Assessing G‑Quadruplex Stability and Nonspecific Interactions in Living Cells Using ¹⁹F NMR

Abstract

G-quadruplex (G4) structures play key physiological roles and serve as functional building blocks in DNA/RNA-based aptamer biosensors. Their function critically depends on their structural stability and nonspecific interactions in cells. However, the behavior of exogenously delivered G4s in living cells remains poorly understood. Here, using ¹⁹F NMR, we monitored three representative G4 sequences, HT, hVEGFP, and TBA, in intact Xenopus laevis oocytes, HeLa cells, and derived lysates. We found that G4 structural stability, interactions, and NMR signal detectability are cell-type-dependent. In oocytes, all three G4s maintain detectable folded signals, with TBA showing slight time-dependent degradation. In HeLa cells, folded signals are absent for all three G4s due to intracellular interactions. HT and TBA undergo degradation in the cytoplasm rather than in the nucleus, whereas hVEGFP uniquely remains protected from degradation owing to both its higher thermal stability and protective interactions. These findings highlight the non-negligible effect of the cellular environment and the necessity of studying G4s in their native cellular context. Furthermore, we demonstrated that cell lysates can serve as a practical platform for reflecting G4 stability in intact cells. Collectively, this work deepens our understanding of G4 behavior in cells and provides a basis for improvement of G4-based agents.

Introduction

G-quadruplexes (G4s) are noncanonical, higher-order nucleic acid structures formed by guanine-rich sequences. These structures are stabilized through the planar arrangement of four guanines into a G-tetrad via Hoogsteen hydrogen bonding, with subsequent vertical stacking of these tetrads coordinated by monovalent cations such as potassium (K⁺). , Bioinformatic analyses have identified approximately 376,000 potential DNA G4-forming sequences in the human genome. Prevalent in functionally critical genomic regions like gene promoters, telomeres, and regulatory elements, G4 structures play significant roles in regulating gene expression and maintaining genomic integrity, making them promising therapeutic targets. − In parallel, serving as functional building blocks in DNA/RNA-based aptamer biosensors, G4-forming aptamers (e.g., AS1411) have emerged as potential tools for in vivo biosensing, targeted delivery, and theranostics. −

The intracellular environment is complex. In living cells, macromolecules (at 200–400 g/L) can occupy 20–40% of the cellular volume. , As a result, nonspecific interactions (e.g., electrostatic forces between the highly negatively charged G4 backbone and intracellular biomacromolecules) are widespread, and G4s simultaneously face a risk of nuclease degradation in living cells. − Importantly, unlike endogenous G4s that are dynamically regulated by specialized helicases and binding proteins, exogenously delivered G4-forming oligonucleotidesused as therapeutic agents or biosensing probeslack such regulatory mechanisms. Thus, for these exogenous G4s, the overall cellular environment may compromise their structural stability, potentially leading to decreased efficacy of G4-based drugs, off-target effects, and consequently increased toxicity. Therefore, monitoring the stability of exogenous G4s and quantifying their nonspecific interactions within the native cellular environment is critical for the application and optimization of nucleic acid therapeutics.

In-cell nuclear magnetic resonance (in-cell NMR) spectroscopy is a powerful tool to monitor the structure, dynamics, interactions, and stability of specific biomolecules in living cells in a nonperturbative manner. − In particular, ¹⁹F in-cell NMR offers distinct advantages owing to the properties of the ¹⁹F nucleus: 100% natural abundance, high NMR sensitivity (approximately 83% that of ¹H), negligible background signal in biological systems, and chemical shift sensitivity to its microenvironment. , Consequently, ¹⁹F NMR has been widely applied in the in situ study of biomolecules in living cells. To date, several ¹⁹F and ¹H in-cell NMR studies have characterized the conformations of G4-forming oligonucleotides, such as telomeric DNA G4, RNA-DNA hybrid G4 and their specific interactions with ligands in living Xenopus laevis oocytes (X. laevis oocytes) and HeLa cells, demonstrating that exogenous G4s can exist in living cells. ,− However, it remains unknown what the stability of exogenous G4s and their nonspecific interactions are like in living cells, and whether these properties are cell-type dependent.

