, a 5′–3′ exoribonuclease (48), to the nascent RNA of a co-transcriptional R-loop facilitating XRN2-dependent termination (44). Senataxin plays a role in transcriptional termination by resolving co-transcriptional R-loops formed at RNAPII pause sites downstream of the poly(A) site. This allows XRN2 access to the nascent RNA strand. Then XRN2 degrades the 3′ product of poly(A) site cleavage of the nascent RNA strand until it “catches up” with RNAPII and promotes displacement from the template DNA (44, 49). The idea that XRN2 helps resolve R-loops at transcriptional termination sites was further supported with evidence that XRN2 knockdown in human cell lines leads to increased R-loop formation (48).

DDX5, a DEAD-box RNA helicase, has been reported to have a DNA:RNA hybrid unwinding activity (50, 51). A recent study using human cell lines found that DDX5 interacts with XRN2 at transcriptional termination regions through its RGG/RG motif to resolve R-loops (51). Although it is not essential for the DDX5 helicase activity in vitro, methylation of the RGG/RG motif of DDX5 by protein arginine methyltransferase 5 (PRMT5) was required for DDX5 to interact with XRN2 and promote R-loop resolution. Dbp2, the S. cerevisiae ortholog of DDX5, also possesses a robust RNA helicase activity in vitro, However its DNA:RNA hybrid unwinding activity is weaker than pure RNA duplexes (52). Additionally, loss of Dbp2 was reported to be associated with increased R-loop accumulation (53). However, its in vivo role in R-loop resolution is still to be confirmed.

RNases H, a class of ribonucleases that cleave the RNA portion of DNA:RNA hybrids, are also key to removal of R-loops in vivo (54). RNases H are ubiquitous enzymes found in bacteria, archaea, yeasts, higher eukaryotes, and humans (54–57). RNases H may exhibit their activity of degrading the RNA strand of DNA:RNA hybrids in various cellular processes because their substrates could originate from different sources like RNA primers forming during DNA replication, nascent RNA molecule hybridizing to its template strand during transcription, or ribonucleotides misincorporating into DNA molecule during DNA synthesis (54, 56). Loss of RNase H has been found to increase R-loop abundance in living cells (18, 58). In eukaryotes, two types of RNases H, RNase H1 and RNase H2, have been well-characterized (54). However, there are still unanswered questions regarding their biological roles and substrate specificity.

In addition to helicases and nucleases, some factors can prevent R-loop formation by controlling parameters that make R-loop formation more likely. For example, DNA topoisomerases are enzymes that can alter the topology of DNA by adding or removing supercoiling (59). Early experiments in E. coli suggested that bacterial TOPO I reduced R-loop abundance by removing transcription-induced negative supercoiling (37, 60), which has been shown to promote R-loop formation (24, 34). The possibility that topoisomerases decrease the likelihood of R-loop formation is consistent with findings that negative supercoiling promotes R-loop formation. Moreover, a recent study in E. coli has found that type 1A topoisomerases, TOPO I and TOPO III, inhibit R-loop formation to prevent unregulated replication from R-loops (61). Loss of RNase H1 in E. coli causes replication to originate from R-loops instead of the chromosomal origin of replication, oriC. This process is termed “constitutive stable DNA replication” (cSDR) (62, 63). Cells lacking type 1A topoisomerases have both increased accumulation of R-loops and increased cSDR, suggesting a major role of TOPO I and TOPO III in controlling genome integrity by inhibiting R-loop formation or accumulation (61).

R-loops and genomic instability

Aberrant or excessive R-loop formation can lead to genomic instability, a hallmark of cancer. Several nucleases can recognize the junction between the DNA:RNA hybrid and the ssDNA in an R-loop leading to impairment of genome integrity. These include the nucleotide excision repair nucleases XPF-ERCC1 and XPG, which can cleave R-loops that form in the switch regions of immunoglobulin genes during CSR (64–66). Furthermore, various factors can lead to DNA damage by modifying nucleotides in the displaced strand of DNA (67, 68), such as AID. AID is an enzyme found in B cells that is involved in immunoglobulin CSR. AID deaminates cytosine residues in the ssDNA strand of an R-loop, thereby converting them to uracil. These residues are then processed by DNA repair factors to DNA nicks and eventually to DSBs, which allow the CSR to occur (5, 8–10). However, inefficient repair of the DSBs generated in the process of CSR can also lead to genomic instability (69). A recent study using CRISPR/Cas9 technology to create DNA DSBs that mimic AID-dependent DSBs showed that not all AID-dependent DSBs in B cells are efficiently repaired (69). In fact, an inability to repair AID-dependent DSBs can lead to chromosomal translocations, a hallmark of B cell lymphomas and other types of cancer.

