Real-time monitoring of DNA G-quadruplexes in living cells with a small-molecule fluorescent probe.

G-quadruplex DNA has been viewed as a prospective anti-cancer target owing to its potential biological relevance. Real-time monitoring of DNA G-quadruplex structures in living cells can provide valuable insights into the relationship between G-quadruplex formation and its cellular consequences. However, the probes capable of detecting DNA G-quadruplexes in living cells are still very limited. Herein, we reported a new fluorescent probe, IMT, for real-time visualization of DNA G-quadruplex structures in living cells. Using IMT as a fluorescent indicator, the quantity changes of DNA G-quadruplex at different points in time during continuous cellular progression responding to Aphidicolin and Hydroxyurea treatment have been directly visualized. Our data demonstrate that IMT will be a valuable tool for exploring DNA G-quadruplexes in live cells. Further application of IMT in fluorescence imaging may reveal more information on the roles of DNA G-quadruplexes in biological systems.

INTRODUCTION

G-quadruplex DNA, a non-canonical secondary nucleic acid structure formed by guanine-rich genomic sequences (1,2), has been viewed as an emerging therapeutic target in oncology due to its potential roles in oncogene transcriptional regulation (3), DNA replication (4) and telomere stability (5,6). Over the past two decades, particular interest has been paid on this structure and the research on this structure has achieved explosive growth in many interdisciplinary fields involving nucleic acid research, genomics, medicinal chemistry and biotechnology (7–12). Only in the development of quadruplex-interactive ligands as an example, over hundreds of ligand molecules have been designed and screened (13), and one of the quadruplex-interactive compounds CX-3543 has even entered Phase II as a first-in-class candidate for multiple types of cancers (14). Despite significant progress, the exact nature of G-quadruplex biological significance remains poorly understood. One of the key reasons is that the intracellular G-quadruplexes, especially the DNA G-quadruplexes located in the nucleus, are still difficult to detect and analyse due to the lack of efficient probes. Therefore, it is indeed of great significance to develop a probe that can explore G-quadruplexes in living cells.

For this reason, intense researches have focused on the development of effective optical probes for detecting the G-quadruplex structures in cells in the last few years. The quadruplex-specific antibodies BG4 (15) and 1H6 (16) were first generated to detect the G-quadruplex location on chromosomes in cells, providing visible evidence for the existence of DNA G-quadruplexes in cells. In a subsequent study, BG4 was used in the chromatin immunoprecipitation and high-throughput sequencing approach (ChIP-seq) to confirm the formation of G-quadruplex structures within the endogenous chromatin context (17), further demonstrating the great value of BG4 in exploring G-quadruplexes. However, question remains as that further study like the dynamic relationships between G-quadruplex formation and its cellular consequences is limited by using the antibody-based approaches because the cells must be fixed and permeabilized to allow the cell-impermeable antibodies enter. It thence appears that cell-permeable small-molecule optical probes may be more desirable for the dynamic monitoring of G-quadruplexes in live cells. Recently, the probes for detecting RNA G-quadruplexes in the cytoplasm (18–22) and DNA G-quadruplexes in the nucleoli (23,24) and mitochondria (25) of live cells have been developed, respectively. But none of them could be utilized to recognize the DNA G-quadruplexes in the chromatin context of live cells. Lately, Vilar’s team has designed a small-molecule probe that uses fluorescence lifetime to detect DNA G-quadruplexes in living cells (26), but the practical application of this probe will be limited by the requirement for special equipment and longer acquisition time. Therefore, the probe that can monitor DNA G-quadruplexes in living cells in real-time is still urgently needed.