In this study, we selected three G4-forming sequences with distinct topologies (Figure A-B): the modified human telomeric DNA (HT) with a hybrid-type fold; the human vascular endothelial growth factor promoter sequence (hVEGFP) with a parallel-type G4; and the thrombin-binding aptamer (TBA) with an antiparallel G4. , Using site-specific incorporation of a single fluorine atom via 2′-fluoro-RNA, we systematically evaluated the structural stability and nonspecific interactions of these G4s in living X. laevis oocytes and human HeLa cells, as well as in various cell lysate environments, by ¹⁹F NMR. Our work reveals that the behavior of G4s in cells, while related to their intrinsic properties such as thermal stability, exhibits significant cell-type dependence.

1.

Sequences, schematic structures, and NMR spectra of modified human telomere (HT), human vascular endothelial growth factor promoter (hVEGFP), and thrombin binding apatmer (TBA). (A) Sequences of HT, hVEGFP and TBA showing fluorine-modified positions. Sugar structure of unmodified DNA and 2′F-RNA (2′-deoxy-2′-fluoro-ribonucleic acid) labeled DNA. (B) Schematic structures of TH, hVEGFP and TBA showing fluorine-modified position. Guanines are colored based on their syn (magenta) and anti (cyan) conformations. (C) ¹⁹F spectra of HT, hVEGFP and TBA in buffer and living cells. The ¹⁹F spectra of three G4s in oocytes were derived from the initial 4-h data of the time-dependent measurements. The resonance at −203.0 ppm corresponding to degraded fragments is marked with a red dashed line.

Materials and Methods

In Vitro Sample Preparation

FAM-labeled, wild-type and ¹⁹F-labeled DNA sequences were synthesized and HPLC purified by Sangon Biotechnology Co., Ltd. (Shanghai, China). The DNA oligonucleotides were further dialyzed to remove some small-molecule impurity before use. DNase I was purchased from Takara Biotechnology Co., Ltd. (Dalian, China). TriDye Ultra Low Range DNA Ladder was purchased from New England Biolabs (NEB, USA).

NMR samples of G4 oligonucleotides in buffer or cell lysates were prepared by dissolving DNA oligonucleotides in buffer or in cell lysates containing 10% D₂O. The DNA concentration was 50 μM. The MC buffer mimicking the cellular environment comprises 25 mM HEPES, 10.5 mM NaCl, 110 mM KCl, 130 nM CaCl₂, 1 mM MgCl₂, pH 7.5, 10% D₂O. ,

The cell lysates from different cells were prepared using different methods. For X. laevis oocytes, the lysates were prepared as described. Briefly, the oocytes were mechanically crushed on ice using a plunger, and centrifuged at 15000 × g for 20 min at 4 °C. For yeast lysates, the cells were harvested and washed twice using ultrapure H₂O. Cell pellets suspended in ultrapure H₂O were lysed using a Nano Homogenizer (ATS Engineering Inc., AH-BASIC) and lyophilized. For E. coli lysates, the cells were harvested, sonicated on ice and lyophilized. For HeLa, HEK293T and A2780 cell lysates, cells were harvested and washed twice with PBS. Cell pellets suspended in ultrapure H₂O were sonicated on ice for 10 min and lyophilized. All lyophilized lysates were stored at −80 °C. The cell lysates of desired concentration were prepared by dissolving lyophilized lysates powder in H₂O or in buffer.

HeLa nuclear lysates were prepared as previously described. Briefly, harvested HeLa cells were pelleted by centrifugation at 300 × g for 5 min at 4 °C, washed twice with ice-cold PBS, and then resuspended twice in 0.1% NP-40/PBS with 10 gentle pipetting strokes. After each resuspension, the suspension was centrifuged at 1,000 × g for 5 min at 4 °C, and the cytoplasmic supernatant was discarded. The nuclei-enriched pellet was subsequently washed three times with ice-cold PBS to remove residual NP-40. For each wash, the pellet was gently resuspended and centrifuged at 1,000 × g for 5 min at 4 °C. The final nuclear pellet was suspended in ultrapure water, sonicated on ice for 10 min, and then lyophilized. HeLa nuclear lysates at desired concentrations were obtained by dissolving the lyophilized powder in H₂O or MC buffer.

In-Cell Sample Preparation

X. laevis oocytes were prepared by microinjection as described. Briefly, each oocyte was injected with ∼20 nL of ∼5 mM ¹⁹F-labeled DNA using a IM-300 microinjector (Narishige Co. Ltd., Tokyo, Japan). The healthy injected oocytes (150 or more) were carefully placed into a Shigemi NMR tube containing ND96 buffer (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES, pH 7.4) plus 10% D₂O.

