Stabilization of Telomeric I‐Motif Structures by (2′S)‐2′‐Deoxy‐2′‐C‐Methylcytidine Residues

G‐quadruplexes and i‐motifs are tetraplex structures present in telomeres and the promoter regions of oncogenes. The possibility of producing nanodevices with pH‐sensitive functions has triggered interest in modified oligonucleotides with improved structural properties. We synthesized C‐rich oligonucleotides carrying conformationally restricted (2′S)‐2′‐deoxy‐2′‐C‐methyl‐cytidine units. The effect of this modified nucleoside on the stability of intramolecular i‐motifs from the vertebrate telomere was investigated by UV, CD, and NMR spectroscopy. The replacement of selected positions of the C‐core with C‐modified residues induced the formation of stable intercalated tetraplexes at near‐neutral pH. This study demonstrates the possibility of enhancing the stability of the i‐motif by chemical modification.

G-quadruplexes and i-motifs are tetraplex structures present in telomeres and the promoter regionso fo ncogenes. The possibility of producing nanodevices with pH-sensitivef unctions has triggered interesti nm odified oligonucleotides with improveds tructural properties. We synthesized C-rich oligonucleotides carrying conformationally restricted (2'S)-2'-deoxy-2'-Cmethyl-cytidine units. The effect of this modified nucleoside on the stabilityo fi ntramolecular i-motifs from the vertebrate telomere was investigated by UV,C D, and NMR spectroscopy.T he replacement of selected positions of the C-core with C-modified residues induced the formation of stable intercalated tetraplexes at near-neutralp H. This study demonstrates the possibility of enhancing the stabilityo ft he i-motif by chemical modification.
[a] Dr.A .AviÇó,P rof. Dr The i-motif structure has two wide and two extremely narrow grooves. The stability of the structure depends on repulsion between adjacent negatively charged backbonep hosphates across the minor grooves. Various backbones ubstitutions have been studied in order to form intermolecular and intramolecular i-motifs. [17,18] Modificationssuch as locked nucleic acid (LNA), [22,23] unlocked nucleic acids (UNA), [24] acyclic threoninol, [25] and peptide nucleic acids (PNA) [26] backbones have been introduced in biologically relevant i-motif structures.

CD spectra
First, the C Me Up i-motif formation wasi nvestigated by circular dichroism (CD) for all sequences at pH 5.5 and 15 8C ( Figure 1; also in Figure S2 in the Supporting Information). We obtained the characteristic i-motif CD signaturew ith positive and negative peaks at 287 and2 62 nm, respectively,f or all the modified oligonucleotides prepared in this study and at different pH values. In all cases, the CD signatures were similart oW TVT (unmodified sequence), thus indicating that the modified nucleoside is well incorporated in the overall structure at all pH valuess tudied.

Acid-base titrations
CD spectra of C Me Up i-motifs against pH were recorded. Acidbase titrationsw ere carriedo ut at 25 8Co ver pH 3-7 ( Figure   Scheme1.C-rich fragment of the vertebrate telomere (C 3 TA 2 ) 3 C 3 T, and chemical structure of (2'S)-2'-deoxy-2'-C-methylcytidine (C Me Up). Table 1. Sequences used in this study. C:  S3), and plots of "foldedf raction" against pH produced sigmoidal curves ( Figure 2). The pH transition midpoint (pH 1/2 ;w here 50 %o ft he oligonucleotide is folded into an i-motif structure) was determined as previously described. [25] In all cases, the titration curvesw ere analyzed by assuming as ingle protonation event in order to extract populations of folded (protonated) or unfolded (deprotonated) states. This assumption wasc hecked by performing am ultivariate analysis (data not shown). The calculated parameters from each curve are summarized in Ta ble 2. In general,t he modifications slightly increased the pH transition midpoint (pH 1/2 ). This increasei sa lmostw ithin the range of experimental uncertainty.T he C Me Up derivativeh as am ethyl group at the 2' position of the ribose. For this reason the introduction of the methyl group wasn ot expected to have as trong impacto np H 1/2 .T he mutations had am ore significant effect on the pH transitional range (R T ), which changed from 0.22 (wild-type)t o0 .5 (VT1CMeUp and VT2C Me Up) and 0.64 (double modification). In other words, the acid-base titration curve of the unmodified sequence is very sharp, thus indicating as trong cooperativity in the deprotonation of the imotif. Once the first proton is removed, it is easier to remove the rest of them, thereby leadingt oq uick destabilization of the i-motif. In contrast the shapes of the curvesf or the C Me Upmodified sequences are wider (lessc ooperative). These differences could explain the stabilizationoft he i-motif at pH values above the midpoint (see below).

