The stability and unfolding of DNA: RNA hybrid G-quadruplexes

Abstract

DNA:RNA hybrid G-quadruplexes (hG4s) are intermolecular G4 structures composed of guanine (G)-rich DNA and RNA strands. These structures play crucial regulatory roles in transcription and telomere stability. However, the absence of methods to produce uniform hG4s has limited systematic studies on their stability and unwinding. In this study, using human telomeric sequences as a model, we developed a PC (photocleavable)-linker-induced cleavage approach to generate hG4s with varying compositions and assessed their relative stability, revealing that hG4 stability increases with higher RNA content. We also showed that various G4-binding proteins recognize and bind to hG4 structures, with G4P exhibiting the highest affinity and effectively stabilizing hG4s. Additionally, the helicases Pif1 and RHAU were confirmed to efficiently unwind hG4 structures. Finally, we validated the roles of stabilizing and unwinding proteins in a transcription system. This study establishes a framework for investigating the stability and unwinding of hG4s and provides insights into modulating hG4 structures to regulate cellular processes.

Graphical Abstract

Graphical Abstract.

Introduction

G-quadruplexes (G4s) are four-stranded, noncanonical secondary structures composed of guanine (G)-rich nucleic acids [1]. They participate in various physiological processes, such as replication, transcription, translation, epigenetics, meiosis, and telomere protection [2]. They are closely associated with human health issues such as cancer, neurodegenerative diseases, and viral pathogenesis [3–5]. Based on their organizational form, G4s can be classified into intramolecular G4s and intermolecular G4s [6]. Depending on their chemical composition, G4s can be categorized as DNA G4 (dG4), RNA G4 (rG4), and DNA:RNA hybrid G4 (hG4) [2, 7, 8]. G4s regulate biological processes through their structural folding and unfolding dynamics [9]. In recent years, G4 dynamics has mainly focused on dG4s and rG4s, and has made many important progresses using both theoretical and experimental methods [10, 11]. hG4s have been proven to be widespread in bacteria [12], yeast [13], higher eukaryotic cells [7, 8, 14], and mitochondria [15, 16]. In addition, since the formation of hG4s does not require four G-tracts on one nucleic acid strand, their abundance is much higher than that of dG4s and rG4s [8, 13]. It is important to highlight that hG4s can still form and even predominate in DNAs containing four G-tracts, despite the fact that these DNAs are theoretically able to independently form intramolecular dG4s [12].

hG4s play crucial roles in transcription [8] and telomere stability [17]. During transcription, the nascent RNA-produced remains base-paired with its template DNA, leaving nontemplate single-stranded DNA (ssDNA). The three-stranded structures called R-loops [18]. In recent years, it was found that hG4s formed between the nascent RNA and the nontemplate DNA strand in R-loops can suppress transcription [19] and stabilize R-loop [15]. At telomeres, telomeric DNA can be transcribed into telomere RNA [20], which can form an hG4 structure with telomeric DNA at the 3' end of the telomere to cap and protect the 3' overhang of the telomeric DNA [14, 21, 22]. Unraveling the structural dynamics of G4s is beneficial for understanding and regulating their intracellular functions. However, there is a lack of systematic research on the dynamics of hG4s.

Currently, at least four key questions require further investigation: First, it is crucial to develop methods to generate pure hG4s. This is one of the main reasons why hG4s have not been well studied in detail. Since hG4 is composed of two pieces of ssDNA and single-stranded RNA (ssRNA), if the two are directly mixed and annealed, hG4, dG4, and rG4 will coexist [17, 23]. For this reason, the generation of uniform hG4s is usually relied on the help of double strands as a scaffold in published literatures [13, 23, 24], or with the help of a click reaction [21, 25, 26]. However, the connected double strands and chemical linkers cannot simulate intracellular conditions and limit the production of hG4 structural diversity. Second, hG4s can form the hG4(2+2) structures containing 2 DNA columns and 2 RNA columns, which are the most studied structures [14, 15, 24]. However, the structure hG4(1+3) with 1 DNA column and 3 RNA columns and the structure hG4(3+1) with 3 DNA columns and 1 RNA column may also form [12, 13]. In addition, intermolecular dG4 and rG4 consisting of two strands also play significant regulatory roles in cells [27–30]. Therefore, the formation of hG4s will be competed by dG4s and rG4s. Obtaining the stability information between different intermolecular G4s is crucial to infer the probability of their existence. However, there is a lack of systematic research in the stability relationships. Third, G4-specific recognition proteins play an extremely important role in studying the formation and dynamics of G4s. In recent years, multiple dG4/rG4-specific recognition proteins have been developed, such as scFv antibodies BG4 [31, 32] and D1 [33], artificial protein G4P [34], and nanobody SG4 [35]. However, binding affinity of these probes to hG4s is unknown. Finally, DNA:RNA hybrids should be processed promptly, otherwise they will affect multiple roles in many cellular processes and threaten genome stability [24, 36, 37]. In cells, the timely processing of DNA:RNA hybrids depends on RNase H and specialized helicases [36–38]. However, RNase H does not degrade stable hG4s [24], meaning that stable hG4s must be resolved by helicases. To date, no hG4 unwinding helicases have been reported [39, 40].

