Dinickel–Salphen Complexes as Binders of Human Telomeric Dimeric G‐Quadruplexes

Abstract Three new polyether‐tethered dinickel–salphen complexes (2 a–c) have been synthesized and fully characterized by NMR spectroscopy, mass spectrometry, and elemental analyses. The binding affinity and selectivity of these complexes and of the parent mono‐nickel complex (1) towards dimeric quadruplex DNA have been determined by UV/Vis titrations, fluorescence spectroscopy, CD spectroscopy, and electrophoresis. These studies have shown that the dinickel–salphen complex with the longest polyether linker (2 c) has higher binding affinity and selectivity towards dimeric quadruplexes (over monomeric quadruplexes) than the dinickel–salphen complexes with the shorter polyether linkers (2 a and 2 b). Complex 2 c also has higher selectivity towards human telomeric dimeric quadruplexes with one TTA linker than the monometallic complex 1. Based on the spectroscopic data, a possible binding mode between complex 2 c and the dimeric G‐quadruplex DNA under study is proposed.


Synthesis of dinickel-salphen complexes
The dinickel-salphen complexes 2a-c were synthesized as outlined in Scheme 2. Compounds 3a-c were reacted with 4amino-3-nitrophenol( 4)i nd imethylformamide at 80 8Ct oy ield compounds 5a-c,f ollowedb yr eduction of the nitro group to give the tetra-amine compounds 6a-c.T hese compounds were reacted with four equivalents of the piperidine-substituted aldehyde 7 in ethanola tr eflux for 2h.T ot his reaction mixture, two equivalents of Ni(OAc) 2 ·4 H 2 Ow ere added to yield the final complexes 2a-c after 16 ho fr eflux. Compounds 2ac, 5a-c,a nd 6a-c werefully characterized on the basis of NMR spectroscopy ( 1 Ha nd 13 C), mass spectrometry (LR and HR), and elemental analysis( see the Experimental Section). Compounds 3a-c and 7 were prepared according to reported protocols. [32,33,36,41] UV/Vis titration to determine DNA affinity The DNA bindinga ffinities of complexes 2a-c and 1 towards the K + stabilized mixed-type monomeric quadruplexG 1a nd dimericq uadruplex G2T1, and the Na + stabilized antiparallel G1 and G2T1 (see Figure 1a nd Table 3), were determined by UV/Vis titrations.T he UV/Vis spectra of thesen ickel(II) complexess howed similarp atterns, with two strong absorption bandsi nt he region 310-330 nm (associated with intraligand p-p * transitions) and in the region 360-390 nm (which involves both the ligand and the metal center). [35] Addition of increasing amountso fG 2T1 to these complexes resulted in considerable hypochromicity (15-34 %) for the two peaks at 310-330nm and 360-390 nm ( Figure 2, plus Figures S28 and S29 in the Supporting Information). Interestingly,t he addition of G2T1 resulted in an oticeable redshift of complex 2c (12 nm in 100 mm NaCl buffera nd 7nmi n1 00 mm KCl buffer;F igure 2b and Figure S28c) andc omplex 1 (16 nm in 100 mm NaCl buffer and 13 nm in 100 mm KCl buffer;F iguresS28d and S29d). These spectral features are indicative of an end-stackingb inding mode rather than groove binding. [35] On the other hand, upon addition of increasing amounts of G2T1, the redshifts of complexes 2a and 2b were considerably smaller (under 4nm; see Figure 2a,F igures S28 and S29), suggesting that the interaction of these complexes with DNA throughe nd-stacking is relativelyw eak. For comparison, the interaction between these nickel(II) complexes and CT DNA was also studied (see Figure S32 in the Supporting Information). Upon addition of increasinga mountso fC TD NA to the corresponding compound, hypochromicity was observed (between 40 and 57 %) but no redshift, suggesting that these complexes are not good duplex DNA intercalators but may possibly act as duplex DNA groove binders.
The intrinsic binding constants of the four complexes towards G2T1, G1, and CT DNA were determined by monitoring  Table 1a nd Ta ble S1.T he di-nickel complexes 2a-c have slightly lower apparentb inding constants (K a values) for the monomeric antiparallel G1 structure than the mono-nickel parent complex 1. On the other hand, complex 2c showed the highest binding affinity towards the dimerica ntiparallel and mixed-type G2T1 structures, followed closely by complex 1.I nterestingly, 2c also displayed the bests electivity for antiparallel G2T1 versus G1 and CT-DNA (30-fold and 297-fold, respectively,T able 1), whereas the selectivity of complex 2c for mixed-type G2T1 versus G1 is only six-fold( Ta ble S1 in the Supporting Information).

