Sugar Puckering Drives G‐Quadruplex Refolding: Implications for V‐Shaped Loops

Abstract A DNA G‐quadruplex adopting a (3+1) hybrid structure was modified in two adjacent syn positions of the antiparallel strand with anti‐favoring 2′‐deoxy‐2′‐fluoro‐riboguanosine (FrG) analogues. The two substitutions promoted a structural rearrangement to a topology with the 5′‐terminal G residue located in the central tetrad and the two modified residues linked by a V‐shaped zero‐nucleotide loop. Strikingly, whereas a sugar pucker in the preferred north domain is found for both modified nucleotides, the FrG analogue preceding the V‐loop is forced to adopt the unfavored syn conformation in the new quadruplex fold. Apparently, a preferred C3′‐endo sugar pucker within the V‐loop architecture outweighs the propensity of the FrG analogue to adopt an anti glycosidic conformation. Refolding into a V‐loop topology is likewise observed for a sequence modified at corresponding positions with two riboguanosine substitutions. In contrast, 2′‐F‐arabinoguanosine analogues with their favored south‐east sugar conformation do not support formation of the V‐loop topology. Examination of known G‐quadruplexes with a V‐shaped loop highlights the critical role of the sugar conformation for this distinct structural motif.

Approachesf or av ersatile and rational G4 designa re expectedt oh eavily support many bio-and nanotechnological G4 applications.Apowerfula nd efficient strategy for the directedmanipulation of the topologyf or ag iven G4-forming sequence involves the deliberate incorporation of G-analogues. [30] Here, the glycosidict orsion angle plays am ajor role as it is correlated with strandorientation and tetrad polarity. Apart from ac omplete disruption of the quadruplex fold, an anti to syn or syn to anti transition is necessarily accompanied by an inversion of either the tetrad polarity or the G-tract orientation. Thus, 8-substitutedGanalogues such as 8-methyl-or 8-bromo-2'-deoxyguanosine have been used to drive anti!syn transitions, often causing at etrad polarity reversal when incorporated into the 5'-tetrado fp arallel G4 structures. [31][32][33][34] Likewise, some 2'-substituted Gs urrogates can be used to enforce anti glycosidic conformations ( Figure 1B). 2'-Deoxy-2'-fluoro-riboguanosine ( F rG) and riboguanosine (rG) prefer a C3'-endo sugar pucker thought to be strongly correlated with an anti conformation.A nother north-favoring Ga naloguew ith a strongp ropensity forthe anti glycosidictorsion angle is the bicyclic 2'-O-4'-C-methylene rG analogue (locked nucleic acid, LNA G), which is strictly lockedi naC3'-endo conformation.I n contrast, 2'-deoxy-2'-fluoro-arabinoguanosine ( F araG) has an anti preference but at endency to adopt as ugar pucker in the south-east pseudorotational domain.
Recently,2 '-substituted Ga nalogues have been incorporated at various sites into a( 3 + 1)-hybrid quadruplex termed ODN featuring ap ropeller,d iagonal, and laterall oop (Figure1A). [35] Substituting two or three syn-residues in the 5'-tetrad with F rG, rG, or F araG caused ar eversal of the tetrad polarity under conservation of the globalq uadruplex fold. [36][37][38][39] On the other hand, incorporating F rG or F araG into the single syn-position of the central tetrad lead to partial refoldingi nto an antiparallel topologya ssociated with an inversion of the first G-tract and of the central tetrad polarity. [40] In both cases, anti-favoring G analogues had very similaro verall effects and differences were only observed in relative thermal stabilities and local structure, e.g.,p osition-dependent sugar conformations and participation of the 2'-substituent in pseudo-hydrogen bonds.
Here, we expand our studies on specific ODN modifications by evaluating dual F rG substitutions at the two syn-positionso f its antiparallel G-tract. These modifications are found to promote an unexpected refolding into aw ell-defined topology with aV -shaped loop connecting the two F rG residues.Adetailed conformational analysis of the V-loop supporting G analoguesa nd additional modificationsi ncluding rG and F araG reveals that the preferred sugar pucker outweighsg lycosidic torsion angle propensities in constituting the major driving force for V-loop formation.

