Stiff‐Stilbene Ligands Target G‐Quadruplex DNA and Exhibit Selective Anticancer and Antiparasitic Activity

Abstract G‐quadruplex nucleic acid structures have long been studied as anticancer targets whilst their potential in antiparasitic therapy has only recently been recognized and barely explored. Herein, we report the synthesis, biophysical characterization, and in vitro screening of a series of stiff‐stilbene G4 binding ligands featuring different electronics, side‐chain chemistries, and molecular geometries. The ligands display selectivity for G4 DNA over duplex DNA and exhibit nanomolar toxicity against Trypasanoma brucei and HeLa cancer cells whilst remaining up to two orders of magnitude less toxic to non‐tumoral mammalian cell line MRC‐5. Our study demonstrates that stiff‐stilbenes show exciting potential as the basis of selective anticancer and antiparasitic therapies. To achieve the most efficient G4 recognition the scaffold must possess the optimal electronics, substitution pattern and correct molecular configuration.


Introduction
G-quadruplexes (G4) are aclass of nucleic acid secondary structure that form from sequences rich in guanine. In contrast to the classical duplex structure stabilized by Watson-Crickb ase pairing, the nucleic acid strand folds to create as tacked arrangement of G-tetrads:s quare-planar ensembles of four guanine residues stabilized by Hoogsteen hydrogenb ondinga nd coordination to ac entral cation,s uch as Na + or K + . [1] These structuresr eceive significant attention as potentialt herapeutic targets. [2] Of particulari nterest is the quadruplexf ormed by the human telomeric sequence;t elomerase expression is upregulated in cancerc ells and partly responsible for cellular immortality by preventing telomere shortening, leading to uncontrolled proliferation. [2][3][4] Furthermore, G4s are found in the promoter regionso fs everal genes associated with the devel-opmento fc ancer( e.g., c-myc, [5] BCL2 [6] and c-kit [7] ), where stabilization of the folded G4 by ligandsi sp roposed to inhibit the binding of transcription factorsl eading to downstream silencing of oncogene expression. [8] More recently, we reported the identification of putative G4forming sequences in the genomes of the protazoanp arasites Trypanosoma brucei and Leishmania major. [9] Both organisms contain frequento ccurrences of the human telomeric sequence [10] in addition to several further unique G4s. For example, the EBR1 sequence occurs 33 times in the T. brucei genome and was subsequently demonstrated to form as table quadruplex under physiological conditions in biophysical studies. The sequence occurs in genomic regionscoding for several proteins including ac ysteinep eptidase and ap urine transporter. [9] G4st herefore present ap otential opportunity as at arget for novel antiparasitic therapies, for which there is an urgent need for further development. [11] The T. brucei parasite, responsible for the Human African Trypanosomiasis (HAT) disease, endangers 69 million people across Sub-Saharan Africa, [12,13] and existing therapies suffer from severe limitations including drastic side effects [14] and emerging drug resistance [15] in the parasitic strains.G 4l igands have long been studied as the basis of anticancer and antiviral therapeutics, but their potential as antiparasitic agents has been neglected until recently. [9,16] The handfulo fc ompounds explored to date are primarilyn aphthalene diimide derivatives, already widely studied as the basis of potentialanticancer drugs. Though the activities are promising, the identification of further DNA-binding chemotypes capable of exerting selectivea ntiparasitic activity is of criticalr elevance to exploring this therapeutic hypothesis.
In ar ecents tudy,w ei dentified an ovel G4-binding chemotype derived from stiff-stilbene, 1,t he first example of aG 4 ligandd erived from this scaffold. [17] Whilst ligand 1 induces high thermals tabilization of the potassium form of human te-lomeric DNA,t he same compound causest he unfolding of the antiparallel form of the same sequence formed in sodium buffer.T hese intriguing activities suggesteds tiff-stilbenea sa promisings caffold for the design of potent and selectiveG 4binding agents, prompting us to consider the potential applications of these derivatives as anticancer and antiparasitic agents. To wards this end, we synthesized further analogues of the previously reported compound and investigated the binding of thesen ew derivatives using FRET thermal melting assay to as mall library of G4 DNA sequences that are relevant in the targetingo fc ancers and parasitic infections. Binding was additionally validated through circular dichroism spectroscopy,U V/ Vis titration studies and NMR spectroscopy.W et hen took initial steps to validate the therapeutic utility of thesec ompounds by examining their toxicity and localization (intracellular uptake) in parasitic cultures and mammalian cells. Our results suggest as trikings electivity of the lead compounds for the target pathologies and recommend this class of DNA-binding molecule as ac andidate for further development of an ew generation of anticancer and antiparasitict herapeutic leads.

