Proc. Natl. Acad. Sci. USA Vol. 93, pp. 12116-12121, October 1996 Chemistry Inter-strand C-H 0 hydrogen bonds stabilizing four-stranded intercalated molecules: Stereoelectronic effects of 04' in cytosine-rich DNA (base-ribose stacking/sugar pucker/x-ray crystallography) IMRE BERGERt, MARTIN EGLIt, AND ALEXANDER RICHt tDepartment of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139; and tDepartment of Molecular Pharmacology and Biological Chemistry, Northwestern University Medical School, 303 East Chicago Avenue, Chicago, IL 60611-3008 Contributed by Alexander Rich, August 19, 1996 ABSTRACT DNA fragments with stretches of cytosine matic cytosine ring systems from intercalated duplexes (Fig. 1A). residues can fold into four-stranded structures in which two Second, unusually close intermolecular contacts between sugar- parallel duplexes, held together by hemiprotonated phosphate backbones in the narrow grooves are observed, with cytosine-cytosine+ (C C+) base pairs, intercalate into each inter-strand phosphorus-phosphorus distances as close as 5.9 A other with opposite polarity. The structural details of this (5), presumably resulting in unfavorable electrostatic repulsion if intercalated DNA quadruplex have been assessed by solution not shielded by cations or bridging water molecules. NMR and single crystal x-ray diffraction studies of cytosine- The close contacts between pairs of antiparallel sugar- rich sequences, including those present in metazoan telo- phosphate backbones from the two interdigitated duplexes are meres. A conserved feature of these structures is the absence a unique characteristic of four-stranded intercalated DNA. of stabilizing stacking interactions between the aromatic ring Indeed, the unusually strong nuclear overhauser effect signals systems of adjacent C-C+ base pairs from intercalated du- between inter-strand sugar Hi' protons and Hi' and H4' plexes. Effective stacking involves only the exocyclic keto protons constitute a fingerprint of this molecule, as such close groups and amino groups of the cytidine bases. The apparent backbone-backbone interactions do not exist in any of the absence of stability provided by stacking interactions between other two-, three-, or four-stranded DNA structures (1, 2, 9). the bases in this intercalated DNA has prompted us to Here we show that these close contacts between hydrogen examine the available structures in detail, in particular with atoms from adjacent sugar moieties in the two strands are the regard to unusual features that could compensate for the lack consequence of stabilizing C-H..0 type hydrogen bonds of base stacking. In addition to base-on-deoxyribose stacking between neighboring deoxyriboses. Accordingly, the 04' ox- and intra-cytidine C-H..0 hydrogen bonds, this analysis ygen atom of a deoxyribose from one strand is found to form reveals the presence of a hitherto unobserved, systematic a hydrogen bond to Cl'-Hi' of a deoxyribose from the intermolecular C-H*--O hydrogen bonding network between neighboring strand in the narrow groove, and vice versa. the deoxyribose sugar moieties of antiparallel backbones in However, close inspection of the relative orientations of such the four-stranded molecule. deoxyribose pairs reveals that these interactions are not com- pletely symmetrical. Thus, a lone electron pair of one 04' is The DNA intercalation motif, a four-stranded arrangement of directed toward the H1' hydrogen from the neighboring sugar, two parallel intercalating duplexes using C-C+ base pairs, is whereas the corresponding lone electron pair of the 04' from adopted by a variety ofcytosine-rich DNA fragments. Detailed the latter appears to be shared among the neighboring Hi' and NMR or x-ray diffraction studies were conducted thus far for H4' hydrogens. Consequently, a systematic C-H..0 hydrogen d(TC5) (1), d(C4) (2), d(CCCT) (3), d(TCC) and d(5MeCCT) bonding network is formed between the antiparallel strands in (4), d(CCCAAT) (5), and d(TAACCC) (6). In the four crystal the narrow grooves of four-stranded intercalated DNA. In structures (2, 3, 5, 6) the intercalated cytosine segments show addition, at some intercalation steps the intercalation motif is a globally similar arrangement, with the remaining nucleotides stabilized by intra-cytidine C-H..0 hydrogen bonds, previ- folded into a variety of motifs, including A-A-T base triplets (5) ously observed in left-handed Z-DNA (10). The crystal struc- and trinucleotide loops (6). Important geometrical features ture of d(CCCAAT) additionally displays stabilizing base-on- displayed by the two parallel-stranded duplexes forming the deoxyribose stacking between- adjacent cytosine and adenine intercalation motif are a slow right-handed helical twist (12- residues. Such stacking interactions appear to be a recurring 20°), a rise of approximately 6.2 A between covalently linked feature in nucleic acid structures and were shown to provide residues, and planar base pairing between cytosines around a stability in Z-DNA, in complexes between DNA and minor 2-fold rotation axis requiring hemiprotonation at the N3 groove