The samples for DNA in HeLa cells were prepared by electroporation as described by Theillet et al. Cells were cultured at 37 °C in humidified 5% (v/v) CO₂ incubators, using high glucose DMEM media supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% penicillin-streptomycin. Cells were detached from culture flasks with 0.25% trypsin, centrifuged at 200 × g for 5 min at room temperature, washed once with PBS and counted on a hemocytometer. FAM- or ¹⁹F-labeled DNA were dissolved in fresh electroporation buffer (EPB: 100 mM sodium phosphate, 5 mM KCl, 15 mM MgCl₂, 15 mM HEPES, 5 mM ATP, 5 mM reduced glutathione, pH 7.0) to a final concentration of 0.3 and 0.6 mM, respectively, and then sterile-filtered. Cells (4 × 10⁶) were pelleted and suspended in 100 μL sterile-filtered EPB containing labeled DNA. The mixtures were transferred into a 100 μL Nucleocuvette Vessel (Lonza) and electroporated using an Amaxa 4D-Nucleofector (Lonza). The pulse program applied to HeLa cells was CN114. Cells were pulsed three times with gentle mixing between each pulse. After electroporation, the mixtures were transferred to a MatTek 35 mm cell culture Petri dish or a T75 cell culture flask filled with 2 or 15 mL of prewarmed culture medium, respectively. The electroporated cells were placed in the incubator to recover for 5 h. For in-cell NMR experiments, the recovered cells were washed three times with prewarmed PBS and detached from the Petri dish or culture flask by 0.125% trypsin. After centrifugation at 200 × g for 5 min at 25 °C, cells were washed once using Dulbecco’s modified Eagle’s medium (DMEM) and counted through a hemocytometer. 2.5 × 10⁷ cells were gently resuspended in DMEM containing 5 mM HEPES (pH 7.2), 90 mM d-glucose, and 10% D₂O to yield a 300 μL of cell suspension for in-cell NMR experiments. After in-cell NMR experiments, supernatants were collected immediately to assess leakage as described previously. Cell pellets were resuspended in 0.5 mL H₂O and sonicated on ice for 20 min to prepare lysates samples.

Circular Dichroism Spectroscopy

The sample was prepared by dissolving DNA oligonucleotides in the MC buffer to a final concentration of 25 μM. CD Spectra and melting curves were acquired on a Chirascan spectropolarimeter (Applied Photophysics Limited. UK) equipped with a Quantum Northwest TC125 temperature controller. The wavelength was varied from 220 to 320 nm at 25 °C. The step size was 1 nm. The dwell time was 0.5 s/nm, and the bandwidth was 1 nm. Melting curves were determined at 288 nm for HT, 294 nm for TBA, and 263 nm for hVEGFP at a rate of 0.5 °C/min.

NMR Spectroscopy

Spectra were recorded in 5 mm Shigemi NMR tubes (Allison Park, PA, USA) on Bruker spectrometers operating at 600 or 700 MHz equipped with a H/F/C/N cryoprobe and on 850 MHz equipped with triple resonance 5 mm HCN-cryoprobe. For in-cell NMR experiments, spectra were acquired at 295 K. For time-resolved ¹⁹F NMR experiments on lysate samples, spectra were acquired at the optimal growth temperature for each cell type: 310 K for bacterial and mammalian lysates (E. coli, HEK293T, A2780, HeLa whole-cell, and HeLa nuclear), 303 K for yeast lysates, and 295 K for oocyte lysates. For studies of G4 nonspecific interactions in cell lysates, whole-cell and nuclear lysates were first prepared at desired concentrations by dissolving lyophilized lysate powder in MC buffer. Subsequently, the G4 oligonucleotides (HT, hVEGFP, and TBA) were then added to the lysate solutions to obtain the NMR samples. All spectra were acquired at 295 K. This reduced temperature, combined with the high K⁺ concentration of the MC buffer, was employed to minimize G4 degradation and ensure sample stability throughout the NMR measurements.

In mammalian cell lysates, the degradation of the three G4s were quantitatively analyzed to compare their degradation rates. Degradation was quantified using the signal intensity of the degraded state at −203.0 ppm, and was considered complete when this signal stopped increasing.

One-dimensional ¹H spectra were acquired with a 2 s relaxation delay using a watergate pulse sequence with gradients (W5) for water suppression. ¹⁹F spectra were acquired with a 2 s relaxation delay and chemical shifts are referenced to trifluorotoluene at −63.72 ppm. Data were processed and analyzed with Topspin 3.1.