Results and Discussion
Design and synthesis of stiff-stilbene ligands To evaluate the effect of ligand structure on the G4 binding properties and antiproliferative activities of stiff stilbenel igands,w ep repared as mall collection of compounds designed to investigate these effects ( Figure 1). These derivatives incorporate av arietyo fs tilbene stereochemical configuration, electronic effects, substitution pattern, and length and nature of the lateral groups.I np articular,w ew ere keen to examine the effect of exchanging the rigidm ethylpyridinium moiety in the previously reportedc ompounds (1 and 2)f or am ore flexible amine-derived side-chain (compounds 4 and 5), previously demonstrated by several groups to confer good G4 affinity by forming electrostatic interactions with the G4 grooves when protonated at physiological pH. [18,19] Meanwhile, the relocation of the pyridiniumg roup to an alternative position on the scaffold (compound 3)s ignificantly alters the ligand shapef rom a more compactS -shaped molecule to an extended linear structure, potentially influencing the ability to interact with G-tetrads as well as modifying the electronics tructure of the molecule. Finally,w ew ere keen to furthere xamine the effectso f the configuration of the central stilbene core on G4 binding and in vitro activity by am ore detailed comparison of the E and Z forms of the ligands( 1 vs. 2 and 4 vs. 5). With routes to compounds 1 and 2 already established, [17] relatedl igand 3 was synthesized using an analogousprotocol (Scheme 1). Briefly,M cMurry couplingo f5 -bromoindanone p-6 afforded bromide p-7,w hich was coupled to 4-pyridylboronic acid 8 using standard Suzukic onditions to afford intermediate p-(E)-9. Methylation proceeded smoothly to afford compound 3,w hich was easily purified by trituration in acetone. Preparation of methylpiperazine analogues 4 and 5 proceeded straightforwardly by Buchwald-Hartwig amination of bromides m-(E)-7 and m-(Z)-7 with 1-(3-aminopropyl)-4-methylpiperazine, and the desired compounds 4 and 5 were obtained as the trifluoroacetate salts following purification by HPLC. Full synthetic procedures and characterization of the compounds is provided in the Supporting Information.

Thermal melting assays
With the ligand series in hand, we initially soughtt ocompare their ability to stabilizeavariety of G-quadruplex and duplex DNA structures with previously reported compounds by means of af luorescence-based thermal melting assay. [20] Briefly,t he DNA sequences of interestw ere conjugated to af luorescence donor (FAM) and to an acceptor (TAMRA) at the 3'-a nd 5'ends, respectively.W hen folded into secondary structures, the proximity of the donor and acceptor results in quenching of the donor fluorescenceb ye nergy transfer (FRET) to the acceptor.U pon raising the temperature, the secondary structure denatures, causing the acceptora nd donor to move apart, observed as an increasei nd onorf luorescence. The resulting Ligands that stabilizet he folded structure cause the T m to increase relative to that of the sequence withouta dded ligand. The resulting difference in meltingt emperature (DT m )i ndicates the ability of the ligand to stabilize the folded structure (see the Supporting Information for full detailso ft he experimental protocol). New ligands 3-5 were evaluatedinthis assay against the telomeric sequence F21T in potassium and sodiumb uffers and the FmycT sequence in potassium buffer.Aduplex DNA hairpin (F10T) was also included to assess the ability of the ligands to discriminate between the different types of DNA secondary structure. Furthermore, all ligands 1-5 were screened againstt he newly available Febr1T G4 sequence found in the genome of T. brucei. [9] The DT m values for all ligands( at 5 mm concentration) are displayed in Table 1, and dependence of DT m on ligand concentration is displayedi nF igure 2a for the potassium form of F21T,t he sequence for which the ligands appear to be most affine. Concentration-dependence curves for the other DNA sequences are provided in the Supporting Information ( Figures S6-S9).