Thermal stability of i-motif
Next, we studied the thermal stabilityo ft he oligonucleotides by CD and molecular absorption spectroscopies. The thermal denaturation curvesw ere obtained by measuring UV absorbance (295 nm) of oligonucleotide solutions over pH 5.0-7.0 ( Figure S4). At this wavelength denaturation of the i-motif afforded ad ecrease in absorbance, from whicht he midpoint (T m )w as determined as the minimum of the first derivative (Table 3a nd FigureS4). The denaturation followed at 260 nm obtained similar results but with increases in absorbance (Figures S5 andS6).
The T m values show that the C Me Up-modified oligonucleotides slightly stabilize the i-motif structure, depending on pH (higher T m values). At pH 5.0, the T m changes (DT m )w ere not significant,t hus indicating that the modification did not change the stability of the i-motif structure. The differences were more pronounced at pH 5.5: DT m increased to 2.3 8Cf or the double-modified variant and around 1.7 8Cf or the single modification (VT2C Me Up). At pH 6.0 there was clear stabilization of the double-modified variant (DT m = 4.6 8C) and VT2C Me Up (DT m = 3.1 8C). The shifts in T m relative to wild-type agree with the behavior of R T above.A st his range widens, the i-motif structure is detected at higher pH values,e ven though as a minor species. Despite the difficulty in analysis of the transitions at pH 6.5, it was still observed for the double-modified variant. Surprisingly,a tp H7,w here almostn oi -motif is observedi nt he natural sequence, the denaturation curve of VT2,14C Me Up at 295 nm ( Figure S4) still shows as mall hypochromic change, thus indicating the presence of the i-motif structure. These results suggestt hat C Me Up modification stabilizes human vertebrate i-motif structures even at near-neutral pH.
Similar results were obtained by CD analysiso fd enaturation of the i-motif sequences (see "Thermal denaturatione xperiments followed by CD"i nt he Supporting Information). At

NMR spectra
We further analyzed the structure of the modified i-motifs by NMR spectroscopy. 1D NMR spectra showed the characteristic cytosine imino protons in hemiprotonated C:CH + base pairs at around1 5-16 ppm ( Figure 3A,C ). At neutral pH i-motif formation was significantly favored for VT2,14C Me Up in comparison to the nonmodified sequence, which did not exhibit the characteristic i-motif signals under these conditions. 2D NOESY spectra exhibit the characteristic crosspeak pattern of an imotif (FigureS7). Although complete assignment of the NMR spectra was beyond the scope of this study,t he 2'-CH 3 signals of the modifiedn ucleotides could be identified (0.46 and 0.81 ppm). Their NOEs with the amino protons of hemiprotonated cytosines ( Figure S7) indicatet hat these methylg roups are oriented towards the i-motif major grooves,a se xpected ( Figure 3E). No stericc lash with neighboring residues is expected in this orientation. In order to confirm this, molecular dynamics calculations were carried out with the AMBER program. After 1ns, the model structure of VT2,14C Me Up remained stable ( Figure S8), with the two 2'-methyl groups well accommodated in the major grooves,and the sugar of the C Me Up residues in North conformation (Figure 4).
These results confirmt hat the arabino sugar configurationi s well tolerated within the i-motif structure, [29,30] and permits different substitutionsa tt he 2' position;t hese are not allowed in the ribose configurations because of bad steric contacts throughout the i-motif minor groove. 2'-Fluorine substitutions provoke the mostd ramatic effect, as the fluorine electronegativity causes favourable changes in the charged istribution within the sugar.T hese changes in charged istribution are not expected in the case of 2'-CH 3 substitutions and, consequently, the stabilization effect is not so pronounced. However,( 2 'S)-2'deoxy-2'-C-methyl-cytidine has as trong tendency to adopt aC 3 '-endo conformation, which is usually preferred in i-motifs. Preorganization of the sugar in the proper conformationi s most probablyt he reason of the observed stabilization of imotif.