In this study, we selected human telomere sequences as the research focus and created a method to employ ultraviolet (UV)-induced PC (photocleavable)-linker fragmentation to conveniently generate pure intermolecular hG4s with varying compositions. Circular dichroism (CD) was used to analyze their structures and stability. Binding experiments revealed that, compared to BG4, D1, and SG4, G4P exhibited stronger binding affinity for hG4s and helped stabilize hG4 structures. Further investigation showed that while RHUA and Pif1 efficiently unfolded hG4, RecQ and DinG did not. This study deepens our understanding of the physical and chemical properties of hG4s and lays the foundation for intracellular regulation of their structures.

Materials and methods

Nucleic acid preparations

All substrates used in the experiments are listed in Supplementary Table S1. Annealing was performed under the following conditions: samples were heated to 95°C for 5 min and gradually cooled to 20°C in annealing buffer [25 mM Tris–HCl, pH 7.5, and 100 mM KCl]. Substrates containing G4 structures were annealed at a concentration of 10 μM. For transcription and double-stranded DNA (dsDNA) unwinding experiments, dsDNA substrates were prepared by mixing two ssDNAs in a 1:1 molar ratio, followed by annealing. For intermolecular G4 formation experiments, UV irradiation (365 nm) was performed in induction buffer [25 mM Tris–HCl, pH 7.5, 100 mM KCl] to induce PC-linker cleavage.

G4 polyacrylamide gel electrophoresis

G4-denaturing polyacrylamide gels were prepared with 100 mM LiCl to prevent G4 structure formation [23, 40]. Denaturing polyacrylamide gel electrophoresis (PAGE) was run at 4°C under a constant voltage of 10 V/cm in Tris-Acetate-EDTA (TAE) buffer containing 100 mM LiCl. Before electrophoresis, substrates were heated to 95°C for 5 min and immediately cooled on ice. Nonfluorescently labeled G4 substrates were loaded at a final concentration of 2.5 μM, with ThT (100 μM) used for detection [23]. Fluorescently labeled substrates were loaded at a concentration of 30 nM. G4-native PAGE was conducted with gels containing 100 mM KCl, under constant voltage at 10 V/cm in TAE buffer at 4°C. All gel images were captured using the ChemiDoc MP Visualization System (BioRad) and analyzed with ImageJ.

Circular dichroism experiments

CD measurements were performed following protocols from our previous work [41]. Samples were prepared at a final concentration of 10 μM. CD spectra were recorded using a Chirascan V100 (Applied Photophysics Ltd.) over a wavelength range of 195–320 nm, with a bandwidth of 1 nm, a step size of 1 nm, and an acquisition time of 1 s per point. Measurements were repeated in triplicate and averaged. For melting curve analyses, the acquisition time was reduced to 0.5 s per point, with temperature gradients from 20°C to 98°C, a heating rate of 1°C/min, and data acquisition every 2 min.

Protein expression and purification

This study expressed and purified four G4-binding proteins (BG4, SG4, D1, and G4P), four helicases [Saccharomyces cerevisiae Pif1 (ScPif1), Bos taurus RHAU (BtRHAU), Escherichia coli DinG (EcDinG), and Escherichia coli RecQ (EcRecQ)], and human RPA. Detailed methods are provided in the Supplementary data. The purity of BG4, SG4, D1, and G4P was confirmed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) (Supplementary Fig. S7A). Results for the four helicases and RPA are shown in Supplementary Figs S9A and S13A. For simplicity, ScPif1, BtRHAU, EcDinG, and EcRecQ are referred to as Pif1, RHAU, DinG, and RecQ, respectively, unless stated otherwise.