Fluorescence Imaging

X. laevis oocytes containing FAM-HT were washed 3 times for 10 min each with PBS, fixed in 4% paraformaldehyde for 12 h, and dehydrated with 15% sucrose for 12 h, followed by dehydration with 30% sucrose for another 30 h. Slices (60 μm) of fixed oocytes were imaged on an Olympus BX61 microscope equipped a 10× objective using green channel excitation.

HeLa cells containing FAM-HT, FAM-TBA, or FAM-hVEGFP were washed three times with PBS, 10 min each wash. Nuclei were stained with 150 μM Hoechst 33342. Confocal images were acquired using a Nikon A1 laser-scanning microscope equipped with a 60× oil objective, with excitation at 405 and 488 nm.

Degradation Experiment Using DNase I

HT was incubated with DNase I (2000 U/mg) in 10% MC buffer at a final concentration of 1000 U per 500 μL reaction. The reaction mixture was maintained at 37 °C for 1 h. For electrophoretic analysis, an aliquot (2 μL) of the digested product was mixed with 6× DNA loading buffer and resolved by 20% native polyacrylamide gel electrophoresis (PAGE) in 1× TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA, pH 8.3). Electrophoresis was performed at a constant voltage of 150 V for 2 h at 4 °C. Following electrophoresis, the gel was stained with ethidium bromide (EB) for 15 min and visualized under UV light to confirm the absence of the intact G4 band. The remaining digestion mixture was then supplemented with 10% D₂O (v/v) for ¹⁹F NMR analysis.

Proteinase K Digestion of Cell Lysates

Proteinase K was prepared as a 20 mg/mL stock solution and stored at −20 °C until use. An appropriate volume of the stock solution was added to the cell lysate to yield a final working concentration of approximately 200 μg/mL. The digestion mixture was incubated in a water bath at 58 °C for 2 h. Subsequently, the mixture was heated at 95 °C for 10 min to inactivate Proteinase K.

Results and Discussion

NMR and CD Analysis of the Three G4s in Buffer

To minimize perturbation of the tetrad unit, we substituted one anti guanine in each sequence with 2′-deoxy-2′-fluoro guanine (^2′FG), − that is, ^2′FG23 for HT, ^2′FG20 for hVEGFP, and ^2′FG11 for TBA (Figure ). The ¹H NMR and CD experiments were performed to investigate the native and ¹⁹F-labeled G4s structure and thermal stability in the MC buffer mimicking the intraoocyte environment. The native three G4 sequences yield almost the same ¹H spectra with those reported previously, respectively. − The CD spectra indicated a negative band at 240 nm and positive bands at 265 and 290 nm for native HT, a negative band at 240 nm and a positive band at 265 nm for native hVEGFP, and a negative band at 265 nm and positive bands at 250 and 295 nm for TBA, suggesting the formation of K⁺-stabilized hybrid, parallel and antiparallel G4s structure in MC buffer, respectively (Figure S1). The ^2′FG-labeled G4s showed the similar imino-¹H NMR, CD spectra and CD melting curves to the corresponding native G4s, indicating that ^2′FG labeling has an insignificant effect on the overall structure and thermal stability.

¹⁹F NMR Spectra of HT, hVEGFP, and TBA in X. laevis Oocytes and HeLa Cells

To probe conformation, stability, and interactions in a cellular environment, three ¹⁹F-labeled G4 oligonucleotides were introduced into X. laevis oocytes by microinjection and into HeLa cells by electroporation, followed by acquisition of in-cell NMR spectra (Figures , S2 and S3). For ^2′FG23-HT in oocytes, the ¹H spectra displayed severely broadened imino proton resonances (∼10–12 ppm) compared with those in dilute solution (Figure S2A), attributable to magnetic susceptibility, viscosity, and weak interactions within cells. , The limited resolution of the in-cell ¹H spectrum is insufficient to provide detailed structural information. In contrast, the ¹⁹F spectra were simplified by the single ¹⁹F label. The time-resolved ¹⁹F resonance of ^2′FG23-HT in oocytes consistently appeared at −199.8 ppm within 12 h (Figure S3A), identical to its chemical shift in buffer, suggesting that the oocyte environment does not perturb the conformation of HT and that the hybrid G4 structure persists stably. Unlike that in oocytes, no signal corresponding to the folded G4 state (at −199.8 ppm) was detected in HeLa cells; instead, a sharp ¹⁹F signal at −203.0 ppm was observed (Figures C and S2B).

