R-loops are nucleic acid structures composed of a DNA–RNA double strand and a displaced single-stranded (ss) DNA. During transcription, the nascent mRNA transcript can anneal to its template DNA strand, which gives rise to these DNA–RNA hybrids [1]. In addition, R-loops can form through the pairing of non-coding RNAs with chromosomal DNA [1]. While transient by nature, R-loops preferentially form in regions at or in close proximity to transcription initiation (Figure 1A) and termination sites [1]. R-loops mediate transcription termination in gene-dense genomic areas, where the formation of R-loop signals for RNA Polymerase II (Pol II) to pause and disengage ^[2][3][2,3]. Disruption of the RNA splicing machinery can also cause increased R-loop formation, as the unspliced mRNA is retained close to its DNA template [4].

R-loops have regulatory functions at pericentromeric and telomeric regions, which are both transcribed by Pol II [1]. Telomeres are protected by R-loops, which form between telomeric repeats and long non-coding RNAs [5]. More specific roles of genomic R-loops include transcription regulation and immunoglobulin class switch recombination ^[3][6][7][3,6,7]. In addition to the genomic DNA, R-loops are also crucial regulatory sequences in mitochondrial DNA (mtDNA), where they prime mtDNA replication [8]. R-loops are found throughout all domains of life—in bacteria and archaea, their formation is part of an internal defense mechanism against viruses, where CRISPR type I triggers R-loop formation and subsequent DNA degradation ^[9][10][9,10]. Their regulated appearance in the mammalian genome indicates that they were co-opted for regulatory activities, in addition to their formation as a seemingly unavoidable side-product of RNA transcription.

In addition to R-loops generated during Pol II activity during the transcription of mRNA, R-loops also arise from RNA transcription through RNA polymerase I and III (Pol I and Pol III, respectively) ^[11][12][11,12]. While Pol I generates rRNA [13], which form the core of ribosomes, which in turn translate mRNAs into proteins, Pol III generates 5S ribosomal RNA as well as predominantly non-coding RNAs, such as tRNAs, snoRNAs, spliceosomal, and Y RNAs [14]. As many of these non-coding RNAs are heavily transcribed, they are prone to R-loop formation, which has been localized to rRNAs, tRNAs, and retrotransposons ^[11][12][11,12]. Interestingly, R-loops at tRNA genes are more resistant to RNAse H treatment [12] and produce both sense- and anti-sense-paired R-loops, reflective of the tRNA’s cloverleaf fold, which is formed through intramolecular complementary regions [15]. In consequence, these highly transcribed areas are especially vulnerable to DNA damage caused by the prolonged persistence of R-loops.

As collisions between R-loops and the replication fork lead to DNA damage, R-loop formation and resolution must be carefully controlled, particularly during replication in the S phase. These collisions can cause single-strand and double-strand DNA breaks and threaten genomic stability [16]. Factors regulating R-loops include RNase H, which degrades the annealed RNA transcript in R-loops, thereby resolving these structures. RNA-DNA mismatches prevent RNAse H1-dependent R-loop resolution [17], and inactive RNAse H can be used to map R-loops [18]. In addition to specific degradation of the RNA strand, the helicase Senataxin (SETX) unwinds R-loops and enables subsequent cleavage [19]. Other helicases, such as RNA helicase aquarius (AQR) and the DEAD Box helicase also resolve R-loops ^{[1][5][9][17][20][21]}[1,5,9,17,20,21]. Similarly, topoisomerase I resolves supercoils, which prevents the formation of transcription-induced R-loops. These have been extensively reviewed elsewhere ^{[8][16][22][23]}[8,16,22,23]. Epigenetic modifications also influence R-loop formation. Open, more accessible chromatin is more susceptible to DNA:RNA hybridization [24], and m6A modifications of the RNA strand stabilizes R-loops [25].

R-loop perturbations are recorded in several auto- and neuroinflammatory disorders (reviewed in ^[26][27][26,27]). In addition to their contribution to neuropathology, R-loops play a critical role in cancer through their complex interplay with known tumor suppressors and oncogenes, as well as due to their central role in maintaining genome stability. However, an excess of R-loops destabilizes the genome, causing senescence in cancer cells and failures in cell division. Elucidating the molecular details of R-loop mitigation and their dysregulation in cancer sheds new light on key players in tumor suppression and highlights new therapeutic avenues by targeting an unavoidable side-product of mRNA transcription and cell proliferation.

In addition to the aforementioned BRCA1/2 and ATR-mediated DNA damage response pathways, the PARP (poly(ADP-ribose) polymerase) family of proteins are central mediators of the DNA damage response. PARP also responds to single-strand DNA breaks, which can occur at the site of R-loop formation. In consequence, PARP inhibition has been explored as an anticancer strategy to cause synthetic lethality in cells with defective homologous recombination repair.

