Glucose–Nucleobase Pseudo Base Pairs: Biomolecular Interactions within DNA

German Edition:DOI:10.1002/ange.201603510 NoncovalentInteractions International Edition:DOI:10.1002/anie.201603510 Glucose–NucleobasePseudoBase Pairs: Biomolecular Interactions within DNA Empar Vengut-Climent, Irene Gýmez-Pinto,RicardoLucas,Pablo PeÇalver,Anna AviÇý, C¦lia Fonseca Guerra, F. Matthias Bickelhaupt, Ramýn Eritja, Carlos Gonzlez, and Juan C. Morales* Abstract: Noncovalent forces rule the interactions between biomolecules.Inspired by abiomolecular interaction found in aminoglycoside–RNArecognition, glucose-nucleobase pairs have been examined. Deoxyoligonucleotides with a6-deoxy- glucose insertion are able to hybridizewith their complementary strand, thus exhibiting apreference for purine nucleobases. Although the resulting double helices are less stable than natural ones,they present only minor local distortions.6- Deoxyglucose stays fully integrated in the double helix and its OH groups form two hydrogen bonds with the opposing guanine.This 6-deoxyglucose-guanine pair closely resembles apurine-pyrimidine geometry.Quantum chemical calculations indicate that glucose-purine pairs are as stable as anatural T-A pair. Noncovalent forces govern interactions among biomolecules,drug–target molecular recognition and assembly processes.Hydrogen bonds, p–pstacking, van der Waals forces, electrostatic forces,and hydrophobic interactions are observed in RNArecognition by drugs such as macrolides, tetracyclines,[1] and new designed RNAbinders.[2] Aminoglycoside antibiotics are awell-known family of RNAbinders which target the 16S rRNAinthe small ribosomal subunit. Apart from electrostatic forces and hydrogen bonds,solution and X-ray structures of aminoglycosides with rRNAshow asingular biomolecular interaction, amonosaccharide–nucleobase stacking interaction [e.g., 2’-amino-2’-deoxyglucose (ring I) of paromomycin stacks over guanine 1491; Figure 1].[3] Recently,wehave studied sugar–DNAstacking interactions using aself-complementary CGCGCG sequence with the carbohydrates directly linked to the 5’-end of DNA. We observed stacking of mono- and disaccharides on top of the terminal DNAbase pairs,which stabilized the duplex between ¢0.5 to ¢1.8 kcalmol¢1.[4] Stacking of sugars on top of the guanine tetrad of aDNA G-quadruplex was also characterized.[4b] Asecond singular biomolecular interaction found when paramomycin binds rRNAisaglycoside-adenine pseudo base pair (Figure 1c).[3] Twohydrogen bonds are formed between ring Iofparomomycin and A1408. Inspired by this interaction we decided to study the possible formation of glucosenucleobase pairs.Since monosaccharides can stack on DNA bases and possess OH groups capable of making hydrogen bonds at the edge of their coinlike structure,like natural bases (Figure 2a), we placed apotential glucose-nucleobase pair inside aDNA double helix to investigate its properties (Figure 2b). This model system also allows us to study possible sugar-sugar pairs.The only non-aromatic nucleobases reported previously are LeumannÏs phenyl cyclohexyl interstrand base[5] and KashidaÏs isopropylcyclohexane and methylcyclohexane DNAbase insulators.