Electrophoretic mobility shift assay

Substrates at a final concentration of 20 nM were incubated with varying concentrations of different proteins (BG4, SG4, D1, G4P, or hRPA) in binding buffer [25 mM Tris–HCl, pH 7.5, 100 mM KCl, 1 mM dithiothreitol (DTT), and 10% glycerol (v/v)] at 30°C for 10 min. Electrophoresis was performed on a 6% native PAGE with 100 mM KCl at 4°C under a constant voltage of 5 V/cm. Gels were imaged using the ChemiDoc MP Visualization System (BioRad). Data were analyzed using ImageJ and fitted to the Hill equation.

Unwinding assay

For G4 unwinding, G4 substrates (40 nM) were incubated with varying concentrations of different proteins (Pif1, RHAU, DinG, RecQ, or RPA) in unwinding buffer [25 mM Tris–HCl, pH 7.5, 100 mM KCl, 1 mM DTT, 5 mM MgCl₂, and 10% glycerol (v/v)] at 30°C for 10 min. The reaction was initiated by adding adenosine triphosphate (ATP) to a final concentration of 1 mM, and helicase activity was allowed to proceed for 10 min at 30°C. Reactions were stopped by adding stop buffer [25 mM ethylenediaminetetraacetic acid (EDTA), 10% glycerol (v/v)] and proteinase K (final concentration: 1 mg/ml) and incubating at 30°C for 20 min. Electrophoresis was conducted on 18% native PAGE containing 100 mM KCl at 4°C under a constant voltage of 12 V/cm. Gels were imaged using the ChemiDoc MP Visualization System (BioRad), and data were analyzed with ImageJ.

For dsDNA unwinding, the final concentration of the dsDNA substrate in the reaction system was 40 nM. Pif1 concentration was varied as indicated. The unwinding reaction was initiated by adding ATP to a final concentration of 5 mM in unwinding buffer. For the experimental group, pyridostatin (PDS) was additionally included at a final concentration of 2 μM. The samples were then incubated at 37 °C for 30 min. To terminate the reaction, stop buffer and proteinase K (final concentration: 2 mg/ml) were added, followed by mixing and incubation at 37 °C for 10 min. The reaction products were then analyzed by electrophoresis on a 10% polyacrylamide gel. Gels were imaged using the ChemiDoc MP Visualization System (BioRad), and data were analyzed with ImageJ.

Transcription experiments

Transcription experiments were conducted following previously published protocols [42]. Substrates at a final concentration of 50 nM were incubated with T7 RNA polymerase (2 U/μl) in reaction buffer [10 mM Tris–HCl, pH 8.0, 50 mM KCl, 2 mM spermidine, 10 mM DTT, 6 mM MgCl₂, 2 mM ATP, 2 mM guanosine triphosphate (GTP), and 40% polyethylene glycol (PEG) 200 (v/v)] at 37°C for 2 h. For hG4 verifying experiments, RNase A (0.4 mg/ml) and RNase H (0.4 U/μl) were added and incubated at 37°C for 1 h. For helicase experiments, diluted helicase samples were added along with 1 mM ATP and incubated at 37°C for 30 min. Reactions were terminated by adding stop buffer [25 mM EDTA, 10% glycerol (v/v)] and proteinase K (1 mg/ml) and incubating at 37°C for 1 h. Electrophoresis was performed on 10% native PAGE containing 100 mM KCl at 4°C under a constant voltage of 12 V/cm. Gels were imaged using the ChemiDoc MP Visualization System (BioRad), and data were analyzed with ImageJ.

Microscale thermophoresis

Microscale thermophoresis (MST) experiments were performed following the protocol established in our previous work [43]. The final concentration of Cy5-labeled, fully intermolecular hG4 was 10 nM, while the concentration of PDS was varied. Samples were prepared in binding buffer [25 mM Tris–HCl, pH 7.5, 100 mM KCl, 1 mM DTT, and 10% (v/v) glycerol], thoroughly mixed, and incubated at 25 °C for 2 min. The samples were then loaded into capillaries and measured using the MST instrument. Each reaction was independently measured in triplicate.