Circular dichroisms pectroscopic studies
Having established that the nickel(II)c omplexes bind to the dimeric G2T1 DNA structures, we were interested in studying the effect of the binding on the structure of the G-quadruplexes. Therefore the interactions of complexes 2a-c and 1 with G2T1 were investigated by CD spectroscopy (Figure 3). We first investigated the Na + stabilized antiparallel dimericq uadruplex G2T1. Upon addition of 2a and 2b,n os ignificant changes in the ellipticity of G2T1 were observed, while addition of 1 and 2c induced minor changes in the negative ellipticity at 265 nm ( Figure 3a). These results suggest that the complexes do not bring about major structuralc hanges in the antiparallel conformation of the G2T1 quadruplex structure. [42] We then investigated the K + stabilized mixed-type quadruplex structure. Though no significant changes were observed in the presence of complexes 2a and 2b,t he addition of 1 and 2c caused am arked increase of the intensity of the positivep eak at approximately 265 nm (associated with the parallel conforma-  , and adecreased intensity of the positive peak at approximately 295 nm (associated with the antiparallel conformation). These resultss uggest that complexes 2c and 1 promote the formationo fp arallel quadruplexi nK + buffer. [30,34,43] CD spectroscopyw as also used to determine the potential templating effects of these nickel complexes on the formation of G2T1 quadruplex DNA. Non-annealed G2T1 in the absence of K + or Na + and without added metal complex, showed the characteristic positive ellipticity at ca. 250 nm consistent with as ingle-stranded DNA sequence ( Figure 4). Upon addition of each of the four nickel complexes under study, the signal centered at 250 nm decreasedw hile the signals associated to the formationo fq uadruplex DNA increased. Interestingly,t he three di-nickel complexes induced mainlyt he formation of an antiparallel quadruplexs tructure (with positive ellipticity centered at ca. 295 nm, Figure 4a-d). Though this is also the case for compound 1 at low concentrations, upon increasing the amount of compound added ap ositive shoulder peak at 265 nm appeared, which suggests the formation of mixed-type quadruplex DNA (Figure 4d). [30,34,35,43] For complexes 2b, 2c, and 1 we noted ad ecrease in the overall intensity of the CD spectra at the highest compound concentrations used, which might be due to aggregation/precipitation of DNA induced by the compounds. [35] Dinickel complexes 2a and 2c displayed induced CD signals in the presence of G2T1 quadruplex DNA:at352 nm (with positive ellipticity) for the former,a nd at 311nm( with negative ellipticity) and 431 nm (with positive ellipticity) for the latter (Figure 4a and c). Complex 1 showedabroad induced CD signal with positive ellipticity at 431 nm ( Figure 4d). No significant induced CD signals were observed for 2b.
CD melting assays were then used to further assess the affinity and thermalstabilization of the nickel(II) complexes towards dimericG -quadruplex G2T1. These experiments were carried out in 10 mm Tris-HCl and 100 mm NaCl buffert oe nsure that the G-quadruplex was present in as ingle antiparallel conformation (rather than mixed conformations as in the case with Table 1. Apparentb inding constants( K a values,M À1 )o fc omplexes 2a-c and 1 for G2T1, G1, and CT DNA in 10 mm Tris-HCl and 100 mm NaCl( pH 7.04) by UV/Vis spectroscopy.   (2) 1equiv;(3) 2equiv;( 4) 4equiv.
K + ;s ee FiguresS33 and S34 in the Supporting Information). Upon increasing the temperature, the signals at 295 nm with positive ellipticity and at 260 nm with negative ellipticity (characteristic of antiparallel conformation), decreased until their disappearance when the G-quadruplex was completely unfolded ( Figure S34 b-e). The melting of G2T1 was then carriedo ut in the presence of the different nickel(II) complexesa nd the results are summarized in Figure 5a.C omplexes 2a and 2b (at a2 :1 molar ratio between complex and G2T1)d isplayed relatively low DT m values:7 .7 and 8.4 8C, respectively.O nthe other hand, the dinickel complex 2c and mono-nickel complex 1 displayed significantly higher DT m values ( Figure 5a): 14.1 8Cf or complex 2c (2:1 complex-to-G2T1 ratio) and 20.8 8Cf or complex 1 (4:1 complex-to-G2T1 ratio). For complex 2c,w ea lso investigated changes in DT m upon increasingt he concentration of the complex;a sc an be seen in Figure 5b,asignificant thermal stabilization (DT m = 19.8 8C) was observed at a4:1 ratio of 2c to G2T1. These resultss uggested that the dinickel complex 2c (with the longest polyether linker) and monomeric nickel complex 1 showedt he higher bindinga ffinitiesa nd thermal stabilization towardG 2T1 as compared to the dinickel complexes 2a and 2b (with the shorter polyether linkers), which is consistent with the resultso btained through UV/Vis titrations. Furthermore,itshould also be noted that complexes 2c and 1 exhibited higher than or comparable affinities/thermal stabilization than other G2T1 bindersp reviously reported in the literature (Table S2 in the Supporting Information). [24,26,27,[29][30][31] We then investigated the binding of 1 and 2c towards dimeric quadruplexes linked by one, two, four,o rs ix TTAs ubunits namedG 2T1, G2T2, G2T4, andG 2T6, respectively (see Figure 1f or schematic representation of these structuresa nd Ta ble 3f or sequences). The DT m values (Table 2a nd Figure S35 for their CD spectra) of these dimeric G-quadruplexesu pon addition of complex 2c decreased with the length of the TTA linkers, indicating that this complexh as highera ffinity for dimers with short TTAl inkers. In contrast, the DT m values of the different dimericG -quadruplexes in the presence of the mono-nickel complex 1 showed little change regardless of the lengtho ft he TTAl inkers( Table 2). These results indicatet hat, although complex 1 has higherb inding affinities and induces higher thermal stabilizationsf or both monomeric and dimeric G-quadruplexes, the dinickel complex 2c has better selectivity: seven-fold higherp reference for the dimericG -quadruplex G2T1 than for the monomericG 1( whereas the selectivity of complex 1 for G2T1 vs. G1 is only two-fold).