To monitor G-quadruplex DNA in living cells, the criteria of an excellent probe should include: good membrane permeability, low toxicity, high selectivity for G-quadruplexes over various nucleic acid structures, fluorescence detectability and high photostability. Thioflavin T (ThT) is a commercial benzothiazole dye with good photophysical stability and membrane permeability (27). It has been reported that ThT exhibits high selectivity for G-quadruplex structures in vitro (28–31). However, in living cells, ThT mainly stains the nucleoli (32), making it impossible to detect the G-quadruplex DNA in chromatin. Nevertheless, the exciting properties of ThT inspire us to develop new benzothiazole family compounds (detailed syntheses and characterization are given in the Supplementary Data) in the hope of obtaining an ideal probe for the direct detection of G-quadruplex DNA in living cells. Through preliminary screening, we found that all these probes exhibit recognition performance for G-quadruplex structures in buffer solution (Supplementary Figure S1), but only the probe IMT (Figure 1) shows better potential for detecting the DNA G-quadruplex structures in living cells (Supplementary Figure S2). IMT is almost non-fluorescent in aqueous environments but displays significant fluorescence enhancement when interacting with G-quadruplex structures. This feature allows IMT to stain DNA G-quadruplexes both in vitro and in living cells with a high signal-to-noise ratio. Through drug-treatment experiments, we further demonstrated that this probe could also be used to track DNA G-quadruplex changes in cellular progression after drug-treatment.

Figure 1.

Schematic representation of the G-quadruplex DNA probed by IMT in live cells.

MATERIALS AND METHODS

Full experimental details for the synthesis, purifications and characterizations of the new compounds including IMT are listed in the Supporting Information. Protocols for general materials and spectral measurement including UV-vis, fluorescence, native-polyacrylamide gel electrophoresis (PAGE), circular dichroism (CD), fluorescence resonance energy transfer (FRET) and nuclear magnetic resonance (NMR) experiments are also given in the Supplementary Data.

Equilibrium dialysis

Equilibrium dialysis experiments were performed according to the reported procedures by Jean-Louis Mergny (33). All the dialysis assays were conducted in 20 mM Tris–HCl buffer solution (40 mM KCl, pH 7.4) for G-quadruplexes and ss-/ds-DNA and 20 mM Tris–HCl buffer solution (40 mM KCl, pH 5.8) for i-motif. A total of 5 ml of the dialysate solution containing 2 μM IMT was used in each competition dialysis assay. A volume of 1 ml at 75 μM monomeric unit (nucleotide, base pair or quartet) of each of the nucleic acids samples was introduced into a separate dialyzer unite (Spectrum, Micro Float-A-Lyzer G2, MWCO 0.5-1 kD). All 16 dialysis units were then placed in the volumetric flask containing the dialysate solution and put on a shaking table to allow continuous shaking for 48 h at room temperature. After the equilibrium period, DNA samples were carefully removed to microfuge tubes and treated with 1% sodium dodecyl sulphate for 4 h to dissociate the DNA–IMT complexes. The IMT concentration in each sample was determined by fluorescence spectroscopy. The amount of IMT in each dialysis unit is directly proportional to the fluorescence signal of the samples.

Cell culture

Hela (cervical carcinoma) and SMMC-7721(human hepatoma) cells were cultured in high glucose DMEM (Dulbecco’s Modified Eagle’s Medium) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. HT1080 (human fibrosarcoma) cells were cultured in MEM-EBSS (minimum essential medium) containing 10% FBS and 1% penicillin-streptomycin. HUVECs (human umbilical vein endothelial cells) were cultured in endothelial cell medium containing 5% FBS. All the cells were cultured 48 h in 5% CO2 at 37°C.

Fixed-cell imaging

Hela cells grown on Petri dish were fixed for 5 min in cold methanol, rinsed twice with phosphate-buffered saline (PBS) and then incubated with 4 μM IMT for 15 min approximately. For enzyme treatment experiments, cells were incubated with DNase I (0.1 mg/ml) or RNase T1 (1 mg/ml) for 1 h at 37°C. Then, cells were incubation with IMT (4 μM) for 15 min. For urea treatment experiment, Hela cells were incubated with 7 M urea solution for 10 min, and then urea was rinsed with PBS buffer solution. The cells were stained with IMT (4 μM) again. Cellular nuclei were co-stained with PI. Then cells were imaged under CLSM. Cell-cycle experiments were performed according to the reported procedures by Shankar Balasubramanian (15). Three phases of cell-cycle progression were explored with IMT: (i) Hela cells were incubated for 24 h in serum-free DMEM with 4 μM IMT (G0/G1); (ii) grown for 16 h in DMEM with 4 μM IMT, 20% FBS and 200 μM mimosine; (iii) for 3 h in DMEM with 4 μM IMT and 10% FBS (S phase). Each cell-cycle stage was confirmed by using a flow cytometer (FACSAria II). The nuclei were co-stained with PI, and CLSM images of IMT (425−525 nm) and PI (570−670 nm) were collected under the excitation wavelength at 405 and 559 nm, respectively.