Results

Design of the light-induced bimolecular G4s

In previous studies, the monomer intramolecular G4 structures were easily obtained by annealing [44, 45]. Inspired by this, we planned to first anneal to form a uniform intramolecular structure, and then induce single-strand breaks to obtain an intermolecular structure. We designed a light-controlled method to generate bimolecular G4s (Fig. 1A, top panel). In this system, PC-linker was used to insert into the middle of G-rich chains to convert intramolecular G4 into intermolecular G4 through UV-induced cleavage (Fig. 1A, top panel) [46, 47]. To verify above design, the well-studied human telomeric (hT) repeat-containing DNA and RNA sequences [17] were selected (Fig. 1A, bottom panel). The three sequences named as hThG4, hTdG4, and hTrG4 are proposed to fold into hG4, dG4, and rG4, respectively. These sequences were annealed in K100 (25 mM Tris–HCl, pH 7.5, 100 mM KCl) to fold into G4 structures as confirmed by CD (Fig. 1D), and then were induced by 365 nm UV light at different time. By running denaturing PAGE, it is clear that, just as expected, as the induction time increases, all the three intramolecular G4 structures can be gradually induced to break (Fig. 1B). Under our experimental conditions, ~95% of the three substrates were split at 1 min, and almost all of them were converted into split substrates at 10 min (Fig. 1B and C, bottom panel). We added “-s” after each substrate name to indicate “split” to distinguish it from the uninduced substrates. In this article, unless otherwise specified, the suffix “-s” indicates that the corresponding substrate has been split by UV-induced cleavage. In order to verify that the break did occur in the PC-linker, we compared the substrates before and after induction (Fig. 1C, top panel). The results showed that hThG4 was indeed induced to form an RNA and a DNA, hTdG4 was induced to DNA, and hTrG4 was induced to RNA, indicating the break does occur at the PC-linker. According to our conventional understanding, shorter nucleic acid fragments should migrate faster during electrophoresis. However, the low-molecular-weight RNA fragments generated by hThG4-induced cleavage were observed to migrate more slowly. We will continue to investigate this anomalous migration in future work. After induction in K100, hThG4-s and hTrG4-s exhibited parallel structures showing an ~260 nm peak, while hTdG4-s exhibited a mixed structure of antiparallel and parallel structures exhibiting a peak at ~290 nm and a small shoulder at ~260 nm, consistent with previous reports [21, 48]. In addition, both the gel electrophoresis (Supplementary Figs S1 and S10) and the CD signal (Fig. 1D) of all three intermolecular structures showed that the breakage of PC-linker did not lead to the collapse of G4 structures. In summary, the above results demonstrate that by UV-induced PC-linker cleavage method, we can conveniently obtain various uniform intermolecular G4 structures.

Figure 1.

Establishment of a method for forming uniform intermolecular G4s. (A) Schematic diagram of the light-induced bimolecular G4s. In this system, a PC-linker was inserted into the middle of G-rich chains, enabling the conversion of intramolecular G4s to intermolecular G4s via UV-induced cleavage. Denaturing G4 gel electrophoresis images of hThG4, hTdG4, and hTrG4 under UV induction at different time points (B), along with the corresponding substrates before and after PC-linker cleavage (C, upper panel) and the quantitative cleavage ratios from three replicates (C, lower panel). The suffix “-s” indicates that the corresponding substrate has been split by UV-induced cleavage. (D) CD spectra of hThG4, hTdG4, and hTrG4 before and after 10 min of induction in K100 buffer (25 mM Tris–HCl, pH 7.5, 100 mM KCl).