The hVEGFP displayed a time-dependent decay of its initial resonance in X. laevis oocytes. The signal intensity at 8–12 h dropped to approximately one-third of that at 0–4 h (Figures C, S2C and S3B). This suggests that hVEGFP may undergo a slow transition to a state with restricted molecular mobility in a cell-context-dependent manner, potentially via macromolecular association or conformational rearrangement, and remains out of equilibrium over observed time windows. In HeLa cells, however, no ¹⁹F signal were detected in either the intracellular or supernatant fractions.

For TBA in X. laevis oocytes, the ¹⁹F spectrum remained dominated by the G4 resonance throughout the 12 h period, accompanied by a gradually increasing peak at −203.0 ppm (Figure C, S2D and S3C). In HeLa cells, however, the spectrum only displayed the sharp peak at −203.0 ppm, with no detectable folded G4 signal. Furthermore, this signal also appeared in the supernatant.

Assignment of the −203.0 ppm Resonance

We initially reasoned that the sharp peak at −203.0 ppm corresponds to a degradation signal. To test this, we performed a degradation experiment using DNase I and HeLa cell lysates. HT was first treated with DNase I and then incubated with HeLa cell lysates. Gel electrophoresis confirmed that HT was degraded by DNase I (Figure A), and the degraded product exhibited a resonance at −201.0 ppm in the ¹⁹F spectrum (Figure B). After subsequent addition of HeLa cell lysates (20 g/L), time-resolved ¹⁹F spectra revealed a narrow resonance at −203.0 ppm that became the dominant signal within 2 h. When the order was reversed, that is, incubating HT with HeLa cell lysates for 5 h before DNase I treatment, the narrow peak at −203.0 ppm that appeared upon lysate addition remained unchanged after DNase I exposure. These observations strongly indicate that the resonance at −203.0 ppm originates from degraded fragments, and that HeLa cell lysates degrade HT into smaller fragments than DNase I alone.

2.

Degradation of HT, GGG, and GGA analyzed by PAGE gel and ¹⁹F NMR spectroscopy. (A) PAGE gel of HT treated with DNase I. (B) Time-resolved ¹⁹F NMR spectra of HT treated sequentially in both orders: DNase I followed by HeLa lysates, and HeLa lysates followed by DNase I (HeLa lysates at 20 g/L). One spectrum was recorded every hour. (C) ¹⁹F NMR spectra of GGA and GGG treated with HeLa lysates (one spectrum per hour). The resonance at −203.0 ppm corresponding to degraded fragments is indicated by a a red dashed line.

To further characterize the size of the degradation fragments, we synthesized ¹⁹F-labeled G^2′FGA and GG^2′FG oligonucleotides and recorded their time-resolved ¹⁹F spectra in HeLa cell lysates (Figure C). In buffer, their ¹⁹F signals appeared near −201.2 and −201.6 ppm, respectively. Upon incubation with HeLa cell lysates, these original signals diminished and were replaced within 1 h by the narrow resonance at −203.0 ppm. These observations suggest that exogenous DNA in lysates may be processed down to single-nucleotide units, whose ¹⁹F signal corresponds to the resonance at −203.0 ppm.

Collectively, these in vitro degradation data, together with the in-cell NMR results, lead to the following conclusion: HT remains stable in X. laevis oocytes but undergoes degradation in HeLa cells, while TBA undergoes degradation in both cell types, although its G4 signal in oocytes remained dominant over the 12 h observation period.

Degradation Occurs in the Cytoplasm

To determine whether degradation occurs in the cytoplasm, nucleus, or both, we delivered the 5′-FAM-labeled G4s into HeLa cells and X. laevis oocytes for subcellular localization, respectively. Fluorescence imaging revealed predominant nuclear accumulation of exogenous DNA (Figures S4 and S5), consistent with previous reports on nuclear localization of other G4 and non-G4 nucleic acids. , Then we performed time-resolved ¹⁹F NMR analysis of them using HeLa cell nuclear lysates (Figure S6). Remarkably, no degradation-associated signals were detected in nuclear lysates throughout the monitoring period, indicating that the three G4s are stable in the nucleus, and its degradation in HeLa cells is primarily mediated by cytoplasmic nucleasesconsistent with the fact that the nucleases responsible for degrading exogenous DNA are predominantly localized in the cytoplasm.

