Characterization of human telomere RNA G-quadruplex structures in vitro and in living cells using ¹⁹F NMR spectroscopy

Abstract

Human telomeric RNA has been identified as a key component of the telomere machinery. Recently, the growing evidence suggests that the telomeric RNA forms G-quadruplex structures to play an important role in telomere protection and regulation. In the present studies, we developed a ¹⁹F NMR spectroscopy method to investigate the telomeric RNA G-quadruplex structures in vitro and in living cells. We demonstrated that the simplicity and sensitivity of ¹⁹F NMR approach can be used to directly observe the dimeric and two-subunits stacked G-quadruplexes in vitro and in living cells and quantitatively characterize the thermodynamic properties of the G-quadruplexes. By employing the ¹⁹F NMR in living cell experiment, we confirmed for the first time that the higher-order G-quadruplex exists in cells. We further demonstrated that telomere RNA G-quadruplexes are converted to the higher-order G-quadruplex under molecular crowding condition, a cell-like environment. We also show that the higher-order G-quadruplex has high thermal stability in crowded solutions. The finding provides new insight into the structural behavior of telomere RNA G-quadruplex in living cells. These results open new avenues for the investigation of G-quadruplex structures in vitro and in living cells.

INTRODUCTION

RNA structure influences the functions of nearly all classes of RNAs (1–3), including RNA G-quadruplexes, which are four-stranded RNA structures that have emerged as potential targets for drug design because of their biological importance (4–6). For example, RNA G-quadruplexes were recently reported to cause protein-dependent oncogene translation in cancer and neurodegenerative diseases (7,8). Recently, we and other groups demonstrated that human telomere RNA, a newly found telomeric repeat-containing RNA (9,10), forms G-quadruplex structures (11–14). We also found that telomere RNA G-quadruplex structures play an important role in providing a protective structure for telomere ends (15). Recently, some studies have also suggested that telomere RNA G-quadruplexes may form polymorphic higher-order G-quadruplexes comprising two stacked G-quadruplex subunits (16,17). However, direct evidence for human telomeric RNA higher-order G-quadruplexes existing in cells has not yet been obtained. Therefore, a more effective chemical approach for obtaining structural information on telomere RNA G-quadruplexes in living cells is desired.

In the present study, we employed fluorine-19 (¹⁹F) NMR spectroscopy of oligonucleotides with fluorine labels to further investigate the structural basis of telomere RNA G-quadruplexes in vitro and in living cells. ¹⁹F NMR has been successfully used to investigate important conformations of nucleic acids due to the high sensitivity of the ¹⁹F chemical shift to the environment, 100% natural abundance of ¹⁹F, absence of any natural background signals in RNA and cells and relative simplicity of ¹⁹F NMR spectra (18–32).

Thus, using the simplicity and sensitivity of ¹⁹F NMR spectroscopy, we directly observed the two-subunits stacked telomere RNA G-quadruplex in living cells, providing the in cell evidence for the presence of the higher-order G-quadruplex in human RNA at first time. The in-cell result suggested that molecularly crowded environment in cells promotes the higher-order G-quadruplex structure formation even at low RNA concentration. It is known that biomolecules function in a crowded intracellular environment. Molecular crowding could affect the structure, stability and conformational transition of biomolecules (33–35). We further demonstrated that telomere RNA G-quadruplexes are converted to the higher-order G-quadruplex under molecular crowding condition and further shown that the higher-order G-quadruplex has high thermal stability in crowded solutions. These results provide valuable information for understanding the structure and function of human telomere RNA.

MATERIALS AND METHODS

Sample preparation

RNA was incorporated a six-ethyl linker and 3,5 bis(trifluoromethyl)phenyl moiety at the 5΄ terminal on a 1.0 μmol scale using an automatic DNA/RNA synthesizer using solid-phase phosphoramidite chemistry. After automated synthesis, the oligomer was detached from the support, deprotected and purified by RP-HPLC using an appropriate linear gradient of 50 mM ammonium formate in H₂O and 50 mM ammonium formate in 1:1 acetonitrile/H₂O. The oligomer was desalted by NAP10 column (GE Healthcare) and identified by a matrix-assisted laser desorption/ionization-time-of flight mass spectrometer (MALDI-TOFMS) on an autoflex III smart beam mass spectrometer (negative mode), Calcd. 4376.58, Found. 4379.88.