The higher the RNA content, the more stable the hG4 appears to be

Besides the above intermolecular (2+2) structures (Fig. 1), intermolecular (1+3) and (3+1) structures can also exist [12, 13]. After establishing a convenient method to produce a uniform hG4 (Fig. 1), we next planned to use this method to study the stability of hG4 with different compositions in order to understand their probability of occurrence in cells. Therefore, we designed hThG4(3+1) and hThG4(1+3) sequences (Fig. 2A, bottom panel). Using the same method as above, we generated hThG4(3+1)-s and hThG4(1+3)-s structures (Fig. 2A, top panel). CD measurements show that both structures are parallel structures, and the structures remain intact after the PC-linker breaks (Fig. 2B and C). Combined with the (2+2) structure (Fig. 1D, left panel), it is not difficult to conclude that hG4s of different compositions all show parallel structures. In order to further understand the stability of different hG4s, we measured their melting temperatures (T_m) by CD melting (Fig. 2D and Supplementary Fig. S2). It is shown that in terms of stability hThG4(1+3)-s > hThG4-s > hThG4(3+1)-s, indicating the higher the RNA content, the more stable hG4 appears to be. This result can explain why hG4 in cells prefers to form a (1+3) structure when there are no less than 3 or even 4 G-quartets [13]. To further verify that this is a solid and general conclusion, we introduced variations in both the loop length and the number of layers in the hG4 constructs (Supplementary Figs S3 and S4). In all cases, the results consistently supported our hypothesis. Furthermore, we checked the stability of hTdG4 and hTrG4 to reveal which dimeric G-quartets preferentially pair to form G4 structures. The results showed that the stability order was hTrG4 > hThG4 > hTdG4 (Supplementary Fig. S5). This result suggests that if DNA and RNA dimeric G-quartets are annealed, hTrG4 will be preferentially formed rather than hThG4, further confirming the infeasibility of obtaining intermolecular hG4 by direct annealing [17, 23]. It can also explain why the opening of hG4 requires a greater external force than the opening of dG4 in previous single-molecule experiments [49]. It is also worth noting that even for intermolecular G4s with the same chemical composition, differences in strand break positions may affect the structural stability of the G4 (Supplementary Fig. S6). In the genome, adjacent G-quartets are the most likely to occur, resulting in the formation of a (2+2) hG4 structure. Consequently, previous studies frequently focused on the (2+2) structure, using it as a model [15, 19, 21, 24], and it has also been observed at telomere positions [14]. Therefore, our research prioritizes the (2+2) structure. For simplicity, unless otherwise specified, hThG4 in this article refers specifically to the (2+2) form.

Figure 2.

Stability of hG4s with varying RNA and DNA compositions. (A) Schematic diagrams and sequences of hThG4(3+1)-s and hThG4(1+3)-s. The suffix “-s” indicates that the corresponding substrate has been split by UV-induced cleavage. (B, C) CD spectra of hThG4(3+1) and hThG4(1+3) before and after 10 min of UV induction in K100 buffer. (D) Temperature-dependent CD signals at 265 nm for hThG4(3+1)-s, hThG4-s, and hThG4(1+3)-s, with calculated melting temperatures (T_m) of 42.3°C ± 0.3°C, 56.9°C ± 0.2°C, and 58.8°C ± 0.2°C, respectively.

G4P binds to and stabilizes hG4

To identify G4 structures in cells and study their dynamics, various G4-specific binding proteins, including antibodies and engineered proteins, have been developed [31–35]. However, the binding affinity of these proteins to hG4 remains unclear. To identify the optimal hG4 recognition protein, we first expressed and purified BG4, D1, SG4, and G4P (Supplementary Fig. S7A), and confirmed their ability to recognize dG4 structures using electrophoretic mobility shift assay (EMSA) (Supplementary Fig. S7B). Notably, this study represents, to the best of our knowledge, the first direct comparison of these G4 recognition proteins under identical conditions. Our results revealed that G4P exhibited the highest affinity for dG4 structures (Supplementary Fig. S7C). We labeled hThG4 with Cy5, hereafter referred to as hThG4-Cy5, and subsequently generated hThG4-s-Cy5 via UV-induced photolysis. Using the same method, we assessed the binding of these proteins to hThG4-s-Cy5. The results demonstrated that, in addition to exhibiting the strongest binding to dG4, G4P also displayed the highest affinity for hG4 (Fig. 3 and Supplemenatry Table S2). This finding explains why a substantial amount of hG4 is detected when using G4P to identify genomic dG4 in living cells [34]. Moreover, T_m measurements indicated that G4P binding enhances the stability of hThG4 (Supplementary Fig. S8). The ability of these proteins to bind to hThG4-s-Cy5 indicates the need for caution in imaging experiments aimed at detecting dG4s in chromosomes, as interference from hG4s must be carefully identified [31, 33].

Figure 3.

