G-quadruplex structures as regulators of cellular processes and drivers of genome instability in cancer

Abstract

G-quadruplexes (G4s) are essential regulatory structures whose biological functions are inseparable from their potential to destabilize the genome. They play critical roles in transcription, replication and chromatin architecture, yet they also contribute to the genomic instability that fuels cancer. This dual role is an inherent consequence of where G4s form in the genome. G4s are enriched at highly active regulatory regions, including promoters, replication origins and topologically associated domain boundaries, where their controlled formation and resolution by helicases and topoisomerases support normal genome function. When this control is lost, the same features that make G4s functional become harmful, leading to R-loop and G-loop accumulation, replication fork stalling and increased conflicts between transcription and replication. Topoisomerase activity, which normally relieves supercoiling stress, can instead generate the DNA double-strand breaks that characterize cancer genomes. Persistent G4 structures also promote micronuclei formation and cytoplasmic DNA accumulation, activating the cGAS-STING innate immune signaling pathway. Here, we discuss these mechanisms and present pan-cancer genomic analyses showing that these processes operate broadly across human tumors. Therapeutically targeting G4s requires balancing their essential regulatory roles with their pathological effects. Understanding this tension is therefore essential for exploiting G4s as therapeutic targets across cancer types.

Introduction

Genome instability is a hallmark of cancer and aging, and refers to an increased propensity for alterations in cellular genetic material, ranging from single-nucleotide variants to large-scale chromosomal rearrangements, structural abnormalities and aneuploidy (Maslov and Vijg 2009; Hanahan and Weinberg 2011; Hanahan 2022). Under normal conditions, cells employ complex repair and surveillance systems to maintain genome integrity, including DNA damage response signaling pathways, mechanisms that ensure DNA repair and replication fidelity, telomere maintenance pathways, and checkpoint systems that ensure accurate chromosome segregation (Sancar et al. 2004). These mechanisms can induce cell cycle arrest to allow sufficient time for repair; however, when the damage is too extensive, the cell may enter programmed cell death (apoptosis) (Kerr et al. 1972; Maslov and Vijg 2009; Nössing and Ryan 2023). In cancer, genome instability is a fundamental driver of tumorigenesis. The increased frequency and accumulation of somatic mutations, particularly when cells escape apoptosis, promote uncontrolled proliferation, oncogene activation, and tumor suppressor gene inactivation, often through impaired DNA repair or replication stress (Abbas et al. 2013; Tubbs and Nussenzweig 2017; Salmaninejad et al. 2021). Considerable effort has been devoted to identifying and characterizing the sources of genome instability (Aguilera and García-Muse 2013; Tubbs and Nussenzweig 2017). Endogenous contributors often arise from dysregulation of tightly controlled cellular processes, including replication errors, replication stress, DNA repair pathway defects, exposure to reactive oxygen species, spontaneous base deamination, transposable elements, and more recently, the pathological formation of non-canonical or alternative DNA secondary structures (Abbas et al. 2013; Aguilera and García-Muse 2013; Yao and Dai 2014; Mustjoki and Young 2021; Duardo et al. 2023).

G-quadruplexes (G4s) are among the most highly abundant and evolutionarily conserved DNA secondary structures, forming spontaneously from single-stranded DNA in a sequence-dependent manner (Huppert and Balasubramanian 2005; Spiegel et al. 2020; Tu et al. 2021; Matos-Rodrigues et al. 2023; Chantzi et al. 2025). The canonical G4-forming motif, or putative quadruplex sequence (PQS), is defined as G_≥3N_1–7G_≥3N_1–7G_≥3N_1–7G_≥3, in which guanine bases within each G-tract are stabilized by Hoogsteen hydrogen bonding to form a G-quartet, the basic unit of a G4 (Huppert and Balasubramanian 2005; Burge et al. 2006; Puig Lombardi and Londoño-Vallejo 2020). The most stable G4s contain at least three stacked G-quartets with a centrally coordinated monovalent cation (Islam et al. 2020; Price et al. 2020). Experimental methods combining DNA polymerase stalling assays with massively parallel sequencing under G4-promoting conditions have identified more than 700,000 potential G4-forming sequences across the human genome, with a subset validated in vivo using a structure-specific antibody (Biffi et al. 2013; Chambers et al. 2015; Hänsel-Hertsch et al. 2018). Sequencing analyses further show that G4s are enriched in nucleosome-depleted regions associated with active promoters, enhancers and replication origins, supporting a regulatory role rather than serving as a passive structural byproduct (Hänsel-Hertsch et al. 2016; Esnault et al. 2023). G4s are formed in a cell-cycle-dependent manner, and localize to specific genomic loci where they influence transcription initiation, replication timing, RNA processing, and genome stability (Biffi et al. 2013; Guiblet et al. 2021b; Lyu et al. 2022). Their non-canonical topology can modulate local chromatin accessibility, stabilize transcription-associated DNA:RNA hybrids called R-loops, and recruit specific DNA-binding proteins (Hänsel-Hertsch et al. 2016; Reina and Cavalieri 2020; Lyu et al. 2022; Wulfridge and Sarma 2024).

In this review, we examine the multifaceted roles of G4s in genome regulation, emphasizing both their physiological and pathological functions. We summarize recent advances demonstrating how tightly regulated G4 formation supports normal cellular processes, including gene expression, DNA replication, DNA repair and activation of immune signaling pathways. We then highlight studies demonstrating that aberrant G4 accumulation or dysregulation contributes to increased mutational burden, genome instability and cancer development (Figure 1). By integrating these perspectives, this review underscores the dual nature of G4s as essential regulatory elements in normal biology and as potential drivers of cancer when dysregulated. Although RNA G4s also play significant roles in pre-mRNA maturation, translation and miRNA targeting and regulation (Dumas et al. 2021; Ruzi et al. 2025), this review primarily focuses on DNA G4s.

Figure 1.

Dual role of DNA G-quadruplex structures in genome regulation and tumorigenesis. Under normal conditions, G4 structures form transiently at guanine-rich genomic regions to modulate key processes, including transcription, DNA replication, and higher-order chromatin organization (left panel). Disruption of G4 homeostasis can promote aberrant oncogene activation, replication fork collapse, genome instability, and mutagenesis, ultimately driving tumor development (right panel).

DNA G4s as essential regulatory elements

Factors regulating G4 stability

DNA topology

G4 formation is a dynamic and spontaneous process modulated by multiple factors, including DNA sequence context, cellular G4 helicases, G4-binding proteins and endogenous G4-stabilizing ligands. The DNA sequence influences G4 formation by affecting the number, length and arrangement of guanine tracts and loops, which in turn determine the DNA’s folding topology and structural stability (Tippana et al. 2014; Monsen et al. 2022). Due to sequence variation, G4s can adopt a wide range of conformations and topologies, from unimolecular to tetramolecular assemblies, with intramolecular forms being the most biologically relevant in single-copy loci due to being kinetically favored under physiological conditions (Farag and Mouawad 2024; Jurkowski et al. 2025). The topology of a G4 is determined by the strand orientation (parallel, antiparallel, or hybrid), loop length and type, and the glycosidic bond angle (Farag and Mouawad 2024; Wen et al. 2025). G4 folding occurs spontaneously but is highly sensitive to physiological conditions, such as ionic environment and molecular crowding (Tan et al. 2020). Molecular crowding can induce structural transitions from antiparallel to parallel conformations, and recent studies suggest that G4 formation may be more favored under highly crowded conditions such as the mitochondrial environment, than non-crowding conditions (Miyoshi et al. 2002; Liu et al. 2025). Likewise, nuclear DNA G4 formation on chromatin undergoes molecular crowding that can trigger phase separation, a process in which molecules condense into a liquid-like droplet with unique biochemical properties from the surrounding cellular space (André and Spruijt 2020). Phase-separated condensates stabilize G4 structures, and the inhibition of phase separation destabilizes G4s across the genome (Gao et al. 2023). The stability of G4 structures is further influenced by centrally located monovalent cations within the stacked G-quartets, with K⁺ providing the greatest stabilization and Li⁺ the least (Zaccaria et al., 2016; Cucchiarini et al. 2025).