[6] Figure 1. a) Structure of paromomycin. b) Detail of the solution structure of paromomycin binding a16S RNA model sequence. c) Drawing of the glycoside-adenine 4108 pseudo base pair. [*] Dr.E.Vengut-Climent, Dr.R.Lucas, Dr.P.PeÇalver, Dr.J.C.Morales Department of Bioorganic Chemistry Instituto de Investigaciones Qumicas CSIC—Universidad de Sevilla Am¦rico Vespucio 49, 41092 Sevilla (Spain) E-mail:j[email protected] Dr.I.Gümez-Pinto, Prof. C. Gonzlez Instituto de Qumica Fsica “Rocasolano”, CSIC 28006 Madrid (Spain) Dr.R.Lucas, Dr.P.PeÇalver,Dr. J. C. Morales Department of Biochemistryand Molecular Pharmacology Instituto de Parasitologa yBiomedicina,CSIC, Parque Tecnolügico Ciencias de la Salud 18016 Armilla, Granada(Spain) Dr.A.AviÇü, Prof. R. Eritja Instituto de Qumica Avanzada de CataluÇa (IQAC), CSIC CIBER—BBN NetworkingCentre on Bioengineering, Biomaterials and Nanomedicine, 08034 Barcelona (Spain) Dr.C.Fonseca Guerra, Prof. Dr.F.M.Bickelhaupt Department of Theoretical Chemistry and AmsterdamCenter for Multiscale Modeling, Vrije Universiteit Amsterdam 1081 HV Amsterdam(The Netherlands) Prof. Dr.F.M.Bickelhaupt Institute of Molecules and Materials (IMM), Radboud University 6525 AJ Nijmegen (The Netherlands) Supportinginformation and the ORCID identification number(s) for the author(s) of this article can be found under http://dx.doi.org/10. 1002/anie.201603510. A ngewandte Chemie Communications 8643Angew.Chem. Int.Ed. 2016,55,8643 –8647 Ó2016Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Glucose (glc) was linked to the phosphodiester backbone of DNAthrough aflexible glycerol linker through its anomeric position. Thus,weprepared the DMT-glucose phosphoramidite derivative 9(Scheme 1) for introduction at the required position of the oligonucleotide,similar to astandard natural DNAbase phosphoramidite.Tomodulate the high polarity of glucose,wealso prepared the 6dglc amidite 10.Aflexible T(T*) was synthesized for comparison of glc with apyrimidine base in the same environment (Figure 2c;see Figure S1 in the Supporting Information). We then measured thermal denaturation of DNA duplexes containing our modifications.Firstly,the introduction of aflexible linker,such as in T*, with respect to anatural Tleads to a10.2–15.988Cdecrease in stability (Table 1). The decrease in stability of DNAcontaining single or multiple acyclicnucleosides has been previously reported.[7] Secondly, the DNAduplexes with glc and 6dglc opposite to the four natural nucleosides were destabilized in comparison to the duplexes with T-Aand T*-A. Considering the penalty of the flexible linker,destabilization of glc pairs compared to T*-A ranges from 6.4 to 9.288C. Interestingly,the presence of the more apolar 6dglc improved the DNAstability,when compared to glc, by 1.2–3.988C. Still, the more stable pair, 6dglc-A, led to aDNA duplex which is 4.588Cless stable than that with T*-A, and 15.788Cless stable than that with T-A. The larger volume of the pyranose ring, in comparison with an aromatic ring, may distort the nearby DNAbase pairs and provoke this decrease in DNAstability.Thirdly,weobserved certain selectivity for glc-nucleobase and 6dglc-nucleobase pairs.Sugar-purine pairs were more stable than sugarpyrimidine pairs (from 1.9 to 3.388C). This difference may be aconsequence of direct hydrogen bonding between the OH groups of glc and 6dglc and the donor and acceptor groups in Aand G. Alternatively,purines may be preferred in front of glucose in the sequence