Results
Resonance assignmentsfor the F rG-modifiedODN Initially,t he ODN sequence was modified with two F rG substitutionsa tp osition1 4a nd 15 and analyzed by 1D 1 HNMR spectroscopy.T his F(14,15) variant folds into am ajor G-quadruplex as indicated by the presence of at least ten wellr esolved imino resonances in the 10.8-12.0 ppm region typical for Hoogsteen hydrogen bonding ( Figure 2A). Individual 15 Nl abeling of all Gr esidues except for G22 at the 3'-terminus and the two F rG analogues was used for the assignment of both guanine iminoa nd H8 resonances through spectral editing with one-dimensional 1 H- 15 NH MQC experiments ( Figure 2B). Surprisingly,G 17 being al oop residue in the native structure participates in G-core formation of the new fold as clearlyd emonstrated by its imino signal at 11.1 ppm. In contrast, G3 does not show any imino resonancea nd is thus identified as al oop residue. Assignments of H8 and H1 resonances were addition- Anti and syn G-core residues are represented by grey and red rectangles, m, n, and wd enote medium, narrow,and wide grooves, respectively. Modification sitesa re highlighted in bold. (B) Anti-favoring Ga nalogues in their typical sugar conformation; F rG and rG favor a C3'-endo (north)sugar pucker whereas F araG has apropensity to adopt a south-type conformation. ally validated by their intra-base correlation to 13 C5 as observed in a 1 H-13 CH MBC experiment ( Figure S1). Sequential NOE contacts can be traced between all unmodified Gs and NOE walks furthere xtendi nto the short intervening sequences followinga nd precedingt he terminal G-tracts( Figure S2). These allowed for the assignment of non-exchangeable protons of unlabeled G22 at the 3'-end of the sequence as well as of loop residues A4, T5, A18, and C19. G1, G6, and G20 were identified as syn residues by their strong H8-H1' intranucleotide NOE crosspeak and their characteristicd ownfield 13 C8 chemicals hift ( Figure S2 and S3). H1' and H2' resonanceso f F rG can easily be detected by their characteristic 1 H- 19 Fs calar coupling to F2' (see below). It should be noted, however,t hat only one F rG residue shows up at 25 8Cw ith resonanceso ft he other analoguea pparently broadened beyond detection at this temperature, also explainingt he reduced number of imino signals ( Figure 2A).

F(14,15) adopts aV -loop topology
Upon increasing the temperature to 40 8C, sugar and H8 resonances of the second F rG resonance are clearly observable (Figure S3 and S4). Also, an additional upfield shifted imino reso-nance emerges at 10.84 ppm and another slightly broadened signal at 11.41 ppm becomes resolved ( Figure 3A). Supported by completed assignments at 25 8Ca nd only minor changes of the crosspeak patterns in 2D NOE and 1 H-13 CH SQC spectra at 40 8C( Figure S3 and S4), non-exchangeable base and sugar resonances werel ikewise assigned through sequential NOE contactsa tt he elevated temperature. Also, correlations with H8 protons through 13 C5 in a 1 H-13 CHMBC experiment allowed for the unambiguous assignment of H1 resonances at 40 8C ( Figure S5). Thus, the two newly emerging imino signals were easily allocated to G22 and F rG14. Surprisingly,t he latter adopts the unfavored syn conformation as indicated by its downfield 13 C8 chemical shifta nd by its weak H8-H2' and strong H8-H1' intra-residualN OE contact ( Figure S3 and S4). Unfortunately,w ith only am odest thermals tability in the buffer used for the NMR experiments( Ta ble S1), the structure is partially unfoldeda te levated temperatures as reflected in some poorly dispersed peaks from single-stranded speciesi n the 1 H-13 CH SQC spectrum ( Figure S3). Nevertheless, NMR experiments on samples in al ow-salt buffer at highert emperatures provided the best spectralq uality andw ere therefore selected for am ore detailed structural characterization (see also FigureS6).  (14,15)( 1mm) at 40 8Cin1 0mm KP i ,pH7.i) NOE contacts of the four iminoprotonsint he top tetradt oT 5Mei nthe first lateral loop. ii) Exchangec rosspeaks between water and imino protonso fthe outer tetrads. iii) Inter-tetradH 1-H1c ontacts, reflecting relative tetrad polarities (markedb ys quares). Noteadditional exchange crosspeaksd ue to minors pecies. iv) H1(w 1 )-H8(w 2 )c ontacts, reflecting hydrogen bond directionalities within tetrads (magenta, green,a nd orange squares for top,c entral, and bottom tetrad, respectively), NOE contacts between tetrads( black squares traced by horizontal and verticall ines), and NOE contacts between outert etrad imino resonances and base protonso fr esidues in the two lateral loops, namely A4, T5, A13, and C12( blue squares). Adenosine H2 resonances are labeled in italic. (C) Folding topology of F (14,15). Anti and syn residues are showni ng rey and red;m,n ,a nd wd enotem edium, narrow, and wide grooves, respectively.
