Cationic helicenes as selective G4 DNA binders and optical probes for cellular imaging

The important role that G-quadruplex DNA (G4 DNA) structures play in regulating biological processes is becoming widely recognised. These structures have also been proposed to be attractive drug targets. Therefore, there has been significant interest in developing small molecules that can selectively bind to G4 DNA over other topologies. In this paper we investigate the interaction between DNA and helical compounds (helicenes) based on a central carbocation trisubstituted with aromatic rings. We show that the non-planar structure of these helicenes results in a significantly reduced affinity for dsDNA when compared to their planar analogues, whilst maintaining a high affinity for G4 DNA. Additionally, the right- and left-handed enantiomers of one of these helicenes recognise the chiral DNA environments of G4 and dsDNA differently. We show that upon DNA binding the helicenes display a fluorescence switch-on effect, which we have successfully used for cellular imaging in live and fixed U2OS cells, staining mitochondria and the nucleus, respectively.


Introduction
It is increasingly recognised that non-canonical DNA structures (i.e. non-duplex DNA) have important biological functions in living organisms. 1 One such structure is the guaninequadruplex (G4), a tetra-stranded helical assembly that forms in guanine-rich sequences of DNA. Bioinformatic studies as well as growing experimental evidence indicate that G4s are involved in a number of biological processes including telomere maintenance, replication and regulation of gene expression. 2,3 Because of their proposed biological roles, G4s have been intensively studied as potential targets for the development of drugs, particularly for cancer. 4 While G4s are thermodynamically stable and form readily in vitro, their presence in a cellular environment is proposed to be transient. This is due to the double stranded structure being the predominant topology in coiled DNA as well as the presence of dedicated helicases to unfold G4 structures. 5 However, the formation and stability of G4s can be signicantly enhanced in the presence of small molecules that preferentially bind to these tetra-stranded structures over other DNA topologies. [6][7][8][9] Thus, G4 DNA binders can shi the duplex-quadruplex equilibrium and prevent helicases from resolving G4s. 5 It has also been shown that small molecules can modify the interaction between G4s and proteins, particularly in the telomeres. 10 This, together with the proposed biological roles of G4s, has prompted the development of a large number of small molecules designed to bind selectively to G4s over other topologies. 6,8,9,11 Some of these molecules have been shown to trigger a number of biological responses in cells that are consistent with G4 stabilisation, while others have been successfully used as optical probes to visualise and detect G4 structures in vitro and, in a few cases, in live cells. [12][13][14][15][16] Most G4 DNA binders are based on planar, polyaromatic molecules featuring positively charged substituents. As discussed extensively elsewhere, the planar core binds to the guanine tetrads via p-p interactions, while the substituents provide means to increase solubility and DNA affinity (e.g. with protonated amines), as well as selectivity for G4s over duplex DNA (dsDNA). 6,8,9,11 While this strategy has yielded some very strong G4 DNA binders, alternative non-planar structural motifs have been explored with the aim of improving selectivity for a specic G4 topology, not only over duplex DNA, but also over other G4 topologies. 8 One such strategy has been to exploit the explicit chiral environment formed in both duplex and G4 DNA, designing enantiomerically specic DNA binding molecules with stereoselectivity originating from octahedral metal-centres or from steric hindrance. For example, Thomas and co-workers studied the DNA binding properties of the di-ruthenium complex [{Ru(bipy) 2 } 2 (tpphz)] 4+ where each of the two metal centres is in a chiral octahedral environment. They showed that the LL isomer has ca. 40 times higher affinity for human telomeric G4 DNA than the DD isomer. 17 Another class of chiral G4 DNA binders are metallohelicenes, where two octahedral metal centres are bridged by three chelating ligands. For example, Qu and co-workers reported that one of the two enantiomers of a di-nickel metallohelice recognised HTelo G4 DNA with high affinity and selectivity over duplex DNA. [18][19][20] Binding of related di-iron metallohelices to right and lehanded HTelo G4 DNA was recently investigated. The D, and L-enantiomers showed selectivity for right, and le-handed G4 DNA, respectively. 21 Organic helical molecules such as foldamers have also been shown to bind G4s with high affinity and selectivity. 22 It was proposed that the binding mode of these helical structures is different to that displayed by planar compoundsand likely to involve interactions with the backbone of G4 DNA. The M enantiomer of a cyclic helicene showed selectivity for B-DNA over the P isomer [K d (P)/K d (M) ¼ 2.0], whereas the reverse trend was observed when binding Z-DNA 23 Following on from this, a series of related chiral helicenes (with varying dihedral angles) showed enantioselective recognition of the M isomer to neighbouring G4s in the telomeric region. 24 We have previously reported that a triangulenium, DAOTA-Morph (also known as DAOTA-M2see Fig. 1), has good affinity for DNA and switches on its uorescence upon binding. 12,15 Its binding affinity towards different DNA topologies is ca. 2-fold higher for G4 than for duplex DNA (K d values ca. 1.7 mM and 1.0 mM for duplex and G4 DNA structures respectively). 12,25 Interestingly, the uorescence lifetime of this probe is highly dependent on the DNA topology it binds to, which has allowed us to use DAOTA-Morph to probe the formation of G4 structures via uorescence lifetime imaging microscopy (FLIM) in live cells. 12,15 With the aim of improving further the selectivity of this type of probe for specic G4 structures over other DNA topologies, herein we present studies with the helical analogues of trianguleniums, namely cationic helicenes. This type of compound, previously reported by Lacour and co-workers, 26 can oen be resolved into their stereoisomers and they are emissive over a wide range of wavelengths, depending on bridging atoms and substituents. These helical trianguleniums derivatives have been used as dyes for cellular imaging and shown to accumulate in mitochondria of live cells, and in one report, they were shown to bind to duplex DNA. 27 Herein we report the synthesis of the new racemic (R) helicences HL-OH(R) and HL-Morph(R) (the latter being a direct analogue of the planar DAOTA-Morph compound which has been successfully used as a lifetime-based optical probe for G4s 12,15 ) and their DNA binding properties, which are compared with those of the previously reported helicene HL-CH 3 (R). 26,[28][29][30][31][32] Further, the racemic mixture of HL-OH(R) was partially resolved into the corresponding M and P stereoisomers. We show that these compounds have high selectivity for G4 DNA over duplex DNA; this is particularly the case for the HL-OH(M) stereoisomer. We also show that in live cells HL-Morph(R) is cell permeable and localises in mitochondria, whereas in xed cells nuclear staining of DNA is possible, as conrmed using FLIM.