G4 helicases

The stability of G4s can also be regulated by specialized ATP-dependent helicases, including DHX36 (RHAU/G4R1), FANCJ, BLM, WRN, RECQ1 and PIF1, which maintain cellular homeostasis by selectively recognizing and resolving G4s (Mendoza et al. 2016). These helicases exhibit high affinity and specificity for G4s, often mediated by unique domains and motifs, like the N-terminal DHX36-specific motif (DSM) in DHX36 or the AKKQ amino acid motif in FANCJ, enabling them to distinguish G4s from other nucleic acid structures and bind with exceptionally low K_d values (Creacy et al. 2008; Chen et al. 2018; Lowran et al. 2019; Banco et al. 2025; Campbell et al. 2025). By unwinding G4s, these helicases prevent stalling of RNA polymerase II (RNA Pol II) during transcription and facilitate the progression of the replication fork. The high affinity and selectivity of these helicases are essential for the dynamic regulation of G4 stability, ensuring faithful DNA replication and accurate transcription.

Small molecules

Given the enrichment of G4-forming sequences in promoters of oncogenes and other cancer-associated loci, substantial effort has been devoted to develop small molecules that target G4s and alter their stability to reduce the proliferation of cancer cells (Bidzinska et al. 2013; Chen et al. 2022). Such compounds, including telomestatin, pyridostatin (PDS) and CX-5461, bind G4 structures and increase their stability and persistence within cells (Masud et al. 2021; Seimiya et al. 2021). As a result, they can impair transcription initiation, alter chromatin accessibility and promote DNA damage, particularly at telomeric and ribosomal DNA (rDNA) regions (Rodriguez et al. 2012; Karna et al. 2024; Waisertreiger et al. 2025). One such compound, CX-5461 has advanced into clinical trials for various diseases (H.-X. Li et al. 2025). It targets G4s in rDNA and promoter regions, in addition to inhibiting RNA polymerase I and topoisomerase II (TOP2), leading to replication stress, especially in BRCA1/2-deficient tumors (Bruno et al. 2020; Teng et al. 2021; Hilton et al. 2022; H.-X. Li et al. 2025). Small molecules that destabilize G4s represent an emerging complementary approach to modulating gene expression by disrupting rather than stabilizing G4 structures (Waller et al. 2009; Mitteaux et al. 2021, 2025; Tateishi-Karimata and Sugimoto 2024), particularly beneficial in G4-helicase-deficient conditions. PhpC compounds have been shown to selectively unfold G4 structures, leading to an increase in the qPCR stop assay output that relies on Taq DNA polymerase amplification of the DNA strand with a central G4-forming sequence acting as a roadblock (Mitteaux et al. 2021). In patient-derived primary astrocytes, PhpC treatment upregulated genes and proteins involved in RNA transactions, mirroring the inverse effect of PDS on G4 stabilization (Mitteaux et al. 2025). Bidirectional chemical modulation of G4 dynamics is emerging as a powerful tool to dissect G4 biology and offers therapeutic opportunities in the context where pathological G4 stabilization or destabilization leads to aberrant gene silencing or activation.

Genomic regions enriched in G4s

G4 motifs are not randomly distributed throughout the genome but are preferentially positioned at genomic elements critical for cell regulation and function (Hänsel-Hertsch et al. 2016; Guiblet et al. 2021b). Multiple genome-wide studies have shown enrichment of G4s at promoters, CpG islands, untranslated regions (UTRs), replication origins, enhancers and topologically associating domain (TAD) boundaries (Hänsel-Hertsch et al. 2016; Guibletet al. 2021b; Lyu et al. 2022; S. Li et al. 2025). Over 40% of human gene promoters contain at least one G4 motif, typically located at or near transcription start sites (TSSs) and in regions of high transcriptional activity (Huppert and Balasubramanian 2007). G4s frequently colocalize with CpG islands, especially within 5’ UTRs, and correlate with open hypomethylated chromatin when located within 150 bp of CpG sites (Chambers et al. 2015; Mao et al. 2018; Jara-Espejo and Line 2020). G4s also mark replication origins, interact with Rif1 to regulate replication timing, and are enriched at sites bearing active histone modifications (e.g. H3K4me3, H3K27ac) and CTCF-bound TAD boundaries, linking them to higher-order chromatin architecture (Hou et al. 2019; Masai et al. 2019; Masai and Tanaka 2020; Guiblet et al. 2021b). Well-characterized examples of genomic G4s include human telomeres, where the repetitive (TTAGGG)_n sequence forms stable G4 structures, and rDNA loci, where G4s adopt parallel conformations (Hänsel-Hertsch et al. 2017; Datta et al. 2021; Chen et al. 2022; Green et al. 2022; Dell’Oca et al. 2024). Collectively, these observations support a model in which G4s function as dynamic regulatory elements across the genome, shaping transcription, replication, chromatin organization and epigenetic regulation.

G4s in regulating transcription

G4s play a significant role in transcription regulation, consistent with the enrichment of G4 motifs in nuclear speckles (Huppert and Balasubramanian 2007; Komůrková et al. 2021; Esnault et al. 2025). At promoters, G4s regulate transcription at multiple levels by influencing nucleosome positioning, transcription initiation and RNA Pol II progression along the DNA templates. Mechanistically, G4 folding and nucleosome assembly are structurally incompatible: histone octamer wrapping requires ~147 bp of flexible duplex DNA capable of adopting a tight superhelical curvature (Luger et al. 1997), whereas a folded G4 is a rigid, non-canonical single-stranded structure that cannot accommodate this bend. By promoting nucleosome exclusion, G4s generate nucleosome-free regions that are more accessible to the transcriptional machinery (Figure 2A), and genome-wide studies confirm that G4-induced chromatin opening correlates with higher transcriptional output (Du et al. 2008; Hänsel-Hertsch et al. 2016; Lago et al. 2021; Esnault et al. 2023; 2025). G4s can also act as transcriptional repressors, with the effects depending on their strand context. When formed on the template strand, a G4 can inhibit transcription by impeding RNA Pol II progression during elongation (Figure 2B). In contrast, G4s on the non-template strand can couple with R-loops to form G-loop structures that further stabilize these intermediates and modulate the transcription (Figure 2C) (Lee et al. 2020; Lim and Hohng 2020; Wulfridge et al. 2023). The transcriptional output then depends on the type and loop length of G4 structure formed: intramolecular DNA G4s with short loops within R-loops inhibit transcription, whereas intermolecular DNA:RNA hybrid G4s with long loops enhance transcription by promoting successful rounds of R-loop formation (Lee et al. 2020; Yang et al. 2025). G4s additionally cooperate with transcription factors (TFs), such as SP1 and AP-1, to enhance cell-type-specific gene activation (Lago et al. 2021). G4s are strongly associated with RNA Pol II promoter-proximal pausing, functioning as cis-regulatory elements that modulate the duration and release of Pol II pausing, thereby fine-tuning gene expression (Eddy et al. 2011; Esnault et al. 2025). Additionally, while phase separation can stabilize G4s, G4s can also facilitate condensate formation to regulate transcription. For example, the CCND1 promoter’s G4 binds the MAZ TF, promoting phase separation to compartmentalize coactivators that are necessary for CCND1 transcription (Wang et al. 2024). Beyond promoters, G4s mark active enhancers and help distinguish active from poised regulatory states, influencing enhancer-promoter interaction and gene activation (Hänsel-Hertsch et al. 2016; Lyu et al. 2022; DeMeis et al. 2025; Doyle et al. 2025; S. Li et al. 2025). G4s also participate in long-range chromatin looping at TAD boundaries, contributing to 3D genome organization (Figure 2D) (Hou et al. 2019; Reina and Cavalieri 2020; Yuan et al. 2023; Raimer Young et al. 2024; Roy et al. 2024). Collectively, these activities position G4s as integrators of epigenetic, transcriptional and structural signals essential for cellular homeostasis.