Immunofluorescence experiment

BG4 with a His-tag was purchased from Absolute Antibody Ltd (Oxford, UK), which had been manufactured using AbAb’s Recombinant Platform with variable regions (i.e. specificity) from the phage display derived scfv BG4 and purified by Immobilized Metal Affinity Chromatography. For immunofluorescence of BG4 antibody, cells grown on glass coverlips were fixed in cold methanol for 8 min and permeabilized with 0.1% triton-X100/PBS for 30 min. After treated by RNase T1 (1 mg/ml) for 1 h at 37°C, the cells were blocked with 5% bovine serum albumin/PBS at room temperature for 1 h. The coverlips were incubated with BG4 antibody at 37°C for 2 h, then incubated with anti-His-tag antibody (#12698S, Cell Signalling Technology) at 37°C for 2 h and then incubated with anti-rabbit Alexa 594-conjugated antibody (A11037, Thermo Fisher Scientific) at 37°C for 30 min. After each antibody incubation procedure, the coverlips were rinsed five times with 1% tween-20/PBS buffer. Last, cells were stained with 4 μM IMT for 20 min. The emission of IMT was collected under excitation at 405 nm, and that of BG4 (570−670 nm) was collected under excitation at 559 nm, sequentially.

Live-cell imaging

For the time-dependent images of IMT staining Hela cells, the imaging was immediately collected within 0−10 min upon titration of IMT (4 μM). For the three-dimensional (3D) confocal microscopy imaging, Hela cells, SMMCF-7721 cells, HUVECs and HT1080 cells were incubated with IMT (4 μM) and SYTO®59 (2 μM) for 10 min in 5% CO2 at 37°C, then 3D imaging measurement was taken directly without any processing steps, respectively. For the pyridostatin (PDS) competitive experiment, Hela cells were seeded in confocal dishes and incubated with IMT (4 μM) and different concentrations of PDS for 4 h at 37°C. SYTO®59, a red fluorescent dye that is commonly used to stain nucleic acids in live cells and has no side-effects on cellular viability, proliferation or cell migration (34), was then introduced to label the nucleus for 30 min before CLSM imaging. CLSM images of IMT and SYTO®59 (650−750 nm) were collected under the excitation wavelength at 405 and 635 nm, respectively. The fluorescence of IMT before binding with target molecules is very weak and thus all images were collected directly in the presence of the culture medium.

Real-time imaging of living cells

Hela cells were seeded in confocal dishes and cultured in high glucose DMEM containing 10% FBS, 1% penicillin-streptomycin and IMT (4 μM) for 10 min at 37°C. Aphidicolin (5 μM) and Hydroxyurea (HU, 15 μM) was titrated in the culture medium, then fluorescence images were acquired every 15 min. The nucleus were co-stained with SYTO®59 (2 μM), and CLSM images of IMT and SYTO®59 were collected under the excitation wavelength at 405 and 635 nm, respectively. To quantify the fluorescence intensity in the cell nuclei, the digital images were analysed using the high content screening studio bio-application from Cellomics ArrayScan Vti (Thermo Fisher Scientific, USA) high-content imaging analysis platform. The data were acquired from ∼300 cells per sample and the standard error of the mean calculated from three replicates. Statistical analyses and P-value were calculated using the Student’s t-test.