Results and discussion
Design and synthesis of helicenes as DNA binders As stated above, the planar aromatic molecule DAOTA-Morph [ Fig. 1(b)] binds to dsDNA and G4 DNA with similar affinities. 12 We aimed to reduce the probe's affinity for dsDNA, whilst maintaining strong G4 binding, thus resulting in improved selectivity for G4 over dsDNA. To achieve this, we designed and synthesised a cationic helicene molecule HL-Morph(R)   aromatic system, causes distortion away from a planar structure. This nonplanar helical topology should in principle be prevented from intercalating into adjacent bases in duplex DNA, favouring a weak groove binding arrangement [ Fig. 1(d)]. On the other hand, the helical molecule still has the appropriate structural features to bind to G4 DNA via end-stacking. We also synthesised HL-CH 3 (R) as a control, and HL-OH(R) which we were able to partially separate into the enantiomers HL-OH(P) and HL-OH(M) as conrmed by CD spectroscopy and chiral HPLC (see below).

HL-CH 3 (R), HL-Morph(R) and HL-OH(R) binding affinities to ctDNA and c-Myc
We rst set out to investigate the binding affinities of HL-CH 3 (R), HL-Morph(R), and HL-OH(R) to dsDNA (ctDNA) and G4 DNA (c-Myc). As indicated above, DAOTA-Morph, the planar aromatic analogue of HL-Morph(R), binds to ctDNA and G4 DNA with similar binding affinities. 12 By distorting the aromatic structure of DAOTA-Morph to make HL-Morph(R), we aimed to maintain strong p-p stacking to the G4 quartet, whilst reducing affinity for dsDNA. These helicenes are uorescent and their emission intensity in aqueous buffered media is switched-on upon DNA binding. Therefore, we were able to study the DNA binding affinity of HL-CH 3 (R), HL-Morph(R) and HL-OH(R) through titrations with ctDNA and c-Myc G4 DNA (see Fig. 2  show that the interaction of these helicenes with dsDNA is at least 50 times weaker than that displayed by DAOTA-Morph [K d ¼ 1.3 mM]. G4 binding was conrmed using a uorescent intercalator displacement (FID) assay, 33 with all three helicene complexes showing a similar ability to displace TO from c-Myc (Fig. S9 †).
We next investigated the uorescence lifetime (s w , a concentration independent parameter) of the helicenes upon binding to DNA. When free in aqueous buffer at pH 7.3, the helicenes' s w ranges between 2.6 and 5.1 ns (Fig. S12 †) while, upon DNA binding, a signicant increase in lifetime to between 8 and 12 ns is observed (see Fig. 2 and S12 †). Although the helicenes' uorescence lifetime cannot be used to discriminate between different DNA topologies (unlike DAOTA-Morph 12,15 ), the signicant increase of s w when bound to DNA was useful for cellular imaging studies in order to conrm that the dyes are bound to DNA in cellular organelles (see below).

Binding modes of HL Morph(R) and DAOTA-Morph to dsDNA
Given that HL-Morph(R) is ca. 100 times weaker binder towards ctDNA than DAOTA-Morph (Table 1), we investigated if this difference could be a result of a different binding mode. Structural perturbations of DNA induced by dye binding can be monitored using CD spectroscopy and used to investigate the dye binding mode. The characteristic CD spectrum of ctDNA (in its B form) shows an increase in the band at 277 nm upon DAOTA-Morph binding [ Fig. S13(a) and (b) †]. Similar spectral changes have been observed following the intercalation of planar aromatic dyes into dsDNA, tentatively assigned to the unwinding of the helical DNA structure to accommodate the intercalated dye. 34,35 Conversely, when the same number of HL-Morph(R) molecules per base pair are bound to ctDNA, a small decrease in band intensity at 277 nm is observed [ Fig. S13(a) and (b) †]. This minor change in the CD signal is characteristic of groove binding which results in minimal disruption to the double helical structure. 