Binding of BG4, SG4, D1, and G4P to hThG4-s-Cy5. EMSA results and binding quantifications. The data from three replicates were fitted using the Hill equation, with values summarized in Supplementary Table S2.

RHUA and Pif1 can efficiently unfold hG4, but RecQ and DinG cannot

The accumulation of DNA:RNA hybrids, including hG4 structures, is toxic to cells and must be addressed promptly by RNase H and specialized helicases [36–38]. Since RNase H is ineffective against stable hG4s [24], helicases are likely to play a critical role. However, no proteins capable of handling these structures have been reported so far [39, 40]. The core component of G4P originates from the RSM domain of the RHAU helicase [34], suggesting that RHAU may be pivotal in unwinding hG4 structures. Furthermore, given that helicases capable of unwinding G4 structures and DNA:RNA hybrid duplexes are essential for transcription and telomere stability (functional overlap with hG4s [8, 17]), we selected helicases from the Pif1 [50–53], RecQ [54], and Fe-S [55–57] families to assess their ability to unwind hG4 structures.

We first expressed and purified four representative proteins, referred to as RHAU, Pif1, RecQ, and DinG (Supplementary Fig. S9A), and verified their activity by dG4 unwinding (Supplementary Fig. S9B and C). Considering the differing unwinding polarities of these helicases, we designed hThG4 substrates with 15T tails at either the 5' or 3' end, naming them 15T-hThG4-Cy5 and Cy5-hThG4-15T, respectively (Supplementary Fig. S10A). Using the same method as described in Fig. 1, we performed denaturing PAGE and observed that these substrates underwent UV-induced cleavage, with the cleavage ratio exceeding 95% within 2 min (Supplementary Fig. S10B). CD spectroscopy revealed that the intermolecular G4 structures remained intact after split, retaining parallel configurations (Supplementary Fig. S10C), indicating that the tail sequences did not affect the topology of hThG4. Native PAGE analysis further confirmed that the G4 structure remained stable after induction (Supplementary Fig. S10D). Moreover, once the hG4 structure was disrupted by heat, it was detectable in native PAGE (Supplementary Fig. S10D). These findings demonstrate that the designed substrates are suitable for helicase unwinding experiments.

With the above preparations, we used Pif1 and DinG to unwind the 15T-hThG4-s-Cy5 substrate with a 5' single-stranded tail, and RHAU and RecQ to unwind the Cy5-hThG4-15T-s substrate with a 3' single-stranded tail, considering their unwinding polarity [58–61] (Fig. 4A). The experimental results showed that Pif1 and RHAU effectively unwound the hThG4 structure, while DinG and RecQ exhibited very weak helicase activity toward the hThG4 structure (Fig. 4A and B), though DinG and RecQ can efficiently unwinding dG4 structure (Supplementary Fig. S9B). This observation suggests that hG4 may represent a structural form distinct from dG4 in cells. Further analysis revealed that the unwinding of hG4 by Pif1 and RHAU is an ATP-dependent active process (Supplementary Fig. S11A) and relies on the tail strand to load (Supplementary Fig. S11B). Although Pif1 cannot bind and translocate on ssRNA [51], RHAU can bind ssRNA and unwind hThG4 structures (Supplementary Fig. S12). Non-helicase proteins also play important roles in unwinding G4 structures [40], for example, replication protein A (RPA) is particularly crucial in G4-related processes via efficiently unfolding dG4 [62]. Our study revealed that RPA can bind to 15T-hThG4-s-Cy5 and Cy5-hThG4-15T-s but cannot unfold either of these structures (Supplementary Fig. S13).

Figure 4.

Pif1 and RHAU effectively unwind hG4 structures, while DinG and RecQ do not. (A) PAGE images showing the unwinding of 15T-hThG4-s-Cy5 by Pif1 and DinG, and Cy5-hThG4-15T-s by RHAU and RecQ. Heat was used to disrupt the intermolecular structure, confirming that the substrate was successfully cleaved upon induction. (B) Quantitative unwinding fractions based on three replicates from panel (A).