Scheme1.Generalsynthetic route to stiff-stilbene G4 ligands.   From our initial screen,t he most striking feature is that the pyridiniuml igand class (1-3)s how superior binding when compared to the flexible methylpiperazine-derived ligands (4)(5). For example, ligand 4 only displays significant stabilization in the quadruplexs equences at 5 mm ligand concentration and above,w hilst all three pyridinium derivatives 1-3 remain potent down to 1 mm ligand concentration ( Figure 2a and the Supporting Information, Figure S6-S8). Given that many potent G4 ligands, such as the well-explored naphthalene diimide (NDI) family,feature flexible basic side-chains of the typeexemplified by 4 and 5, [18,19,21] it is somewhat surprisingt hat such a design feature does not appear to confer high G4 stabilizing ability to the stiff-stilbene scaffold, despite this core displaying ac lear ability to serve as the basis of selective G4 ligands (as exemplified by compounds 1-3). Compounds in the NDI series functionalized with alkyl chains bearing methyl piperazine termini have been found to bind G4 by forming stacking interactions between the NDI core and external tetrads of the G4, with the basic side chains residing in the G4 grooves. [19] The comparatively poor performance of the stiff-stilbene analogues of these compounds suggestst his scaffold is comparatively ineffective at achievings uch ab inding mode and perhaps the high stabilization displayed by ligands 1-3 may originate from alternative binding modes, such as groove binding rather than end-stacking (see furtherd iscussionb elow).
Both Z ligands( 2 and 5)a re comparatively weak G4 binders in comparison to their E counterparts. Whilst the E ligand 4 induces appreciable stabilization in the potassium form of F21T K + (DT m at 5 mm = 6 8C), Z isomer 5 inducesanegligible stabilization at the same concentration. Asimilareffectwas previously observed for ligands 1 and 2.G iven the significantly different molecular geometricalc onfigurations of the Z and E stiff stilbene core, these resultss uggest that the olefin configuration of the central scaffold is critical in determiningt he activity in this class of compound, and is independent of then ature of the lateral groups.M eanwhile, the DT m values observed for pyridinium ligand 3 are, in general, significantly lower than those for compound 1.T his suggests that the more compact arrangement of 1 proves superior for G4 binding when compared with the extended linear configuration of ligand 3.
It is important to highlightt hat all ligands 1-5 induce only negligible to minor stabilization on the duplex DNA model F10T at concentrations that induces ignificant stabilization of G4. This indicates the stiff-stilbene scaffold possesses an inherent selectivity for binding to G4 over duplex sequences. Given the particularly high activity of compounds 1-3,w ew erek een to verify the observed selectivity of these compounds by running ac ompetition experiment in whicht he effect of increasing concentrationso fu nlabelled duplex DNA (ds26) on the DT m of the G4 sequences is measured. Under such conditions, off-target binding is observeda sr eduction in the DT m value relative to that obtainedi nt he absence of the competing species. All three pyridinium ligands 1-3 discriminate moste ffectively between the F21T K + sequence and duplex DNA (Figure 2b). Even at 25 molar equivalents of ds26, over 50 %o ft he thermals tabilization induced by ligands 1-3 is retained for this sequence. The ligandsa ppear to be significantly less selective againstt he same sequence in sodium-containing buffer (see the Supporting Information, Figure S10), particularly ligand 2, for whicht he thermals tabilization inducedb yt he ligand is lost entirely at 12.5 molar equivalents of ds26c ompetitor and above.T hisb ehavior is reflectiveo ft he lower affinity of the ligands for F21T in sodium-containing buffer observed in Ta ble 1. Both ligands 1 and 3 (the more potent G4 ligands) retain over 50 %o ft he induced thermal stabilization of the FmycT sequence (see the Supporting Information, Figure S11) at 25 molare quivalents of ds26, confirming ah ighd egree of selectivity for thesel igands. On the other hand, ligand 2 is again less selective,and agradualerosion of the inducedstabilization on increasingt he concentration of duplex competitor is observed. Againstt he FebrT found in the T. brucei genome, ligand 1 outperforms ligands 2 and 3 for selectivet argeting of the G4 structure over duplex DNA (see the Supporting Information, FigureS12). Taken together,t he resultso fT able1 and the competitiona ssays across the panel of G4s indicatec ompound 1 as the lead ligand in the series in terms of both the magnitude of the inducedt hermals tabilization of G4, and its general selectivity for G4 DNA in favor of the double-stranded secondary structure.

