Cellular Visualization of G-Quadruplex RNA via Fluorescence- Lifetime Imaging Microscopy

Over the past decade, appreciation of the roles of G-quadruplex (G4) structures in cellular regulation and maintenance has rapidly grown, making the establishment of robust methods to visualize G4s increasingly important. Fluorescent probes are commonly used for G4 detection in vitro; however, achieving sufficient selectivity to detect G4s in a dense and structurally diverse cellular environment is challenging. The use of fluorescent probes for G4 detection is further complicated by variations of probe uptake into cells, which may affect fluorescence intensity independently of G4 abundance. In this work, we report an alternative small-molecule approach to visualize G4s that does not rely on fluorescence intensity switch-on and, thus, does not require the use of molecules with exclusive G4 binding selectivity. Specifically, we have developed a novel thiazole orange derivative, TOR-G4, that exhibits a unique fluorescence lifetime when bound to G4s compared to other structures, allowing G4 binding to be sensitively distinguished from non-G4 binding, independent of the local probe concentration. Furthermore, TOR-G4 primarily colocalizes with RNA in the cytoplasm and nucleoli of cells, making it the first lifetime-based probe validated for exploring the emerging roles of RNA G4s in cellulo.


Photophysical characterization
Table S1 -Photophysical properties of TOR-G4 in aqueous buffer and in the presence of a 10-fold excess of G4 (c-MYC).

NA
Absorption spectra were recorded on an Agilent 8453 UV-Visible spectrophotometer across a range of 190-1100 nm in 1 nm intervals.Emission spectra were recorded on a Fluoromax4spectrofluorimeter (Horiba Jobin-Yvon) with excitation at 470 nm.Excitation spectra were recorded on a Fluoromax4-spectrofluorimeter (Horiba Jobin-Yvon) with detection at 660 nm and excitation from 345 -650 nm or detection at 540 nm and excitation from 275 -530 nm.
In vitro fluorescence lifetime measurements were made via time correlated single photon counting using a DeltaFlex modular lifetime system (Jobin-Ybon, Horiba), coupled to a 467 nm NanoLED diode laser (pulse width <200 ps, Horiba) as an excitation source.To measure the instrument response function (IRF), prompt measurements were made using a dilute LUDOX solution at the laser excitation wavelength.Emission was monitored at 540 nm with a 32 nm bandpass, across a 100 ns time scale (split between 4096 time bins) until 10,000 counts were reached in the maximum.A long-pass filter at 495 nm was used to block the detection of scattered excitation light during the measurements.The time resolved decays were fitted to a biexponential decay function (equation 1) with deconvolution from the IRF, using Horiba DAS6 lifetime analysis software.Intensity-weighted average lifetimes (w) were calculated from the individual amplitudes (A) and lifetimes () values extracted from each decay according to equation 2.
(1) Y(t) = A1e -t/τ1 + A2e -t/τ2 (2) Fluorescence switch-on was calculated by dividing the emission of TOR-G4 (at 540 nm) in the presence of a given nucleic acid by that of the probe alone in aqueous buffer.
DNA/RNA binding titrations were performed by recording the fluorescence intensity of TOR-G4 (2 M) at 540 nm following 470 nm excitation, with varying concentrations of each nucleic acid (0-150 g/mL).Fluorescence intensity was normalized for absorbance of each solution at 470 nm and plotted against nucleic acid concentration.The resulting curves were fitted using Graphpad prism one site -total model.The G4/totRNA titration was performed by measuring the lifetime of TOR-G4 (2 M) bound to yeast totRNA (100 g/mL) and increasing amounts of G4 RNA (TRF-2).The subsequent response curve was then fit in Graphpad prism using the [agonist] vs response -variable slop (4 parameters) model.