36,37 We next studied the accessibility of ctDNA bound DAOTA-Morph and HL-Morph(R) to uorescence quenching by iodide. If intercalated, the proximity of base pairs above and below the dye should protect against quenching, whereas this shielding will be less for groove binding dyes which are still exposed to the solvent environment. Once bound, DAOTA-Morph is almost completely protected from quenching by iodide [K SV ¼ 21.8 (free) and 3.1 (bound) M À1 ], whereas for HL-Morph(R) the quenching is in fact enhanced [K SV ¼ 36.1 (free) and 48.7 (bound) M À1 , Fig. S13(c) †]. This seemingly unusual enhancement in quenching when bound to dsDNA has been observed for other positively charged, weak groove binding dyes, 36,38 which was assigned to weakened electrostatic HL-Morph(R)-dsDNA interaction through increasing ionic strength of the solvent upon the addition of KI. As a result, HL-Morph(R) is released into solution allowing more efficient quenching. Molecular docking studies provided further evidence that DAOTA-Morph intercalates into dsDNA whilst HL-Morph(R) interacts via groove binding. These studies showed that HL-  Morph(R) is too wide to t into an intercalation site (Fig. S14 †), whereas DAOTA-Morph ts well. Taken together, this evidence indicates that distortion of the aromatic surface in HL-Morph(R) away from a planar structure did indeed result in a change in binding mode, which favours a weak groove binding mode over intercalation into dsDNA. This in turn accounts for the high selectivity of the helicenes for G4 DNA (over ctDNA) conrming our original hypothesis.
HL-OH(P) and HL-OH(M) binding affinities to c-Myc, BCL2, HTG4 and ctDNA As discussed in the previous section, HL-CH 3 (R), HL-Morph(R), and HL-OH(R) show high selectivity for G4 over dsDNA (Table  1)  open face of these quadruplex structures presents a modestly chiral environment for binding. 39 Using the published NMRstructure of c-Myc incorporating a G4-binder, 39 we performed docking studies on the ligand free structure [ Fig. 3(e)]. This conrmed the independent biding preferences for both HL-OH(P) and HL-OH(M) into a chiral pocket caused by the 5 0 overhang of the G4. We next investigated binding of HL-OH(P) and HL-OH(M) to a mixed parallel/antiparallel quadruplex structure that forms in the promoter region of the BCL2 gene, as it contains loop regions that interact with both G-tetrad faces, potentially forming an increased chiral environment for binding. 40 Indeed, the difference in binding between HL-OH(P) and HL-OH(M) [K d ¼ 7.5 mM and 5.1 mM, respectively, Fig. 3 and S14 †] is slightly increased when interacting with BCL2, as compared to c-Myc and HTG4.
A much bigger difference in affinity between the two isomers was observed when binding to ctDNA [ Fig. 3 and S18 †]. HL- Fig. 3(c)] shows a reduced interaction with ctDNA compared to HL-OH(M) [K d ¼ 110 mM, Fig. 3(c)], with selectivity for c-Myc over ctDNA of 134 and 29, respectively. As binding to BCL2 is weaker, the selectivity over ctDNA is decreased for both HL-OH(P) and HL-OH(M) (84 and 22, respectively) compared to c-Myc.
Given the large difference in binding to ctDNA between the P and M isomers, we investigated if this difference could be observed by CD spectroscopy when HL-Morph(R) is bound to a large excess of ctDNA. Racemic HL-Morph(R) shows no bands in the CD spectrum, however, once added to a large excess of ctDNA, negative CD bands at 314, 374 and 472 nm develop [ Fig. S19(a) †]. The appearance of a CD signal in a region outside of any ctDNA absorption implies the enrichment of one isomer bound to ctDNA compared to one isomer free in solution. Based on the measured spectra of free HL-OH(M), and HL-OH(M) bound to ctDNA, we calculated CD spectra that would be expected if either the M or the P isomers had the stronger association constant [ Fig. S19(b) †]. The spectrum expected for the M isomer strongly bound to ctDNA, closely matches the experimental spectrum, conrming the preference of the M isomer in binding to ctDNA.