Circular dichroism spectroscopya nd UV/Vist itration studies
To further examinet he nature of interaction of ligands 1-5 with G4 topologies, we employed ac ombination of circular dichroism (CD) spectroscopy and UV/Vis titration studies (techniquesc ommonlyu sed for the study of G4/ligand interactions). [22] Both approaches have the advantage that modification of the oligonucleotide sequence with artificial fluorophores is not necessary,a llowing validation of binding effects against natural sequences where the folding topology is wellvalidated by structural studies. These studies thereforea llow inferenceo fp ossible bindingm odes along with quantification of binding affinity under physiologically-relevant conditions. Prior to undertaking these studies, we first validated that all ligands were stable to photoisomerization/photodegradation under the assayc onditions (see the Supporting Information, FigureS13). In this study,w ec hose to examinet he binding of the ligandst ot he hybrid (telo23, potassium) form of the telomeric sequence, [23] since this sequence occursi nb oth human and parasitic genomes and is the most relevant topology of this G4 in vivo owing to the high concentration of potassium inside cells. [24] The circulard ichroism spectrum of telo23 is characterized by ap ositive band at 290 nm, and aw eaker shoulder band at 260 nm indicative of ap redominant hybrid G4 topology. [25] Binding of the pyridinium derivatives 1-3 is evidenced by hyperchromicity in the positiveb and at 290 nm (Figure 3a,cand e) and the increased intensity of the negative band at 240 nm. The effect is mosts triking for compound 1 (as previously reported [17] ), which is consistent with this compound being the more potent of the three pyridinium ligands investigated in the current study.T hese changes indicatea no verall stabilization of the native G4 fold by compound 1.H owever,w hilst the 260 nm shoulder band is preservedu pon titration with  (Figure 3a), this band disappears upon titration with ligand 2 (Figure 3c), suggestingashift in folding equilibrium in favor of an alternative G4 topology.W hilst it is not possible to draw detailed conclusions from CD data alone, subsequent NMR studies (see below) indicate that the equilibrium mixture of major and minor speciesformed by telo23 under the experimental conditions shifts to favor as ingle specieso na ddition of ligand 2.T he CD spectralf eatures present in the telo23/ ligand 2 complex are superficially representative of an antiparallel-folded G4, but more detailed studies are necessaryt o truly establish the precise structure of the telo23/2 complex. Only very weak induced CD signals are observed in the ligand region for compound 3,i nc ontrast to the more evident induced CD signals inducedb yl igand 1,s uggesting that whilst ligand 1 binds throughg roove binding modes, [17,26] end stacking modes are available for ligand 3.L esser (thoughs till significant) spectral perturbationsw ere observed for ligand 4 (see the SupportingI nformation, Figure S14a) indicating the weaker affinity of this ligand for G4. No perturbation of the CD spectra of telo23 waso bserved for ligand 5 (see the Supporting Information, Figure S14b),c orroborating the results from the FRET assay,w here no stabilization of G4 was observedo ver the range of concentrationsstudied. . Circular dichroism and UV/Vis titrationso fl igands 1, 2, 3.CDspectra of telo23 (4.2 mm,black traces) titrated with 0( black trace) to 7equiv (dark gray trace) of the ligands a) 1,c)2,and e) 3.Intermediate titration points are shown in light gray.Induced CD signals in the ligand regionsa re marked with a square bracket. UV/Vis spectraofl igands b) 1,d)2,and f) 3 (10 mm)titrated with telo23.The insets show fitting of the data to independent and equivalent sites binding models to yield the dissociation constants. Data for compound 1 were reported previously. [17] Chem.E ur.J.2020, 26,6224 -6233 www.chemeurj.org Quantificationo ft he binding affinity of ligands 1-5 to hybrid G4 was performed using UV/Vis titration studies ( Table 2). The resultingt itrations andr epresentative isotherms are shown in Figures 3b,da nd f, and in the Supporting Information ( Figures S15-S19), in which the absorbance spectrum of the ligand was measured upon titration with telo23 G4. Apparentbinding isothermsand stoichiometries were determined by fitting the observed hyperchromic shifts to an independent-and-equivalent-sites binding model (see the Supporting Information). [27,28] Strikingly,t he observed affinitiesm irror the trends observed in the thermalm eltinga ssays, with compound 1 again emerging as the most potent