The quadruplex topology as depicted in Figure 3C is unambiguously established through H8-H1 NOE contacts within the tetrads,s howing the followingh ydrogen bond directionalities: 2!6!20!15;1 !16!21!7; 14!17!22!8( Figure 3B). G1 is placed into the central tetrad and az ero-nucleotide Vshaped loop connects F rG14, located in the bottom tetrad to occupy the free positionw ithin the first G-tract, and F rG15, located in the top tetrad within the third G-tract.R elative tetrad orientations result in both heteropolar and homopolar stacking interactions in line with H1-H1 NOE contacts betweena nd within G-tracts, respectively.O fn ote, strand inversion occurs within the third G-tract between F rG15 and G16. Also, NOE contactsb etween T5 Me within the first laterall oopa nd the imino resonances of G2, G6, G20, and F rG15 residues in the top tetrad are conspicuous. To gether with imino signals of four other residues, namely G8, G17, G22, and the F rG14 analogue, the same four iminos ignals exhibit exchange peaks with water due to their location within outer tetrads and their faster solvent exchange.

V-loop supporting F rG analogues adopt an N-types ugar conformation
Although F rG14 is forced into an unfavored syn conformation, refoldingo fm odified ODN is apparently driven by the two fluorine-substituted analogues being anchor points for the newly formed V-loop. In an attempt to better understand the driving force for such ar earrangement and to evaluateV -loop conformational features, sugar conformationso ft he two F rG residues weres ubjectedt oamore detailed analysis. Conspicuously,H 1 '-H2' crosspeaks are clearly observable in 2D NOE but unobservable in DQF-COSY spectra for both residues, pointing to small 3 J H1'H2' scalar couplings with cancellation of antiphase COSY crosspeaks and sugar puckers in the north domain of the pseudorotation cycle ( Figure 4). Facilitated by the E.COSY-type pattern of H1'-H2' and H2'-H3' crosspeaks as ar esult of additional 19 Fp assivec ouplings, 3 J F2'H1' and 3 J F2'H3' could be directly extracted from corresponding 2D NOE and DQF-COSY correlations without resorting to simulations. Also, following the unambiguous assignment of 19 Fr esonances to F rG14 and F rG15 in aH OESY experiment through heteronuclear 19 F-1 Hd ipolar couplings ( Figure S7), 3 J HF scalar couplings were independently determined by the selective 1 Hd ecoupling of 19 Fs pectra (Figure S8). Of note, the rather unusualc onformation of F rG14, combining a syn glycosidict orsion angle with a north sugar pucker,i sa ssociated with an extremelyd ownfield shifted H3' resonance almosti sochronous with its H1' resonance at 6.05 ppm (see below), preventing the extraction of reliable 3 J HF couplings in F rG14 from selective decoupling experiments. Using the observed 19 F-1 Hs calar couplings (Table S2) together with aK arplus-type relationship between vicinal 1 H- 19 Fc oupling constantsa nd H-CÀC-F torsion angles for the 2'-fluoro sugar, [41] major pseudorotamers of both analogues were found to adopt a north-east pucker with ap hase angle of pseudorotation 408 < P < 708.
