The positions of low-Z cations in crystal structures of nucleic acids are often difficult to determine even at high resolution.12,17 One reason for this difficulty is that monovalent ions such as NaI and KI can have irregular coordination geometries, and NaI has an electron density that is similar to that of a water molecule.18 Disorder and low occupancies also contribute to the difficulty of localizing cations in oligonucleotide structures, and have been a particular challenge for structural characterization of NaI ions. These observations are consistent with an overall picture of monovalent cations predominantly contributing to a mobile counterion atmosphere in oligonucleotides. There are, however, a few examples demon- strating localized monovalent ion sites in DNA and RNA structures and in small nucleotide complexes. In addition to X-ray crystallography, a number of techniques including NMR spectroscopy and dynamics calculations have been used to characterize cation binding in nucleic acid duplexes. In some cases it has been suggested that the long residence times of NaI ions in certain sites has an influence on duplex structure, which in turn may influence biological function.19,20
8.29.4.1 Sodium
Sodium ions have been proposed to bind to N3 of guanine or adenine and the O2 of cytosine or thymine in the minor groove of the DNA dodecamer referred to as the Dickerson-Drew-B-DNA dodecamer.21 This sequence, the subject of several X-ray crystal structures extending to sub- A˚ngstrom resolution, contains a run of A:T base pairs known as an A-tract. All monovalent cation sites are only partially occupied in these structures. Because sodium ions are difficult to distinguish from water molecules using crystallographic techniques, a bond-valence sum approach was employed in the analysis by Williams and co-workers, who proposed that the DNA minor
groove was partially occupied by sodium and water molecules.21 Analysis of a higher-resolution crystal structure of the same B-DNA dodecamer by Egli and co-workers found no evidence for sodium ions in the minor groove.20However, a series of other monovalent ions including KI, CsI, RbI, and TlI have been shown to partially occupy similar sites at the ApT step of this dodecamer (vide infra).
G-quartets are an interesting class of supramolecular structures formed by certain oligonucleo- tide sequences containing runs of guanines. These structures can form stacked quadruplexes or tetraplexes of guanine bases. Each tetraplex consists of four nearly planar hydrogen-bonded guanine bases, with the sugar and phosphodiester groups remaining on the outside of the structure (see Figure 7). The structure contains a central pore in which guanine O6 ligands point towards the center, creating specific ion binding sites. Monovalent ions stabilize and can cause conformational changes in G-quartets.22,23
Sodium ions bind to O6 of guanines in the centers ofd(G4T4G4) and [d(TG4T)]4G-quartets.22–24The 0.95 A˚ resolution structure of [d(TG4T)]4shows a line of seven dehydrated NaIions positioned in the central core of the eight stacked guanine tetraplexes (Figure 7). Most of the NaIions are positioned between tetraplexes, garnering eight O6 ligands from tetraplexes above and below the ion with NaI–O distances of 2.2–3.7 A˚. At the end of the quadruplexes the NaIions sit in the center of the plane of the guanine bases and have square pyramidal geometries, with four O6 ligands and an additional aqua ligand. Coordination to the O2 atom of thymine has also been observed in G-quartets.23
Several crystal structures of nucleotide model complexes have revealed Na ions coordinating to base, sugar, phosphate, and aqua ligands with distances of 2.2–2.9 A˚.25 A complex with 5-bromocyditine monophosphate shows dehydrated NaI in a distorted octahedral environment, bridging two nucleotides through the cytosine O2 and N3 positions from each base with additional coordination to phosphate oxygen ligands.26In the case of purines, sodium can coordinate directly to the N7 of adenine or guanine or the O6 of guanine at distances ranging from 2.5 A˚ to 2.7 A˚.25,27 Coordination of NaI to the O2 of uracil was observed in the early structure of Na2ApU (sodium adenylyl-30-50-uridine), which found Watson–Crick base pairing between dinucleotide units and a resulting pseudohelical structure. Foreshadowing results from later structures of full duplexes, partially hydrated NaI coordinates to uracil O2 positions that are 4 A˚ apart in the
‘‘minor groove’’ of this mini-helical structure.25‘‘Minor groove’’ binding was not observed in the related structure of sodium GpC, suggesting steric hindrance by the guanine N2 exocyclic amino groups.
Figure 7 NaIions in central channel of DNA G-quartets or tetraplexes (d(TGGGT)) (PDB 352D, 0.95A˚
resolution, after Phillips et al.24). Two molecules from the asymmetric unit are shown. NaI, KI, PbI, and other ions coordinate multiple guanine O6 positions in the stacked quartets.
Coordination of NaIto phosphate oxygens is commonly observed with bond distances in the range of 2.3–2.8 A˚.25,27 Additional ligands can include the sugar O20 and O30positions,25 as well as one example in (50-ATPNa2) of coordination to the sugar O40.28 In nearly all of these crystal structures, the sodium ions are either partially or completely surrounded by water oxygen atoms demonstrating the difficulty of dehydrating this ion.
