Direct visualization of both DNA and RNA quadruplexes in human cells via an uncommon spectroscopic method

Guanine-rich DNA or RNA sequences can fold into higher-order, four-stranded structures termed quadruplexes that are suspected to play pivotal roles in cellular mechanisms including the control of the genome integrity and gene expression. However, the biological relevance of quadruplexes is still a matter of debate owing to the paucity of unbiased evidences of their existence in cells. Recent reports on quadruplex-specific antibodies and small-molecule fluorescent probes help dispel reservations and accumulating evidences now pointing towards the cellular relevance of quadruplexes. To better assess and comprehend their biology, developing new versatile tools to detect both DNA and RNA quadruplexes in cells is essential. We report here a smart fluorescent probe that allows for the simple detection of quadruplexes thanks to an uncommon spectroscopic mechanism known as the red-edge effect (REE). We demonstrate that this effect could open avenues to greatly enhance the ability to visualize both DNA and RNA quadruplexes in human cells, using simple protocols and fluorescence detection facilities.

. A, Jablonski diagram depicting the ground (S 0 ) and excited states (S 1 ) of a fluorophore (red rectangle) in dilute conditions, in which solvent (blue circles) relaxation occurs more rapidly (τ R ≈ 10 -10 s) than light emission (τ F ≈ 10 -9 s). B, Jablonski diagram corresponding to the same fluorophore in a constrained environment The Vavilov's rule is pursuant to the Kasha's rule, which stipulates that photon emission occurs only "from the lowest excited state of a given multiplicity" (IUPAC definition), that is, that the S 1 "S 0 transition occurs only from the lowest energy sublevel of S 1 , accessed through vibrational relaxation (dashed arrows, Figure S1A). However, some exceptions to the classical applications of the Kasha-Vavilov rules exist. In dilute solution, the S 1 "S 0 transition (emission) usually occurs in the 10 -9 to 10 -6 -s time scale (τ F ), that is, slower than the solvent relaxation (τ R ≈ 10 -10 s) that leads to an alignment of solvent dipoles opposite to the fluorophore's dipole to decrease the interaction energy ( Figure S1A). In highly ordered environments, solvent mobility is restricted and solvent relaxation becomes slower, that is, τ R ≥ τ F : this leads to various intermediate states between the initial (S 1 ) and the final, relaxed excited state (S 1Rel. ), with partially aligned solvent dipoles ( Figure S1B). This phenomenon is of critical importance in polar environments in which dipole-dipole interactions have significant contributions; gradual solvent relaxation lowers the energy level of each intermediate, thus decreasing the S 0 n Rel. "S 1 n Rel. transition cost that could be achieved with photons of lower energy, i.e., with red-shifted absorbance wavelengths (λ n ). Therefore, absorption spectra of fluorophore in constrained environments are the sum of a set of new and lower transition energies (E 1 , E 2 .. E x ), but individual contribution cannot be distinguished,

S3
making the absorption poorly-defined on the red edge of its spectrum: this is called the inhomogeneous broadening. Inhomogeneous broadening is thus responsible for the REE; this effect does not violate the Kasha-Vavilov rules per se, it solely originates in their application in very specific conditions.

II. Confirmation of REE with TERRA.
The REE is characterized by both the dependence of the wavelength of the emission maximum on the excitation wavelength, λ em max = f(λ ex ), and the reciprocal dependence of the wavelength of the excitation maximum on the emission wavelength, λ ex max = f(λ em ).

S5
consequently confirm the suitability of N-TASQ for the visualization of both DNA and RNA quadruplexes by confocal microscopy.
b/ Reverse titrations with c-Myc: to confirm also the generality of the fluorescence response, reverse titrations were subsequently performed with c-Myc (10µM) and increasing amounts (0 to 10µM) of N-TASQ under excitation at wavelengths that correspond to lasers of standard confocal microscopes, that is, λ ex = 408, 488 and 555nm. As seen in Figure S4, the N-TASQ emission maxima at 474, 519, 592, 607 and 647nm increase in a doseresponse manner, thus confirming again the suitability of N-TASQ for the quadruplex detection by confocal microscopy.    This issue can be conveniently addressed on the basis of the difference between the quantum gains of REE probes versus conventional fluorophores. Here, we have stained fixed MCF7 cells with either N-TASQ (100µM), or BG4 (detected through incubation with a secondary antibodies conjugated to an Alexa594 probe), or both concomitantly. As seen in Figure S9A,B, images collected upon irradiation at 555nm and collected through the Alexa filter (585nm and above) show that the pattern obtained in the presence of both N-TASQ and BG4 (right panels) correspond to that of BG4 alone (center panels) only, owing to the weak emissive nature of REE probes. This highlight that co-staining experiments can be conveniently performed with both REE and conventional stains. It should be noted that BG4 results are somewhat surprising in MeOH-fixed cells, since nucleoli seems to be strongly labeled while such staining has not been reported in the two initial articles (G. Biffi et al., Nature Chem. 2013, 5, 182 & Nature Chem. 2014. This was further investigated via two-photon investigations that provides higher-resolution images (experiments were performed on a Nikon A1-MP scanning microscope (Nikon, Japan) with a ×60 Apo LWD objective (NA: 1.27, Water Immersion, Nikon, Japan)). As seen in Figure S9C,D, collected images confirm that BG4 labels quadruplexes mostly in cytoplasmic sites (RNA quadruplexes) after PFA-triton fixation and in nuclear sites (DNA quadruplexes) after MeOH fixation, and that bright foci seen in Figure S9B (center and right panel) actually correspond to isolated nuclear foci and not to nucleoli labelling. BG4 + AlexaFluor594-tagged secondary antibodies (referred to as "2Ab" and "AF-594-3Ab"), and 3-N-TASQ and AF-594-3Ab only. Conclusions drawn from the images seen in Figure S10 are the following: 1-this series of images confirms that the DAPI and FITC channels are completely reflective of N-TASQ staining, while the Alexa channel is not ( Figure S10A); 2-the S10 labelling pattern seen in Figure S10B is in agreement with the one seen in Figure 5B (main manuscript) albeit with a lower contrast due to the lower N-TASQ concentration used herein; 3-this series of images also shows that BG4 incubation precludes efficient N-TASQ labelling when it is used at low concentrations possibly due to a competition between the antibody and the ligand ( Figure S10A versus S10B); and 4-that the low-intensity red cytoplasmic background is an artefact of the 2AB+AF594-3Ab antibody per se ( Figure S10C).
Collectively, these experiments demonstrate that N-TASQ and BG4 can be used concomitantly in a reliable manner and beyond this, that they have similar cellular targets therefore lending credence to the capability of N-TASQ to detect quadruplexes cells.