Three-dimensional structure of F (14,15) For ad eeper structurali nsight, we determined the high-resolution structure of the modified F(14,15) quadruplex from NMRderived distance and torsion angle restraints. Whereas A9 H8 could be assigned based on sequential NOE contacts to G8 ( Figure S5), other adenosine and cytidine residues of the long and flexible 5-nt lateral loop preceding the V-loop escaped their unambiguous identification. In trying to complementa ssignments, corresponding nucleotides were individually labeled by 15 N-dA and 5-methyl-dC. Thus,C 10 and C12 were easily identified by the disappearance of the corresponding H5 resonance and as hift of H6 and 13 C6 resonances in the 5methyl-dC modified samples ( Figure S9). Also, 15 Ne dited spectra of 15 N-dA labeled samples allowedf or as traightforward assignment of A13 H8 and H2 resonances at 7.89 and 7.56 ppm, respectively,b oth exhibiting severalN OE contacts to imino protons of the bottom tetrad and providingv aluable restraints for subsequents tructure calculations ( Figure 3B). Unfortunate- ly,a ssignment of H2/H8 resonances for residue A11w as hampered by intermediate exchange processes at 25 8Ca nd signal overlap with unfolded speciesat4 08C(FigureS10).
As uperposition of ten final structures determined by molecular dynamics calculations in explicit water (Table S3) together with ar epresentative structure is shown in Figure 5A and B. RMSD values of 0.84 and 2.96 for the G-core and the overall structure are similart ov alues reported for the unmodified (3 + 1) hybrid G4. [35] All G-core residues are welldefined, including the 5'-terminal guanosine in the central tetrad as well as the two V-loop flankingGa nalogues. While the first lateral loop is well structured with G3 and T5 loosely stacked onto the upper tetrad, the other two loops experience higher flexibility.I np articular, C19 within the 2-nt propeller loop as well as C10 and A11i nc entralp ositionso ft he long 5-nt lateral loop are very dynamic. In contrast, the two laterall oop residues directly preceding the V-loop are well defined with A13 stacked onto the bottom tetrad.
The structure features two medium grooves,o ne wide groove bridged by two laterall oops at either side, and a narrow groove spanned by the 0-nt V-loop ( Figure 3C). Ap eculiarity derives from the F rG15 analogue located at the V-loop 3'-end and being part of the third G-column. By adopting af avored anti glycosidic torsion angle, its opposite base orienta-tion when compared to the following anti-G16 demands inversion of the 5'!3' strand orientation between the two residues within the same G-tract. [5] This is recognized by the sharpt urn of the sugar-phosphate backbone between F rG15 and G16a nd seemst or ely on aC 4 '-exo sugar pucker of F rG15 (Figure5C). Notably,t he anti conformationo ft his G-column 5'-residue as a consequence of its flipped backboneo rientation is fully compatible with the narrow groove geometry between the first and third G-tract.
G-corer esidues in a syn conformation include G1, G6, G20, as well as F rG14 which supports the V-loop at its5 '-end. The latter adopts this disfavored conformation in spite of its known anti preference ( Figure 5D). The north-type sugar conformation in combination with a syn glycosidic torsion angle positions the F rG14 H3' proton in close proximity to the deshielding regiono fi ts guanineb ase, accounting for its unusual downfield shift ( Figure 5D). Also, the non-conventional (H8 i -H2' iÀ2 ) contact between G16 and F rG14 is reflected in the 3D structure with corresponding distances in the range 2.8-3.9 (Figure 5C). As another consequence of the north sugar,b othf luorine atoms are oriented away from the quadruplex narrow groove in F (14,15). Avoiding the positioning of 2'-substituents within aG 4n arrowg roove has been observed beforea nd attributed to unfavorable interactions. [37,39] Of note, the two F2' substituents of the V-loop flanking residues are not involved in any pseudo-hydrogen bond as has frequently been observed