8.29.4.2 Potassium, Thallium
Like NaI, KI ions bind to O6 of guanines in d(G4T4G4) and [d(G3T4G3)]2 G-quartets.22–24 Coordination to the O2 atom of thymine has also been observed in these G-quadruplexes.23 Unlike NaI, the ionic radius of KI is too large to coordinate in the plane of the terminal G-quartet, which may affect the stability of the structure.22
TlI has been shown to be a useful KI mimic because both ions have similar ionic radii and irregular coordination geometries, but the higher electron density of TlI allows it to be located more readily by X-ray crystallography.18 The TlI crystal structure of the Dickerson-Drew-B- dodecamer revealed 13 partially occupied (10–35%) TlI sites.18 These sites were distributed between the minor groove in the A-tract part of the helix and the major groove of the G-tract segment. In the G-tract major groove of the dodecamer, TlIbinds both N7 and O6 atoms of each guanine with ligand distances of 2.3–3.1 A˚. In the A-tract minor groove TlIbinds the adenine N3 and O2 of thymines, and the sugar O40 positions, with bond lengths of 2.5–3.2 A˚. A terminal -cation interaction of TlI with the face of a cytosine base is suggested.18 In general, the coordination of the TlI ion in these sites is a distorted octahedral with as many as four inner- sphere or direct contacts to the nucleic acid. Structural analysis of the dodecamer is consistent with DNA bending towards localized cations within the crystal structure.18
Specific sites for KI have been observed in at least two complex RNA structures, and are considered important for function in both. In the 160-nucleotide P4–P6 domain of the Tetra- hymena group I intron, a KI ion is coordinated below the AA platform within the tetraloop receptor.29Thallium was also used as a potassium mimic to determine the coordination of the ion.
TlIbinds directly to the N7 and O6 of G227, the O4 of U228, the pro-Rpphosphate oxygen of A226, and the O20 of the sugar of A225, with bond lengths in the range of 2.4–2.7 A˚.29 This monovalent ion appears to organize the AA platform of the tetraloop receptor in order to enhance tertiary interactions with the tetraloop.29 Since potassium, rather than thallium, is a physiologically relevant ion, it is believed that potassium binds in this sitein vivo.
A crystal structure of a 58 nucleotide ribosomal RNA fragment was found to contain a KIsite buried in a pocket of the RNA, bound to six phosphate oxygen atoms with an average ligand distance of 3.0 A˚ (Figure 8).30To confirm the potassium ion site, again TlIwas used as a mimic.
The TlI-substituted structure also has six phosphate oxygen atoms from four nucleotides. The site
Figure 8 A KIion buried in a pocket of a 58-nucleotide ribosomal 23S RNA domain binds six phosphate oxygen atoms with an average distance of 3 A˚. Dotted lines indicate distances of 3.0–3.3 A˚ (PDB 1HC8, 2.8 A˚
resolution, after Connet al.30).
is shielded from solvent, creating an unusual mainly dehydrated monovalent cation near to six RNA-derived ligands. It has been shown that the monovalent cation binding site must be occupied for this 58 nt rRNA to fold.30 This site represents an excellent example of an RNA–
cation interaction that has been both structurally and functionally characterized.
A crystal structure of (nitrato)(1-methylcytosine)thallium(I) demonstrates TlI interactions with N3 and O2 atoms of cytosine with distances ranging from 2.9 A˚ to 3.1 A˚, and is different from the complex made with AgI (vide infra).31 Each TlI ion is chelated by an N3/O2 pair from two cytosines, and an additional O2 of two additional cytosines. The TlI has three additional oxygen ligands from nitrate making it an eight-coordinate complex.31
8.29.4.3 Rubidium, Caesium
Like KI and presumably NaI, RbI ions bind to the central ApT step in the minor groove of the Dickerson-Drew-B-DNA dodecamer.32The RbIsite is partially occupied, and has a coordination sphere that consists of seven oxygen atoms, with four DNA-derived ligands and three aqua ligands. The ion is coordinated to the O(2) atoms of two thymines and to two O40sugar positions with distances ranging from 2.8 A˚ to 3.5 A˚.
Analysis of the CsI form of the Dickerson-Drew-B-DNA dodecamer suggests lower occupan- cies of CsIfor this ApT site.20,32
An anomalous diffraction analysis of an A-form-DNA decamer demonstrated success of that method in locating RbI, CsI, and KI in three types of sites.33 This study also demonstrated that the occupancies of the monovalent ion sites can be highly dependent on crystallization conditions.
Two sites including coordination to phosphate O atoms and to 30-OH positions from more than one duplex, with average cation to O distances of 2.9 A˚ to 3.2 A˚, were observed. The third site was found in the major groove of the decamer, with apparent direct contacts to thymine O4 and guanine N6 positions as well as at least five additional aqua ligands.