RESULTS

Photophysical properties and cytotoxicity of IMT

As a potential probe for live-cell imaging, the unique characteristics including sufficient photophysical stability and low cytotoxicity are crucial. Photodecomposition experiment was first performed to study the photostability of IMT at an excitation wavelength of 405 nm, which corresponds to the maximum absorption wavelength of IMT and is also the same as the excitation wavelength used in the CLSM measurement. As a result, the maximum fluorescence intensity of IMT showed no any noticeable degradation at different time points within 1 h during continuous irradiation (Supplementary Figure S3), indicating high photostability of IMT. The pH and ionic strength sensitivity were then examined to assess the photophysical stability of IMT for use in biological environment. In the fluorescence spectra of IMT, no considerable difference is observed over the pH range of 4−9 and the K+ range of 0.02−1 M (Supplementary Figure S4), confirming that IMT is insensitive to both pH and ionic strength.

The cytotoxicity of IMT was evaluated using a standard methyl thiazolyl tetrazolium assay. The living Hela cells, SMMC-7721 cells, HT1080 cells and HUVECs were exposed to different concentrations of IMT for 48 h and then the percentage of the viable cells was quantified. The results showed that the cell viability with regard to four cell lines still exceeded 60% even when the concentration of IMT was as high as 60 μM in the culture medium (Supplementary Figure S5), indicating that the cytotoxicity of IMT is negligible.

Selective recognition of G-quadruplexes by IMT

To evaluate the specificity of IMT for G-quadruplex structures, the binding behaviour between IMT and various nucleic acid models including G-quadruplex DNA/RNA, i-motif (35,36), ss/ds-DNA and ss/ds-RNA (Supplementary Table S1) was compared through absorption and fluorescence emission spectroscopy under identical salt conditions. With the gradual addition of G-quadruplexes, the absorption peak of IMT bathochromically shifted by 7−19 nm and a neat isosbestic point appeared at 415 nm (Supplementary Figure S6). However, with the titration of other nucleic acid models, the IMT absorption peak did not shift, which means that IMT binds more strongly to the G-quadruplex structures than other nucleic acid structures. On the other hand, the emission intensity of IMT displayed significant enhancement with the increase in the concentration of G-quadruplex structures, but only exhibited subtle variation with the titration of other nucleic acid structures (Supplementary Figure S7). As the samples were excited at the isosbestic point (∼415 nm) to maintain similar absorbance in all cases, it is believed that the fluorescence light-up observed here should be attributed to the specific binding of IMT with G-quadruplexes. Figure 2A shows a comparison of the overall fluorescence intensity enhancement (F/F0) monitored at 500 nm for different nucleic acid forms. The fluorescence enhancement is in the range of 259- to 630-fold for G-quadruplexes but <8-fold for other nucleic acid models. Clearly, the striking dominance of fluorescence enhancement in the G-quadruplexes versus other nucleic acid structures unambiguously establishes the highly selective fluorescence recognition of IMT toward G-quadruplex structures. To further confirm the good fluorescence identification qualities, competitive fluorescence titrations were also performed in the presences of various amounts of duplex-DNA. The results show that the fluorescence enhancement induced by G-quadruplex structures is only poorly affected by the presence of 10 molar equiv of duplex-DNA (Supplementary Figure S8). This indicates that IMT exhibits an exquisite quadruplex versus duplex selectivity.

Figure 2.

(A) Dependence of IMT (4 μM) fluorescence intensity at 500 nm on a variety of oligonucleotides (8 μM). λex = 415 nm. IMT with G4 (G-quadruplex) DNA, G4 RNA, ss/ds-DNA and ss/ds-RNA were measured in 20 mM Tris–HCl buffer solution (40 mM KCl, pH 7.4); IMT with i-motif DNA were measured in 20 mM Tris–HCl buffer solution (40 mM KCl, pH 5.8). (B) Non-denaturing PAGE images of G4s and non-G4s stained with SYBR Gold (left) and 4 μM IMT (right), respectively. All DNA samples (8 μM) were prepared in 20 mM Tris–HCl (40 mM KCl, pH 7.4). The images were collected on GE Typhoon Trio.

The selective recognition of G-quadruplexes by IMT was also determined in the gel system. In the control group, all of the oligonucleotides could be stained by SYBR Gold (Figure 2B), an extremely sensitive fluorescent dye that binds to nucleic acids, corresponding to the non-selectivity of SYBR Gold on nucleic acid structures. In the treatment group, only the G-quadruplex structures were stained by IMT while other nucleic acids are invisible under the same conditions. This result further confirms the selective fluorescence identification performance of IMT on the G-quadruplex structures.