in F rG, but also F araG and rG modified G4 structures. [37-40, 42, 43] Structural impact of other modifications at position 14 and 15:Importance of the sugar pucker To more generally assess the role of a north sugar pucker for Vloop formation, we studied additional 14,15-disubstituted ODN quadruplexes. Thus, sequences r (14,15) and FA (14,15) were modified with two riboguanosines and two 2'-fluoro-2'-arabinoguanosines, respectively.W hereas rG favors north conformers, the F araG analogue hasapropensity for a south-east sugar pucker.J ust like F(14,15),b oth sequences exhibit aC Ds ignature typical of ah ybrid-type G4 with both homopolar and heteropolars tacking interactions ( Figure S11). However,t he imino protons pectral region of FA (14,15)r evealed that it does not fold into am ajor quadruplex species, suggesting ad etrimental effect for folding into either the native or the rearranged Vloop structure of the two F araG substitutions with their favored south-east sugar pucker and anti conformation ( Figure 6). In fact, 19 FNMR spectrai ndicateapolymorphic mixture with several coexisting species( Figure S12). In contrast, imino resonances for r (14,15) clearly point to am ajor folded species with ap attern of iminos ignals similar to F(14,15). 1 H-13 CH SQC spectra of r(14,15)s how an almostp erfect overlap with corresponding spectral regionso fF (14,15) ( Figure S13), suggesting that they share the same topology and allowing for the assignment of most r (14,15) resonances. Sequential contacts can be traced along all G-tractsa nd along the first lateral loop. Analogous to F(14,15), G1, G6, G20, as well as modified rG14 can be identified as syn residues from their downfield 13 C8 chemical shift and their strong intra-residualH 8-H1' NOE contact (Fig-Figure 5. Three-dimensionalstructure of F (14,15). (A)Superposition of 10 final low-energy structuresi nasideview with the V-loop in front. For loop residues only the backbone is shown. (B) Representativestructure with labeled loop residues. G-core syn residues are colored in red;the first and second lateral loop and the propeller loop are shown in cyan, blue, and yellow,respectively.( C) Detailed view on the V-loop and the third G-tract of arepresentative structure in stick representation.Short interproton distances between F rG15H 8a nd G16 H1' as well as between G16 H8 and F rG14 H2' are traced by brokenl ines.( D) Stick representation of anti F rG15and syn F rG14,b othadoptingaC4'-exo conformation. In the latter, H3' is positioned in close proximity to the 6-membered ring of the guanine base,r ationalizing its downfield chemicals hift.
Conformational features of the rG nucleosides in r (14,15) are very similar to the 2'-fluoro-G analoguesi nF (14,15). In both G4s, the modified residuesa dopt a north-type sugar pucker as revealed by the presence and absence of H1'-H2' correlations in a2 DN OE andD QF-COSYs pectrum, respectively (see Figure S16). Ap articularly strong H1'-H4' contact points to a north-east pucker for rG15. Remarkably,t he (syn,north)c onformation of residue 14 is very similara mong the two structures as shown by unusually downfields hifted H3' resonances at 5.84 ppm for r (14,15). Again, the deshielding of H3' can be attributed to its positioning close to and almost in plane with the guanine base (see also Figure 5D). Other diagnostic interactions sensitive to the V-loop conformation involve the H8 i -H2' iÀ2 NOE contact between G16 and rG14 and an unusuals e-quentialN OE contact between rG H8 at position 15 and G16 H1' (see Figure 5C and S17). It should be noted, however, that the latter H8 i -H1' i + 1 NOE crosspeake scapes unambiguous identification in F(14,15) due to nearly isochronous H1' resonances of G16and F rG15.
Ta ken together,d ual modificationsw ith south-east favoring F araG analogues seem incompatible with V-loop formation of the ODN sequence. In contrast, north favoring Gs ubstitutes induce rearrangements into the same well defined V-loop structuret ermed ODN (14,15) for the following discussion.