G4 ligand. Ligand 5,w hich displayed negligible effect on the stability of G4, induced only subtle perturbations to the UV/Vis spectrum of telo23 (see the Supporting Information, Figure S19), indicating weak interaction and meaning the dissociation constant could not be reliably determined. These results confirm that the critical nature of the stilbene configuration in G4 recognition, with E compounds 1 and 3 exhibiting micromolar G4 affinity,w hilst Z ligand 2 displays affinity two orders of magnitude lower.N otably,astriking bathochromic shift (ca. 30 nm) is observed in the spectrum of ligand 3 (Figure 3f). This effect is indicative that an end-stacking ligand binding mode is present, where the energy of the p-p*t ransition responsible for the Soret band is lowered by the interaction of the ligand chromophore with the G-tetrad. [29] Such marked shifts are not observed for relatedl igand 1,f or which the CD data is indicative of groove binding modes. [17] To further investigate the G4/duplexs electivity observed for lead compound 1,w eu ndertook af urthert itration study of this compound with duplexD NA (ds26) to examinet he origin of the selectivity between the two DNA species observed in the FRET assay ( Figure S20). Changes in the UV-region of the ligand 1 absorbance spectrum also occur upon titration with ds26, albeit to al esser degree than with the G4 species. We propose thesep erturbations arise from electrostatic interactions between the cationic ligand and the negativelyc harged DNA-phosphate backbone, whicha re likely to be relatively independent of DNA secondary structure when compared to specific steric binding modes. More strikingly,t he induction of hyperchromicity in the shoulder band in the visible region of the ligand spectrum (centered on 430 nm) by telo23, is barely observed upon titrationw ith duplex DNA (Figure 4). Therefore, it appearst hat the telo23 G4 provides af urtherl igand binding mode that is not accessible in the duplex sequence. We propose that this structure-specific binding mode is responsible for the overall G4 selectivity observed in the melting assays.

NMR studies
To provide initial structurali nsights into the data obtained in the UV/Vis and circulard ichroism titratione xperiments, and provide preliminary validation of the proposed binding modes of ligands 1-3 to telo23 G4, we undertook 1D 1 Hi minoN MR experiments ( Figure 5). Spectra werei nterpreted using assignments of the resonances of the major hybrid-fold species previously reported by Patel et al. [23] Significant line broadeningo f the imino resonances can be observedu pon titration with ligand 1 ( Figure 5b). Such broadening effects could be attributed to the strong binding of this ligand, resulting in intermediate-to-slow exchange between free andb ound ligand states on the NMR timescale. All imino signals broaden to a similar degree and remaind istinguishable, suggesting interactions with specific G-tetrad residues do not dominate in the association of ligand 1 with G4 and providing additional evidence for the groove binding mode inferred from the CD titrations. Ligand 3 (Figure 5d)a lso induces spectral line broadening, thoughs ignals associated with the lower G-tetrad (G15/ G23) disappear entirely,s uggesting stacking (or possibly intercalative) interactions with this parto ft he G4 are important in the binding of ligand 3.M eanwhile, line broadening is much less significant for ligand 2.T his ligand exhibits weaker binding to the G4 (see above)a nd therefore faster exchange between bound and unbound ligand states on the NMR timescale can be expected. This allows significant chemical shift perturbations to be observed in the imino resonances of the G4 upon addition of the ligand,i ndicating that this ligand may also interact with the G4 target through association with the G-tetrads. Interestingly,w hile the unboundG 4s equence exists as a mixture of major and minor conformations, complexation with ligand 3 appearst of avor as ingle conformation, with only 12 distinct imino signals visible at 2:1l igand:G4 stoichiometry (indicatedi nF igure 5c)v ersus the more complex spectrumf or the G4 sequence in the absence of the ligand, which repre-  sents am ixture of major and minor folded species, as previously reported. [23] This structuralp erturbation may explain the disappearance of the shoulder band in the CD spectrum of telo23 upon titration with ligand 3.I nc onjunction with the data obtained in the CD and UV/Vis titrations, we infer that ligand 1 does interact primarily throughg roove binding, whereas stacking interactions are more important in the binding of ligands 2 and 3.