Apart from the absorption and fluorescence assays described above, equilibrium dialysis assay (33,37) was also used to evaluate the structural selectivity of IMT for various nucleic acid models. The competitive dialysis experiments were performed using 16 nucleic acid structures against a common IMT (2 μM) solution for 48 h. More IMT accumulated in the dialysis tube containing the structural form with the highest IMT-binding affinity. By measuring the amount of IMT in the dialysis tube, we can estimate the percentage of IMT bound. The equilibrium dialysis results for IMT with various nucleic acids are shown in Figure 3 as a bar graph. The data show that IMT binds well to all of G-quadruplex forms but binds poorly to other nucleic acid forms, fully consistent with the fluorescence and PAGE results.

Figure 3.

Selectivity of IMT to various nucleic acid topologies in equilibrium dialysis. The i-motif samples were performed in 20 mM Tris–HCl (40 mM KCl, pH 5.8) and other samples were performed in 20 mM Tris–HCl (40 mM KCl, pH 7.4). The value of IMT without nucleic acids is defined as 1.0.

Taking into account that amino acids and other bioactive substances may also interact with IMT and interfere with G-quadruplex detection in cells, some amino acids (Glu, Arg, Ser, Ala, Asn, Cys, Asp, Gly), Vitamin C, glucose, proteins (nucleolin, human serum albumin, transferrin, holoprotein) and various metal ions (Mg2+, Ca2+, Zn2+, Cu2+, Fe3+) were also chosen for investigation. In sharp contrast to the remark fluorescence enhancement with the addition of G-quadruplexes, only very slight changes of the emission intensity were observed upon addition of the above substances (Supplementary Figure S9), pointing to the negligible interaction between IMT and these substances. This implicates the potential utility of IMT for detecting G-quadruplex structures in biological samples.

Effect of IMT on formation and stability of G-quadruplexes

The satisfactory selectivity has implicated a possible use of IMT to recognize G-quadruplex structures. In fact, in order to offer an unbiased visualization of G-quadruplexes, it is expected that the probe will not interfere with the folding of G-quadruplexes. Since DNA G-quadruplex structures have distinctive CD spectra (38,39), CD was thus conducted to assess the effect of IMT on the formation and stability of G-quadruplex structures. In the freshly prepared solution without metal ions, the G-rich oligonucleotides are mainly present in a random single-stranded form (Supplementary Figures S10 and 11). With the gradual addition of IMT, the CD spectra of these oligonucleotides showed no obvious change (Supplementary Figure S12). In addition, the melting curves of these oligonucleotides were almost overlapping in the presence and absence of IMT (Supplementary Figure S13). Both of these results suggest that IMT cannot induce G-quadruplex formation. To assess whether IMT could stabilize G-quadruplex structures, the G-quadruplex structures were previously induced by potassium ions and then their melting curves in the presence and absence of IMT were tested (40,41). We found that the melting transition temperature of the G-quadruplexes in the presence of IMT was the same as that in the absence of IMT (Supplementary Figure S14), indicating that IMT did not increase the thermal stability of G-quadruplex structures. The FRET melting assay, another popular method of characterizing the thermal stability of G-quadruplex structures (42,43), exhibited the same result as above (Supplementary Figure S15), further precluding the stabilizing effect of IMT on G-quadruplexes.

Binding mechanism

As a cationic dye, IMT may be embedded into the grooves of the quadruplex DNA, intercalated between the G-quartet planes (44,45), or stacked onto the top/bottom quartet. The major binding mode of ThT to G-quadruplexes has been revealed as end-stacking (28). In addition to end-stacking, there are also secondary binding modes including groove-binding and partial intercalation, as evidenced by the induced CD signals of ThT (28). By comparing the CD spectra of ThT and IMT with G-quadruplexes, we found little evidence of induced CD signals of IMT (Supplementary Figures S16 and 17), implicating that the binding patterns adopted by ThT and IMT may be somewhat different. We speculate that, like most G-quadruplex ligand molecules, IMT may simply stack on the terminal quartet planes.

