In recent years it has been well established that DNA does not always adopt the canonical right-handed B-DNA configuration that is depicted in textbooks (Figure 1A). This DNA structure is based in two linear antiparallel DNA strands that twist together around the same axis forming a double helix that contains a major groove and a minor groove. Albeit this is the form that the most DNA acquires in the cell, there are other non-canonical conformations of DNA. Such structures have been shown to exist in vitro and in vivo, and to be related to many biological processes although their roles remain to be fully characterised. There are many non-canonical DNA structures, and exhaustive recent reviews can be found at (Kaushik et al., 2016; Saini et al., 2013). Among these alternative structures we will focus on G-quadruplexes (G4s, secondary structures arising in repetitive guanine rich areas of either DNA or RNA) (Figure 1B) (Bochman et al., 2012; Kaushik et al., 2016) and R-loops (three-stranded structures that harbour a DNA-RNA hybrid) (Figure 1C).

As previously stated, G4s are non-canonical nucleic acid secondary structures formed in guanine-rich areas. The connection of four guanines by Hoogsteen hydrogen bonding generates a square planar arrangement known as G-quartet (Figure 1B). These planar G-quartets can stack on top of each other generating four-stranded helical structures. G4 structures can adopt a variety of conformations resulting from different arrangements of strand directions. Thus, G4 structures can be intramolecular (formed within one strand) or intermolecular (formed from multiple strands), parallel or antiparallel (Figure 1B) (Spiegel et al., 2020). The human genome contains thousands sequences with the potential to form such structure, known as PQS (Putative Quadruplex Sequence) (Chambers et al., 2015; Hänsel-Hertsch et al., 2016; Zheng et al., 2020). G4s are formed naturally in vivo (Kouzine et al., 2019; Lipps and Rhodes, 2009) but their presence can be stabilized by different drugs, known as G4 ligands (Table 1; more ligands can be found at www.g4ldb.com) (Drygin et al., 2009; Xu et al., 2017). Although G4s are widely distributed across the genome, their appearance seems to be enriched in promoters, telomeres, 5′ UTRs and splicing sites (Chambers et al., 2015). The presence of these structures in such pivotal areas for gene regulation has implicated them in a variety of biological processes.

In many species, telomeric DNA consists of repetitive short G-rich sequences that fold into G4s. These structures have been implicated in the maintenance of telomeres by protecting their degradation and regulating telomere length (Zahler et al., 1991). Indeed, the use of G4 stabilizing compounds inhibits telomerase activity leading to telomere shortening (Sun et al., 1997), although not every G4 ligand is able to inhibit telomerase since this enzyme seems to be able to elongate through parallel G4s, but not antiparallel ones (Moye et al., 2015; Paudel et al., 2020; Zhang et al., 2010). The enriched presence of PQS in the promoters of transcriptionally active genes seems to highlight G4 implication in regulating gene expression (Fleming et al., 2018; Lee et al., 2020). In addition, the use of G4-ligands, such as pyridostatin (PYR) (Table 1), modulates transcription of the BRCA1 gene in neurons (Moruno-Manchon et al., 2017). Furthermore, the deficiency of the helicase RTEL1, involved in G4 dissolution (see below), leads to an altered transcription of genes possessing potential G4 sequences in their promoters (Kotsantis et al., 2020). Also, the presence of PQS have been confirmed genome-wide at mapped replication origins of higher eukaryotes (Besnard et al., 2012; Langley et al., 2016). Notably, G-quadruplexes are of functional importance for replication origin activity since the deletion of a G-rich element known as Origin G-rich Repeated Element (OGRE) strongly reduced origin activity in mouse cells and the introduction in an ectopic origin-free area stimulated the re-establishment of a new functional origin (Prorok et al., 2019). Conversely, G4 formation also show negative effects in the normal cellular metabolism. For example, the presence of these structures might block DNA polymerases progression and, in consequence, collapse the replication fork (Sarkies et al., 2010). Also, as discussed below, G4s are known to promote genomic instability. Due to these negative effects, several G4 ligands (Table 1) are currently in clinical trial for cancer treatment.