CD measurement

Circular dichroism (CD) spectra were measured using a Jasco model J-820 CD spectrophotometer. The spectra were recorded using a 1 cm path length cell. Samples were prepared by heating the oligonucleotides at 95°C for 5 min and gradually cooling to room temperature. Solutions for CD spectra were prepared as 0.3 ml samples at a 10 μM strand concentration in the presence of 50 mM KCl, 10 mM Tris-HCl buffer (pH 7.0).

In vitro ¹⁹F-NMR spectroscopy

RNA samples of 0.2–5.0 mM were dissolved in 150 μl of designed solvent which containing 10 mM Tris-HCl buffer (pH 7.0) and 50 mM KCl. ¹⁹F NMR spectra were measured at a frequency of 376.05 MHz on a Bruker AVANCE 400 MHz spectrometer and were referenced relative to external CF₃COOH (−75.66 ppm). Experimental parameters were as follows: ¹⁹F excitation pulse 15.0 μs, spectral width 89.3 kHz, acquisition time 0.73 s, relaxation delay 1 s, number of scans 64 or 512 or 2048, an exponential window function with lb = 0.3 Hz was used, no zerofilling was used. At temperature-dependent experiment, the sample is kept for 10 min at each temperature.

In-cell ¹⁹F-NMR spectroscopy

In-cell NMR sample was prepared by direct microinjection 50 nl aliquot of the stock solution (3 and 5 mM of 12-mer RNA) into the oocyte cell (∼150 and ∼250 μM intracellular concentration). Approximately 150 Xenopus laevis oocytes were injected by using an IM-30 Electric Microinjector (NARISHIGE, Tokyo). After injection, the oocytes were transferred to a disposable dish, washed carefully with oocyte stocking buffer (15 mM Tris-HCl (pH7.6), 88 mM NaCl, 1 mM KCl, 0.4 mM CaCl₂, 0.3 mM Ca(NO₃)₂, 0.8 mM MgCl₂ and 2.4 mM NaHCO₃). The injected oocytes were transferred to a Shigemi tube (Shigemi 5 mm Symmetrical NMR microtube) and kept in an oocyte stocking buffer containing 10% of D₂O. Sample for measurement in lysate was prepared as follows: after the in-cell NMR measurements, oocytes were mechanically crushed and used for NMR measurement.

RESULTS AND DISCUSSION

In vitro ¹⁹F NMR analysis of telomere RNA G-quadruplexes

The present study is based on the concept that ¹⁹F NMR signals are strongly dependent on the structural environment of the ¹⁹F label. Thus, it should be possible to distinguish different RNA structures of the same sequence by the corresponding resonances of the different structures, such as single strands and G-quadruplexes (Figure 1). To achieve this goal, a 3,5-bis(trifluoromethyl)benzene moiety was introduced into 5΄ termini of oligonucleotide using phosphoramidite chemistry (Figure 2A and Supplementary Figures S1–S6). The ¹⁹F labeling of RNA with six equivalent ¹⁹F atoms was expected to afford the high ¹⁹F intensity. First, we examined the conformation of the oligonucleotide ORN-1 with the ¹⁹F label at 5΄ termini by CD spectroscopy. The CD spectrum of ORN-1 in the presence of K⁺ at 20°C showed a positive band at 265 nm and a negative band at 240 nm (Supplementary Figure S7), which are the characteristic CD signatures of a parallel dimeric G-quadruplex structure of RNA consistent with previously reported (11–14), suggesting that the ¹⁹F label at 5΄ termini does not affect the folding of G-quadruplex. Next, we performed a concentration-dependent experiment to investigate the structural behavior of ORN-1 for the formation of RNA G-quadruplex by ¹⁹F NMR (Figure 2B). The ¹⁹F NMR spectrum was firstly obtained at 0.2 mM RNA concentration. One signal was observed at −62.91 ppm in the presence of 50 mM KCl, indicating dimer G-quadruplex formation consistent with CD result. As shown in Figure 2B, a new signal appears as the RNA concentration increases (−62.57 ppm). The new signal is clearly observed at an RNA concentration of 1.5 mM, and at 3.0 and 5.0 mM its intensity becomes remarkably greater than that of the initial peak. Each ¹⁹F NMR signal arises as a result of the unique fluorine environment; thus, the presence of the two signals confirms the existence of two conformers of the telomere RNA. Accordingly, the two signals were assigned to the dimeric and two-subunits stacked G-quadruplexes in accord with previous studies that suggested two RNA G-quadruplex subunits are most likely to form a two-subunits stacked G-quadruplex (16,17).