Pif1 and RHAU can unwind hG4 structures under transcriptional conditions, but with differential sensitivity to G4 stabilizers

Through the above experiments, we demonstrated that G4P can recognize and stabilize hG4 structures (Fig. 3 and Supplementary Fig. S8), while the helicases RHAU and Pif1 can unwind hG4 structures (Fig. 4). To validate these findings in biological systems, we selected a commonly used transcription system for our studies (Supplementary Fig. S14A) [16, 19, 24, 42]. We did not employ the same DNA and RNA sequences used in Figs. 1–4 in the transcription experiments (Fig. 5). This decision was based on preliminary attempts to use those sequences under transcription conditions, during which we found that the resulting hG4 structures were susceptible to degradation by RNase A and RNase H. In the presence of KCl, this system primarily produces two types of hG4 structures during transcription, both of which are resistant to RNase A and RNase H (Fig. 5A, red arrow), consistent with previous reports [24]. The slower-migrating band may correspond to a three-layer hG4 structure, while the faster-migrating band may correspond to a two-layer hG4 or some other intermediate state. Furthermore, to provide additional evidence that the observed transcriptional bands correspond to the formation of hG4 structures, we replaced the post-transcriptional proteinase K treatment with G4P addition. Without proteinase K treatment, only a single prominent band was observed, which may be due to RNA polymerase binding masking the mobility difference between the two structures (Supplementary Fig. S14B, lane 2). The resulting higher-molecular-weight band observed after G4P adding in EMSA strongly supports the formation of hG4 (Supplementary Fig. S14B, lane 3). As the concentration of RHAU increases, the hG4 structure is gradually unwound [Fig. 5A (top panel) and 5B], indicating that RHAU can resolve hG4 structures formed during transcription. As a control, RecQ was unable to unwind the transcription-generated hG4 structure under the same conditions (Supplementary Fig. S14C).

Figure 5.

hG4 structures formed during transcription are unwound by RHAU and Pif1 and stabilized by G4P and PDS. RHAU unwinding of transcription-induced hG4 structures (A) and quantitative unwinding fractions from three replicates (B). Black arrows indicate dsDNA substrates for transcription (labeled as dsDNA, lane 1) and red arrows indicate hG4-containing nucleic acids produced in KCl (labeled as transcription in KCl, lane 3), which are resistant to RNase A and RNase H digestion (labeled as treatment with RNase A/H, upper panel, lane 4). The hG4s are absent in LiCl (labeled as transcription in LiCl, lane 2). Lanes 5–9 show electrophoretic results with increasing RHAU concentrations. The middle and bottom panels show the effects of G4P (1 μM) and PDS (2 μM) on unwinding, respectively. The label “G4P” indicates that G4P was added to the transcription system in KCl without the addition of RHAU. Similarly, the label “PDS” represents the condition in which PDS was added in KCl, but RHAU was not included. Pif1 unwinding of transcription-induced hG4 structures (C) and quantitative unwinding ratios from three replicates (D). The experiments in panel (C) used the same conditions as panel (A) but replaced RHAU with Pif1.

To investigate the function of G4P in the transcription system, we added 1 μM of G4P under the same conditions. This concentration is sufficient to bind nearly all hThG4 (Fig. 3). The results showed that G4P inhibited the helicase activity of RHAU at various concentrations [Fig. 5A (middle panel) and 5B]. Additionally, we observed that the G4 stabilizer PDS not only binds to hG4 structures (Supplementary Fig. S15) but also locks the two hG4 structures into a single species (Fig. 5A, bottom panel, lanes 3 and 4). The concentration of PDS added in the transcription system was 2 μM. This concentration was selected based on our MST results (Supplementary Fig. S15), which showed that hG4 structures are nearly saturated by PDS at this level. Under the same experimental conditions, PDS exhibited stronger inhibition than G4P (Fig. 5B). It is worth noting that G4 stabilizers, which can induce double-strand breaks through an R-loop-dependent mechanism [63], may also bind to hG4 structures, particularly considering the higher propensity for hG4 formation compared to dG4s [12, 13]. We then used the same approach to study Pif1’s role in hG4 unwinding during transcription, as well as the inhibitory effects of G4P and PDS on Pif1-mediated hG4 unwinding (Fig. 5C and D). The results showed that Pif1 unwinds hG4 structures formed during transcription. PDS inhibits this unwinding, while G4P does not. The inability of G4P to inhibit Pif1’s helicase activity may stem from Pif1’s ability to efficiently displace nucleic acid-bound proteins [64, 65], even those with strong interactions such as biotin–streptavidin [58]. To clarify whether the inhibitory effect of PDS on helicase activity stems directly from its interaction with G4 structures—rather than through some other aspect of the reaction system independent of G4 [66]—we examined the effect of PDS on Pif1-mediated unwinding of dsDNA, and the results showed that PDS had minimal impact on this process (Supplementary Fig. S16). These experiments demonstrate that both proteins and small molecules can directly modulate the unfolding and folding of hG4 structures, thereby influencing biological processes.