Figure 2.

Biological functions of DNA G-quadruplexes. (A) G4 formation upstream of the transcription start site (TSS) recruits transcription factors (TF) that activate or repress transcription and promotes nucleosome exclusion, resulting in open chromatin and increased recruitment of transcription machinery. (B) During transcription, G4 formation on the template strand can directly impede RNA polymerase II (RNAP II) progression. (C) G4s formed on the non-template strand can associate with R-loops generated during transcription to create G-loop structures and recruit G4 helicases. (D) G4s can bridge enhancer and promoter elements, influencing 3D chromatin architecture and topologically associating domains through recruitment of CTCF, thereby regulating transcription. (E) G4s contribute to DNA replication initiation by recruiting the origin recognition complex (ORC). (F) G4s can stall DNA replication by blocking DNA polymerase progression.

The interplay between DNA supercoiling and G4s

Many biological processes in eukaryotic cells involving motor proteins, such as polymerases and helicases, produce torsional stress as they translocate along DNA. Movement of these complexes forces the DNA helix to rotate around its axis, generating positive supercoiling (over-winding) ahead of and negative supercoiling (under-winding) behind the moving machinery. Consistent with the twin-supercoiled-domain theory in transcription (Liu and Wang 1987; Giaever and Wang 1988), supercoiling accumulates locally at TSSs of highly expressed genes as well as at TAD boundaries, reflecting the processive activity of transcription machinery and cohesin complexes (Racko et al. 2019; Achar et al. 2020; Gittens et al. 2024; Hall et al. 2025; Yao et al. 2025). Local negative supercoiling in regulatory regions is crucial for gene expression and DNA replication. The underwound state weakens the hydrogen bonds between base pairs in canonical B-form DNA, increasing accessibility to DNA-binding proteins and facilitating the formation of alternative secondary structures. Studies have demonstrated that transcription-generated negative supercoiling behind RNA Pol II promotes the formation of G4 structures from putative quadruplex sequences, which are frequently enriched in these regulatory regions (Zhang et al. 2013; Selvam et al. 2014; Zheng et al. 2017; Li et al. 2019; Hwang et al. 2025). This indicates that negative supercoiling plays a role in G4 stability in vivo beyond DNA sequence composition alone. The stabilization of G4 structures at highly active genes can create a positive feedback loop: RNA Pol II-generated negative supercoiling promotes G4 folding, which in turn excludes nucleosomes and maintains an open promoter state, facilitating recruitment of the transcription machinery and additional rounds of Pol II initiation. This self-enforcing loop potentially explains the enrichment of G4-forming sequences often found at the promoters of highly expressed genes and oncogenes.

G4s in regulating replication

DNA G4s form in a cell-cycle-dependent manner, with higher levels of G4s detected by immunofluorescence during S phase, correlating with active DNA replication (Biffi et al. 2013). Live-cell imaging studies further demonstrate that G4s dynamically transition between folded and unfolded states, with peak abundance occurring during S phase (Zhang et al. 2018; Di Antonio et al. 2020). G4s have also been implicated in replication initiation (Besnard et al. 2012). Prorok et al. demonstrated that replication origins are preferentially associated with G-rich repetitive elements capable of forming G4 structures (Figure 2E) (Prorok et al. 2019). Treatment with the G4-stabilizing ligand PhenDC3 generated new G4-associated replication origins while suppressing non-G4 origins, supporting a functional role for G4s in replication initiation (Prorok et al. 2019). Beyond initiation, G4s influence replication fork progression. During unwinding of the parental DNA duplex, a subset of active replication forks can spontaneously form G4 structures on the newly exposed single-stranded DNA behind the helicase (Lee et al. 2021). Replication-coupled G4 formation on the leading strand template can stall the CMG helicase complex (Cdc45-MCM2–7-GINS) (Figure 2F) (Sato et al. 2021; Batra et al. 2025). Resolution of these structures requires coordinated action of specialized helicases, including DHX36 and FANCJ, which function sequentially to bypass and unwind G4s at the replication fork (Sato et al. 2021). Overall, these findings indicate that G4s can both promote origin firing and, if unresolved by specialized helicases, impede replication fork progression, linking G4 dynamics to genome maintenance and the prevention of instability (Lee et al. 2021; Guiblet et al. 2021b; Sato and Knipscheer 2023).

Mechanisms of G4-induced genome instability

Transcription- and replication-associated genome instability at G4s

Multiple studies have examined the transcription-associated DNA damage linked to G4s (De Magis et al. 2019; Komůrková et al. 2021; Bartosik et al. 2025; Sato et al. 2025). Immunofluorescence staining showed strong colocalization of G4s with transcription factories marked by RNA Pol II phosphorylated at Ser5, as well as an association between G4s and the DNA damage marker γH2AX following γ-irradiation (Komůrková et al. 2021). Notably, γ-irradiation did not affect colocalization between G4s/γH2AX and 53BP1, or MDC1, suggesting that G4-associated DNA damage is not solely a consequence of exogenous genotoxic stress (Komůrková et al. 2021). During transcription, unwinding of the DNA duplex generates a transient single-stranded DNA (ssDNA) region that favors R-loop formation, in which nascent mRNA hybridizes with the template DNA strand, while leaving the non-template strand exposed. When this exposed strand contains a G4-forming sequence, it becomes prone to G4 folding (Figure 2C) (Lee et al. 2020; Lim and Hohng 2020; Varshney et al. 2020; Yang et al. 2025). R-loop formation helps to stabilize G4 structures, and reciprocally, G4s further promote stabilization and accumulation of R-loops, creating hybrid G-loop intermediates that are established as drivers of genome instability (De Magis et al. 2019). Moreover, PDS-induced DNA damage, measured by γH2AX immunostaining, occurs through an R-loop-dependent mechanism and is significantly reduced by overexpression of RNase H1, which degrades the RNA component of R-loops (De Magis et al. 2019). Genome-wide DNA break mapping in DHX36-depleted cells showed enrichment of double-strand breaks (DSBs) at G4s and TSSs (Bartosik et al. 2025). Highly expressed genes and histone modifications like H3K27ac and H3K4me3, linked to active enhancers and promoters, show significantly more DSBs upon DHX36 loss, underscoring G4 structures as hotspots of transcription-associated fragility (Bartosik et al. 2025). Several studies have shown that G4s are enriched at RNA Pol II promoter-proximal pausing sites, where they function as regulatory elements (Eddy et al. 2011; Szlachta et al. 2018; Esnault et al. 2025). Importantly, these RNA Pol II promoter-proximal pausing sites are enriched for DNA DSBs and activate DNA damage signaling pathways with evidence linking topoisomerase activity to DSB generation at these sites (Bunch et al. 2015; Szlachta et al. 2018; Singh et al. 2020). The frequency of DSBs at pausing sites correlates with pausing strength, independent of distance from TSS (Singh et al. 2020). In addition, stabilization of the G4 structure through loss of DHX36 helicase leads to an increase in the level of DSBs at those pausing sites (Bartosik et al. 2025). Cells tolerate increased fragility as an unavoidable consequence of maintaining high transcriptional output, which reflects a fundamental paradox: transcription itself is both essential for cellular functions and, at the same time, a major source of DNA double-strand breaks. Thus, cells evolved to harness controlled DNA breakage as a regular mechanism of gene expression, allowing for local genomic instability as the cost of transcriptional plasticity. Such vulnerability becomes particularly consequential when G4-mediated replication stress converges with transcription-associated DNA fragility at promoter regions of oncogenes.