Molecular modelling
Molecule geometry was optimized with Gaussian 09 software using DFT calculations utilizing the B3LYP functional and 6-31g basis set.The duplex DNA structure was taken from the RCSB protein data bank (pdb108D) based on the solution NMR structure of a DNA sequence in complex with homo-dimeric thiazole orange (TOTO). 3The G-quadruplex structure was taken from the protein data bank structure of the c-Myc promoter in complex with a quindoline ligand (pdb2L7V). 4The dimensions of the active site box were selected using AutoDock tools and were set to cover the entirety of the DNA structure.AutoDock Vina was used to run molecular dynamic measurements using default settings, energy range=9 and exhaustiveness=9. 5Output files were visualized with PyMOL.

Cellular characterization
U2OS cells (ECACC) were plated (2x10 4 cells per well, 250 L, 0.8 cm 2 ) 24 hours before each experiment and grown in DMEM media (Gibco) supplemented with 10% FBS.For fixed cell experiments, cells were washed three times with ice cold PBS before fixation with 4% paraformaldehyde for 10 min.Cells were then washed again three times with ice cold PBS, before probe staining.
Cell lysate was extracted from 1x10 7 U2OS cells by first scraping and pelleting cells.Cell pellets were then lysed with ice-cold Chromatrap hypotonic buffer (500 L, 10 min).Further nuclear lysis occurred via treatment on ice with Chromatrap lysis buffer (100 L, 10 min).Lysate was then diluted in nuclease-free water (1 mL) to yield a final concentration of 90 ug/mL of RNA and 30 ug/mL of DNA quantified using a qubit 4 fluorimeter and the qubit RNA and DNA broad range kits.Nucleic acids were removed by treating lysate with benzonase (1,000 units, 24 hours, 37 °C, Millipore).
RNA was extracted from U2OS cells (4 x10 6 ) using Qiagen's RNeasy extraction kit, following the manufacturer's instructions.Briefly: adherent cells were trypsinized, counted and aliquoted.Cell pellets were then lysed with the RNeasy lysis buffer and ran through Qiagen's QIAshedder homogenizing columns.Cell lysate was then washed with RNeasy wash buffer and an on-column DNase I digestion was performed, followed by RNA purification according to the kit's instructions and RNA elution with RNase-free water (50 L).The final concentration of RNA obtained was measured on a NanoDrop One UV-Vis spectrophotometer.
To measure cellular toxicity of the probe, MTS assays were conducted using the Promega CellTiter 96 Aqueous One Solution Cell Proliferation Assay kit, following manufactures instructions.Cells were incubated with TOR-G4 over 6 hours at varying concentrations from 5-2,000 nM.Absorbance of the MTS reagent was then measured at 490 nm.The experiment was conducted in triplicate and survival curves were fit using the GraphPad Prism Inhibitor vs. response -Variable slope model.
FLIM images were obtained on a Leica SP5 II confocal microscope coupled to a TCSPC module (Becker & Hickl GmbH) following excitation with a pulsed diode laser at 477 nm (Becker & Hickl GmbH, 20 MHz).For two-photon excitation FLIM images, a femtosecond Ti:sapphire laser (Coherent, 80 MHz) at 760 nm was used.Fluorescence was collected in a 550-700 nm window with a PMC-100-1 photomultiplier tube detector (Hamamtsu) for 500 s.Images were acquired at 256 x 256-pixel resolution.High resolution zoomed images were acquired with 516 x 516 pixels, this resulted in larger distinctions in lifetime being seen between nucleoli and rest of the nucleus and a corresponding change in the pseudo pixel coloring.
FLIM images were analyzed with FLIMfit software (Imperial College London) 6 : whole cells were manually segmented and fluorescence decays were fitted pixel-wise to a bi-exponential decay function using the maximum likelihood algorithm, with deconvolution from the IRF.Scattering of light and peak offset were fitted locally using FLIMfit default settings.The reported cellular lifetimes are the intensity-weighted average fluorescence lifetimes as calculated in equation 2.
Nuclease experiments were performed by monitoring TOR-G4 fluorescence following the incubation of fixed cells with either DNase I (200 units/well, Qiagen), RNase H (0.2 U/L, New England BioLabs), Ambion RNase A (0.1 µg/L, Invitrogen) or RNase T1 (2 U/L, Life Technologies) at 37 °C for 30 mins prior to probe incubation.
For the G4 displacement assay, Ni-Salphen or PhenDC3 (1 µM) were added to cells following probe incubation and imaged by FLIM over approximately 7 hours.
All cell imaging experiments are an average of at least two independent biological repeats.Instances where images of three biological repeats were used are highlighted in relevant Figure captions.
Statistical significance of perturbation experiments was assessed with the Welch t-test in Graphpad prism.We note that it is not possible to fully deconvolute the effects of binding to nucleic acids from that of aggregation, which affect the absorbance and emission of the probe (see Figure S20).Error bars are standard deviation of experiments performed in triplicate.Associations constants calculated by fitting binding curves to Graphpad prism one site -total model.