This is of particular importance in BRCA1/2-driven breast and ovarian cancer cells. The E3 ubiquitin ligase RNF168 is a double-strand break responder that promotes non-homologous end joining of double-strand breaks by recruiting BRCA1 and other repair factors to sites of damage, including R-loops ^[28][88]. There, RNF168 directly ubiquitylates the R-loop helicase DHX9, which causes its recruitment to R-loops and their subsequent resolution ^[29][89]. Decreased expression or loss of RNF168 correlated with a lower incidence of tumors in BRCA1/RNF168 double knockout mice as well as a better survival outcome in patients with homologous repair deficient tumors ^[29][89].

BRD4 inhibition also downregulates Topoisomerase II binding protein 1, a DNA damage response protein, which inhibits activation of the ATR pathway ^[30][90]. As a result, ATR does not induce cell cycle arrest, leading to proliferation despite severe DNA damage and causing replication stress. Cancer cells treated with BRD4 inhibitors subsequently suffer not only damage arising from double-strand breaks and R-loop accrual at a subset of BRD4-controlled genes but also mitotic catastrophe, all leading to cell death ^[30][31][90,91].

Chemosensitivity and cell death through R-loops are especially relevant in cancer cells that already harbor intrinsically elevated R-loop levels. As discussed earlier, Ewing sarcoma cells contain an increased burden of R-loops due to the mechanism underlying their tumorigenicity and are therefore highly sensitive to ATR inhibition ^[32][80]. In addition to ATR inhibitors, Ewing sarcoma cells are quite vulnerable to transcription blockade through Topoisomerase and PARP inhibitors as well ^[32][80].

Elevated R-loop levels also potentiate the cytotoxicity of anticancer drugs in triple-negative breast cancer ^[33][92]. This aggressive breast cancer subtype is associated with poor prognosis and high rates of chemotherapy and radiotherapy resistance ^[34][93]. In a subset of triple-negative breast cancer cells, the double-strand break repair protein MRE11 is mutated ^[35][94]. Physiologically, Mre11 senses transcription-induced double-strand breaks and initiates a DNA damage response to defy genomic instability. However, breast cancer cells that are deficient in Mre11 accumulate R-loops and, in turn, R-loop-dependent DNA damage ^[33][92]. This increases their vulnerability to further DNA damage induced by PARP and ATR inhibitors ^[33][92].

Another malignancy with increased R-loop formation is Embryonal Tumor with Multilayered Rosettes (ETMR), which are aggressive tumors that occur in the brain. Comparison between ETMR and other brain tumors, as well as healthy brain tissue, suggests that mutations induced by R-loops are causative ^[36][95]. Mutational patterns are similar to those found in Ewing Sarcoma, and an increased number of R-loops were found surrounding the most common mutation site ^[36][95]. R-loop sites also coincided with mutation and breakpoint hot spots in these tumors ^[36][95]. The resulting genomic instability renders ETMR sensitive to DNA damaging agents, and administration of PARP and TOP1 inhibitors results in the synergistic killing of ETMR cells resistant to conventional platinum therapy ^[36][95]. Interestingly, the most common amplification and fusion event predisposing patients to ETMR affects a microRNA (miRNA) cluster, suggesting a connection between miRNA processing and R-loop formation ^[36][95]. Drosha, a key enzyme in miRNA processing, has both been previously associated with the formation of R-loops as it stabilizes RNA-DNA hybrids at DNA break sites and recruits repair factors ^[37][96]. In plants, R-loops arise at miRNA loci, initiating co-transcriptional processing of miRNAs, directly linking miRNAs, and stabilizing R-loop formation ^[38][97].

Another approach to induce R-loop-dependent anticancer effects is by impeding the transcription machinery. As described earlier, Pol II stalls on the transcribed sequence, waiting for a transcription elongation signal ^[39][53]. This is signaled by the phosphorylation of a serine in the C-terminal domain of Pol II, allowing Pol II release and transcription elongation ^[40][98]. One of these elongation signals is triggered by Bromodomain-containing protein 4 (BRD4), which directly interacts with cyclin-dependent kinase 9, which in turn phosphorylates Pol II ^[41][99]. Due to its pro-oncogenic role in leukemia, BRD4 inhibitors have been tested and show promising pre-clinical results ^[42][43][44][100,101,102]. As BRD4 inhibitors induce cancer cell death by promoting stalling of Pol II, their mechanism of action also promotes the subsequent annealing of the transcribed pre-mRNA strand to its template, hence forming R-loops ^[31][91]. Accrual of transcriptional R-loops in the S phase results in collisions between the transcription and the replication apparatus and causes double-strand breaks ^[31][91].

JTE-607, a cytokine inhibitor with promising outcomes in the treatment of acute myeloid leukemia and lymphoma ^[45][46][103,104], was recently shown to perturb R-loop homeostasis. In an interesting mode of action, JTE-607 inhibits pre-mRNA release during transcription, leading to elevated R-loop levels ^[47][105]. This halts tumor growth in mouse xenografts and induces apoptosis ^[47][105], suggesting that JTE-607 may be effective in the treatment of tumors with increased R-loop levels.