CTCF (CCCTC-binding factor) can function as a TFor a chromatin insulator, anchoring the high-order chromatin interactions and blocking the communication between enhancers and promoters (Cuddapah et al. 2009; Phillips and Corces 2009; Dixon et al. 2012; Hanssen et al. 2017). CTCF is a haploinsufficient tumor suppressor that is frequently altered in cancer through multiple mechanisms, including mutations, allelic loss, or epigenetic silencing (Kemp et al. 2014; Marshall et al. 2017). Analysis of human breast tumors in the Cancer Genome Atlas database showed that 60% of breast cancers exhibit CTCF copy number loss (Damaschke et al. 2020). CTCF binds to specific DNA motifs and functions as an insulator at TAD boundaries, preventing cohesin complex sliding. Lebeau et al. demonstrated that a 50–60% reduction in CTCF protein expression causes many sites to lose CTCF binding, but sites located at TAD boundaries can maintain their insulating ability (Lebeau et al. 2022). We have identified these CTCF-lost sites as being over-represented with G4s (Raimer Young et al. 2024). Furthermore, G4s associated with R-loops are known to promote CTCF binding to DNA (Wulfridge et al. 2023). The role of G4s in maintaining TADs under limited CTCF binding is attractive for several reasons: (i) G4s are enriched at TAD boundaries (Hou et al. 2019; Wulfridge et al. 2023; Raimer Young et al. 2024), which may explain why many TAD boundaries lack clear CTCF consensus sequences (Fang et al. 2020); (ii) The strand orientation of these G4s displays a convergent configuration, which can accurately establish loops (Hou et al. 2019); (iii) TAD boundaries are enriched with negative supercoiling and G4s can assist in resolving these topological stresses; (iv) Many transcription regulators that bind to G4s (L. Li et al. 2021; Su et al. 2021; Pipier et al. 2021; Zhang et al. 2021) could significantly impact TAD formation. Previously, we demonstrated that under conditions of abundant CTCF, CTCF binding strength defines regions of genome fragility facilitated by TOP2 (Szlachta et al. 2020; Raimer Young et al. 2024). To investigate a potential role for G4s as backup insulators at TAD boundaries under a limited CTCF environment, we knocked down CTCF protein (~50%) in the nonmalignant breast cell line, MCF10A, mimicking the CTCF copy number loss observed in breast cancer. We found that TAD boundary sites that lost CTCF exhibited significantly more DSBs compared to the lost sites within the TAD loops. These CTCF-lost TAD boundaries were enriched for TOP2B, and importantly, overrepresented with G4s (Raimer Young et al. 2024). Furthermore, knocking out DHX36, a major G4 resolvase, increased DSBs at these sites (unpublished data). Based on these observations, we propose a mechanistic model for DNA fragility at CTCF binding sites, in which, under CTCF-limited conditions, alternative DNA secondary structures, such as G4s, generated from DNA unwinding by topological stresses, can act as backup insulating structures that maintain TAD boundaries and facilitate TOP2 recruitment. Following TOP2 recruitment, G4-mediated TADs are susceptible to DNA fragility (Figure 3). Our study provides insight into how reduced CTCF protein levels contribute to genomic instability and oncogenic processes through the generation of TOP2-mediated DNA breaks.

Figure 3.

G4s as backup insulating elements at topologically associating domains. In a normal environment, CTCF, together with the cohesin complex and TOP2, organizes chromatin loops that define topologically associating domains (TADs) and regulate chromatin architecture and gene expression (top panel). Under conditions of limited CTCF, increased topological stress promotes local DNA unwinding, facilitating G4 structure formation. These G4s, as alternative insulating elements, can rescue TAD insulation, albeit at the cost of elevated TOP2 activity and increased levels of DNA breaks (DSBs) at these sites (bottom panel).

During DNA replication, G4s can form at the replication fork within the single-stranded region generated immediately behind the MCM helicase and before nascent DNA synthesis begins (Lee et al. 2021; Batra et al. 2025). Using purified human and yeast replisomes, Batra et al. demonstrated that a single G4 on the leading strand template is sufficient to arrest the CMG helicase, causing replication fork stalling in vitro (Batra et al. 2025). When not efficiently resolved by specialized helicases such as DHX36 and FANCJ, G4-induced fork stalling can trigger replication stress and compromise fork restart, posing a significant threat to genome stability (Sato et al. 2021). The effects of G4 structures on replication are strongly strand-dependent. G4 formation on the leading-strand template preferentially destabilizes fork progression and produces distinct polymerase inhibition patterns, whereas G4s at lagging-strand templates do not effectively stall CMG helicase activity (Lopes et al. 2011; Sato et al. 2021; Batra et al. 2025; Hile et al. 2025). In addition to helicase arrest, G4s locally perturb replisome organization by limiting replication protein A (RPA) binding to the exposed ssDNA and interfering with coordinated DNA synthesis, thereby attenuating normal checkpoint signaling (Lee et al. 2021). Failure to resolve G4 structures can lead to delayed fork processing and accumulation of DNA double-strand breaks (Pepe et al. 2025; Sethi et al. 2025). Persistent G4-associated replication stress promotes the formation of ssDNA gaps that may subsequently convert into DSBs, activating error-prone DNA repair pathways leading to microhomology-mediated recombination and chromosomal instability (Pepe et al. 2025; Sethi et al. 2025). The cellular response to G4-induced replication stress is context-dependent: replication fork stalling at sites with formed G-loops and high transcription is prominent during early S phase, whereas the translesion synthesis-driven bypass mechanism is preferentially engaged in late S phase to avoid immediate DNA break formation (Pepe et al. 2025). Genome instability is further exacerbated when replication forks encounter the transcriptional machinery complex. The severity of transcription-replication conflicts depends on collision orientation, with head-on encounters generating greater mechanical stress and replication fork stalling than co-directional encounters (Lang and Merrikh 2021). At G4 sites, these collisions are particularly disruptive because transcription-induced G4 structures can directly block the advancing replication fork. Resolution requires coordinated action of helicases, such as RTEL1 and FANCJ, along with topoisomerases and other factors, increasing the likelihood of DSBs and chromosomal instability when such resolution fails (Kumar et al. 2021; Sato et al. 2021).