Toxicity studies
Having demonstrated ar ange of binding affinitiesf or stiff-stilbene ligands 1-5 to G4 DNA,w ewere keen to make an initial assessment of the performance of these compounds as therapeutic agents. In particular,w ew ere interested to examine whether the in vitro biological activity of the compounds correlated with their DNA binding properties. We measured the viability of parasitic and mammalianc ell cultures in the presence of increasing doses of compounds 1-5.A ctivity was first measured against both T. brucei and L. major, with the MRC-5 line chosen as an on-tumoral cell model for comparison. To our delight, compounds 1-5 exhibited potent toxicityt oT. brucei with GI 50 values in the nanomolar range for all com-pounds( Ta ble 3a nd the SupportingI nformation, Figure S21). Critically, the compounds are significantly less toxic against the non-tumoral MRC-5 cells with selectivity indexo fu pt o7 00fold, suggesting ap romising therapeutic window for these compounds as anti-trypasanomal agents. However, the fact that ligand 5 also exhibits high toxicityd espite its poor affinity for G4 suggests the mechanism of toxicity is unlikely to be re-  [a] Measured by Alamar bluea ssay (MRC-5, T. brucei)o rM TT assay (L. major). See the SupportingI nformation for full details and dose-response curves lated to G4 recognitioni nt he case of this organism. Further studies are underway to investigate the origin of the high selectivity index of these compounds for T. brucei,w hich are significantly higher than those observed in relateds tudies. [9,16] Though appearing to exert al ower toxicityt ot he L. major parasite (Table 3a nd Supporting Information, Figure S22), ligands 1 and 3 (whichd emonstrated the strongest affinity for G4), show sub-micromolar efficacy against this organism. The weaker G4 ligands (compounds 2, 4 and 5)w ere significantly less efficacious, suggesting ap otentialr ole for G4 recognition in their mechanism of action in this case. The selectivity index for L. major is significantly lower than observed in the case of T. brucei,b ut the values are comparable to those observed for other G4 ligands screened against this organism. [9] We next examined the potential of the compounds to serve as anticancer agents. At three-day exposure, all compounds exhibited low toxicity (GI 50 = 10-100 mm)t oH eLa cervical cancer cells (see the Supporting Information, Ta ble S1 and Figure S23a). However,p revious work by others has identified that as ignificantly longer exposure time to G4-binding compounds is often necessary to observe toxic effects, since a mechanism of action involvingt elomere shortening theoretically requiress everalp opulation doublings to take effect. [30] We therefore subjected both HeLa and MRC-5 cells to a longer-term (seven day) exposure of compounds 1-5 (Table 4 and the Supporting Information, Figures S23b and S24b). Strikingly,w hereas the toxicityt owards MRC-5 was comparable with that observed at shorter-term exposure (GI 50 = 10-100 mm), the cancerous cells becames ignificantly more susceptible to lower doses of compound 1 (GI 50 = 62 nm), which, as we demonstrated above, is the most potent G4 ligand in our compound series.I ndeed, this increasei np otency returnsaselectivityi ndex of 29 at long-terme xposure. We rationalize that the factt hat the toxicity to MRC-5 cells does not significantly depend on exposure time for 1 could be attributed to the telomerase negative nature of this cell line, resulting in less susceptibility to the effects of telomeric G4 binding ligands. This ligand therefore shows significant promise as the basis of apotential G4-mediated selective cancert herapeutic and is worthy of future extensive screening and mechanism of action studies. The toxicity of the weaker G4 ligands 2, 4 and 5 are comparable (within one order of magnitude)t ot he values observed at short-term exposure. This indicates that although these compoundsd oe licit somec ytotoxicity to mammalian cells, al ong-term mechanism of action, as observed for ligand 1,i sn ot present.T his is perhaps explained by the relatively lowG 4a ffinity of these compounds observed in the titration studies and thermalm eltinga ssays. Compound 3 also displayed significantly stronger toxicity to HeLa cells at long-term exposure, but its acute toxicity to MRC-5 cells (retainedo nl ong-term exposure)y ields only av ery modests electivity index of 4. We propose that the poor discrimination between the cancerous and non-cancerous cell lines might result from binding to duplex DNA, since the FRET assaysd etermined this compound to be less selective for G4-DNAt han ligand 1.