Discussion
Because of the potential applicationso fi -motif structures, there is increasing interest in designing modified-cytosine derivatives capable of forming stable i-motifs, particularly at nearneutral pH. [13][14][15] One of the first modificationsw as the backbone-modified derivativeP NA. [26] Some stabilization of the imotif at neutral pH was observed, but these resultsw ere obtained for short tetrameric structures and are difficult to extrapolate to biologically relevant intramolecular i-motifs. As cytidine is subject to epigenetic modifications, the presence of 5-methyl-C and 5-hydroxymethyl-C in the C-rich telomeres e-  quenceh as been studied, thus showings mall changes in the stabilityo ft he modified i-motifs. [19][20][21] Interestingly,i th as been observedt hat the inclusion of ribocytidines destabilizes imotifs, [27] buta rabino-cytidines have the opposite effect. [29] It has also demonstrated that the presence of 2'-fluoro-araC produces very stable i-motifse ven at neutral pH, [30] mainly as ar esult of favorable electrostatic interactions as ar esult of the electron-withdrawing properties of the fluorine atom.
In this work we studied the effecto f( 2 'S)-2'-deoxy-2'-Cmethylcytidine derivatives on the stability of the i-motif, and demonstrated that substitution of -H for -CH 3 gives rise to more-stable i-motifs, as can also be observed at near-neutral pH. We replaced dC by C Me Up in different positions of the Ccore stretches of the C-rich fragment of the vertebrate telomere (C 3 TA 2 ) 3 C 3 T, by selecting singlesubstitutions at 5'-terminal externalp osition (VT1C Me Up), as well as single (VT2C Me Up) and double replacements (VT2,14C Me Up) at the next internal position. Thermal denaturation studies followed by UV and CD analysiss howed that the most stable variant at mildly acidic pH was theVT2,14C Me Up, followed by VT2 Me Cup. The singlemodifieds equence at an external position (VT1C Me up) was the least stable but was still more stable than the unmodified sequencea tm ildly acidic pH. The most evidentd emonstration of the existence of i-motif at neutral pH for the double-modified variant arises from the presence of exchangeable protons at pH 7.0i nt he NMR spectra of VT2,14C Me Up up to 35 8C. Moreover,U Va nd CD spectroscopies confirmt he distinct behavior of the double-modified variant at pH 7.0 in accordance with the presenceo fa ni -motif structure for VT2,14C Me Up at low temperatures.
Ar emarkable observation is the smalli nfluence of the C Me Up-modification on the pH midpoint transition values, which varied between 6.03 and 6.10 ( Table 2). In contrast, other cytidined erivatives, such as 5-methyl-C [20] and 2'-Fluoro-araC, [30] induce larger variations in i-motif pH 1/2 values, thus explaining the changes in stability.Inour case the C Me Up modification induced an alteration in the acid-base profile of imotifs, thus indicating that protonation andd eprotonation of cytidinesa re less cooperative. Therefore, the observed trend is that at pH above the midpoint,t he stability of the i-motifsi ncreases, whereas at lower pH it decreases.
It is interesting to compare these results with the stabilization induced by the 2'-F-araC derivatives. Both studies confirm that i-motifst olerate well2 '-substitutions in the arabino sugar configuration. [30] However,2 '-F-araC derivatives stabilized the imotif at all pHs, whereas (2'S)-2'-deoxy-2'-C-methyl-cytidine did not stabilizet he i-motif at pH 5.0. In contrast to the 2'-F substitution,t he methyl group at 2' should not alter the charged istribution in the sugar and, therefore, the favorable electrostatic interactions involving 2'-F-arabinoses are absent. Consequently, the stabilization observed for (2'S)-2'-deoxy-2'-C-methyl-cytidine substitution is lessp ronounced. In this case, the preference of C Me Up derivatives for the North-sugar conformation is probablyt he main factor in the stabilization. These results open the door to the design of other stabilizing Cd erivatives.