Native gelelectrophoresis
Based on previouslyr eported protocols, [24,30,32] the selectivity of complex 2c for G2T1 over G1 was investigatedb yg el electrophoresis ( Figure 6a nd Figures S37 and S38 in the Supporting Information). The gel shown in Figure 6a indicates that addition of 2c to antiparallel G1 (in the presence of Na + )d id not lead to the appearance of any new band (lane 2), suggesting that this compound does not form astable complex with monomeric G1 under the gel electrophoresis conditions. By contrast, the presence of complex 2c increased the mobility rate of the antiparallel dimericquadruplex G2T1, which could be rationalized by the formation of am ore compactG 2T1 upon interaction with 2c (lane 4), as has been previously proposed for other G-quadruplexb inders. [30] To furtherv erify the preference of 2c forG2T1 over G1, the complex wasincubated with amixture of G1 and G2T1 and the mixture analyzed by gel electrophoresis. As ac ontrol, am ixture of G1 and G2T1 in the absence of 2c wasa lso analyzed by gel electrophoresis;a sc an be seen in Figure 6( lane 5), this sample gave the characteristic  Table 2. Quadruplex DNA stability measurements from CD-melting curves with complexes 2c and 1 in 10 mm Tris-HCl and 100 mm NaCl (pH 7.04). The amounto fc omplex added was such that all samples had a2:1 ratio of nickel-salphen with respect to each G-quadruplex unit (e.g.,  bands corresponding to intramolecular monomeric (G1) and dimeric( G2T1)G -quadruplexes.A fter addition of complex 2c to the G1 and G2T1 mixture, an ew band correspondingt o complex G2T1-2c appeared. This band became more intense upon addition of increasing amounts of 2c to the G1/G2T1 mixture, but no changes were observed for the band associated with G1 (lanes 6t o9in Figure 6a). An analogous experiment was carriedo ut with the mononickel complex 1 (Figure 6b). The presence of 1 increased the mobility rate of the dimeric quadruplex G2T1, which is analogous to what we observed with 2c.I nterestingly,w ea lso observed an ew band for the monomeric G1 structure upon addition of complex 1,s uggesting that this compound does not discriminate between the monomeric and dimeric G-quadruplex structures (see Figure 6b,l anes 2a nd 4). Thisw as further confirmedu pon addition of complex 1 to am ixture containing G1 and G2T1:t wo new bands corresponding to complexes G2T1-1 and G1-1 were present (lanes 7-9).
The same set of gel electrophoresise xperiments were carried out with the K + stabilized parallel/antiparallel mixed-type G2T1 and G1 structures ( Figure S38). The behaviors for both 2c and 1 are analogoust ow hat was observed with the Na + stabilized parallel G2T1 and G1 structures,n amely 2c has higher binding selectivity for G2T1 versusG 1t han complex 1. An interesting observation was that, when the molar ratio of complex 2c andG 2T1 reached 4:1, the whole G2T1 structure could not be converted into complex G2T1-2c ( Figure S38a, lane 8). However,a sd escribed above,a ll the antiparallel G2T1 structure could be converted into complexG 2T1-2c when the molar ratio of complex 2c and G2T1 was 2:1( Figure 6a, lane 8). The result impliesthat complex 2c might have apreference for the antiparallel conformation of quadruplex G2T1.