Figure 1.

Concept for the detection different RNA structures by a ¹⁹F label. Three ¹⁹F resonances of different chemical shifts are expected according to single-stranded, dimeric G-quadruplex and two-subunits stacked G-quadruplex.

Figure 2.

¹⁹F NMR and ¹H NMR spectra of ¹⁹F labeled RNA. (A) Chemical structure of ¹⁹F labeled telomere RNA bearing 3,5 bis(trifluoromethyl)phenyl group at the 5΄ terminal. (B) ¹⁹F NMR spectra of ¹⁹F labeled telomere RNA at different concentrations. The peaks of the dimeric and two-subunits stacked G-quadruplex are marked with red and green spots, respectively. Concentrations of RNA indicated on the right. (C)¹⁹F NMR of ¹⁹F labeled RNA at different temperatures. Red and green spots indicated dimer and two-subunits stacked G-quadruplex. The peaks of single strand are marked with black spots. Temperatures indicated on the right. Condition: 3 mM RNA in 50 mM KCl and 10 mM Tris-HCl buffer (pH 7.0). The sample is kept for 10 min at each temperature. (D) ¹⁹F NMR of ¹⁹F labeled RNA at different temperatures and concentrations. The peaks of dimer G-quadruplex are red spots. The peak of two-subunits stacked G-quadruplex is green spot. The peaks of single strand are marked with black spots. (E) ¹H imino proton NMR of ¹⁹F labeled RNA corresponding to ¹⁹F NMR at different temperatures and concentrations. The peaks characteristic of the dimeric and two-subunits stacked G-quadruplex are marked with red and green spots, respectively.

As shown in Figure 3A, profiles of the relative peak areas of the ¹⁹F resonance signals versus concentration revealed that a higher RNA concentration promotes the formation of the two-subunits stacked G-quadruplex, suggesting that the of stacking two G-quadruplex subunits is a concentration-dependent manner.

Figure 3.

¹⁹F NMR shift versus concentration and temperature profiles. (A) Profiles of the relative peak areas of the ¹⁹F resonance signals versus concentration. Dimer and two-subunits stacked G-quadruplex conversions followed by ¹⁹F NMR spectroscopy. (B) Profiles of the relative peak areas of the ¹⁹F resonance signals versus temperature. Two-subunits stacked G-quadruplex/dimer G-quadruplex/single strand conversions followed by ¹⁹F NMR spectroscopy. To obtain the relative peak signal of each conformation, the total value of the relative peak signal for three conformations was estimated to be 1.0. Plotting the values of relative peak signal against temperature results in two melting curves for the two-subunits stacked Gquadruplex and dimer G-quadruplex.

A temperature-dependent experiment was then performed to confirm the assignment of the two signals observed in the ¹⁹F NMR spectrum (Figure 2C). As the temperature increased from 23°C to 75°C, the intensity of the signal at −62.57 ppm for the two-subunits stacked G-quadruplex decreased, while that of the signal at −62.91 ppm for the dimeric G-quadruplex increased, indicating the conversion of the two-subunits stacked G-quadruplex to the dimeric G-quadruplex at higher temperatures. Upon heating to 50°C, a new peak at –62.64 ppm corresponding to the unfolded single strand appeared, and at 75°C, only this peak remained with a strong intensity, whereas the peaks of the dimeric G-quadruplex (at −62.91 ppm) and two-subunits stacked G-quadruplex (at −62.57 ppm) completely disappeared. This result indicated that at high temperatures, both the dimer and two-subunits stacked G-quadruplexes unfolded to a single strand.