Discussion

hG4s have been identified in both eukaryotic and prokaryotic cells, where they play regulatory roles in transcription [8] and telomere stability [17]. However, limited research has been conducted on the stability and dynamic regulation of hG4s. In this study, we developed a method to obtain uniform intermolecular hG4s, investigated the stability relationships among different hG4 components, and identified factors that either promote or inhibit hG4 dynamics during transcription.

Due to the absence of established methods for obtaining intermolecular hG4s, we introduced a PC-linker-induced cleavage approach to generate uniform intermolecular G4s (Fig. 1). Using this technique, we analyzed the topology and stability of hG4s at varying DNA: RNA ratios (Figs 1 and 2). Our findings revealed that these hG4s predominantly adopt parallel structures (Figs 1 and 2). With further research on hG4s of diverse sequences, other topologies may emerge in the further. As rG4s were initially believed to fold exclusively into parallel structures, subsequent studies revealed additional topologies [67]. For hG4s, we demonstrated that increasing the RNA proportion enhances thermal stability (Fig. 2D and Supplementary Figs S3 and S4). This trend was also observed in (2+2) intermolecular structures of rG4s, hG4s, and dG4s (Supplementary Fig. S5C). Admittedly, this method has certain limitations. For instance, it is not suitable for studying the folding dynamics of hG4s. It can only be applied to investigate the stability of hG4s and their interactions with proteins or ligands in vitro and is not applicable to intracellular studies. However, the method we established for generating intermolecular hG4s can be applied to investigate the thermodynamics of various intermolecular G4s and provides a framework for obtaining other intermolecular structures.

hG4-recognition proteins are essential tools for studying hG4s both in vitro and in vivo. We confirmed that multiple G4-binding proteins can bind to hG4s (Fig. 3), underscoring the need to account for hG4 interference when using these proteins in dG4 and rG4 studies. Among the proteins tested, G4P exhibited the highest affinity for hG4s and effectively stabilized their structures (Fig. 3 and Supplementary Fig. S8). Helicases are crucial for resolving G4 structures and preventing their accumulation [40], yet no helicases specific to hG4s have been reported. We developed a method to measure hG4 unwinding activity and evaluated several candidate helicases (Fig. 4), including Pif1, RHAU, RecQ, and DinG. Our results showed that Pif1 and RHAU efficiently unwind hG4 structures, while RecQ and DinG do not (Fig. 4). Further studies are needed to validate the hG4-unwinding activity of additional helicases, particularly those capable of unwinding both hybrids [36, 38] and G4s [40]. This system establishes a foundation for future research on the hG4-unwinding activity of other proteins. Our next step will not only confirm their unwinding ability in cells [13] but also focus on using single-molecule techniques and structural methods to uncover the mechanisms underlying helicase-mediated hG4 unwinding. Identifying natural hG4-specific binding proteins through omics approaches and investigating their effects on hG4 stability will also be the focus of our future research.

Finally, we used a transcription system to evaluate the effects of stabilizers and helicases on hG4s. We found that Pif1 and RHAU facilitate hG4 clearance in the transcription system, while G4P and PDS act as stabilizers (Fig. 5). This system provides a platform for identifying hG4 stabilizers and unwinding proteins and paves the way for regulating biological processes through the modulation of hG4 structures within cells.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was sponsored by National Natural Science Foundation of China (32171272), Natural Science Foundation of Henan (242300421103), Guizhou Province Science and Technology Support Program—Novel Coronavirus Special Project (QKHJC(2020)4Y204), Guizhou Provincial Science and Technology Plan Foundation (QKHJC-ZK[2023]-520 and QKHJC-[2024]QingNian325), and Zunyi Medical University Doctoral Start-up Fund (F939).

Data availability

The data underlying this article are available in the article and in its online Supplementary data.