Discussion
AV-shaped loop was first reported for at wo-layered G-quadruplex forming an interlockedd imer. [18] In the past years, ag rowing number of such V-shaped loops have been found in various three-layeredb i-or monomolecular G4s, making the Vloop am ore recurrent structural motif in G-quadruplexes. [11,15,[19][20][21][22][23][24] It links two adjacent antiparallel G-columns through residues located on opposite outer faces of the Gcore. Startingw ith ag uanosine in the bottom tetrad of a broken G-tract it is easily distinguished from ap ropeller-type loop by the upwardo rientation of this Gn ucleotide at the Vloop 5'-end ( Figure S18). Of note, the particularV -loop topology adopted by ODN (14,15) with two lateral loops followed by aV -loop and a propeller loop was also reported for CHL1 and HPV52 sequen-ces from the 5'-intron of the human CHL1 gene and the G-rich region of the human papillomavirus type 52. [11,21,22] Interestingly, HPV52 and ODN are derived from the same sequence encompassing fiveG-runs.W hereas HPV52 comprises G-tractsI -IV, the ODN sequence includes G-tracts II-V with additional inversion of the 5'-3' orientation. [21] ODN (14,15), HPV52,a nd CHL1 differ in the length and sequence of loops, but they all feature al ateral loop adenosine directly precedingt he V-loop. Also, the V-loop 3'-supporting Gr esidue adopts an anti glycosidic conformation while a syn Ga nchors the V-loop at its 5'-end in all structures.
Only considering glycosidict orsion angle propensitieso ft he introduced anti-favoring F rG surrogates, aV -loop structure is anticipated to be favored with respect to the native (3 + 1) hybrid fold with syn conformers at both modification sites. Yet, the observed rearrangement of modified ODN (14,15) into the V-loop quadruplex with its single syn F rG analogue is not easy to account for.I nf act, a syn conformation is highly unusual for F rG nucleotides but not withoutp recedence. Thus, a syn F rG residue was found in the native fold of a1 5-F rG mono-substituted ODN sequence F(15) (see Figure 1A)t hat coexists with another rearranged topology comprising F rG in an anti conformation. [40] Likewise, a syn rG in r(14,15) seems highly unfavorable considering RNA'spreference for an anti glycosidic torsion angle. In fact, rGs in the syn conformation have rarely been reported with af ew notable exceptions in non-parallel G4 structures [44][45][46] and are otherwise primarily knownf rom Z-RNA. [47] This suggests, that it is the propensity for a north-type sugar pucker rather than glycosidict orsion angle preferences of the modified nucleosides that critically determines the folding of ODN (14,15). Accordingly, F araGs ubstitutions favoringasoutheast sugar pucker seem to disrupt the native fold without promoting V-loop formation.T his is in striking contrast to previously studied quadruplexes, where F rG and F araG analogues showeds imilar effectsw hen incorporated at corresponding positions. [38,40,43,48,49] Notably, F araG modifications imparted highert hermals tabilities to most of the structures and were more effective in inducing rearrangements to an ew topology or in the selectiono faparticular G4 conformation. Consequently,o bserved substitution effects and conformationald etails of the ODN (14,15) quadruplex indicate that mostly north pseudorotamers are compatible with its 0-nt V-loop geometry, which is likely restricted to an arrow conformationalr ange of backbonetorsion angles.
Perusal of all G4s with V-shaped loopst hat have been reported to date reveals ap ronounced prevalenceo fnorth-east pseudorotamers within the V-loop linked segment for the monomolecular quadruplexes (Figure 7A-D, G). The HPV52 G4 features ap seudorotation angle for the syn residue preceding the V-shaped loop of about 3208,b eing outside the typical north pseudorotational range and even more remote from a C2'-endo pucker usually favored by the Gn ucleotide (Figure 7C). [22] However,t he conformation of this residue in HPV52 and ODN (14,15) is very similara sd emonstrated by the H3' protonw hich is unusually downfields hiftedi nb oth cases due to its close proximity to the guanine aromatic ring. North sugarsa re also present in ab imolecular V-loop structure formed by an LNA modified sequence, where the transition from the native antiparallel to aV -loop scaffold is driven by a locked LNA Gm odification fixing the V-loop at its 3'-end (Figure 7E,F). [50] Because LNA Gi ss trictly locked in a C3'-endo conformation it again corroborates af avorable conformational match of a north-type sugar pucker with the V-loop architecture. Whereasi nt he latter case as ingle modification at the V-loop 3'-end sufficesf or ac orresponding refolding, this does not apply for the single F rG modification in ODN at position 15, [40] also pointing to as ignificant role of a north-type pucker for the V-loop preceding residue at position 14 in ODN (14,15).I n fact, additional 2D NOE spectra acquired on a1 4-F rG monosubstituted ODN sequence F (14) showedanumber of crosspeaks almost perfectly superimposable with some strong NOE contactso bserved forF (14,15) ( FigureS19). Thiss uggests its partial folding into aV -loop architecture even in the absence of as econd Ga nalogue at the V-loop 3'-flanking position. However,w eaks ignals indicatef olding to only occur to as mall extent and emphasize the impact and possibly synergistic effect of both consecutive Gsurrogates in V-loop formation.