HL-Morph(R) staining in live and xed U2OS cells
We next investigated how HL-Morph(R) stains live and xed U2OS cells. We chose HL-Morph(R) due to low toxicity to live cells (Fig. S20 †), synthetic ease and the ability to compare against our previous results with structurally related lifetimebased G4 probe, DAOTA-Morph. 12 At the concentrations used in our live cell experiments (10 mM, 24 h), the cytotoxicity of HL-Morph(R) is negligible. Similarly to other previously reported helicene compounds, 27a HL-Morph(R) accumulates in mitochondria of live cells, conrmed by co-localisation with mitotracker green [MTG, Fig. 4(a)]. Again, this contrasts with DAOTA-Morph, which localises predominantly in the nucleus of live U2OS cells. 12 In xed U2OS cells, HL-Morph(R) passes the nuclear membrane, revealing both nuclear and non-nuclear staining [ Fig. 4(b) and (c)]. Recording FLIM maps of these xed cells, intensity weighted average lifetimes recorded within the nucleus are represented by a histogram with a peak maximum at ca. 9.5 ns, consistent with in vitro experiments for HL-Morph(R) bound to DNA [9.8 ns regardless of topology, Fig. 2(f)]. We note that signicantly higher uorescence intensity can be observed in the nucleoli [ Fig. 4(b), le panel], although the concentration and topology-independent lifetime measurement does not reveal any reason for this.

Conclusions
The vast majority of G4 DNA binders reported in the literature are based on planar polyaromatic systems. While planar molecules have high affinities to G4 DNA due to efficient p-p endstacking, they also tend to intercalate efficiently into duplex DNA. Herein we have demonstrated that by breaking the planarity of polyaromatic systems, it is possible to generate much more selective G4 binders. More specically, we have shown that the uorescent helicene compounds, HL-CH 3 (R), HL-Morph(R), and HL-OH(R) have high selectively for G4 over duplex DNA, and further selectivity can be introduced through partial chiral resolution of HL-OH(R) into HL-OH(P) and HL-OH(M). The distorted core of the helicene compound reduces affinity for dsDNA compared to the planar analogue DAOTA-Morph, whist maintaining strong affinity for G4 DNA [selectivity ¼ 134 for HL-OH(P)]. We also show that HL-OH(P) and HL-OH(M) bind differently to G4 and dsDNA topologies, with HL-OH(M) consistently displaying a higher affinity. We have also used the increased uorescence intensity and lifetime of HL-Morph(R) upon DNA binding to enable cellular imaging studies. In live U2OS cells this helicene accumulates in the mitochondria, whereas in xed cells, HL-Morph(R) passes the nuclear membrane and binds to DNA, as conrmed using uorescence lifetime imaging microscopy, FLIM.