Stable R-loops can also lead to genomic instability by interfering with the processivity of the DNA replication machinery. In both E. coli and human cells, R-loop formation leads to replication fork stalling and increased DNA recombination, consistent with genomic instability (70). Resolution of R-loops by RNase H overexpression leads to decreased rearrangement and recombination of DNA, suggesting that the genomic instability seen is directly due to formation of R-loops (70). Moreover, it was found that this R-loop–mediated genomic instability requires the presence of active replication. These observations suggest R-loops compromise genomic integrity by promoting replication fork stalling.

In a recent study, researchers performed experiments in E. coli to study the link between R-loops and DNA damage (71). They proposed a model whereby the RNA portion of an R-loop can serve as a primer for DNA replication. If the ensuing replication fork encounters a nick in the template strand, a dsDNA end will be produced. Various attempts by cells to remedy the issue of a dsDNA end can lead to chromosomal rearrangements, point mutations, and other forms of mutations (71). It is apparent that aberrant R-loop stabilization poses a considerable risk to genome integrity, which may be of specific interest for researchers studying etiology of human diseases.

R-loops as beneficial structures

For quite some time, CSR was the only process where R-loops were believed to be beneficial. This is quite true. However, more recent research has provided evidence of diverse beneficial roles of these nucleic acid structures. For instance, studies have found connections between R-loop formation and chromatin modifications (17, 72, 73) suggesting that R-loops may have a role in regulating gene expression. R-loops are associated with alteration of epigenetic marks, such as DNA methylation and histone modification.

R-loops have been shown to be linked to histone modifications like histone H3 Ser-10 phosphorylation (H3S10P), which is a known mark of chromatin compaction (74). Using budding yeast, this study found R-loop formation at a certain genomic region to be associated with increased H3S10P marks and, consequently, decreased chromatin accessibility at this region. Although this study did not establish the causal relationship between R-loops and chromatin condensation, it showed that overexpression of RNase H led to suppression of H3S10P, which supports a proposed model in which R-loop formation leads to formation of highly condensed chromatin and not the opposite. This study pointed out the potential importance of R-loops in genome function and dynamics. R-loops may also play a role in DNA modification. In one example, the DNA methyltransferase DNMT3B1 binds DNA:RNA hybrids less efficiently than dsDNA (15, 75). Therefore, R-loop formation in promoter regions can promote gene expression by blocking DNA methylation.

Moreover, the types of chromatin interactions tend to be different between R-loops that form in promoter regions and those that form in terminator regions. Promoter region R-loops are enriched with histone marks like H3K4me2, H3K4me3, H3K9ac, and H3K27ac, which are typically associated with increased chromatin accessibility and active transcription (17, 73). Terminator regions where R-loops are likely to form tend to overlap enhancer regions and are frequently associated with increased H3K4me1 and recruitment of the p300 acetyltransferase (17), which are events characteristic of an enhancer state, a chromatin state associated with gene activation (76, 77). However, the mechanism is not clear, as nucleosome formation is likely mutually exclusive with R-loop formation. One possibility is that R-loops and chromatin marks are spatially correlated but are temporally distinct. More work is needed to determine the mechanisms linking chromatin modifications and R-loops.

In addition, published studies from our laboratory provided the first evidence that long noncoding RNAs (lncRNAs) can function as transcriptional inducers via R-loop formation (53). Using the S. cerevisiae model system, those studies established that the GAL gene cluster-associated lncRNAs (GAL lncRNAs) form R-loops across the GAL gene cluster that promote transcriptional induction. It was also shown that regulated formation of these lncRNA:DNA hybrids by the DEAD-box RNA helicase, Dbp2, allows for faster adaptation to carbon source shift. These studies highlight the role of R-loop formation in the regulation of gene expression, where it is critical for cellular fitness and environmental adaptation.

Furthermore, several studies have demonstrated a connection between R-loop formation and DSB repair. There are two main pathways for DSB repair—homologous recombination (HR) and nonhomologous end joining (NHEJ). HR is used to repair DSBs using a homologous template strand as a guide (78). NHEJ is used to repair DSBs without use of template DNA as a guide. Therefore, NHEJ tends to be more error-prone than HR (79). A process known as end resection is a key step in pathway choice for determining how a DSB will be repaired (78). End resection is the generation of 3′ overhangs at DSB sites, thereby preparing the DNA for downstream events in the HR pathway.