Notablee xceptionst ot he north-east sugar conformation typicallya ssociatedw ith V-shaped loops relate to (i)a monomolecular quadruplex formed by aG -rich HIV-1 long terminal repeat sequence, exhibiting a south sugar for the V-loop 5'-flanking syn guanosine ( Figure 7G) [23] and to (ii)two bimolecular G4s with experimentally determined south sugars at both flankingr esidues ( Figure 7H,I). [19,24] Remarkably,t he former G4 exhibits ad ifferent type of V-loop conformation with at urn of the sugar phosphate backbone occurring between its two Vloop linked residues rather than within the following G-tract. As ac onsequence,t he syn glycosidic torsion angle at the 3'flankingg uanosine follows standard rules that link syn/anti patterns to G-tract directionalities (Figure S18 Ba nd C). In case of the bimolecular G4s, C2'-endo sugar puckersf or the two consecutive Gs in the linked segmentc an be rationalized by a potentially less constrained V-loop. Also, glycosidict orsion angles for the V-loop 3'-anchoring residue in both G4s are found in the far high-anti region, likely causing as lightly different backbone conformation.
contrast, quantum-mechanical calculations have determined similar energies for (north,syn)a nd (south,syn)c onformers of free deoxyribonucleosides, suggesting that conformational propensities ande ven corresponding correlations may critically depend on the particular structural context. [51] Notably, F rG and rG analogues incorporated into aD NA G-quadruplex have previously been found to enforce as tructural rearrangementb y adoptinga nanti glycosidic torsion angle, yet with a C2'-endo rather than af avored C3'-endo sugar pucker. [37][38][39][40] Here, ODN (14,15) exemplifies the reverses ituation:S election for the most favorable north-type sugar pucker apparently outweighs energetic penalties expected for an associated syn conformation at the V-loop 5'-flanking residue. In this respect, the very malleable ODN sequence has provenapowerful tool in studying the impact of 2'-substitutedGmodificationsi nd ifferent conformational and topological environments. Depending on the deliberate selection of substitution sites the introduction of Gs urrogates enabled its refolding into different quadruplexes including (i)a5'-tetrad polarity inversion upon conservation of the globaln ative fold, [36][37][38][39] (ii)acombined G-tract and tetrad flip resulting in an antiparallel G4, [40] and (iii)the here reported rearrangement to aV -loop structure.
While substitution-induced rearrangementsa nd availableVloop structures suggestt he sugar conformation to play ac ritical role in V-loop stabilization, capping structures like base triplets and stable GNA loops have also been recognized as stabilizing elements for individual V-loop forming sequences. Thus, both CHL1 and HPV52 sharing the same topology with ODN(14,15) feature a3 '-terminal thymidine involved in ab ase triplet with residues of the second laterall oop that stackso nto the bottom tetrad. [11,22] Accordingly,t he relativelyl ow thermal stabilityo ft he F rG andr Gm odified ODN(14,15) quadruplexes may partially be attributed to the lack of such additional interactions.C ombining capping structures and favorable GNA loops with north-affine G-analoguesa ta ppropriate positions may allow the design of sequences adoptingv ery stable Vloop architectures. This could be of interestf or an umber of technological applications, as the V-loop represents av ery unique G4 structural element andm ay serve as as pecific receptor for various interacting ligands.