General synthetic procedures
All chemicals were purchased from commercial sources and used as received, unless stated otherwise. 1 H and 13 C NMR spectra were recorded using either a 400 or 500 MHz Bruker Avance Ultrashield NMR spectrometer at 296 K. Spectra were referenced internally by using the residual solvent ( 1 H d ¼ 3.34 and 13 C d ¼ 49.86 for CD 3 OD-d 4 relative to SiMe 4 ). ESI-MS spectra were recorded by Dr L. Haigh (Imperial College London) on a Bruker Daltronics Esquire 3000 spectrometer. HL-CH 3 (R) was synthesised and characterised according to published procedures. 26

Synthesis of HL-Morph(R)
Compound 1 (0.1 g, 0.198 mmol) and 4-(2-aminoethyl)morpholine (0.644 g, 4.96 mmol) were mixed with anhydrous 1methyl-2-pyrrolidinone (NMP) (1 mL) under an argon atmosphere in a microwave (MW) tube. This reaction mixture was stirred in a MW synthesizer for 10 min at 170 C then allowed to cool down to room temperature. CH 2 Cl 2 (5 mL) was added to the reaction mixture and was washed with a 1 M aqueous solution of HBF 4 (2 Â 2 mL). The organic phase was separated and dried over anhydrous Na 2 SO 4 and evaporated under vacuum. The resulting crude material was puried using ash chromatography to yield HL-Morph(R) (0.015 g, 11%). 1  removed under reduced pressure. The solid obtained was dissolved in 3 mL acetone and kept overnight at 0 C. The solid deposited at the bottom of the vial was separated from the mother liquor. The solvent of the latter was removed in a rotary evaporator to yield a second solid. CH 2 Cl 2 was added to each solid fraction (the original precipitate and the solid obtained aer evaporation) followed by HPF 6 (9 mg) and 0.6 mL KPF 6 (0.1 M solution in water) and stirred for 1 h. The organic layers were separated and puried using ash chromatography.
To conrm the enantiomeric resolution of HL-OH(R), samples were analysed by chiral HPLC as follows: a 100 mM stock solution of the corresponding compound (i.e. HL-OH(R), HL-OH(P) or HL-OH(M)) was prepared in MeOH. 1.5 equivalents of NaBH 4 were added, causing an immediate colour change from blue to colourlessdue to reaction of the helicene carbocation as described previously. 26 The solution was le for 6 h to allow for full NaBH 4 hydrolysis. Next, the corresponding solution was diluted to 50 : 50 hexane : MeOH before injection onto a Chiralpak AD-H 250 Â 4.6 mm column. An isocratic 60 : 40 hexane : isopropanol gradient was run and the peaks corresponding to each stereoisomer integrated to give an ee of 96% for HL-OH(M) and 32% for HL-OH(P) (see Fig. S22(b) † for chromatograms). The two enantiomers were analysed by CD spectroscopy and the spectra are shown in Fig. S22(a). †

Time-correlated single photon counting (TCSPC)
Time-resolved uorescence decays were obtained using an IBH 5000F (Jobin Ybon, Horiba) time-correlated single photon counting (TCSPC) device equipped with a 635 nm NanoLED as an excitation source (pulse width <200 ps, HORIBA) with a 100 ns time window and 4096 time bins. Decays were detected at l em ¼ 655 nm (AE4 nm) aer passing through a 645 nm long pass lter to remove any scattered excitation pulse. Decays were accumulated to 10 000 counts at the peak of uorescence decay.
A neutral density lter was used for the instrument response function (IRF) measurements using a Ludox solution, detecting the emission at the excitation wavelength. Decay traces were tted by iterative reconvolution to the equation I(t) ¼ I 0 (a 1 e Àt/s1 + a 2 e Àt/s2 ) where a 1 and a 2 are variables normalised to unity. The intensity-weighted average lifetime (s w ) was calculated using the equation: A prompt shi was included in the tting to take into account differences in the emission wavelength between the IRF and measured decay. The goodness of t was judged by consideration of the deviations from the model via a weighted residuals plot.