By a Job’s plot analysis (46), we know that the binding stoichiometry of various G-quadruplex structures with IMT is 1:1 (Supplementary Figure S18), hinting that IMT may only bind to one terminal plane. To understand the actual binding pattern, we monitored the1H-NMR spectra of the c-kit G-quadruplexes with increasing concentrations of IMT. In the absence of IMT, 12 peaks belonging to exchangeable imino protons of the c-kit G-quadruplexes were observed from upfield to downfield (47). In the presence of IMT, almost all of these peaks shift toward upfield (Figure 4A). Comparing the chemical shifts of all imino protons, we found that the imino proton (G6) shifted the most, suggesting the top quartet is the most favored site for IMT stack. Besides, we also conducted an IMT-induced fluorescence quenching assay to further confirm the binding site. In this assay, differentially Cy5 end-labeled oligonucleotides were used to measure 5′-quartet versus 3′-quartet quenching caused by a ligand, thereby probing the ligand binding at either end of a G-quadruplex structure (48). As shown in Figure 4B, the emission intensity of Cy5 at 5′-quartet is more pronouncedly quenched by IMT compared with that of Cy5 at 3′-quartet, revealing that IMT indeed exhibited much higher binding selectivity for the 5′-end quartet, consistent with the NMR result.

Figure 4.

(A) 1H-NMR spectra of the c-kit G-quadruplex DNA (0.5 mM) with increasing concentrations of IMT in 20 mM PBS (pH 7.0, 90% H2O/10% D2O) at 25°C. (B) (left panel) Schematic representation of G-quartet selectivity analysis. (right panel) The ratio (F/F0) of fluorescence intensity of Cy5-labeled c-kit (0.1 μM) as a function of IMT concentration in 20 mM Tris–HCl (pH 7.4) with 40 mM KCl. F0 and F represent the fluorescence intensity of Cy5-quadruplex before and after interacting with IMT, respectively.

Staining of fixed cells

To further determine the specific staining of IMT on the DNA G-quadruplex structures in fixed cells, the fixed Hela cells were incubated with IMT and then measured using a fluorescence microscope. Homogeneous fine granular fluorescence foci were found in the nuclei of Hela cells (Figure 5A and Supplementary Figure S19), which disappeared after DNase I treatment but still existed after RNase T1 treatment (Figure 5B), suggesting that these fluorescence foci are derived from the interaction of IMT with DNA but not with RNA. On the other hand, the fixed cells were incubated with urea, which can denature the secondary structures of nucleic acids by forming multiple hydrogen bonds with nucleic acid bases (49,50). We observed that the fluorescent foci disappeared with the presence of urea but recovered after urea was simply rinsed (Figure 5B), supporting the selective binding of IMT to the secondary structures of DNA.

Figure 5.

(A) Confocal imaging of fixed Hela cells stained with IMT (4 µM), nucleus dye PI and Merge images. (B) Fluorescent microscopy analysis of Hela cells stained with IMT (4 µM) and PI before and after treatment of cells with DNase I, RNase T1, urea, rinsing to remove urea, and PDS (8 µM). (C) (left panels) Co-localization images of fixed Hela cells stained with IMT and BG4, and merge images demonstrate that co-localization of IMT/BG4. (right panel) Pixel fluorogram from the section displayed in left panels. (D) (left panels) Confocal imaging of synchronized fixed Hela cells at the G0/G1 and G1/S boundaries, as well as during S phase stained with IMT (4 µM). (right panel) relative pixel intensity (R.P.I.) corresponding to each phase (∼300 cells were quantified for each phase). The pixel intensity of S phase is defined as 1.0. For clarity, the images were presented in pseudocolors of red (IMT), blue (PI) and green (BG4).