just for the better stacking as observed in abasic sites,[8] but the NMR spectroscopy and theoretical calculations shown below indicate that hydrogen bonding between glc and 6dglc with Aand Gispossible. Fourthly,the stability of DNAduplexes containing sugarsugar pairs ranges from 29.6 to 32.388C. Unexpectedly,pairs containing two flexible spacers and two bulky pyranose rings do not show further DNAdestabilization in comparison to sugar-natural nucleobase pairs.For example,a6dglc-6dglc pair is as stable as a6dglc-purine pair. We also incorporated glc and 6dglc into aglycerol nucleic acid (GNA) double helix. GNAstrands have an acyclic backbone of propylene glycol nucleosides which incorporate Scheme 1. Synthesis of the carbohydrate DMT phosphoramidites. Glycosyl trichloroacetimidate donors were used as starting material. Reagents and conditions: a) TMSOTf, CH2Cl2for glc;BF 3·OEt2,CH 2Cl2 for 6dglc;b)AcOH/H2O(4:1), 8088C; c) DMTCl, DMAP, CH2Cl2; d) DIPEA, CH2Cl2.DIPEA=diisopropylethylamine, DMAP=4-(N,N- dimethylamino)pyridine, DMT=4,4’-dimethoxytrityl, Tf =trifluoromethanesulfonyl, TMS=trimethylsilyl. Table 1: Melting temperature (Tm)for DNA duplexes containing T*, glc, and 6dglc, and GNA duplexes containing 6dglc. DNA duplexes 5’-d(GATGACXGCTAG)[a,b] 3’-d(CTACTGYCGATC) X-Y Tm[88C] X-Y Tm[88C] X-Y Tm[88C] T*-A 37.7 glc-A 30.6 6dglc-A 33.2 T*-T 32.8 glc-T 28.5 6dglc-T 29.9 T*-C 32.0 glc-C 28.7 6dglc-C 29.9 T*-G 35.6 glc-G 31.3 6dglc-G 32.7 T*-T* 31.1 glc-glc 29.6 6dglc-6dglc 32.0 GNA duplexes 3’-TAAAATTTAXATTATTAA[b,c] 2’-ATTTTAAATYTAATAATT X-Y Tm[88C] X-Y Tm[88C] T-A 57.3 A-6dglc 45.4 T-T 48.0 T-6dglc 37.4 6dglc-T 38.7 6dglc-6dglc 43.0 [a] The natural DNA duplex containing X-Y=T-A results in aT mvalue of 47.988C. [b] Conditions for DNA duplexes:10mmNaH2PO4,150 mm NaCl, pH 7.0. ForGNA duplexes:10mmNaH2PO4,200 mmNaCl, pH 7.0. Estimated errors are: 0.488C(in DNA, except for 6dglc-6dglc: 1.088C) and 1.088C(GNA). Average value of three experiments measured at 1.2 mmconc (DNA) and 0.7 mmconc (GNA). [c] GNA monomers in italics. Figure 2. Description of the carbohydrate derivatives under study. a) CPK models of thymine and glucose. b) Schematic drawing of aDNA double helix containing aglycoside-nucleobase pair.c)Structures of the two sugar units prepared, (S)-2,3-dihydroxypropyl glucose (glc) and (S)-2,3-dihydroxypropyl6-deoxyglucose (6dglc), thymidine (T), (S)-3,4-dihydroxybutyl thymine or flexible T(T*), and glycol T (T-GNA). A ngewandte Chemie Communications 8644 www.angewandte.org Ó2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Angew.Chem.Int. Ed. 2016,55,8643 –8647 the natural nucleobases in the strand and are connected by phosphodiester bonds (Figure 2c).[9] In this context, both natural bases and our sugar modifications are linked through flexible linkers to the skeleton and no energetic penalty is expected. TheT mvalues measured showed that the GNAduplex with an A-6dglc pair was 11.988Cless stable than with T-A(Table 1) and only 2.688Cless stable than with amismatched T-T. It is quite notable that the selectivity between T-Aand T-Tpairs (9.388C) is similar to that between 6dglc-A and 6dglc-T pairs (6.7–8.088C). Lastly,a6dglc-6dglc pair results in aGNA duplex stability of 4388C, thus pointing to the formation of hydrogen bonds between the OH groups of each 6dglc unit. We next