Oligonucleotide titrations
The helicene under study (2-6 mM) was dissolved in 10 mM lithium cacodylate buffer (pH 7.3) supplemented with 100 mM KCl and the UV/visible/uorescence/CD spectra and/or TCSPC lifetime recorded (where applicable). Increasing amounts of the oligonucleotide under study were added, maintaining a constant concentration of the helicene compound. Aer each addition, the mixture was le to equilibrate for >30 s (a time determined to be sufficient for equilibrium to occur) before the corresponding photophysical measurements were recorded. Fluorescence spectra (l ex ¼ 580 nm, l ex ¼ 590-850 nm) were integrated between 600 and 800 nm and the integrated intensity normalised against the absorption at the excitation wavelength. Titrations are plotted as DF, the difference between the normalised emission, and the normalised emission of the free helicene (rst point of the titration). For HL-OH(M) in aqueous buffer, broadening of the absorbance spectral bands can be observed, consistent with compound aggregation. In this case, DF was calculated as the difference between the normalised emission, and the normalised emission from the rst addition of oligonucleotide (second point of the titration).
Titration curves were tted to either a simple binding (which assumes the 2 stereoisomers are equivalent) 41,42 or competitive binding (which assumes independent binding for each stereoisomer) 43,44 models using a modied form of the MatLab script reported previously (Fig. S10 †). 41,42 Based on the titration data, a binding stoichiometry of two compounds to one G-quadruplex and two compounds per ve base pairs for ctDNA was used allowing for direct comparison with DAOTA-Morph. 12 No cooperativity between binding sites and no oligonucleotide uorescence response was assumed. Titrations of racemic mixtures of HL-CH 3 (R), HL-Moph(R) and HL-OH(R) were tted independently to the simple binding model to solve for the association constant (K H ) and the uorescence change on binding (k DG H ). Titrations with HL-OH(R) (0% ee), HL-OH(P) (32% ee) and HL-OH(M) (96% ee) were tted simultaneously to the competitive binding model solve for K P , K M , k DG P and k DG M . In the case of ctDNA, a complete titration was not possible due to a low binding affinity, which resulted in a less accurate t. To account for this, the values of k DG P and k DG H were xed from the results of competitive binding to c-Myc. Reported K d values are the reciprocal of the association constant. The uorescence 'switch-on' values for HL-CH 3 (R), HL-Morph(R), and HL-OH(R) were calculated as F/F 0 , where F is calculated from the asymptotic value of the binding t, and F 0 is the initial point on the binding curve, before the addition of oligonucleotide. N.B., this value includes correction for absorption at the excitation wavelength. The selectivities of G4 over dsDNA was calculated as ratios of K d values (Table 1).

Thiazole orange uorescence indicator displacement assay (TO-FID)
Experiments were carried out using a BMG CLARIOstar® Microplate reader with Greiner Bio-One half volume (100 ml well) plates using a method adapted from the literature. 33 Fluorescence titrations (l ex ¼ 475 nm, l ex ¼ 520 nm) were carried out using c-Myc G4 DNA (100 mM KCl, 10 mM Tris-HCl, pH 7.3). The nal concentrations in the plate were 1 mM G4 DNA, 2 mM thiazole orange (TO) and 0-20 mM helicene (0, 0.94, 1.25, 1.88, 2.50, 3.75, 5.0, 7.5, 10,15,20). Sample preparation was carried out by rst preparing double concentration stocks of helicene (40 and 30 mM) and a DNA/TO mixture (2 mM and 4 mM respectively). Helicene concentrations were prepared using serial dilutions at 50 mL per well, which was followed by the addition of 50 mL of the DNA/TO mixture and gently shaken for 5 minutes. Percentage displacement curves were calculated from the measured uorescence intensity (F), using: displacement ¼ 100 À [(F/F 0 ) Â 100], where F 0 is TO uorescence from the probe bound to c-Myc without added helicene. Displacements were tted to a Hill function which was used to calculate the DC 50 .