To further confirm that two subunit G-quadruplexes stack together to form a two-subunits stacked G-quadruplex structure, ¹H imino proton NMR analysis was performed. Figure 2E shows the spectrum for a 0.5 mM RNA solution. Six peaks assignable to the dimeric G-quadruplex are observed at 11.0–12.0 ppm that correspond to the peak of the dimeric G-quadruplex in the ¹⁹F NMR spectrum (Figure 2D) and are consistent with the results of previous studies (11–14). In the ¹H NMR spectrum of a 3 mM RNA solution at 23°C, new signals were observed in addition to major ones (Figure 2E), which was in agreement with the presence of two signals for the dimeric and two-subunits stacked G-quadruplexes in the ¹⁹F NMR spectrum (Figure 2D) and suggested that the new signals were due to the two-subunits stacked G-quadruplex structure. Upon heating to 60°C, the new signals disappeared and only the dimeric G-quadruplex peaks were detected in the ¹H NMR spectrum (Figure 2E). These observations agreed well with the ¹⁹F NMR spectrum at 60°C (Figure 2D), in which signal for the dimeric G-quadruplex appeared. Although signal for the unstructured single strand was observed at ¹⁹F NMR at 60°C (Figure 2D), the unstructured single strand does not exhibit any resonance peaks at 11.0–12.0 ppm of the ¹H NMR because this region only corresponds to the imino protons of the G-quartet. After complete denaturation (75°C and 85°C), the signals for the dimeric and two-subunits stacked G-quadruplexes were not observed in the ¹H NMR spectrum, which is in accordance with the ¹⁹F NMR spectrum, in which only the peak for the unstructured single strand was observed.

Encouraged by the ability to use ¹⁹F NMR spectroscopy to monitor the conformational transition behavior of the RNA telomere via observation of pronounced changes in the ¹⁹F resonances as a function of temperature, the melting process was characterized by plotting the relative peak areas of the ¹⁹F resonance signals at various temperatures (Figure 3B). The T_m values for conversion of the two-subunits stacked G-quadruplex to dimeric G-quadruplex and finally to the single strand were estimated to be 52.4°C and 67.8°C, respectively, suggesting stable two-subunits stacked G-quadruplex formation. Importantly, using the corresponding ¹⁹F signal curve, a highly precise T_m value was obtained for the two-subunits stacked G-quadruplex, which is not easy to do using other methods, such as CD and ultraviolet spectroscopy, because CD and ultraviolet spectra could not discriminate the higher-order G-quadruplex among different structures.

The presence of resonances for different conformations in the same spectrum at different temperatures allowed quantitative characterization of the thermodynamics of the process (Table 1, and Supplementary Figures S8–S10). The enthalpy change (ΔH) indicates that the stacking interaction of two G-quadruplex subunits is of an enthalpic origin (ΔH = −61.8 kJ/mol). Although conversion from the dimeric G-quadruplex to the single strand involved an unfavorable entropic gain (ΔS = −764.9 J/mol K), large enthalpic contributions to the stability of the G-quadruplex were attributed to the formation of three core G-tetrads (ΔH = −260.6 kJ/mol). Accordingly, the ΔH value for the formation of a single G-tetrad was estimated assuming that the three G-tetrads contribute equally (ΔH = −86.9 kJ/mol = −260.6 ÷ 3). Notably, the ΔH value (ΔH = −61.8 kJ/mol) for the stacking interaction of two G-quadruplex subunits was close to that for the formation of the single G-tetrad, reflecting the significant contribution of the stacking interaction of the two G-quadruplex subunits to the stability of the two-subunits stacked G-quadruplex. These data provide the first evidence that the stacking interactions between two G-quadruplex subunits provide nearly the same energetic contribution as a single G-tetrad.

Table 1. Thermodynamic parameters for RNA conformational transition.

RNA conformational transition	−ΔH (kJ/mol)	−ΔS (J/mol K)
Two-subunits G-quadruplex → Dimeric G-quadruplex	61.8	186.5
Dimeric G-quadruplex → Single strand	260.6	764.9

^aThe thermodynamic parameters were determined from van't Hoff plots (Supplementary Data). The experimental errors for enthalpy (ΔH) and entropy (ΔS) were ±5 kJ/mol and ±10 J/mol K, respectively.

In-cell ¹⁹F NMR analysis of telomere RNA G-quadruplexes

Although in vitro¹⁹F NMR spectroscopic methods have made very valuable contributions to our understanding of human telomere RNA structures, in vivo observations of RNA G-quadruplex structures are required to gain better insight into the structural basis of their functions inside cells. Indeed, direct evidence for the formation of higher-order G-quadruplexs by human telomere RNA within cells has not yet been obtained.

Recently, developments in NMR technology have made it an effective method for the investigation of biological macromolecules in living cells (in-cell NMR) (36–39). However, the stronger background in the cellular environment often leads to complicated or poor-quality in-cell NMR spectra. For example, ¹H imino proton NMR analysis of a telomere DNA G-quadruplex in the cellular environment yields a low-resolution spectrum, thus distinguishing the resonance signals for molecules of interest from the background noise due to the other molecules present in cells remains a challenge (40).