G4s are dynamic structures, so cells have evolved different mechanisms to resolve these structures. The main proteins associated are helicases (Mendoza et al., 2016) (see Table 2). Several helicases have been reported to be involved in G4-resolution. The main ones can be divided in several superfamilies (SF), including PIF1, BLM, WRN, RecQ, FANCJ, DNA2 and the aforementioned RTEL1 (Mendoza et al., 2016). Precisely, RTEL1 is important for the maintenance of telomeres integrity helping dismantling telomeric DNA secondary structures to allow efficient telomere replication (Vannier et al., 2012). Although those are the most studied helicases involved in this process, there has been a growth in the number of helicases with a role in G4 processing. The knock out of the helicase DHX36, for example, increases the stress response due to the stabilization of an RNA G4 (Sauer et al., 2019). Indeed, these helicases usually collaborate with other proteins to unwind G4s. That is the case of DDX1 that has been lately reported to be recruited by Timeless to the replication fork to ensure processive replication nearby G4 structures (Lerner et al., 2020).

Another non-canonical secondary structure that can be formed on DNA is the R-loop. R-loops are three-stranded structures in which a DNA-RNA hybrid is formed in a dsDNA context, thus creating a displaced ssDNA region (Figure 1C). They arise usually as a consequence of negative supercoiling of the DNA template behind the transcription complex, especially in highly transcribed regions. As well as G-quadruplexes, R-loops are both implicated in physiological DNA transactions such as class switch recombination of immunoglobulins (Yu et al., 2003), telomere maintenance (Tan et al., 2020a), gene regulation (Sun et al., 2013; Grunseich et al., 2018) or in double-strand breaks (DSBs) repair (Liu et al., 2021) but also are associated with negative effects such an increase of DNA damage.

The most studied pathway in which “scheduled” R-loops have been implicated is in regulating gene expression. The enriched presence of these structures at promoters or termination regions might facilitate the modulation of this process at different steps. First, R-loops have been described to affect chromatin dynamics principally by preventing methylation of CpG islands (Ginno et al., 2012) which would favour transcription. In addition, they can also act in gene silencing since their presence correlates with higher levels of the chromatin condensation mark phosphorylated histone H3 S10 (H3S10P) (Castellano-Pozo et al., 2013). On a second level, R-loops generated from long non-coding RNA can also recruit or displace transcription regulators to the gene promoter modulating gene expression (Boque-Sastre et al., 2015; Arab et al., 2019). Finally, the presence of R-loops at the 3′end of some genes can modulate transcription termination by stalling RNAPII until resolution by RNA-DNA helicases and the nascent RNA is release and degraded (Skourti-Stathaki et al., 2014).

Although R-loops have usually been conceived as drivers of DNA damage, a new associated role in double-strand breaks (DSBs) repair is emerging for these structures. Indeed, the presence of DNA-RNA hybrids in transcriptionally active regions seems to be important for the recruitment of RAD52 and the later RAD52-dependent activity of BRCA1 in antagonizing RIF1-53BP1 blockade of DNA end resection (Yasuhara et al., 2018). The correct processing of R-loops is a key step in determining if their presence may hamper DNA repair or, on the contrary, favour the process. To regulate this, BRCA2 seems to be the responsible protein by recruiting factors involved in R-loop degradation, i. e RNase H2, and, in consequence, ensuring proper repair of the damage by Homologous Recombination (HR) (D’Alessandro et al., 2018).

The above-described beneficial roles of R-loops in DNA metabolism imply that it is not the structure per se but the deregulation of their processing what causes genome instability. Consistent with this idea, the mutation or loss of factors involved in R-loops resolution, such as SETX, RNase H or Fanconi Anemia (FA) factors, leads to persistent R-loops that may result in DNA damage (Wimberly et al., 2013; Herrera-Moyano et al., 2014; Cohen et al., 2018). One of the best-known pathways hampered by non-regulatory R-loops is DNA replication. The “unscheduled” presence of these structures can block replication fork progression which could result in transcription-replication collisions and, eventually, in DNA breakage (Gan et al., 2011). Furthermore, this increased rate of DNA breaks caused by R-loops has been associated with the hyper-recombination phenotype observed in a transcription impaired background that might lead to genomic instability (Huertas and Aguilera, 2003).