binding drugs, as well as in large RNA molecules (ref. position. The two duplexes are intercalated with opposite 10 and references therein]. polarity and form a quasi two-dimensional array. The structure It is now widely accepted that short C-H..0 contacts has two broad and two narrow grooves with the antiparallel constitute electrostatically stabilized attractive interactions, backbone pairs from intercalated duplexes in van der Waals which can be considered hydrogen bonds (11-13). A survey of contact in the narrow grooves. Thus, the intercalation motif is 113 accurate neutron diffraction crystal structures clearly dramatically different in its architecture compared with A-, B-, indicated their widespread occurrence (11). They were ob- or Z-form duplex DNA and also the guanine quadruplex (7, 8). served in the crystal structures of nucleosides and nucleotides It is characterized by two features, both ofwhich one is inclined (14) and may contribute to the stability of nucleic acid base to attribute as destabilizing influences. First, there is the pairs (15). C-H-.0 hydrogen bonds between water molecules almost complete absence of overlap between adjacent aro- and purines, pyrimidines, amino acids, alkaloids, and others were found in the crystal structures of hydrates of these The publication costs of this article were defrayed in part by page charge molecules (16) and are thought to stabilize the conformations payment. This article must therefore be hereby marked "advertisement" in of the anticodon loop in tRNA (17) and 13-sheets in proteins accordance with 18 U.S.C. §1734 solely to indicate this fact. (18). The energy of the C-H-..0 interaction was calculated to 12116 Downloaded by guest on October 2, 2021 Chemistry: Berger et al. Proc. Natl. Acad. Sci. USA 93 (1996) 12117 A 3. T6 3- / uT16 B 3, 05 _ -R1 5 T4 / R 3' 5' 5S .5 C3 - C13 C2 -2-- 12 1 C 1 3 L'.:; c3 ooo~ T112 5 1 5 3' 31 fR1C1l4 ::':3-05=i-CR4, Tl 6 T6/$ 3' d(CCCT) d(CCCAAT) FIG. 1. Architecture of the intercalation motif. (A) Characteristic stacking pattern between adjacent cytosine-cytosine+ (C-C+) base pairs in four-stranded intercalated cytosine-rich DNA molecules. Stacking is confined to exocyclic keto and amino groups and stacking is not observed between the aromatic heterocycles of the bases. The top C C+ base pair from one parallel duplex is drawn with solid bonds and hydrogen bonding interactions between the cytosine bases are shown with solid lines. The lower C C+ base pair from the second intercalated duplex is depicted with open bonds, and hydrogen bonding interactions are drawn with broken lines. Nitrogens are highlighted by stippling. (B) Schematic view of four-stranded intercalated d(CCCT) (3) and d(CCCAAT) (5) illustrating the overall architecture of the intercalation motif and the nucleotide numbering scheme. Covalent bonds are drawn with solid lines. Base pairing interactions are illustrated with tapered bonds. The intercalated segment is highlighted by stippled tapered bonds. In the d(CCCAAT) structure, a crystallographic dyad axis (solid oval) exists in the center of the molecule. Symmetry related nucleotides are marked with asterisks. be around 2 kcal-mol-1 (19). Although C-H -0 hydrogen drawings, vectors describing the 04' lone electron pairs were bonds were previously noticed between bases (e.g., refs. 20 and assigned a hypothetical length of 1.4 A (10). The lone electron 21) and between bases and backbones (e.g., ref. 10) in the pair pointing to the same side ofthe furanose ring as the glycosidic crystal structures of oligonucleotides, their systematic forma- bond is termed (3 lone pair, the other a lone pair (10). Similarly, tion between the deoxyriboses of two adjacent oligonucleotide to analyze the sugar-base interactions in the two structures, the backbones described here constitutes the first example of a hydrogen positions of ring carbons were calculated on the basis tertiary nucleic acid folding motif that is extensively stabilized of ideal sp2 hybridizations for the carbon atoms and C-H bond by this type of interaction. lengths of 1A. METHODS RESULTS AND DISCUSSION The coordinates for the crystal structures ofintercalated cytosine- Stacking of Exocyclic Base Atoms from Adjacent Base rich nucleic acid fragments determined thus far are deposited in Pairs. In four-stranded intercalated molecules, neighboring the Nucleic Acid Database (22), and are accessible via NDB ID C-C+ base pairs from intercalated duplexes are arranged in a codes UDD023 for d(CCCF), UDD024 for d(C4), UDF027 for way that virtually no overlap exists between their six- d(TAACCC), and UDF043 for d(CCCAAT). Since the confor- membered aromatic ring systems (Fig. 1A). Instead, pairs of mations of the intercalated cytosine segments in the crystal exocyclic keto groups and pairs of exocyclic amino groups structures are sufficiently similar with respect to the observations overlap in an antiparallel fashion with occasional direct inter- described here, only two of them were selected for compiling the actions of the amino group hydrogen atoms with the hetero- data given in Tables 1 and 2.