Bindingmode of complex 2c toward G2T1
The results presented in the previouss ections clearly indicate that complex 2c is av ery good binder for antiparallel dimeric G-quadruplex G2T1. We were therefore interested in furtheri nvestigating its binding mode. In the UV/Vis titration experiments described above, the noticeable redshift observeda t 360-390nms uggests that complexes 2c and 1 interact with G2T1 through an end-stacking mode. [35] Both the UV/Vis titrations and CD meltings tudies clearly indicated that complex 2c has higherb inding affinities toward G2T1 than complexes 2a and 2b with shorter polyether linkers (Table 1a nd Figure 5a). Moreover,t he binding affinity of complex 2c towards the dimeric quadruplexes becomes progressively lower as the TTAlinker becomesl onger ( Table 2). These results imply that the distance between the two nickel-salphen units in complex 2c matches the distance from the centero fo ne G-quartet plane to the centero fa nother G-quartet in G2T1, suggesting that this compound is likely to interact with the two G-quadruplexes in G2T1. [24] To study this possibility further, we carried out emission spectroscopic studies with G-quadruplexes modifiedw ith 2aminopurine (Ap), af luorescenta denine isomer that has been previously used to study the interaction of ligands with Gquadruplexes. [24,[45][46][47] In particular, we modified G2T1 with as ingleA pb ase at positions 7, 13, 31, or 37 (named as Ap7, Ap13, Ap31, and Ap37, respectively;s ee Figure 7a and Ta ble 3). These positions weres elected because they are located on the four different exposed G-quartets in G2T1. Addition of complex 2c to the four different modified-G2T1 sequences significantly decreasedt he fluorescent intensities of Ap7, Ap13, Ap31, and Ap37 (Figure 7a), indicating that, upon binding with G2T1, complex 2c has considerable contact with the four Aps. This in turn suggests that the complex interacts with the four different tetrads in G2T1.
To investigate furtheri f2c has ap reference for G2T1'se xternal or internal tetrads, we modified the sequence with two Ap bases namedA p7 + Ap31 and Ap13 + Ap37 (Table 3). As can be seen in Figure 7b,a ddition of 2c to either of the two doubly labeled G2T1 sequences, led to equallyh igh quenching of Ap's emission. This observation would be consistent with two molecules of 2c interacting equally with each of the four G-quartets of the two G-quadruplex units in G2T1 as schematically shown in Figure7d. Therefore, we investigatedt he binding stoichiometry by titratingt he Ap31-labeled G2T1w ith complex 2c,k eepingt he concentration sum of complex 2c and G2T1 constant, whilev arying the [2c]/([2c] + [G2T1]) ratios from 0t o1 .0. The Job's plot resulting from this titration (Figure 7c)c learly shows a2 :1 bindingb etween 2c and G2T1. [45] This stoichiometry is consistent with our observations in the electrophoresis titration experiments(see Figure 6a,l ane 8).
Ta ken together,t he UV/Vis titrations, CD-melting, and fluorescence studies with Ap-labelled G2T1, indicate that two moleculeso fc omplex 2c are likely to stack on the four end Gquartetsi nG 2T1. This binding mode-ifc onfirmed by future structurals tudies-is different to most previously reported di- meric G-quadruplex bindersw here the ligandsi nteracta tt he cleft between the two G-quadruplexes. [25][26][27]29]

Conclusion
In summary,t hree new dinickel-salphen complexes have been prepared and fully characterized. Using ac ombination of UV/ Vis titrations, CD spectroscopy,C D-melting assays,a nd electrophoresis, we have demonstrated that complex 2c (with the longestp olyether linker)h as high binding affinity towards dimeric quadruplex G2T1. This compound also displays the highest selectivity for G2T1 over G1, as compared to complexes 2a and 2b (with the shorterpolyether linkers) and the monomeric nickel complex 1.F luorescent titration assays using Ap-modi-  fied G2T1 suggest that two molecules of complex 2c may stack on the four end G-quartets of the two well-matched Gquadruplexu nits in one G2T1. This work provides new insights into the bindingp roperties of dimetallic complexesw ith dimeric quadruplex structures.