To overcome these limitations, in-cell ¹⁹F NMR spectroscopy was employed to investigate the structural features of human telomere RNA in living cells. Because there is no natural intracellular concentration of fluorine, fluorinated oligonucleotides do not suffer from high background signals. The in-cell NMR study was performed by injecting ¹⁹F-labeled telomere RNA into Xenopus laevis oocytes and detecting their in-cell conformations using ¹⁹F NMR spectroscopy. The strategy used for in-cell ¹⁹F NMR spectroscopy is shown in Figure 4A. Nuclei of Xenopus oocytes are transferred directly by inserting the glass nuclear transfer pipette into the centre of the animal pole of the oocyte (41,42). The reference in vitro spectrum was compared to the in-cell ¹⁹F-NMR spectrum, enabling reliable determination of the intracellular telomere RNA conformation. Figure 4B shows a comparison of the in vitro and in-cell NMR spectra for the RNA in the pure form (top panel) and upon oocyte injection (bottom panel). Only one signal was observed in the bottom panel NMR spectrum, for which the chemical shift is identical to that observed for the corresponding higher-order G-quadruplex in the in vitro¹⁹F NMR spectrum. We further performed an in-cell experiment using low concentration RNA and showed that only one signal, which is same as the chemical shift with high RNA concentration (Supplementary Figure S11), indicating that the two-subunits stacked G-quadruplex structure is still formed at relatively low RNA concentrations. These results demonstrate that the higher-order G-quadruplex structure is present in living cells. The line width of the signal increased in the in-cell spectrum compared to that in the in vitro spectrum, partially because of the higher viscosity of the cellular environment (43,44). NMR signals depend on whether the molecule of interest is freely available for tumbling in solution. In general, molecules display small tumbling rates due to their sizes, intermolecular interactions, intramolecular interactions with subsets of residues, and the high viscosity environments of intracellular materials, leading to fast relaxation and broad NMR lines with reduced overall signal intensities. In addition, an inherent sample inhomogeneity due to the in-cell environment is also is the cause of the broad line width of in-cell signal (43,44). A control experiment in oocyte stocking buffer at 0.5 mM RNA concentration was performed to confirm that the formation of high order G-quadruplex does not result from the surrounding oocyte stocking buffer of in-cell experiment. As shown in Supplementary Figure S12, the chemical shift in oocyte stocking buffer is same with dimeric G-quadruplex, indicating that the NMR signal of higher-order G-quadruplex results from inside the cells rather than the surrounding oocyte stocking buffer.

Figure 4.

In-cell ¹⁹F NMR of telomere RNA G-quadruplex. (A) Schematic overview of in-cell ¹⁹F NMR experiments. Xenopus laevis oocytes are sorted and collected for microinjections. For in-cell ¹⁹F NMR applications in Xenopus oocytes, telomere RNA sample was injected into the oocyte cells. Comparison with the position of reference in vitro spectrum provides a reliable determination of intracellular telomere RNA conformation. (B) Comparison of ¹⁹F NMR spectra of in vitro sample of telomere RNA (up) with Xenopus egg lysates (middle) and in Xenopus oocytes (bottom).

For the further investigation of the influence of molecular crowding, cellular lysates represent a good molecular crowding since they provide a native-like condition. We crushed the oocytes containing the injected RNA and recorded ¹⁹F NMR spectrum. Notably, the NMR spectral pattern for the lysate of the telomere RNA is similar to those observed for both the corresponding in vitro and in-cell samples (Figure 4B), indicating higher-order G-quadruplex formation in a cell-like condition.

Therefore, by using ¹⁹F NMR spectroscopy, it was demonstrated that the telomere RNA G-quadruplexes preferentially adopt a stacked two-subunits G-quadruplex conformation rather than remaining as single-unit G-quadruplex in living cells. These findings provide the first insights into the structures of telomere RNA G-quadruplexes in the cellular environment.