Intracellular Interactions Mask G4 Signal While Protecting It from Degradation

In HeLa cells, no ¹⁹F signal was detected for hVEGFP in either the intracellular or supernatant fractions (Figures C and S2). We reasoned that this signal absence was not due to inefficient delivery of hVEGFP into HeLa cells, but rather resulted from strong interactions with intracellular components. These interactions both broaden the folded G4 signal beyond detection and protect hVEGFP from nuclease degradation. Additionally, its high thermal stability (Figure S1) is also responsible for the resistance of hVEGFP to nuclease degradation.

To test this hypothesis, we lysed the HeLa cells and recorded the ¹⁹F spectrum of the lysate using the same acquisition time as for the in-cell measurement (Figure A). We found that the G4 signal remained undetectable and a very weak degradation peak appeared near −203.0 ppm, indicating slight degradation occurred during cell lysis and NMR acquisition. We then treated the lysate with proteinase K to digest interacting proteins and release hVEGFP. After that, the degradation peak increased markedly, and meanwhile a weakly resonance corresponding to the G4 structure became visible, confirming that the hVEGFP G4 was indeed present in cells but NMR-invisible due to broadening and that no degradation had occurred in HeLa cells. Furthermore, time-dependent ¹⁹F spectra in lysates at different concentrations showed that higher lysate concentrations correlated with less degradation and greater G4 signal recovery (Figure B), indicating that stronger intracellular interactions confer greater resistance to degradation. Given the high concentration of macromolecules in the cellular environment, this concentration-dependent protection accounts for the resistance of hVEGFP to degradation in living HeLa cells.

3.

NMR spectra of hVEGFP in HeLa cell lysates. (A) ¹⁹F spectra of the lysates prepared from HeLa cells for hVEGFP in-cell NMR experiments. (B) Time-resolved ¹⁹F and ¹H spectra of hVEGFP in various concentrations of HeLa cell lysates at 310 K, showing that the interactions protect hVEGFP G4 structures. One ¹⁹F spectrum was recorded every 2 h.

A similar signal-masking phenomenon was observed for HT. In-cell ¹⁹F spectra acquired immediately after electroporationskipping the 5-h incubation to minimize degradationrevealed only minimal HT degradation (Figure S7). Yet upon subsequent lysis and proteinase K treatment, both pronounced degradation signals and a weak folded G4 resonance emerged, indicating that, like hVEGFP, the HT G4 signal was broadened beyond detection due to intracellular interactions in HeLa cell. Unlike hVEGFP, however, HT remained susceptible to degradation in HeLa cells, likely due to its relatively lower T _m and weaker intracellular interactions. TBA may follow the similar mechanism.

Lysates Recapitulate Relative G4 Stability

In-cell NMR results for the three G4s revealed that both stability and nonspecific interactions are cell-type-dependent and also vary among different G4 sequences within the same cell type. A key question, therefore, is whether these differential behaviors can be recapitulated in cell lysates. To address this, we employed time-resolved ¹⁹F NMR to systematically assess the stability and nonspecific interactions of three G4 sequences (HT, hVEGFP, TBA) in lysates derived from X. laevis oocytes and HeLa cells. Agreement between lysate and intact-cell results would validate the lysate model. Such validation would offer two major advantages: first, it would greatly reduce experimental costs, as in-cell NMR requires large quantities of ¹⁹F-labeled oligonucleotides and cells due to low delivery efficiency and poor sensitivity; and second, it would enable studies in cell types that are not amenable to in-cell NMR (e.g., those refractory to electroporation or streptolysin O treatment). Thus, the lysate system offers a broadly accessible platform for evaluating G4 stability and interactions across diverse cellular contexts.

We therefore examined the three G4s in 20 g/L oocyte and HeLa lysates using time-resolved ¹⁹F NMR (Figures , S8 and S9). Over 9 h of monitoring, no degradation signals were detected for HT or hVEGFP in oocyte lysates, whereas TBA underwent degradationa result fully consistent with observations in intact oocytes, where HT and hVEGFP are stable while TBA is relatively unstable. In HeLa lysates, all three G4s exhibited degradation, which appears to differ from the in-cell results (HT and TBA degrade, but hVEGFP does not). However, analysis of degradation rates revealed that hVEGFP degrades more slowly than HT and TBA in HeLa lysates, and this stability order is in complete agreement with the observation in intact HeLa cells. Thus, despite its inability to fully recapitulate the intact cellular microenvironment for quantitative rate comparisons, the lysate system preserves the relative stability hierarchy of G4s established in intact cells, validating its utility as a surrogate model for ranking G4 stability.