Topoisomerase and supercoiling mediated genome instability at G4s

Topoisomerase 1 (TOP1) and TOP2 resolve excess negative and positive supercoiling in the nucleus (McClendon et al. 2005; Salceda et al. 2006; Pommier et al. 2016; 2022) through binding DNA substrates, generating transient single-strand or double-strand breaks, respectively, and creating cleavage complexes. In the TOP1 cleavage complex, torsional strain is relieved by controlled rotation of DNA prior to religation (Koster et al. 2005). In contrast, TOP2 functions primarily as a homodimer, binding two DNA helices, introducing a DSB in one segment, and transferring the second DNA segment through the break to resolve topological stress (Pommier et al. 2016; 2022). The binding and cleavage activity of TOP1 and TOP2 are only weakly sequence-dependent (Spitzner and Muller 1988; Porter and Champoux 1989; Capranico et al. 1990; Jaxel et al. 1991; Mueller-Planitz and Herschlag 2007). Pommier and colleagues proposed that non-B form secondary structures, including G4s, may inhibit topoisomerase activity to preserve biologically required negative supercoiling (Pommier et al. 2016). Consistent with this model, multiple studies show that TOP1 binds G4 structures with high affinity and is catalytically inhibited by them (Marchand et al. 2002; Shuai et al. 2010; Liang et al. 2024). However, the impact of TOP1 on G4 formation and stability is less clear. While studies have shown that TOP1 promotes G4 formation and stability (Marchand et al. 2002; Keller et al. 2022), others revealed that TOP1 depletion increases G4 formation, likely due to excess negative supercoiling (Bossaert et al. 2021).

Although TOP1 relaxation is strongly inhibited by G4s, TOP2 appears capable of cleaving DNA at or near G4 structures. In vitro experiments by Chung et al. (Chung et al. 1992) demonstrated that TOP2 cleaves a G-rich oligonucleotide preferentially when it adopts a four-stranded parallel DNA structure. We previously demonstrated that TOP2-mediated DNA breaks are enriched at regulatory regions containing alternative DNA secondary structures, especially G4s (Szlachta et al. 2020; Raimer Young et al. 2024). We also revealed that both TOP1 and TOP2 are actively recruited to RNA Pol II promoter-proximal pausing sites, where they contribute to DSB generation (Singh et al. 2020), and treatment with topoisomerase poisons, camptothecin or etoposide, results in dose-dependent increases in DSB formation at these sites (Singh et al. 2020). More recently, we showed that loss of G4-resolving helicase DHX36 leads to increased DSBs at G4-containing loci. Treatment of DHX36 knockout cells with the TOP2 catalytic inhibitor, merbarone, reduced γH2AX levels, indicating that DNA damage caused by G4 stabilization is at least partly dependent on TOP2 activity (Bartosik et al. 2025). Consistent with these findings, recent studies suggest that TOP2 contributes to genomic instability caused by G4 stabilizing ligands, such as CX-5461 and PDS (Bruno et al. 2020; Olivieri et al. 2020; Bossaert et al. 2021; Pan et al. 2021). Analysis of drug screening datasets reveals that the primary cytotoxicity mechanism of CX-5461 resembles that of TOP2 poisons. It stabilizes and traps the TOP2 cleavage complexes on DNA, leading to γH2AX accumulation. Moreover, a CRISPR-Cas9 screen against 27 genotoxic agents revealed that PDS cytotoxicity depends on a gene set similar to that required for cellular response to TOP2 poisons, suggesting that PDS promotes TOP2 trapping following G4 stabilization (Olivieri et al. 2020). Overexpression of TOP2A or loss of the repair enzyme TDP2, which resolves trapped TOP2 cleavage complexes, hypersensitized cells to both PDS and CX-5461 (Olivieri et al. 2020). Bossaert et al. further showed that siRNA-mediated depletion of TOP2 reduces γH2AX accumulation following PDS and CX-5461 treatment, reinforcing a role for TOP2 in G4 ligand-induced DNA damage (Bossaert et al. 2021). Pang and colleagues reported that CX-5461 increases TOP2 cleavage complex formation and TOP2B and the chromatin remodeler ATRX cooperatively resolve G4s (Pang et al. 2025). Taken together, these findings indicate that the genomic instability caused by PDS and CX-5461 may arise from a combination of G4 stabilization and direct modulation of TOP2 activity. The relative contributions of these mechanisms remain unresolved and warrant further investigation (Figure 4).

Figure 4.

DNA fragility at G4s due to TOP2 activity. Excess negative supercoiling at promoter regions of highly expressed genes can promote G4 formation, which is further stabilized by small molecules such as PDS and/or CX-5461, as well as by loss of DHX36. In this model, G4 formation can serve as a structural signal for TOP2 to recognize, cleave, and relieve supercoiling stress, thereby facilitating transcription progression. However, TOP2 activity at G4 sites also increases the formation of DNA double-strand breaks (DSBs), representing a tradeoff required to sustain high transcriptional output.

Moreover, topoisomerase 3B (TOP3B), which preferentially binds single-stranded DNA substrates, has recently been shown to act at R-loops (Saha et al. 2022). Because R-loop formation can stabilize G4s on the opposite strand, further investigation is warranted to clarify the respective roles of TOP2 and TOP3B in relaxing negative supercoiling at G4-rich regulatory sites. Differential interactions between G4 structures and TOP1, TOP2 and TOP3B can explain how cells maintain localized negative supercoiling at regulatory loci without allowing torsional stress to reach pathologic levels.

G4-induced DNA damage response and therapeutic implications

G4s are well-established activators of the DNA damage response (DDR) pathways. By promoting replication fork stalling and DNA double-strand break formation during DNA replication and transcription, G4s trigger recruitment and activation of canonical DDR kinases, such as ATM, ATR and DNA-PK (De Magis et al. 2019; Lee et al. 2021; Linke et al. 2021; Masud et al. 2021; Rider et al. 2022; Liu et al. 2023; Pang et al. 2025). Stabilization of G4s, either through endogenous factors or exogenous small molecules, increases DSB accumulation and the activation of DDR markers, including γH2AX and 53BP1, along with ubiquitin-mediated signaling factors, RNF168 and UBE2N (Masud et al. 2021; Liu et al. 2023). Small molecules that stabilize G4s can induce synthetic lethality in cells deficient in homologous recombination or ubiquitin-mediated DDR pathways, highlighting a functional dependence on intact repair mechanisms (Masud et al. 2021; Liu et al. 2023; Yan et al. 2025). Repair of G4-induced lesions involves homologous recombination and fork resolution proteins, such as RAD51, FANCJ, XPF and MUS81 (Lee et al. 2021; Rider et al. 2022; Liu et al. 2023). In cancer cells, ligand-stabilized G4s can further promote R-loop formation, exacerbating the accumulation of DSBs, genome instability and sustained DDR activation (De Magis et al. 2019). In addition to break-driven signaling, G4 structures may directly interface with DNA repair pathways. PARP-1, a key DNA damage sensor and poly(ADP-ribose) polymerase, binds several parallel G4 conformations, and a subset of these interactions stimulates PARP-1 enzymatic activity, connecting G4s to poly(ADP-ribose)-mediated DNA repair signaling (Edwards et al. 2021). The dynamic balance between G4 formation, stabilization and resolution is therefore critical for maintaining genome integrity. Disruption of this regulatory network sensitizes cells to G4-stabilizing ligands, underscoring their therapeutic potential in cancers characterized by DNA repair deficiencies (Masud et al. 2021; Liu et al. 2023; Yan et al. 2025).