determined the three-dimen- sional high-resolution structure of DNA duplexes containing a6dglc-G and a6dglc-T pair using restrained molecular dynamics methods based on experimental NMR distance constraints (Figure 3). Theexchangeable proton region of the NMR spectra exhibited 11 imino proton signals between d=12.5 and 14.5 ppm, thus indicating the formation of aduplex structure with Watson–Crick base pairs (see Figure S4). Additional imino proton signals are observed in the d=10–11 ppm region corresponding to the nucleotide located opposite to the carbohydrate. Full proton assignment of the DNAand sugar units was performed with only afew exceptions (see Tables S1 and S2). DNAchemical-shift differences between the conjugates and the natural DNA control duplexes are limited to the surroundings of the carbohydrate moieties,thus indicating the overall duplex structure is not distorted (see Figure S5). The3Dstructures obtained are B-form helices without global distortions (Figure 3; see Figures S7 and S8). The carbohydrates and the nucleobases (G or T) located in the opposite position remain intercalated between the surrounding base-pairs maintaining extensive contacts at both sides.As aconsequence,alarge number of NOE crosspeaks (see Table S3) between the linker and the carbohydrate protons with the DNAare observed. In all cases the double helices are slightly unwound and the rise between flanking residues is increased, as usually occurs in intercalation complexes. Thestructures of helix 6dglc-G and helix 6dglc-T are deposited in the PDB (2N9Fand 2N9H, respectively). 6dglc is well located opposite to both guanine (Figure 3b)and thymine (Figure 3f), thus orienting its alpha face towards its 3Ï-neighboring nucleobase and its methyl group towards the major groove.Inthe helix 6dglc-G the sugar-nucleobase pair forms two hydrogen bonds.Insix of the ten resulting structures,these hydrogen bonds are H1 G–O4 6dglc and HN2 G–O3 6dglc (Figure 3b). Theother structures show different conformers,including the pattern obtained in the quantum calculations (see below). All orientations are experimentally supported by anumber of intra- and interstrand NOEs,like those between the methyl group of 6dglc with H5 and H6 C18, with amino protons of C5, and with H1 G20 (Figure 3c). In the helix 6dglc-T the monosaccharide forms only one hydrogen bond with the opposing T. In most structures (7 out of 10), this hydrogen bond is H3 T–O4 6dglc (Figure 3f), and in the other cases the hydrogen bond is HO4 6dglc–O2 T. Theorientation is supported by NOEs,for example,between the methyl group of 6dglc with methyl and H6 T7;with H1 and H8 G19;and with H5 and H6 C17 (Figure 3g). Theformation of an extra hydrogen bond, when comparing 6dglc-G and 6dglc-T,may contribute to the enhanced thermal stability and it may explain the observed selectivity for purines. Thepairing properties of glc and 6dglc in the gas phase and water were investigated quantum chemically by means of Figure 3. Solution structure of the helix 6dglc-G and helix 6dglc-T.a)Sequences of the helix 6dglc-G. b) Refinedsolution structure of the helix 6dglc-G and zoom views of the 6dglc-G pair.c)NOE interactions between 6dglc and its 5’-neighboring base C18 and its 3’- neighboring base pair G20-C5. d) NOESY trace showing key NOEs with the methyl group of 6dglc. e) Sequences