4.

Structural stability of human telomere in cell lysates. (A) Time-resolved ¹⁹F spectra of 50 μM ^2′FG23-HT in various cell lysates. Spectra were acquired at the optimal culture temperature for each cell type (E. coli, HEK293T, A2780 and HeLa cell lysates, 310 K; yeast, 303 K; oocytes, 295 K). One ¹⁹F spectrum was recorded every 1 h. (B) ¹H spectra of ^2′FG23-HT in cell lysates acquired after the time-resolved ¹⁹F experiments. (C) Comparison of ^2′FG23-HT degradation in various mammalian cell lysates.

Having validated the lysate system, we extended our stability study to lysates from multiple cell types to evaluate G4 behavior in diverse cellular contexts (Figures , S8 and S9). HT was stable in E. coli and Pichia pastoris (yeast) lysates, similar to that in oocyte lysates, but underwent degradation in mammalian HEK293T and A2780 cell lysates, with degradation rates following the order: HeLa > A2780 > HEK293T (Figure C). These results were corroborated by ¹H NMR, in which characteristic G4 imino signals (10–12.5 ppm) remained strong in E. coli, yeast, and oocyte lysates, were weaker in HEK293T lysates, and were absent in A2780 and HeLa lysates.

Like HT, hVEGFP was stable in E. coli and yeast lysates but degraded in mammalian cell lysates, following the same order of degradation rates: HeLa > A2780 > HEK293T (Figure S8). However, due to stronger intracellular interactions, its degradation was markedly slower than that of HT (Figure S10). TBA, by contrast, degraded in all lysates tested (Figure S9), with degradation rates exceeding those of HT and hVEGFP under identical conditions, reflecting its substantially lower stability (Figure S10). The degradation rates across lysates followed the consistent order: HeLa > A2780 > HEK293T > oocyte > E. coli > yeast.

Across all three G4s, a strong correlation emerged between structural stability and cell type: stability was consistently higher in E. coli, yeast, and oocyte lysates than in mammalian cell lysates. Notably, stability in lysates from cancer cell lines (HeLa and A2780) tended to be lower than in the noncancerous HEK293T cell line, a difference that may be partially attributed to variations in nuclease type or abundance across cell types, leading to distinct degradation capacities.

G4 Nonspecific Interaction in Cells and Cell Lysates

Ideally, the full width at half maximum (FWHM) offers a convenient readout of nucleic acid interactions in cells, provided that the signal of interest remains detectable and stable over time. In X. laevis oocytes, HT G4 exhibited a time-invariant ¹⁹F signal over 12 h, whereas TBA G4 exhibited a ¹⁹F signal with an invariant FWHM, although its intensity gradually decreased due to degradation (Figure S3). These behaviors support using FWHM to assess their nonspecific interaction strength, with measured values of ∼300 Hz for HT and ∼480 Hz for TBA, suggesting that TBA interacts more strongly than HT in the oocyte environment. By contrast, the hVEGFP G4 signal decayed continuously and remained out of equilibrium over 12 h without detectable degradation, consistent with a slow transition to an NMR-invisible state through restricted molecular mobility. Under these nonequilibrium conditions, FWHM is no longer a reliable indicator, as the residual signal reflects only the free or weakly bound population and thus underestimates the true interaction strength. The behavior of hVEGFP described abovecontinuous signal decay without degradationalready points to a stronger interaction than that observed for both HT and TBA.

We next tested whether this interaction hierarchy would be recapitulated in cell lysates. Notably, FWHM measured at a single, arbitrarily chosen concentration is unreliable for comparing interaction strengths across different G4s, because the underlying G4–lysate interactions are inherently nonlinear. At low concentrations, where interacting partners are limiting and G4s remain largely unbound, FWHM values substantially underestimate interaction strength and fail to resolve differences among distinct G4 sequences. By contrast, the lysate concentration at which signal loss occurs provides a direct and robust readout of interaction strength. Stronger G4–lysate interactions cause more extensive line broadening, leading to signal disappearance at a lower concentration threshold. Applying this criterion to oocyte lysates, we found that the hVEGFP signal broadened to near invisibility at 40 g/L, whereas TBA nearly disappeared at 100 g/L, and HT remained clearly detectable even at 100 g/L (Figure ). This establishes an interaction order of hVEGFP > TBA > HT, consistent with the hierarchy inferred from oocytes, indicating that the lysate model recapitulates the relative hierarchy of G4 interactions in X. laevis oocytes.