¹⁹F NMR analysis of telomere RNA G-quadruplexes in molecular crowding conditions

It is known that an intracellular environment was occupied with biological molecules of the cellular volume. A crowded environment (molecular crowding) has reported to influence the structural transition and thermodynamic stability of DNA G-quadruplexes (45,46). Recently, it is reported that ligands can specifically distinguish DNA multimeric G-quadruplexes from monomeric ones under molecular crowding conditions (47,48). To test whether the higher-order RNA G-quadruplex structure is promoted by the crowded environment, we used ¹⁹F NMR spectroscopy to investigate the structural feature of telomere RNA G-quadruplex in K⁺-containing crowded conditions, simulated by different cosolutes. We performed a concentration-dependent experiment to investigate the effect of molecular crowding conditions on the RNA G-quadruplex in crowded solution simulated by three organic solvents acetonitrile (ACN), dimethyl sulfoxide (DMSO) and ethanol (EtOH) at different concentrations (45,46). In the dilute solution (50 mM KCl water solution), one signal was observed at −62.91 ppm, indicating a dimeric parallel G-quadruplex formation, which is consistent with previously reported results (11–14). In crowded solutions of ACN, DMSO and EtOH, a new signal appeared when increasing the volume of cosolutes as compared with the dilute solution (Figure 5A–C). Furthermore, we employed a widely used cosolute PEG 200 to mimic steric crowing, showing the same result that one new signal appears (Figure 5D). This new signal was clearly observed, and its intensity increased remarkably at a cosolute volume of 40–50% (v/v) in four different types of crowded solutions. Each ¹⁹F NMR signal arises as a result of the unique fluorine environment; thus, the presence of the two signals confirms the existence of two conformers of the telomere RNA. Therefore, we assigned the new signal to the two-subunits stacked G-quadruplex structure based on the previous reports that molecular crowding could induce the formation of a multimer complex (44,49–54). It is notable that although the PEG 200 and organic solvents have different dielectric constant, polarity and other properties, all of these promote the formation of stacked dimeric G-quadruplex. Recently, several studies demonstrated the dehydration of organic solvents leads to the formation of parallel DNA G-quadruplex from other species (55–57). To combine our results and previous studies, we suggested that the exclusion effect and/or dehydration effect can promote the formation of high-order RNA G-quadruplex.

Figure 5.

¹⁹F NMR of telomere RNA G-quadruplex in molecular crowding conditions. (A–D) ¹⁹F NMR spectra of 12-mer telomere RNA (0.5 mM) in crowded solutions induced by (A) ACN, (B) DMSO, (C) EtOH and (D) PEG 200. ¹⁹F NMR spectra were recorded under 30% (v/v), 40% (v/v) and 50% (v/v) of ACN, DMSO and PEG 200. The crowding condition of EtOH was only simulated by the addition of 30% (v/v) and 40% (v/v) EtOH, due to 50% (v/v) EtOH leading to a RNA precipitation. Red and green spots indicated the dimeric G-quadruplex and two-subunits stacked G-quadruplex. ¹⁹F NMR reference spectrum of ¹⁹F labeled 12-mer telomere RNA in dilute solution is shown in each case for comparison.

To further understand the effect of molecular crowding on RNA G-quadruplex structure, we measured the specific NMR signals by diluting high concentration RNA to 0.5 mM in diluted solution and molecular crowded condition. We found the transition of two-subunits stacked G-quadruplex to dimeric G-quadruplex is fast in diluted solution, while two-subunits stacked G-quadruplex is still stabilized at ∼0.5 mM under molecular crowded condition (Supplementary Figure S13), indicating that the observation of high-order G-quadruplex in living cells was due to the molecular crowding environment.

A temperature-dependent experiment was then performed to further confirm the assignment of the two signals that were observed in the ¹⁹F NMR spectrum. Figure 6 shows that as the temperature increased from 23°C to 80°C in crowded solution of 40% and 50% (v/v) ACN, the peak intensity of the higher-order G-quadruplex decreased, whereas that of dimeric G-quadruplex increased, indicating the conversion of the higher-order to the dimeric G-quadruplex. In addition, upon heating to 55°C, a new peak corresponding to the unfolded single strand appeared and, at 80°C, only this peak remained with a strong intensity, whereas the peaks of the dimeric G-quadruplex and higher-order G-quadruplex nearly disappeared. These results indicate that at high temperatures, both G-quadruplexes unfolded to an unstructured single strand. Notably, the peak of the higher-order G-quadruplex in 50% (v/v) ACN had a higher intensity than that in 40% (v/v) ACN and still appeared at 75°C but not in 40% (v/v) ACN, indicating a higher thermal stability of the higher-order conformation in higher ratio of cosolutes. NMR data (Supplementary Figure S14) showed that a similar conformational transition could also occur in DMSO, PEG 200 and EtOH molecular crowding conditions under increased temperature. These observations suggest that molecular crowding induced the conformational transition from a dimeric G-quadruplex to a higher-order G-quadruplex structure. The results of T_m value show that molecular crowding substantially stabilizes the higher-order G-quadruplex structure in all types of crowded solutions, leading to an increase in T_m (ΔT_m = 2.6–23.8°C) as compared with the dilute solution (Figure 7 and Table 2). A greater stabilization of the higher-order G-quadruplex may result from the structural conversion of RNA G-quadruplex that is induced by the crowded solutions.