A recent study showed that prolonged pharmacological G4 stabilization dramatically alters transcriptomes and induces epigenetic changes that extend beyond the effects of endogenous G4 regulation (Karatayeva et al. 2025). This study also showed that PhenDC3 did not compromise genome integrity, assessed by insertion, deletion, or chromosome rearrangement rates, in cells with intact DNA repair machinery, suggesting that G4 stabilization preferentially exposes vulnerabilities in repair-deficient contexts. Consistent with this idea, multiple studies report that PhenDC3 and CX-5461 exhibit selective toxicity in tumors lacking homologous recombination and non-homologous end-joining pathways (Zimmer et al. 2016; Xu et al. 2017; Bruno et al. 2020; Hilton et al. 2022; Jin et al. 2023). As discussed in the previous section, CX-5461 causes genotoxic stress through combined G4 stabilization and topoisomerase inhibition, resulting in replication-dependent DNA damage, disruption of TOP2B-G4 interactions, replication fork stalling, and MRE11-dependent fork degradation (Bruno et al. 2020; Bossaert et al. 2021; Pang et al. 2025). PhenDC3 promotes recombination-dependent genome instability at G4s by inducing replication fork stalling, generating ssDNA breaks and interfering with Pif1 helicase-mediated G4 unwinding in yeast (Piazza et al. 2010; Obi et al. 2020). Another G4 ligand, PyBI, preferentially binds to parallel G4s, increasing their stability and activating a DNA damage response marked by γH2AX accumulation. PyBI exposure induces replication stress, G2/M cell-cycle arrest and apoptosis (Yan et al. 2025). Yan et al. further showed PyBI-dependent synthetic lethality in DNA repair-deficient systems through genetic depletion of RAD51 or Ku80, or pharmacological inhibition of RAD51 and DNA-PK, which corresponded with increased DNA damage signaling (Yan et al. 2025). Collectively, these findings indicate that G4 ligands stabilize DNA secondary structures at the replication fork, promoting fork stalling and inducing DNA breaks. Resolution of these breaks requires homologous recombination and non-homologous end-joining pathways, creating therapeutic vulnerabilities in cancers with compromised DNA repair capacity and supporting synthetic lethality-based treatment strategies.

Helicase deficiency-driven G4 instability

The dynamic formation and resolution of G4 structures are tightly regulated by specialized helicases, ensuring timely resolution of G4 structures and maintenance of genome integrity. Several G4-resolving helicases, including Bloom helicase (BLM), Werner syndrome helicase (WRN), FANCJ and DHX36, are critical for resolving G4s and maintaining genome integrity (Wu et al. 2008; Babbe et al. 2009; Mao et al. 2010; Paeschke et al. 2013; Croteau et al. 2014; Lee et al. 2021; Sato et al. 2021; Zhou et al. 2022; Mizumoto et al. 2023; Orren and Machwe 2024; Bartosik et al. 2025). Deficiencies in these helicases lead to the accumulation of unresolved G4s, which promote DSBs, replication stress and aberrant recombination, all of which are hallmarks of cancer development (Sato and Knipscheer 2023; Jonchhe et al. 2025; Menon and Gueble 2025).

Among G4-resolving helicases, DHX36 exhibits the highest specificity and affinity for parallel G4 structures (Giri et al. 2011; Heddi et al. 2015; Chen et al. 2018; Banco et al. 2025). During the S phase, DHX36 coordinates with FANCJ to bypass the CMG helicase on the leading strand, preventing replication fork stalling and the associated damage (Sato et al. 2021). DHX36 also resolves G4s, limiting the accumulation of DSBs. Mizumoto et al. demonstrated that the loss of DHX36 results in G4 accumulation, accompanied by an increase in γH2AX levels detected by immunofluorescence in IMR-90 cells (Mizumoto et al. 2023). Genome-wide mapping of DSBs in DHX36-depleted Jurkat cells further showed an increase in DNA breaks at G4 sites, highlighting their role as critical mediators of genome integrity at regulatory regions (Bartosik et al. 2025). Additionally, loss of DHX36 slows cell growth and enhances sensitivity to G4-stabilizing ligands in lung cancer cell lines (Cui et al. 2021; Mizumoto et al. 2023). DHX36 also exerts context-dependent effects on gene expression in cancer cells. Loss of DHX36 produces bidirectional transcriptional changes rather than uniform gene activation or repression, and differentially expressed genes following DHX36 loss include numerous TFs and oncogenes linked to various cancers (Esnault et al. 2023; Gumina et al. 2024). This bidirectionality is consistent with the strand- and context-dependent roles of G4s in transcription regulation described in Figure 2. In pediatric B-cell acute lymphoblastic leukemia (B-ALL), higher DHX36 expression correlates with improved overall survival (Bartosik et al. 2025). Similarly, DHX36 acts as a tumor suppressor in non-small cell lung, testicular, breast and colon cancers, where its downregulation accelerates migration and tumor growth (Matsumura et al. 2017; Zeng et al. 2020; Cui et al. 2021; Wang et al. 2023). However, a recent study in breast cancer showed that DHX36 overexpression can enhance cell proliferation, migration and invasion, with higher DHX36 levels being associated with poor recurrence-free survival (Wei et al. 2025). Together, these findings indicate that loss of DHX36 drives genome instability through G4-mediated DSBs, while its dysregulation plays a more complex and context-dependent role in cancer development, acting either as a tumor suppressor or, in some cases, as an oncogene.20232023

BLM and WRN, members of the RecQ helicase family, are well-established suppressors of G4-driven genome instability. BLM unwinds DNA G4 structures in an ATP-dependent manner (Wu et al. 2015). Using single-cell DNA template strand sequencing (Strand-seq), Van Wietmarschen et al. mapped sister chromatid exchanges (SCEs) in human and murine cells lacking functional BLM. They showed that BLM helicase suppresses SCEs at G4 motifs within transcribing genes, thereby maintaining genome stability (Van Wietmarschen et al. 2018). In Bloom syndrome cells, which lack functional BLM, SCEs are highly enriched at G4 motifs in the coding regions of actively transcribed genes, supporting a specific role for BLM in preventing G4-driven recombination during transcription (Van Wietmarschen et al. 2018). Similarly, loss of WRN in Werner syndrome is associated with increased chromosomal instability. WRN helicase activity is required for efficient replication through G4-rich telomeric DNA, where G4 formation impedes lagging-strand synthesis (Crabbe et al. 2007; Li et al. 2017). Loss of WRN leads to replication-associated telomere loss, resulting in chromosomal fusions, dicentric chromosomes and complex translocations characteristic of Werner syndrome (Orren 2006; Ariyoshi et al. 2007; Crabbe et al. 2007; Poot 2025). Additionally, in a Fanconi anemia patient-derived cell line harboring an FANCJ mutation that abolishes G4 unwinding activity, the loss of functional FANCJ correlates with the accumulation of large genomic deletions near G4 motifs (London et al. 2008). Collectively, deficiencies in DHX36, BLM, WRN and FANCJ compromise G4 unwinding, drive genome instability and contribute to cancer development through distinct but overlapping mechanisms. Each helicase protects the genome at specific loci and during distinct biological processes, but their loss converges on G4-mediated DNA fragility, replication stress and aberrant recombination.