Conclusion
We studied the effects of replacement of dC by C Me Up on the ability to form intramolecular i-motifs. The stabilities of the modified structures were analyzed by UV,C D, and NMR spectroscopies, and showed that C Me Up residues stabilize the imotif, with the C Me Up:C Me Up pair more stable than the C Me Up:dC pair.T his stabilization could be used to modulate the stability of i-motif structures at mildly acidic to neutralp H.
UV Melting studies: UV thermal denaturation data [41] were obtained on aV -650 spectrophotometer (JASCO) equipped with aP eltier temperature control. Oligonucleotides were resuspended (5 mm) in citrate/phosphate buffer (20 mm, pH 5.0 and 5.5), or phosphate buffer (pH 6.0, 6.5, and 7.0) containing KCl (50 mm). These solutions were heated at 85 8Ca nd slowly cooled to room temperature for annealing, then stored at 4 8Co vernight. The thermal denaturing curves were obtained by following the change in absorption at 295 nm from 10 to 80 8C( 0.5 8Cmin À1 ). The dissociative melting temperatures were determined as the midpoint of the transition (T m )byc alculating the first derivative of the experimental data. T m values are averages of three experiments.
CD spectroscopy: Oligonucleotides were resuspended (1.3 mm) in citrate/phosphate buffer (20 mm, H5.0 and 5.5) containing KCl (50 mm) and in phosphate buffer (pH 6.0, 6.5, and 7.0). The solutions were heated at 85 8Ca nd slowly cooled to room temperature for annealing and stored at 4 8Co vernight. Thermal denaturing curves were obtained in as ingle experiment by following the change in absorption (280 nm) from 5t o8 0 8C( 0.5 8Cmin À1 )o n aJ -815 Vs pectropolarimeter (JASCO) equipped with aJ ulabo F-25/ HD temperature control unit. Spectra were recorded at 15 8Co ver ar ange of 210-350 nm (100 nm min À1 ), response time 4s,d ata pitch 0.5 nm, and bandwidth 1nm. CD absorption data (280 nm) were analysed as af unction of temperature, as described elsewhere. [42] Acid-base titrations: Acid-base titrations [25,43] of the sequences were monitored in the JASCO spectropolarimeter with aH ellma quartz cell (10 mm path length, 3mL): 25 8C, phosphate buffer (20 mm) with KCl (50 mm),2mm oligonucleotide. Titrations were carried out by adjusting the pH (stepwise pH). The ellipticity values at 280 nm were converted into "fraction of folded DNA" by using where ellipticity pH is the ellipticity at 295 nm at ag iven pH, and BS fold and BS unfold correspond to the baseline values of the folded and unfolded species, respectively.T hese chosen baseline values correspond to pH values of around 3.5 and 7.5, respectively.T he transition midpoint of each transition (pH 1/2 )w as obtained by plotting fraction of folded DNA against pH, and fitting to as igmoidal by using the Curve Fitting tool in the Optimization To olbox (Matlab;M athWorks, Natick, MA). From the calculated values, the pH transitional range (R T ,1 0-90 %u nfolded structure) was also determined. The validity of this two-state transition process was validated prior to the calculations by using multivariate analysis.
NMR spectroscopy and molecular modeling of structures: Samples were dissolved (0.4 mm) in sodium phosphate buffer (10 mm; either D 2 Oo rH 2 O/D 2 O( 9:1)), and annealed before NMR analysis. Experiments were carried out over pH 5.0 to 7.0 (adjusted with concentrated DCl or NaOD). All NMR spectra were acquired on AvanceA V-600 and AvanceA V-800 spectrometers (Bruker) operating at 600 and 800 MHz, equipped with cryoprobes, and processed with TOPSPIN software. NOESY spectra were acquired with mixing times of 150 and 250 ms. TOCSYs pectra were recorded with the standard MLEV-17 spin-lock sequence and am ixing time of 80 ms.
In most of the experiments in H 2 O, water suppression was achieved by including aW ATERGATE module [44] in the pulse sequence prior to acquisition. Molecular structures were calculated with the sander module of AmberTools (University of California, CA).