of helix 6dglc-T.f)Refined solution structure of helix 6dglc-T and zoom views of the 6dglc-T pair.g)NOE interactions between 6dglc and its 5’-neighbouring base C17, and with its 3’-neighboring base pair G19-C6 and opposite base T7. h) NOESY trace showing key NOEs with the methyl group of 6dglc. Color code:sugar and spacer (purple), nucleobases (green and blue) and surrounding natural base pairs (light blue). A ngewandte Chemie Communications 8645Angew.Chem. Int.Ed. 2016,55,8643 –8647 Ó2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim www.angewandte.org dispersion-corrected density functional theory (DFT) at the BLYP-D3(BJ)/TZ2P level of theory.[10] Thebinding energies (DE)ofthe Watson–Crick base pairs A-T and G-C agreed with those reported in literature (Table 2).[11] Sugar-purine pairs show greater hydrogen-bonding energy than any other combination for both for glc and 6dglc.Infact, the G-6dglc pair shows greater energy of bonding than anatural A-T base pair, with two hydrogen-bonds between the OH3 and OH4 of 6dglc with the NH1 and C=O, respectively,ofG(Figure 4). Twohydrogen bonds are also observed for A-6dglc,G-glc, and A-glc pairs (see Figure S9). Differences in the hydrogenbonding pattern in comparison to the NMR structures (such as in 6dglc-T) may be ascribed to factors such as base–base stacking or distance restrictions resulting from the DNA skeleton, and these are factors which are not included in these calculations. Small differences between glc and 6dglc were observed when pairing with purines.Infact, the hydrogen bond formed between the OH4 of glc and the C=OofGis reinforced through athird hydrogen bond which is formed between the OH6 and OH4 (see Figure S9), thus yielding abetter energy of bonding for glc-G than for 6dglc-G.Note that glc and 6dglc form significantly more stable pairs with pyrimidines than with purines.The only exception is the glc-T pair which has abond energy of ¢9.55 kcalmol¢1. Interestingly,sugar-sugar base pairs show similar bond energies to that of sugar-pyrimidine base pairs.One,two,and three hydrogen bonds are formed in 6dglc-6dglc,6dglc-glc, and glc-glc,respectively,where the OH groups involved can act as either hydrogen-bond donors or acceptors (Figure 4d). In contrast to the experimental data found for DNA, the 6dglc-6dglc pair appears to be less stable than the glc-glc base pair. These differences may be ascribed to interactions with the surrounding bases,[11] which are not considered in the present calculations. In conclusion, we have shown glycoside-nucleobase pairs form within aDNA double helix. They cause some destabilization of either aDNA or GNAduplex but they display selective pairing with purines versus pyrimidines.This selectivity can be explained by the formation of hydrogen bonds between either glc or 6dglc with the natural bases as shown by quantum chemical calculations and NMR studies.Moreover, 6dglc stacks within the interior of aduplex when paired with either GorTand does not stick out of the helix seeking better hydration. These 6dglc-G and 6dglc-T pairs infer only minor distortion in the double-helix structure. Acknowledgments We thank the Ministerio de Economa yCompetitividad (CTQ2011-15203-E, CTQ2012-35360, CTQ2014-52588-R, BFU2014-52864-R), the Netherlands Organization for Scientific Research (NWO-CW and NWO-EW), and