To provide more direct evidence that the DNA secondary structure targeted by IMT is G-quadruplex, we further conducted a competition displacement assay using PDS as a competitor because PDS has been proved to show high specificity for the G-quadruplex structures in cells (15,26,51–53). Besides, PDS also adopts an end-stacking mode to bind with G-quadruplexes (48,51,54), in the same way as that IMT binds with G-quadruplexes. The binding constants of PDS with G-quadruplexes were determined to be of the order of 106 M−1 (48), which are relatively higher than the binding constants of IMT with G-quadruplexes ((0.77−1.6) × 105 M−1, Supplementary Figure S20). Thus coexistence of IMT and PDS will lead to effective displacement of IMT from the quadruplex moiety by PDS, resulting in obvious fluorescence quenching of IMT in solution (Supplementary Figure S21). In the cells co-incubated with IMT and PDS, we also observed remarkable fluorescence quenching of IMT (Figure 5B), confirming the specific targeting of IMT to G-quadruplex structures. Moreover, we also employed the highly specific G-quadruplex antibody BG4 in the immunofluorescence experiment to compare the localization of BG4 with IMT. As shown in Figure 5C and Supplementary Figure S22, the IMT foci well co-localized with BG4 staining in the nucleus with an overlap coefficient (OLC) of 0.87. Collectively, the results described above well support that IMT can be used to selectivity stain G-quadruplex structures in cells.

G-quadruplexes are known to be modulated in a manner dependent on whether or not DNA is being replicated during the cell cycle (15). G-quadruplex structures most probably form during DNA replication when duplex strands become separated. To further confirm that the IMT-targeted DNA secondary structure is related to G-quadruplexes, we also followed the continuous cell-cycle progression. As expected, the number of fluorescence foci in nucleus shows significant difference dependent on cell-cycle phases (Figure 5D). From G0 to S phase, the number of fluorescence foci increased about 5-fold, which is consistent with the replication-dependent formation of G-quadruplex structures (15).

Real-time monitoring of DNA G-quadruplexes in living cells

Having known that IMT showed a high specificity to DNA G-quadruplexes in fixed cells, we next asked whether IMT was able to probe the DNA G-quadruplex structures in living cells. For this reason, we analysed the movement of IMT foci in a time-dependent experiment. When IMT was just added, the background shown Figure 6A and Supplementary Movie S1 indicates that the probe is almost non-fluorescent in the cell culture media. Over time, the fluorescence foci in the nuclei gradually increased and reached saturation within 7 min. These results suggest that IMT can easily penetrate the living cell membrane into the nuclei in a short period of time. We further found that the binding of IMT to G-quadruplexes was almost instantaneously completed (Supplementary Figure S23), so the staining of G-quadruplexes may be achieved as soon as IMT enters the nuclei.

Figure 6.

(A) Time-lapse CLSM images of Hela cells treated with IMT (4 µM). IMT fluorescence images were continuously collected upon IMT being titrated in the cell culture medium. (B) CLSM images of living Hela cells stained by IMT (4 µM) with increasing concentrations of PDS. (C) Three-dimension (3D) confocal imaging of the live Hela cells incubated with IMT (4 µM). For clarity, the images were presented in pseudocolors of red (IMT). The nuclei were counterstained with SYTO®59 (blue).

To validate the fluorescence foci originate in the staining of G-quadruplexes by IMT, we then conducted a competition displacement assay using 2 molar equiv (a concentration range that is nontoxic to cells, Supplementary Figure S24) of PDS. The fluorescent foci in living Hela cells were significantly weakened in the presence of PDS (Figure 6B), confirming the specific staining of IMT to DNA G-quadruplexes. We also performed IMT-stained 3D confocal imaging of Hela cells and three other cell lines including HUVEC, SMMC-7721 and HT1080 cells, and intense and discrete foci were visible in the nuclei of all these living cells (Figure 6C and Supplementary Figure S25), confirming that IMT specifically stains DNA G-quadruplexes of different cell types.

Because of good biocompatibility, IMT may also be applicable for real-time monitoring of G-quadruplex DNA during continuous cellular progression. To explore this application, real-time imaging experiments were performed. IMT was first incubated with Hela cells at 37°C. After 10 min of incubation, the cells were treated with Aphidicolin (5 μM), a specific inhibitor of eukaryotic DNA polymerase α (55,56) and examined with fluorescence microscopy to obtain real-time fluorescence images. As the incubation time elapses, the fluorescence of the control cells remained constant at different times (Figure 7). However, the fluorescence of the cells with Aphidicolin treatment diminished gradually and decreased by about 50% at 90 min. Furthermore, the cells were treated with HU (15 μM), a replication inhibitor that can induce replication stalling at G-quadruplex sites (57) and the IMT fluorescence in these cells increased ∼1.5-fold at 90 min. These results clearly demonstrate that IMT will be a useful probe for real-time monitoring of G-quadruplex DNA in living cells.