G quadruplexes (G4) are nucleic acid structures that form primarily in GC-rich sequences. Repetitive G sequences induce the formation of a planar “G-tetrad” that can stack on each other, forming a helical structure. Notably, G quadruplexes can stabilize regulatory R-loops when formed on the displaced ssDNA (reviewed in ^[48][49][106,107]). Small molecules that bind G quadruplexes (G4 binders) have shown great promise in cancer treatment due to their cytotoxicity ^[49][50][107,108]. Monohydrazone-based G4 binders induce cancer cell death by accumulating G4 and R-loops in the genome of cancer cells ^[51][109]. In BRCA2-mutant cancer cells, G4 binders induced R-loop accumulation followed by double-strand break and micronuclei aggregation ^[52][110]. It is worth noting that G4 binders may provide therapeutic benefits to a broad spectrum of malignant diseases due to their general mode of action. An analysis of 22 patient-derived breast cancer xenografts showed that G4-forming sequences are enriched at promoters of highly expressed genes, which leaves highly proliferating tumor cells vulnerable to G4 binders ^[53][111].

Histone deacetylase (HDAC) inhibitors such as romidepsin are clinically approved for the treatment of T cell lymphomas and multiple myelomas but are less efficacious against solid tumors. Romidepsin induces histone hyperacetylation, which leads to more open chromatin at which R-loops accumulate ^[54][113]. This, in turn, threatens the genome integrity, which is rescued by the upregulation of several DNA repair enzymes, including PARP1 ^[54][113]. Notably, administration of the PARP inhibitor Olaparib potentiated R-loop dependent DNA damage leading to increased double-strand break and decreased cell viability ^[54][113]. Therefore, HDAC inhibitor activity could be potentiated in solid tumors by combination with inhibitors of DNA damage, such as PARP inhibitors, to provide synergistic cytotoxic effects through R-loop build-up.

Resistance to anticancer drugs has been described as “molecular chess”, which reflects the evasiveness of cancer cells to both classical chemotherapies (which include broadly DNA damaging agents) as well as newer targeted therapeutics ^[55][114]. It is estimated that out of every 10 cancer deaths, 9 will be attributed to anticancer drug resistance ^[56][57][115,116], highlighting the importance of circumventing anticancer drug resistance. One prominent mechanism of resistance is the upregulation of DNA repair mechanisms—in consequence, inhibition of more than one DNA repair pathway confers synthetic lethality. Inhibiting the resolution of R-loops is therefore an alternative pathway that can promote efficaciousness in known anticancer drugs.

Topoisomerase I TOP1 inhibitors, such as camptothecin, are approved chemotherapeutic agents used for the treatment of solid tumors ^[58][117]. Resistance against TOP1 inhibitors remains a challenge in clinical settings ^[59][60][118,119]. TOP1 aids in the resolution of R-loops, particularly at transcription termination sites ^[61][29]. Hepatoma cells resistant to the TOP1 inhibitor camptothecin showed an upregulation of the DNA repair protein PARP, which initiates a pathway to promote R-loop resolution ^[62][112]. PARP can thereby rescue cells from camptothecin-induced cell death ^[62][112], suggesting a clinical benefit to combination therapy to evade chemoresistance to TOP1 inhibitors.

Fast-growing solid tumors frequently experience a lack of oxygen (hypoxia) due to limited blood supply. Hypoxia is, in turn, associated with chemotherapy and radiotherapy resistance ^[63][120]. Under hypoxic conditions, cancer cells experience an increase in R-loops formation, likely due to transcriptional stress, and upregulate the expression of the R-loop resolving helicase Senataxin ^[64][121]. Interestingly, the expression of SETX was controlled through the unfolded protein response and the main regulator of the cellular integrated stress response, the transcription factor ATF4 ^[64][121]. Knockdown of SETX in hypoxic cells led to the persistence of co-transcriptional R-loops, resulting in lower replication rates and apoptosis ^[64][121]. Selective inhibition of Senataxin in hypoxic cancer cells might therefore provide an effective strategy.

BRCA2 mutated ovarian cancer cells that resist platinum chemotherapy revealed yet another mechanism that involves R-loop interacting proteins. These cells overexpress the microRNA miR-493-5P, which downregulates several R-loop processing genes ^[65][124]. These include an RNAse H, which cleaves the RNA in R-loops directly, and Flap Structure-Specific Endonuclease 1 (FEN1), which cleaves trinucleotide repeats in R-loops, resulting in overall R-loop build-up ^[65][124]. Notably, miR-493-5P also decreased Mre-11 activity, which not only impairs homologous recombination but may further induce R-loop accrual, as discussed earlier in triple-negative breast cancer cells ^[33][92].

RNAse H1 and RNAse H2 function differently from each other. RNAse H1 is the major nuclease to resolve R-loop-induced cell stress regardless of the cell cycle, while RNAse H2 activity has a housekeeping function post replication to avoid the persistence of R-loops ^[66][125]. RNAse H1 does not induce double-stranded nicks, while RNAse H2 cleavage sites necessitate repair ^[66][125]. In consequence, the reduction of RNAse H2 led to cell cycle arrest and double-strand breaks in leukemia cells directly while also sensitizing cancer cells to radiation and other DNA damage-inducing agents ^[67][126].