5.

¹⁹F NMR spectra of the three G4s in oocyte (A), HeLa (B), and HeLa nuclear lysates (C) at various concentrations recorded at 295 K.

In HeLa lysates, the same critical concentrations were observed40 g/L for hVEGFP and 100 g/L for TBA, with HT remaining detectable at 100 g/Lyielding an interaction order of hVEGFP > TBA > HT. In HeLa nuclear lysates, however, both HT and TBA nearly disappeared at 100 g/L, suggesting that HT interactions were markedly enhanced in the nuclear fraction, reaching a level comparable to that of TBA; the overall order (hVEGFP > TBA ≈ HT), however, remained unchanged (Figure ). Of note, because none of the three G4s yielded detectable structural states in intact HeLa cells, direct comparison with a living-cell hierarchy was not feasible. Although this prevents direct validation, the enhanced HT interaction in the nuclear fraction, relative to whole lysates, raises the possibility that G4 interaction hierarchies may differ between whole-cell and organelle-specific contexts.

Conclusions

In summary, using in-cell and lysate-based ¹⁹F NMR, we systematically compared the stability and nonspecific interactions of three G4 sequences (HT, hVEGFP, TBA) in X. laevis oocytes, HeLa cells, and derived lysates. We found that G4 behavior, including degradation, and NMR signal detectability, is cell-type-dependent. In oocytes, all three G4s maintain detectable folded signals, with TBA showing slight time-dependent degradation. In HeLa cells, by contrast, folded G4 signals are absent for all three sequences due to intracellular interactions. HT and TBA undergo degradation in the cytoplasm rather than in the nucleus, whereas hVEGFP is uniquely protected from degradation, owing to both its higher thermal stability (T _m) and protective interactions. These findings reveal that the cellular environment profoundly influences G4 NMR visibility and fate, and that the same sequence can behave differently across cell types, underscoring the necessity of studying G4s in their native cellular context. Nevertheless, in-cell NMR remains technically challenging and is currently limited to a few amenable cell models, including oocytes, HeLa, A2780, and HEK293T cells. Broadening its scope to more pathophysiologically relevant contexts will require further methodological advances in sample delivery, signal enhancement, and cell viability preservation.

The lysate-based approach offers a valuable complement to in-cell studies. Based on comparative analysis of in-cell and lysate data, we found that the lysate-based model can rank the stability of a given G4 across different cell lines, as well as compare stability and nonspecific interactions among different G4s within the same cell type. These findings provide both a rationale and a practical NMR-based method for using the lysate model as a low-cost, high-throughput initial screening platform for G4-based agents in early stage drug discovery. Collectively, this work expands our understanding of how cellular environments influence G4 fate and provides a valuable tool for assessing G4 stability in a cellular context.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c01086.

• NMR, CD, and melting curve of modified and wild-type G4s in MC buffer; fluorescence image of FAM-labeled G4s in X. laevis oocytes and HeLa cells; NMR spectra of G4s in living X. laevis oocytes, HeLa cells, and cell lysates (PDF)

Qiong Wu: Investigation, Methodology, Validation, Visualization, Writingoriginal draft; Wenwen Zheng: Investigation, Conceptualization, Validation, Visualization, Writingoriginal draft; Jiajing Zhao: Investigation; Chen Wang: Investigation; Xiaoli Liu: Investigation; Qing Liu: Investigation; Zhaofei Chai: Investigation; Jinbo Yu: Investigation; Yansheng Ye: Investigation; Guohua Xu: Conceptualization, Investigation, Validation, Visualization, WritingReview and Editing, Supervision, Funding Acquisition, Project administration; Maili Liu: Supervision, Funding Acquisition; Conggang Li: Conceptualization, Validation, WritingReview and Editing, Supervision, Funding Acquisition, Project administration.

This work is supported by National Natural Science Foundation of China grants (22474148, 21874149, 21605157); the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0540000, YSBR-068, and XDB0750100); and Ministry of Science and Technology of China grant (2021YFA1302600).

The authors declare no competing financial interest.