Figure 7.

(A) Confocal imaging of live Hela cells stained with IMT (4 µM) before and after treatment of Aphidicolin (5 µM) and Hydroxyurea (HU, 15 µM) at different time point. (B) Relative pixel intensity (R.P.I.) corresponding to each group displayed in right panels (∼300 cells were quantified for each group). The pixel intensity measured at 0 min is defined as 1.0. The standard error of the mean was calculated from three replicates.*P<0.001.

Increasing evidences have demonstrated the important biological roles of G-quadruplexes in oncology. In contrast to the linear structures of single-stranded and duplex DNA, G-quadruplex is a globular structure with large planes which allows G-quadruplexes to be targeted with a high degree of selectivity. The potential biomolecular roles in disease pathology and unique structural characteristics make the G-quadruplex structure a promising target for drug design. In recent two decades, the research activity on G-quadruplex structures has greatly increased. Despite this, since the effective probe tools that can be commonly used to explore G-quadruplex DNA in living cells are really limited, a clearer definition of the dynamic relationships between G-quadruplex formation and its cellular consequences remains difficult to reveal.

To develop a fluorescence probe that can directly stain G-quadruplex DNA in living cells, we synthesized three new benzothiazole fluorescence probes. Through cell imaging experiments, we confirmed that only IMT could stain DNA G-quadruplexes in the nucleus, while other probes mainly stain nucleoli. Differences in the localization of these probes within the cells may involve a variety of factors (58,59), such as the water/octanol partition coefficient (log P), substituent bulk parameter and differential fluorescence identification of DNA and RNA G-quadruplexes. IMT is almost non-fluorescent before binding to a target molecule or after binding to single/double-stranded nucleic acids, but exhibits over 200-fold fluorescence enhancement after binding with G-quadruplexes. This allows IMT to light-up the DNA G-quadruplex structures in cells directly without the interference from background fluorescence. The number of fluorescent foci is positively correlated with the quantity of G-quadruplexes in the cells. However, since the adjustment of parameters such as brightness and contrast can affect the number of fluorescent foci and cause big error, it is more accurate to analyse the relative change of G-quadruplexes under the same parameters.

We also noticed that there is no fluorescent focus in the cytoplasm of the cells. Since IMT does not show a significant difference in fluorescence after binding to the G-quadruplex structures of DNA and RNA, theoretically the fluorescence foci derived from the labeling of RNA G-quadruplexes should be observed in the cytoplasm. According to the quantitative structure activity relations (QSAR) models (58,59), the physicochemical characteristics of a probe molecule determine its localization within the cells. As a permanent cationic species, the physicochemical characteristics of IMT make it more likely to localize to the nuclear chromatin rather than the cytoplasm. In the nuclear chromatin, IMT further selectively binds to G-quadruplex structures and exhibits significant fluorescence enhancement, allowing us to observe the fluorescence foci originated from the DNA G-quadruplexes.

In addition, IMT enjoys good photophysical properties and biocompatibility, both of which facilitate real-time monitoring of DNA G-quadruplexes in living cells. To date, the clear definition of the frequency of DNA G-quadruplexes at different points in continuous cell progression and the true response of DNA G-quadruplexes to drug treatment remains far from clear. Real-time monitoring of DNA G-quadruplexes in living cells will be an efficient way to help answer these questions. The use of IMT to track DNA G-quadruplexes changes with the treatment of Aphidicolin and HU well demonstrates the potential of IMT as an effective tool for exploring DNA G-quadruplexes and further understanding the relationship between G-quadruplex formation and its cellular consequences. From the experimental studies involving PDS and drug treatment, it can be speculated that IMT is also promising for screening quadruplex-specific ligands and evaluating quadruplex-associated drug efficacy.

Click here for additional data file.