In some cases, R-loop formation may play a beneficial role in DSB repair. Perhaps the relationship between R-loops and DNA damage response is more nuanced than once thought. R-loops have been shown to accumulate at sites of DNA damage. A potential role for RNA in DSB repair was demonstrated when two studies showed that both the human and yeast orthologs of the recombinase Rad52 can perform inverse strand exchange using ssRNA, as well as ssDNA. This suggests that RNA may be used in the DSB repair process and that Rad52 has a role in this RNA-templated DNA DSB repair (80, 81). Consistent with this, removal of the RNA portion of DNA:RNA hybrids by RNase H overexpression impairs HR in budding yeast (82). In human cells, removing the RNA portion of DNA:RNA hybrids impairs repair by both HR and NHEJ (83).

Several studies have also suggested that R-loop formation plays a role in end resection. Work in Schizosaccharomyces pombe demonstrated that deletion of RNase H leads to impaired replication protein A (RPA) recruitment (82). RPA recruitment at DSB sites is required for protecting ssDNA overhangs generated by end resection. Overexpression of RNase H was shown to cause excessive strand resection and to increase RPA levels at DSB sites (82). This suggests that R-loops alter the efficiency of end resection. Conversely, it was demonstrated in human cell lines that overexpression of RNase H leads to reduced DNA resection efficiency at break sites (83). These conflicting results may suggest that R-loops play different roles in end resection, depending upon the organism.

There is also evidence that R-loops may recruit factors that participate in DSB repair. Cockayne syndrome protein B (CSB) is likely such a factor (84). CSB participates in DSB repair by recruiting Rad52, which then recruits Rad51, a protein involved in the strand invasion step of homologous recombination (Fig. 4). It has been shown that CSB has a strong affinity for DNA:RNA hybrids. Rad52 itself has an affinity for DNA:RNA hybrids, albeit weaker than that of CSB.

Figure 4.

Proposed model for the role of R-loops in DSB repair. Transcription is induced at DSB sites (82). This increase in transcription leads to an increase in R-loop formation near the DSB. R-loops recruit HR repair factors due to their preferential binding to R-loops over other nucleic acid substrates (73). Whereas CSB is likely one such factor, there are possibly more. Missing from this model is any potential mechanism by which R-loops could influence end resection. A, a DSB is formed in the genome. B, transcription near DSB sites leads to R-loop formation in these regions (C). D, the R-loops recruit various factors involved in DSB repair, such as Rad52 and CSB, which allow for efficient DSB repair (for a review, see Ref. 86).

It is possible that R-loops could also serve as signals for DSBs, especially in actively transcribed regions, because R-loops are more likely to form in these regions (15). Furthermore, DSB induction has been demonstrated to promote increased R-loop formation (82). Perhaps R-loops serve to increase efficiency of DSB repair in a given locus. Although R-loops may signal damage, their resolution is essential for DNA repair because persistent R-loops can lead to additional DNA damage (82).

Obviously, both R-loop formation and resolution are critical for cellular functions. Accordingly, these nucleic acid structures must be tightly regulated in living cells. However, mechanisms controlling R-loop formation and resolution remain one of the major outstanding questions in the field of R-loop biology.

Concluding remarks and future directions

It has now been demonstrated that R-loops can regulate gene expression by recruiting protein factors. We still need to develop a better understanding of what these factors are and the mechanisms by which they are recruited. It has already been demonstrated that R-loops can recruit factors that have an increased affinity for DNA:RNA hybrid structures.

Aside from regulating gene expression, R-loop–mediated recruitment of protein factors could play a role in various other physiological processes. The possibility of a link between R-loop formation and DSB repair is an emerging area of study with exciting new evidence and questions. R-loops that form at sites of DNA damage could serve as a signal for DSBs. Likewise, R-loop–mediated recruitment of DNA repair factors could be used to localize such factors to regions of repair, thereby increasing the efficiency of repair.

There are still many unanswered questions concerning the role of R-loop formation in double-strand break repair. Research covering the subject describes processes such as transcription-associated homologous recombination repair and transcription-coupled homologous recombination (84, 85), but it is unclear whether these processes are independent or in some way linked—perhaps even to the point of being the same phenomenon.

Furthermore, whether DNA DSBs trigger formation of R-loops at the site of lesion or R-loop formation leads to increased DSBs remains a significant question in the field. Whether DNA:RNA hybrid formation at sites of DSBs promotes or hinders DNA repair still needs more investigation. In addition, the effect of R-loop formation on end resection is unclear. There have been conflicting findings that R-loop formation either promotes or hinders end resection. Another possibility is that an intermediate level of R-loop formation is necessary for end resection, and excessive R-loop formation or resolution could both lead to impairment of end resection.

Finally, developing new techniques to accurately map R-loops genome-wide is still the interest of many researchers. The currently available R-loop–mapping methodologies have generated different profiles of R-loop–forming “hot spots,” which encourages the search for new mapping techniques that can validate the published data. Having a reliable technique for a high-resolution mapping of R-loops would allow researchers to study, in more depth, the cellular mechanisms involved in R-loop biology.