General cell culture
Human bone osteosarcoma epithelial cells (U2OS, from ATCC) were grown in high glucose Dulbecco's Modied Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS) at 37 C with 5% CO 2 in humidied air.

Cytotoxicity of HL-Morph(R) and HL-CH3(R)
Cytotoxicity of HL-Morph(R) and HL-CH 3 (R) in U2OS cell lines were investigated using the MTS assay. The cells were equally distributed (5 Â 10 3 cells per well) in a 96 well plate in a DMEM medium containing 10% FBS and incubated for 12 h under standard condition. The culture media was removed, and fresh media added with compounds at required concentrations (20, 10, 5, 2.5, 1.25 and 0.625 mM) in triplicate. Wells were maintained without compound (only cells) and without cells (only culture medium) as 100% and 0% viability controls, respectively. Aer 24 h incubation the cells were treated with MTS/ PMS solution and incubated for another 4 h before taking an absorption reading using a plate reader.

Confocal imaging
1024 Â 1024 resolution uorescence images were collected using an inverted confocal laser scanning microscope (Leica SP5 II). MitoTracker Green (MTG) emission (500-600 nm) was collected following one-photon excitation from an internal microscope laser at 488 nm, and helicene emission collected at (650-790 nm) following excitation from an internal microscope laser at 633 nm.

Fixed cell experiments
Cells were seeded on chambered coverglass (1.5 Â 10 4 cells, 250 mL, 0.8 cm 2 ) for 48 h. Cells were washed (x3) in ice cold PBS before incubation in ice cold paraformaldehyde (PFA, 4% in PBS) solution for 10 min, and a further wash (x3) with ice cold PBS. Fixed cells were further treated with HL-Morph(R) (20 mM, 0.5 h, 21 C) in PBS before being le under PBS. Cells were le under PBS for imaging to limit the effect of refractive index of the xation medium on the orescence lifetime. 45 Fluorescence lifetime imaging microscopy (FLIM) FLIM was performed through time-correlated single-photon counting (TCSPC), using an inverted confocal laser scanning microscope (Leica SP5 II) and a SPC-830 single-photon counting card (Becker & Hickl GmbH). A pulsed diode laser (Becker & Hickl GmbH, 640 nm, 20 MHz) was used as the excitation source, with a PMC-100-1 photomultiplier tube (Hamamatsu) detector. Fluorescence emission (650-790 nm) was collected through an Airy 1 pinhole for an acquisition time sufficient to obtain signal strength suitable for decay tting. For all live cell imaging, cells were mounted (on chambered coverglass slides) in the microscope stage, heated by a thermostat (Lauda GmbH, E200) to 37 (AE0.5) C, and kept under an atmosphere of 5% CO 2 in air. A 100Â (oil, NA ¼ 1.4) objective was used to collect images at either 256 Â 256 or 512 Â 512 pixel resolution, as stated in the text. The IRF used for deconvolution was recorded using reection of the excitation beam from a glass cover slide.
Lifetime data were tted using the FLIMt soware tool developed at Imperial College London (v5.1.1, Sean Warren, Imperial College London) to a bi-exponential function, and the intensityweighted lifetime (s w ) calculated using eqn (1). 5 Â 5 and 9 Â 9 square binning was used to increase signal strength for images recorded at 256 Â 256 and 512 Â 512 resolution, respectively. A scatter parameter was added to the decay tting to account for scattered excitation light. Before tting, a mask was applied to the images to analyse individual cell nuclei staining, or extra-nuclear staining. A threshold was applied to the average of each nucleus to require a minimum of 175 at the peak of the decay and a goodness-of-t measured by c 2 of less than 2.

Molecular docking
Molecular docking was performed using AutodocVina. 46 Ligand structures of isolated compounds were minimised in Gaussian using Density Functional Theory (DFT), the B3LYP functional, and a 6-31G(d,p) basis. Compounds were docked into the lowest energy form of c-Myc (PDB ID:5W77), 39 or dsDNA (PDB ID:1Z3F), 47 already stripped of their bound ligands. A grid box encompassing the entire quadruplex was used for blind docking. The lowest energy solution was used and the docked structures were visualised using PyMol v2.3.4.

Data availability
The essential spectroscopic characterisation and analytical data is included in the ESI. † Additional data is available from the corresponding authors upon reasonable request.