Conclusions
The growingn umber of reported G-quadruplexes featuring a V-shaped loop attests to their potentials ignificance as important structural elements in vivo but also as powerful tools for technological applications that are based on quadruplex recognition. Disregarding sequence requirements,t he V-loop motif is shown here to be generally supportedb yt wo V-loop flanking north-type conformers, allowing for am ore rational design of these structural elements. Remarkably,t he sugar pucker seems to constitute the major contributor for V-loop formation, overwriting glycosidic preferences of the corresponding residues.W hile glycosidic torsion angle propensities of Ga nalogues have frequently been exploited in the past for ad irected modulation of G4 stabilities, the deliberate use of Gs urrogates for their favored sugar pucker has rarely been employed in the manipulationo fG -quadruplex structures. Clearly,s imple rules assist in predictingg lycosidicc onformations for ap articular G4 topology but sugar conformational properties are often neglected and more difficultt op redict within ag iven structural context. In addition, propensitieso fm ost Ga nalogues fora particulars ugar pucker but also for ag lycosidicb onda ngle are far from being fixed and may dependo np articular topological features. As ac onsequence, the evaluation of modification effects may be challenging in some cases and may set limits to the rational designofG 4structures induced by appropriate Gs urrogates. On the other hand, many Ga nalogues like F rG andr Gm ay adopt aw ide range of conformationst of inally stabilized istinct structural motifs, making them an even more powerful tool for at argeted G-quadruplex design.

Experimental Section
Materials and sample preparation Unlabeled and isotope-labeled F rG, rG, F araG, and 5-methyl-dC modified oligonucleotides were purchased from IBA (Gçttingen, Germany) or Microsynth (Balgach, Switzerland) and quantified based on their absorbance at 260 nm after ethanol precipitation. NMR samples were prepared by dissolving the corresponding oligonucleotide in 10 mm potassium phosphate buffer at pH 7, followed by heating to 80 8Ca nd cooling to room temperature. Concentrations for unlabeled NMR samples ranged between 0.1 mm for F (14) and 1mm for F (14,15) and for 15 No r5 -methyl-dC labeled F (14,15) samples between 0.2 mm and 0.3 mm.F or optical measurements, oligonucleotide concentrations of 5 mm were used in a buffer containing 20 mm potassium phosphate, 100 mm KCl, pH 7.

Optical measurements
Circular dichroism (CD) spectra were acquired with 5a ccumulations, as canning speed of 50 nm min À1 and ab andwidth of 1nm at 35 8Co naJasco J-810 spectropolarimeter (Jasco, To kyo, Japan). All spectra were blank-corrected by subtraction of the buffer spectrum. Melting curves were recorded in triplicate on aC ary 100 spectrophotometer equipped with aP eltier temperature control unit (Varian Deutschland, Darmstadt) with quartz cuvettes of 10 mm path length. The absorbance at 295 nm was measured between 15 and 90 8Ci n0 .5 8Ci ntervals with ah eating rate of 0.2 8Cmin À1 .T he melting point T m was determined from the minimum of the first derivative of the heating phase.

NMR spectroscopy
All NMR spectra were acquired on aB ruker Avance Neo 600 MHz spectrometer equipped with an inverse 1 H/ 13 C/ 15 N/ 19 Fq uadruple resonance cryoprobehead and z-field gradients. For spectral processing and analysis, To pspin 4.0.4 and CcpNmr Analysis 2.4 were employed. [52] Further experimental details are given in the Supporting Information.

Structure refinement
As imulated annealing protocol in Xplor-NIH 2.49 was used to generate 100 starting structures of the DNA sequence. [53] The RED software was used to calculate the partial atomic charges for the modified F rG residues for subsequent calculations with Amber16. [54,55] Ar estrained simulated annealing was performed in Chem. Eur.J.2020, 26,524 -533 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim implicit water using the parmbsc0 force field including the c OL4 , ez OL1 ,a nd b OL1 corrections. [56][57][58][59] Twenty lowest-energy structures were selected, equilibrated for 1nsw ith explicit solvent, and shortly minimized in vacuum. Atomic coordinates of an ensemble of ten final structures with the lowest energy have been deposited in the Protein Data Bank (accession code 6RS3). Details of the calculation process can be found in the Supporting Information. Structural parameters were determined with the 3DNA software package. [60] Accession codes Atomic coordinates of the F(14,15) G-quadruplex have been deposited in the Protein Data Bank (accession code 6RS3).