Figure S28
-Emission spectra of TOR-G4 in aqueous buffer (black) and hexane (blue).Excitation at 470 nm.No significant shift in the emission peak is observed when changing solvent, suggesting the emission is not influenced by charge transfer processes.

Figure S8 -Figure S9 -Figure S10 -
Figure S8-Absorbance spectra of TOR-G4 (2 M) in aqueous buffered solutions and bound to G4, total RNA and duplex DNA sequences (300 g/mL).A clear spectral change is visible upon interaction of TOR-G4 with G4 DNA.

Figure S13 -
Figure S13 -A) Confocal and B) FLIM images of TOR-G4 in U2OS cells before and after transcriptional inhibition with DRB (1 uM), 477 nm excitation 550-600 nm detection.C) Quantification of probe cellular lifetime before and after DRB treatment.

Figure S14 -Figure S15 -Figure S16 -
Figure S14-Emission spectra of TOR-G4 (2 M) recorded in aqueous buffered solution in the presence of c-MYC DNA (100 g/mL) with increasing concentrations of KCl (0-1000 mM).The decrease in the intensity of the 660 nm band is consistent with disaggregation of TOR-G4 upon increasing ionic strength, or K + ions.

Figure S17 -
Figure S17 -Fluorescence lifetime component analysis of TOR-G4 (2 M) bound to various DNA and RNA topologies (100 g/mL), showing lifetime components A) Tau1 and B) Tau 2, and decay amplitudes C) Amplitude 1 and D) Amplitude 2. Excitation at 470 nm, detection at 540 nm.The original time-resolved decays are shown in Figure S11.

Figure S23 -
Figure S23 -Confocal image of TOR-G4 in U2OS cells, excitation at 514 nm and detection at 550-700 nm.Corresponding FLIM image in Figure 3A of main text.

Figure S24 -
Figure S24 -Phasor analysis of TOR-G4 fluorescence decays within U2OS cells.A) Cellular segmentation of high lifetime portion of phasor plot.B) Cellular segmentation of low lifetime portion of phasor plot.Analysis performed in FLIMfit.

Figure S26 -
Figure S26 -A) Confocal intensity and B) FLIM images of TOR-G4 in U2OS cells incubated at 2 M or 5 M for 2 hours.C) Correlation between fluorescence intensity and fluorescence lifetime of cells incubated with either 2 M (blue crosses) or 5 M (black crosses) of TOR-G4.Excitation at 477 nm and detection at 550-700 nm.

Figure S27 -
Figure S27 -Investigating photostability of TOR-G4.A) Confocal image of cells prior to irradiation B) First FLIM image of irradiated cells (500s acquisition time).C) Confocal image of cells after first image acquisition -red box shows the irradiated area.D) Second FLIM image of cells -red box shows the previously irradiated area.E) Fluorescence lifetime of TOR-G4 within U2OS cells across 6 hours of continuous imaging.Excitation at 477 nm and detection at 550-700 nm.

Table S2 -
List of all nucleic acid sequences characterized

Table S3 -
Goodness of fit (chi-sq value) of TOR-G4 decay to a biexponential function when bound to various DNA/RNA structures.