Figure 6.

A temperature-dependent experiment of telomere RNA G-quadruplex in molecular crowding conditions. NMR spectra were recorded in crowded solution simulated by 40% (v/v) and 50% (v/v) of ACN. Condition: 0.5 mM RNA in 50 mM KCl and 10 mM Tris-HCl buffer (pH 7.0). The sample is kept for 10 min at each temperature. Red and green spots indicated the dimeric G-quadruplex and two-subunits stacked G-quadruplex. The peaks of single strand are marked with black spots. Temperatures indicated on the right.

Figure 7.

Melting profiles of the relative peak areas of the ¹⁹F resonance signals versus temperature. The profiles were obtained by plotting the normalized two-subunits stacked G-quadruplex relative peak areas of the ¹⁹F resonance signals in NMR spectra (Figure 2C, Figure 6 and Supplementary Figure S14) as a function of temperature.

Table 2. T _m for higher-order G-quadruplex in dilute and different crowded solutions.

Conditions	T m (°C)	ΔTm (°C)
Dilute solution	52.4	−
40% ACN	58.6	6.2
50% ACN	69.5	17.1
40% DMSO	61.5	9.1
50% DMSO	76.2	23.8
40% PEG 200	56.8	4.4
50% PEG 200	61.9	9.5
40% EtOH	55.0	2.6

^aThe experimental errors for T_m were ±0.5°C.

3 mM RNA in diluted solution and 0.5 mM RNA in molecular crowding solution.

CONCLUSIONS

First, we showed that the simplicity and sensitivity of ¹⁹F NMR spectroscopy effectively enables structural studies of telomere RNA G-quadruplexes in vitro. Using ¹⁹F NMR spectroscopy, we directly observed the two-subunits stacked G-quadruplex, demonstrating that the two-subunits stacked G-quadruplex is formed predominantly at relatively high RNA concentration. Based on the sharp and clear ¹⁹F NMR signals, we determined the thermodynamic parameters of the telomere RNA G-quadruplexes. We demonstrated that the stacking interactions of two G-quadruplex subunits provide the primary energetic contribution to the stability of higher-order RNA G-quadruplex at a level similar to that of a single G-tetrad. These results enable a better understanding of the formation of higher-order RNA G-quadruplexes.

Next, because there is no natural intracellular concentration of fluorine in cells, there is no background noise in in-cell ¹⁹F NMR spectra. Therefore, ¹⁹F NMR spectroscopy is an ideal tool for studying RNA G-quadruplex structures in living cells. We demonstrated that telomere RNA G-quadruplexes preferentially adopt a two-subunits stacked G-quadruplex in living cells using ¹⁹F NMR spectroscopy. To our knowledge, this study is the first direct observation of higher-order RNA G-quadruplex in a cellular environment and provides new insight into the structural behavior of telomere RNA G-quadruplexes in living cells. Our results open new avenues for the investigation of G-quadruplex structures in vitro and in living cells.

Finally we demonstrated that molecular crowding conditions simulated by ACN, EtOH, DMSO and PEG 200 promoted the formation of two-subunits stacked G-quadruplex and stabilized the higher-order structure, which provided unequivocal evidence for the formation of higher-order G-quadruplex structure in living cells.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Grant-in-Aid for Scientific Research (B) and Research Activity Start-up from Ministry of Education, Science, Sports, Culture and Technology (Japan) [26288083 and 25888019]; Takeda Science Foundation. Funding for open access charge: Grant-in-Aid for Scientific Research (B) from the Ministry of Education, Science, Sports, Culture and Technology (Japan) [26288083].

Conflict of interest statement. None declared.