G4-driven immune signaling activation

Stimulator of interferon genes (STING) is a critical adaptor protein in the innate immune response in vertebrates and serves as a hub for sensing cytosolic DNA fragments (Sun and Hornung 2022). Upon recognition of double-stranded DNA in the cytoplasm by cyclic GMP-AMP synthase (cGAS), cGAS is activated and catalyzes the formation of the second messenger cyclic GMP-AMP (cGAMP) from ATP and GTP (Sun and Hornung 2022). The cGAMP then binds to STING, initiating downstream cascades involving TANK-binding kinase 1 (TBK1) and interferon regulatory factor 3 (IRF3), which lead to the production of type I interferon and NF-κB-dependent proinflammatory cytokines (Hopfner and Hornung 2020; Balka and De Nardo 2021; Sun and Hornung 2022). While this pathway evolved primarily to detect pathogen-derived nucleic acids, it can also be activated by self-generated nucleic acids, including cytoplasmic endogenous double-stranded DNA, mitochondrial DNA released during cellular stress, extranuclear chromatin fragments generated by genotoxic damage, and DNA from ruptured micronuclei (Dhanwani et al. 2018; Hopfner and Hornung 2020; Hu and Shu 2020). The innate immune signaling response cannot distinguish between microbial and endogenous nucleic acids; therefore, the subcellular localization, quantity, and structural features of DNA determine whether self-nucleic acids trigger an immune response. Persistent replication stress, especially in cells lacking functional helicases (e.g. FANCJ, PIF1), can lead to fork collapse and the formation of DSBs. Mis-segregated chromosomes or chromosomal fragments may then form micronuclei during mitosis (Fenech et al. 2020; Pepe et al. 2025; Waisertreiger et al. 2025). Genomic instability is particularly pronounced at rDNA and pericentromeric satellite regions, where G4 stabilization induces DNA breaks and rearrangements, directly promoting micronuclei formation (Waisertreiger et al. 2025). Thus, unregulated G4s initiate a cascade of replication stress, error-prone repair and chromosomal missegregation, culminating in micronuclei accumulation (Pepe et al. 2025; Waisertreiger et al. 2025). Micronuclei that rupture into the cytoplasm are recognized by the cGAS-STING signaling pathway, triggering type I interferon production and induction of proinflammatory cytokines (Mackenzie et al. 2017; Li and Chen 2018; MacDonald et al. 2023). In cancer cells treated with G4 ligands, increased micronuclei formation robustly activates the cGAS-STING signaling pathway, upregulating innate immune gene expression, including type I interferons and NF-κB-dependent cytokines (Miglietta et al. 2021; Bartosik et al. 2025). Loss of the G4-resolving helicase DHX36 similarly exacerbates genome instability, partially manifested as cytoplasmic double-stranded DNA accumulation, activation of the NF-κB signaling pathway, and pro-inflammatory gene expression and cytokine production (Bartosik et al. 2025). The immunogenic consequences of G4-dependent cGAS-STING pathway activation are context-dependent, with cancer cells exhibiting more pronounced activation than non-transformed cells (Miglietta et al. 2021).

DNA G4s and genome instability in cancer

G4s in cancer development and as a source of mutations in cancer genomes

G4 motifs in promoter regions are preferentially enriched in oncogenes compared to tumor suppressors, suggesting a pattern shaped by selective pressure that promotes genomic instability in oncogenes. This distribution indicates an active role for G4s in driving tumorigenesis when their formation is dysregulated (Eddy and Maizels 2009). Functionally, these G4s act as modulators of oncogene transcription, often serving as regulatory hubs for TF recruitment and chromatin remodeling. The most extensively studied example is the MYC oncogene, in which the Pu27 G4-forming sequence was described as a transcriptional repressor (Siddiqui-Jain et al. 2002; You et al. 2015; Calabrese et al. 2018). However, a more recent, cell-relevant study using CRISPR/Cas9 gene editing in the endogenous chromatin context has demonstrated the opposite (Esain-Garcia et al. 2024). The MYC G4 motif located in the nuclease hypersensitive III₁, approximately 100 bp upstream of the P1 promoter that controls 10–20% of its expression, positively regulates MYC transcription from the P1 by preventing nucleosome deposition and facilitating the recruitment of the transcription machinery. Loss of G4 structure within this region suppresses MYC transcription through de novo nucleosome deposition and can be rescued by a different G4 structure, such as the KRAS G4 (Esain-Garcia et al. 2024). Stabilizing the G4 structure in the promoter of the oncogene KRAS has been shown to favor transcription, through the binding of MAZ and PARP-1 to the G4 (Cogoi et al. 2010). In contrast, G4 formation within promoter regions has been shown to act as a transcriptional repressor in oncogenes such as c-KIT and BCL2 (Wang et al. 2010; Gao et al. 2015). Interestingly, upon loss of DHX36, expression of c-KIT was significantly downregulated in mouse testis and human cell lines, and this regulation involves the direct binding of DHX36 to G4 in the c-KIT promoter (Gao et al. 2015). Beyond individual oncogene promoters, G4 structures are enriched in super-enhancers across multiple cancer cell lines, where they function as transcriptional regulatory hubs that sustain the malignant state (Huang et al. 2026). In breast cancer-derived tumor xenograft models, G4 motifs were detected not only in the gene promoter but also within the first introns, highlighting their potential influence on both transcription initiation and early elongation (Shu et al. 2025). Dysregulated G4s can further promote recombination and genomic instability by contributing to rDNA copy number variation and chromosomal rearrangements (Datta et al. 2021; Pavlova et al. 2021; Duardo et al. 2023). These regions are highly transcribed and are therefore susceptible to DNA damage and double-strand break formation (Hänsel-Hertsch et al. 2016; Cristini et al. 2021).

Aberrant G4 formation has been associated with higher nucleotide substitution rates (Guiblet et al. 2021a) and the enrichment of mutations linked to cancer, implicating G4s as contributors to somatic mutagenesis (Stein et al. 2022; Zhuk et al. 2024). G4s are frequently found at cancer hotspots due to their localization in genomic regions that are more prone to genomic instability, thereby increasing the likelihood of somatic mutations. High-throughput sequencing and genome-wide mapping of G4s have shown that G4s are particularly abundant in 5’ UTRs of highly transcribed cancer-related genes and in regions of somatic copy number alterations (Chambers et al. 2015; Hänsel-Hertsch et al. 2016). Mechanistically, G4 formation can impede DNA replication and repair, leading to increased local genome instability, which is frequently manifested as single-nucleotide variations (SNVs), deletions and recombination events, all of which are hallmarks of cancer. Recent large-scale analyses have demonstrated a positive correlation between the local G4 density and the density of cancer-associated SNVs, which supports the role of G4s in driving mutational events at cancer hotspots (Richl et al. 2024). Moreover, the presence of G4s in these regions is also associated with elevated transcriptional activity and chromatin accessibility, features characteristic of oncogenic transformation (Hänsel-Hertsch et al. 2016; De-Paula et al. 2024). For example, De-Paula et al. analyzing data from the Catalog of Somatic Mutations in Cancer (COSMIC) database, identified the NOTCH1 locus as a prominent recurrent mutational hotspot associated with G4 formation (De-Paula et al. 2024). Interestingly, the NOTCH1 hotspot was located in the most unstable of the analyzed DNA sequences. Those results highlight the fact that the most mutagenic G4 sequences present in the germline and cancer have a high propensity to form alternative DNA secondary structures in vivo.