Experimental Section
General 1 HNMR and 13 CNMR spectra were recorded on either aB ruker Avance4 00 MHz Ultrashield NMR spectrometer or aB ruker Avance 500 MHz NMR spectrometer.M ass spectrometric analysis was performed on aL CT Premier mass spectrophotometer.A ll chemicals were purchased from Sigma-Aldrich, BDH, or Apollo Scientific and used without further purification.
Oligonucleotides listed in Ta ble 3w ere purchased from Eurogentec (Belgium). Complexes 1 and 2a-c were dissolved in am ixture of DMSO (95 %b yv olume) and 1mm HCl aqueous solution (5 %b y volume) to give 2.0-3.0 mm stock solution. All solutions were diluted to 1mm with DMSO before use. They were then further diluted using suitable buffer to the appropriate concentration.

CD spectroscopy
The oligonucleotides G1 and G2T1 were dissolved in Milli Q. water to yield a1 m m stock solution. They were then diluted using 10 mm Tris-HCl and 100 mm NaCl or KCl (pH 7.04) buffer to 10 mm. Prior to use in the CD assay,t he DNA solution was either annealed or remained nonannealed. The DNA solution was annealed by heating the solution to 95 8Cf or 10 min and then cooling to room temperature overnight. The CD spectra were measured in as trainfree 10 mm 2mmr ectangular cell path length cuvette. The CD spectra were measured in the spectral range of 600-200 nm. The following CD spectra were recorded:( 1) CD spectra of annealed G2T1 (2.5 mm)i n1 0mm Tris-HCl and 100 mm NaCl (pH 7.04) with complexes 1 and 2a-c;( 2) CD spectra of annealed G2T1 (2.5 mm) in 10 mm Tris-HCl and 100 mm KCl (pH 7.04) with complexes 1 and 2a-c;( 3) CD spectra of nonannealed G2T1 (2.5 mm)i n1 0mm Tris-HCl (pH 7.04) with complexes 1 and 2a-c.

CD-melting
The oligonucleotides, G1, G2T1, G2T2, G2T4, and G2T6, were dissolved in Milli Q. water to yield a1m m stock solution. They were then diluted using 10 mm Tris-HCl and 100 mm KCl or NaCl (pH 7.04) to 10 mm.P rior to use in the CD assay,t he DNA solution was annealed by heating the solution to 95 8Cf or 10 min and then cooling to room temperature overnight. The preparation of the solutions was similar to the procedure described for the UV/Vis titrations. CD spectra were measured in the wavelength range of 230-340 nm using aq uartz cuvette with 1.0 nm path length. The scanning speed was 100 nm min À1 ,a nd the response time was 2s. CD-melting was monitored at 295 nm at ah eating rate of 1 8Cmin À1 from 25 to 95 8C. The melting temperature (T m )w as determined from the melting profiles with the software origin 8.0.

Gel electrophoresis
The oligonucleotides G2T1 and G1 were dissolved in Milli Q. water to yield a1m m stock solution. They were then diluted to 20 mm with 10 mm Tris-HCl and 100 mm NaCl (pH 7.04). The DNA solutions were annealed at 95 8Cf or 10 min, gradually cooled to room temperature, and incubated at 4 8Co vernight. The final loading sample was prepared by mixing complex 2c or 1 (100 mm)w ith the annealed DNA samples, followed by incubation at 4 8Cf or 3h. Native gel electrophoresis was carried out on acrylamide gel (15 %), run at 0 8Ci n1 TBE buffer (pH 8.3) and was stained by ethidium bromide. DNA binding selectivity was analyzed with Alpha Hp 3400 fluorescent and visible light digitized image analyzer.

Fluorescence spectroscopy
The Ap-labeled oligonucleotides were dissolved in 10 mm Tris-HCl and 100 mm NaCl (pH 7.04) buffer to yield a5mm solution. The oligonucleotide was annealed by heating to 95 8Cf or 10 min and then cooled to room temperature overnight. Fluorescent measurements were carried out on aP erkinElmer spectrofluorometer at 25 8C. [24,45] The fluorescence spectra were measured at l ex /l em = 305/370 nm with ex/em = 10/10 nm. The DNA solution (5 mm)w as titrated with ac oncentrated solution of 1mm complex 2c or 1 (buffer:1 0mm Tris-HCl, 100 mm NaCl, pH 7.04). For the binding stoichiometry assays between complex 2c and Ap31-G2T1, the spectra were recorded by keeping the concentration sum of complex 2c and Ap31 constant (