Confocal microscopy
As as tep towards validating the mechanism of action of our compounds and more concretely establish the intracellular localization of the ligands within the cell lines studied, we examined the uptake of these compounds by mammalian cellsa nd parasites through microscopy studies. Unfortunately,l ead compound 1 was not sufficiently fluorescent to visualize using this technique. We therefore undertook localization studies on compound 3,w hichh as photophysical properties much better suited to visualization in cells (l em = 550 nm, Supporting Information, Figure S25). After 30 min incubation at 37 8C, significant uptake of compound 3 in both T. brucei and HeLa cells was observed. The ligand was mainly localized in the nucleolus and the cytoplasm of HeLa cells, and partially in the mitochondria (Figure 6a). In T. brucei,l igand 3 was mainly found in the nucleusa nd the kinetoplast (Figures 6d). Similarl ocations patters were observed at longer( 2h)i ncubationt imes (Figure 6b and e), andi nL. major and MRC-5 cells (see the Supporting Information,F igure S26). These resultss uggest that pyridinium stiff-stilbene G4-ligands can reach DNA harboring sites and therefore possibly target G4 structures.

Conclusions
G4 nucleic acids continue to offer exciting potential as at arget against ar ange of disease pathologies. However,n oG 4-targeting drug has reached the clinic to date. The identification and development of new ligand scaffolds that display promising bioactivities is in high demanda nd will form the basis of new G4-based drug discovery projects.W eh ave examined as election of the structuralf eatures that govern the binding of a novel series of stiff-stilbene ligands to nucleic acidt argets. We show that ligand 1 appears to have the optimal molecularconfiguration, electronics and substitution pattern for efficient G4 recognition amongst this first generation of compounds.Moreover,wehave demonstrated that 1 has ahigh level of selectivity for G4 structures, and anticipate that this selectivity could be further improved through the development of subsequent generationso fG 4-binding molecules based on this emergent G4-binding chemotype.F urthermore, we have shown for the first time that stiff-stilbene derivatives demonstrate high toxicity both to parasitic organismsa nd cancerous cell lines, whilst they remainu pt ot wo orders of magnitudel ess toxic to a non-tumoral model.C ritically,i nt he case of the cellular Table 4. Viability assay data for ligands 1-5 against tumoral (HeLa) and non-tumoral (MRC-5)mammalian cells after 7day incubation. [ models, in vitro cytotoxicity to cancerous cellss trongly correlates with G4-binding activity,a nd dose response times indicate long-term mechanisms of action that mayi nclude G4mediated pathways, such as telomerase inhibition. We believe this proof-of-concept study reveals intriguing activities that render stiff-stilbene compounds an exciting lead scaffold for DNA-targeted drug development. Towards this end, further investigationsi nto the therapeutic mechanism of stiff-stilbenel igands is warranted in order to interrogate the true biological targets and obtain more conclusive evidence regarding the po-tential G4-mediated mode of activity.E fforts towards these goals are under active pursuit in our laboratory and progress will be reported in due course.