Pan-cancer genomic analysis of G4-associated instability

Given the established role of DHX36 as the primary G4-resolving helicase, we examined its prognostic significance across The Cancer Genome Atlas (TCGA) cohorts for each recommended cancer-endpoint combination (overall survival [OS], progression-free interval [PFI] and disease-free interval [DFI]), adjusting for age, sex, tumor stage and/or grade (Liu et al. 2018; Smith and Sheltzer 2022). Acute myeloid leukemia (LAML) was the only cancer in which DHX36 expression remained significantly associated with survival within each endpoint family, with higher expression associated with a reduction in the hazard of death (Figure 5A). In contrast to the protective association in LAML, higher DHX36 expression was consistently associated with worse outcomes in lung adenocarcinoma (LUAD), suggesting a context-dependent prognostic role (Matsumura et al. 2017; Zeng et al. 2020; Cui et al. 2021; Wang et al. 2023; Wei et al. 2025). Gene set enrichment analysis also revealed a consistent molecular dichotomy between DHX36-high and DHX36-low tumors across 31 cancer types (Figure 5B). DHX36-high tumors were enriched for cell proliferation programs, including E2F targets and the G2M checkpoint. Because G4 motifs are disproportionately enriched among cell cycle and proliferation-associated genes (Eddy and Maizels 2006), we propose a model in which elevated DHX36 preferentially increases transcriptional activity (via promoter G4 resolution) (Huang et al. 2012; Mizumoto et al. 2023) and translational efficiency (via 5’UTR G4 resolution) (Sauer et al. 2019) of this gene class, including CCNB1, CCND1 and NOTCH1, whose protein levels directly track DHX36 abundance (Wei et al. 2025). In contrast, DHX36-low tumors showed upregulation of oxidative phosphorylation, DNA repair and the p53 pathway, along with robust activation of innate immune signaling: interferon response, inflammatory response and apoptosis. This pattern is consistent with the model in which reduced G4 resolution capacity leads to the accumulation of unresolved G4 structures, increased replication stress and cytoplasmic DNA that activate the cGAS-STING pathway, triggering interferon production and inflammatory signaling (Miglietta et al. 2021; Bartosik et al. 2025).

Figure 5.

G4s as a source of instability in cancer genomes. (A) Covariate-adjusted survival curves (median split) for LAML (overall survival) and LUAD (progression-free survival) illustrate the bidirectional prognostic association of DHX36 expression. Cox proportional hazards regression models using continuous log2(FPKM-UQ) expression and adjusting for age, sex, stage, and/or grade were applied. A sensitivity analysis using restricted mean survival time showed a consistent but non-significant trend. Therefore, the interpretation of the Cox estimate should be made with caution. (B) GSEA analysis using MSigDB Hallmark gene sets. Dot color encodes NES (red = enriched in DHX36-high Q4; blue = enriched in DHX36-low Q1); dot size encodes -log10(Padj). (C) SV breakpoint enrichment at G4 sites. Local z-score profile showing observed overlap of 5,384 breakpoints far exceeding the permutation mean of ~3,314 (z = 35.8, p < 10^-4). (D) Somatic mutation distribution across G4 structural elements, with 69.5% (27,806) within G-runs significantly exceeding the element-count null expectation of 57.1% for G-runs (binomial test, p < 2 × 10^-16). (E) Proportion of destabilizing, neutral, and stabilizing mutations by mutation type. Deletions were the most destabilizing class (13.2% destabilizing, mean Δ = -1.32); SNVs showed a net destabilizing effect (mean Δ = -0.80, Wilcoxon p < 2 × 10^-16); and insertions showed a net stabilizing effect (14.9% stabilizing, mean Δ = +1.73).

While the association between G4s and point mutations has been established (Richl et al. 2024), the relationship between G4s and structural variants (SVs) in cancer genomes remains underexplored. We applied consensus SV calls from 2,527 primary tumor samples in the Pan-Cancer Analysis of Whole Genomes (PCAWG) consortium in permutation testing (1,000 random genome-wide shuffles) and found a highly significant enrichment of SV breakpoints at G4 sites (Figure 5C). This enrichment was consistent across all SV classes (deletions, insertions and translocations), suggesting G4 structures directly contribute to breakpoint formation rather than reflecting broader regional effects.

Another intriguing question is whether somatic mutations at G4 sites show structural selectivity (G-run vs loop regions). Using somatic mutations identified in PCAWG, we identified 136,032 unique mutations that overlapped G4 CUT&Tag sites across six different cell lines (NHEK, HaCaT, U2OS, K562, HeLa and NSC) using ChIP-seq and CUT&Tag data generated with the structure-specific antibody, BG4 (Hänsel-Hertsch et al. 2016; Hänsel-Hertsch et al. 2018; Mao et al. 2018; Hui et al. 2021; C. Li et al. 2021; Zyner et al. 2022; Bartosik et al. 2025). We then used the pqsfinder tool (Hon et al. 2017) to map each mutation to a specific G-run or loop element. Among 40,007 mappable mutations, 27,806 (69.5%) fell within G-runs and 12,201 (30.5%) within loops (Figure 5D). SNVs and deletions showed strong G-run enrichment (~69–70%), whereas insertions were more equally distributed between G-runs and loops (~47% vs ~53%), suggesting that the structural context differentially constrains mutation types. Beyond co-localization, somatic mutations also directionally alter G4 stability (as calculated by G4 folding energy, ΔG). Of the 107,189 scorable mutations (99,368 SNVs, 2,238 multi-nucleotide variants, 1,531 insertions and 4,052 deletions) mappable to G4 sites, destabilizing mutations (n = 10,909) outnumbered stabilizing ones (n = 2,928) by nearly 4:1 overall (10.2% vs. 2.7%). This balance was strikingly dependent on mutation type (Figure 5E): deletions were the most destabilizing class; SNVs showed a net destabilizing effect; and insertions showed a net stabilizing effect potentially by extending G-runs or creating novel G-run-capable sequences. These findings indicate that somatic mutations at G4 sites are not functionally neutral but directionally alter G4 structural properties depending on mutation type, with potential consequences for local gene regulation and genome instability.

Concluding remarks and future perspective

G4s occupy a dual position in genome biology: as essential regulatory elements that shape transcription, replication and chromatin architecture, and as potent drivers of genome instability when their formation or resolution is dysregulated. The evidence reviewed here establishes that G4-induced instability operates through multiple converging mechanisms, including transcription-associated R-loop/G-loop formation, replication fork stalling, topoisomerase-mediated DNA breakage and helicase deficiency, each of which contributes to the elevated somatic mutation burden observed in cancer genomes. The pan-cancer analyses of TCGA and PCAWG data presented in this review provide direct evidence that these mechanisms operate at scale in human tumors: G4 sites are significantly enriched for structural variant breakpoints across all SV classes, and somatic mutations at G4 loci directionally alter G4 stability depending on their structural context and mutation type. Furthermore, the expression level of the G4-resolving helicase DHX36 consistently stratifies tumors into distinct molecular states, with DHX36-low tumors showing coordinate upregulation of DNA repair, p53 and interferon signaling pathways across 31 cancer types.

However, several important questions remain. First, although substantial evidence links G4s to genome instability, much of the mechanistic work has been conducted in vitro or in cell lines. Future studies should focus on investigating G4 dynamics and their interactions with chromatin in physiologically relevant in vivo systems, leveraging emerging live-cell G4 imaging and long-read sequencing technologies. Second, the observation that somatic mutations can either stabilize or destabilize G4 structures, with deletions net-destabilizing and insertions net-stabilizing, raises the possibility that G4 structural changes are subject to selection during tumor evolution, a hypothesis that warrants investigation through longitudinal analysis of tumor genomes. Third, the robust connection between DHX36-low status and interferon pathway activation across cancer types, consistent with the cGAS-STING model of G4-driven immune signaling, opens therapeutic opportunities: G4-stabilizing ligands may enhance anti-tumor immunity by promoting micronuclei formation and innate immune activation, particularly in combination with immune checkpoint inhibitors. Finally, the context-dependent roles of G4 helicases such as DHX36, with bidirectional dysregulation across cancer types and associations with both tumor grade and patient outcomes, highlight the need for cancer-type-specific therapeutic strategies targeting G4 biology.