The similarity between both structures resides not only in their function but also in their genome distribution. In a recent genome-wide mapping, an enrichment of G4 rich regions was reported in potential R-loop forming regions (Kuznetsov et al., 2018) supporting the possibility that G4s and R-loops may act synergistically in their biological function (Lee et al., 2020; Shrestha et al., 2014), in some cases creating a novel structure that combines both and it is known as a G-loop (Duquette et al., 2004) (Figure 1D). Indeed, in a recent study, it has been observed that the formation of an R-loop at the 5’ UTR of a gene leads to the G4 folding that, in turn, stabilizes the R-loop promoting transcription (Lee et al., 2020). The current view is that the formation of an R-loop facilitates the formation of G4s in the now free ssDNA strand. Then, the formation of the G4 in turn stabilizes the R-loop structure in a positive feedback loop (Lim and Hohng, 2020). Along this lines, it has been shown that stabilization of G4s using ligands increases R-loop formation and promotes R-loop-mediated replication stress (Chappidi et al., 2020; De Magis et al., 2019; Kotsantis et al., 2020). These G-loops have been shown to have different outcomes in transcription levels depending on where is the G4 located, having a positive impact in mRNA production if it is in the non-template strand and a negative one when it is located in the template strand (Lee et al., 2020). Additionally, the possibility of the existence of hybrid G4s (HQs) formed between the nascent RNA and the coding DNA strand (that was first described in the mitochondrial R-loop (Figure 1E; Wanrooij et al., 2012)) might also be taken into account. Indeed, the co-transcriptional formation of those structures has been reported to occur and to have a role in transcription acting as a cis element of control (Zheng et al., 2013). However, and despite the inference that G-loops and HQ structures may exist in vivo, it has been proven technically challenging to demonstrate that they do as opposed to a simply cohabitation of G4s and R-loops in close proximity.

As previously stated, R-loops are well known stimulators of genomic instability, as its presence caused an increase in replication problems and hyper-recombination (further reviewed in (García-Muse and Aguilera, 2019; Rondón and Aguilera, 2019)).

On the other hand, recently it has been described that the use of G4 ligands (Table 1) is linked to an increase in DNA damage due to the induction of replication stress at PQS (Rodriguez et al., 2012). Moreover, an increase in cell cytotoxicity is observed when G4 ligands are used in a cell background where HR-factors are impaired (Zimmer et al., 2016; Xu et al., 2017). Although their evolutionary conservation supports a physiological role of the G4-forming sequences in these DNA processes, G4 structures weaken the genome and render it prone to accumulate DNA damage when they are not efficiently regulated. Indeed, the mapping of DSBs at the genome, using the technology DSB capture, showed an enrichment of these structures nearby the DNA breaking point (Lensing et al., 2016). Also, the association of G4s with DNA damage has been usually linked to their tendency to stall replication forks and cause chromosome breakage when not properly resolved (Paeschke et al., 2011).

Additionally, when the mechanisms responsible for G4 resolution fail to do so, replication forks that encounter these structures may stall or collapse leading to DNA DSBs (Lopes et al., 2011). This situation poses an extra challenge to the cell that might employ the pathways necessary to resolve it. Since the most faithful pathway to deal with this kind of breaks is HR, it is not a surprise that this process is also implicated in G4 induced damage resolution. Notably, the stabilization of G4s in cells deficient for HR factors, such as BRCA1/2 and EXO1, leads to an increased lethality (Zimmer et al., 2016; Xu et al., 2017; Stroik et al., 2020). Furthermore, several proteins with G4 unwinding capacity are recruited to DNA damage and facilitate DNA repair. This includes various helicases involved in G4 resolution such as BLM, WRN (Bernstein et al., 2010; Croteau et al., 2014), FANCJ (Wu et al., 2008; Sarkies et al., 2012) or PIF1 (Bochman et al., 2010; Muñoz-Galván et al., 2017). Indeed, we recently demonstrated that the 5′-3′ helicase PIF1, is required for the correct resection processivity by both unwinding G4 and recruiting BRCA1 to the break (Jimeno et al., 2018).

Finally, the role of G4s in DNA repair might not be limited only to their damage inducing capacity. Indeed, in (De Magis et al., 2020), the authors showed that G4s are also positively implicated in the repair of DNA damage. This observation was supported by the fact that the stabilization of G4s by Zuo1 in Saccharomyces cerevisiae stimulated the recruitment of Nucleotide Excision Repair machinery acting as a loading platform which led to a more efficient repair of the damage.