the National Research School Combination—Catalysis (NRSC-C) for financial support. E.V.C.thanks Ministerio de Educaciýn, Cultura yDeporte for aFPU fellowship and Cost Action CM1005 for aSTSM grant. Keywords: DNA ·hydrogen bonds ·NMR spectroscopy · noncovalent interactions ·nucleobases Howtocite: Angew.Chem. Int. Ed. 2016,55,8643–8647 Angew.Chem. 2016,128,8785–8789 [1] A. Blond, E. Ennifar,C.Tisne,L.Micouin, ChemMedChem 2014,9,1982–1996. [2] a) R. Moumn¦, M. Catala, V. Larue,L.Micouin,C.Tisn¦, Biochimie 2012,94,1607–1619;b)S.P.Velagapudi,B.R. Vummidi, M. D. Disney, Curr.Opin. Chem. Biol. 2015,24,97– 103. [3] Q. Vicens,E.Westhof, Biopolymers 2003,70,42–57. [4] a) R. Lucas,I.Gýmez-Pinto,A.AviÇý, J. J. Reina, R. Eritja, C. Gonzlez, J. C. Morales, J. Am. Chem. Soc. 2011,133,1909– 1916;b)I.Gýmez-Pinto,E.Vengut-Climent,R.Lucas,A. AviÇý, R. Eritja, C. Gonzalez, J. C. Morales, Chem. Eur.J.2013, 19,1920–1927;c)R.Lucas,P.Penalver,I.Gomez-Pinto,E. Vengut-Climent, L. Mtashobya, J. Cousin, O. S. Maldonado,V. Perez, V. Reynes,A.AviÇý, R. Eritja, C. Gonzalez, B. Linclau, J. C. Morales, J. Org.Chem. 2014,79,2419 –2429. [5] M. Kaufmann,M.Gisler,C.J.Leumann, Angew.Chem. Int. Ed. 2009,48,3810 –3813; Angew.Chem. 2009,121,3868–3871. [6] H. Kashida, K. Sekiguchi, H. Asanuma, Chem. Eur.J.2010,16, 11554–11557. [7] a) K. C. Schneider,S.A.Benner, J. Am. Chem. Soc. 1990,112, 453–455;b)D.Zhou, I. M. Lagoja, J. Rozenski, R. Busson, A. Table 2: Hydrogen-bond energies (in kcalmol¢1)ofsugar-base pairs in the gas-phase (DEgas)and in aqueous solution (DEwater).[a] X-Y DEgas DEwater X-Y DEgas DEwater A-T ¢18.5 ¢9.4 G-C ¢34.0 ¢13.5 6dglc-G ¢23.8 ¢10.5 glc-G ¢23.3 ¢12.2 6dglc-T ¢10.5 ¢6.1 glc-T ¢15.4 ¢9.5 6dglc-A ¢16.7 ¢10.5 glc-A ¢16.7 ¢10.7 6dglc-C ¢12.9 ¢6.7 glc-C ¢17.7 ¢5.1 6dglc-6dglc ¢8.3 ¢4.1 glc-glc ¢12.7 ¢8.7 6dglc-glc ¢11.4 ¢7.6 [a] Calculated at BLYP-D3(BJ)/TZ2P using COSMO to simulate aqueous solution. Figure 4. A-T,G-6dglc,6dglc-T,and glc-glc pairs calculated at BLYP- D3(BJ)/TZ2P using COSMO to simulate aqueous solution. Hydrogen bonds are shown in green. A ngewandte Chemie Communications 8646 www.angewandte.org Ó2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim Angew.Chem.Int. Ed. 2016,55,8643 –8647 VanAerschot, P. Herdewijn, ChemBioChem 2005,6,2298– 2304. [8] a) P. Cuniasse,L.C.Sowers,R.Eritja, B. Kaplan, M. F. Goodman, J. A. Cognet, M. LeBret, W. Guschlbauer, G. V. Fazakerley, Nucleic Acids Res. 1987,15,8003–8022;b)P.Cuniasse, L. C. Sowers,R.Eritja, B. Kaplan, M. F. Goodman,J.A.Cognet, M. Le Bret, W. Guschlbauer, G. V. Fazakerley, Biochemistry 1989, 28,2018–2026. [9] a) L. Zhang,A.Peritz, E. Meggers, J. Am. Chem. Soc. 2005,127, 4174–4175;b)M.K.Schlegel,A.E.Peritz, K. Kittigowittana, L. Zhang,E.Meggers, ChemBioChem 2007,8,927–932. [10] G. te Velde,F.M.Bickelhaupt, E. J. Baerends,C.Fonseca Guerra, S. J. A. van Gisbergen, J. G. Snijders,T.Ziegler, J. Comput. Chem. 2001,22,931–967. [11] a) J. Poater,M.Swart, C. Fonseca Guerra,F.M.Bickelhaupt, Chem. Commun. 2011,47,7326–7328;b)J.Poater,M.Swart, F. M. Bickelhaupt, C. Fonseca Guerra, Org. Biomol. Chem. 2014,12,4691 –4700. Received:April 11, 2016 Revised: May 19, 2016 Published online: June 22, 2016 A ngewandte Chemie Communications 8647Angew.Chem. Int.Ed. 2016,55,8643 –8647 Ó2016 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim www.angewandte.org