The importance of metal-DNA interactions in living systems, and their potential applications especially in the treatment and diagnosis of diseases has stimulated considerable recent interest in this area. The most important application is the antitumour action of certain heavy metals, which work by binding to and distorting DNA, causing cell death. For example, Cisplatin (Cl2H6N2Pt ) is one of the most potent and widely used anticancer drugs in use today.

This book provides an overview of metal-DNA interactions, the mechanism of their interaction, metal-based drugs and metal ion toxicity. It is essential reading for academic researchers in bioinorganic chemistry, biochemistry, biology, biotechnology and medicine, and pharmaceutical industries involved in developing metal-based drugs.

Nikolaos Hadjiliadis is Professor in the Dept of Chemistry at the University of Ioannina, Greece. Einar Sletten is Professor in the Dept of Chemistry at the University of Bergen, Norway.

Metal Complex – DNA Interactions

John Wiley & Sons

Copyright © 2009 Blackwell Publishing Ltd
All right reserved.
ISBN: 978-1-4051-7629-3

Chapter One

Sequence-Selective Binding of Transition Metal Complexes to DNA

Einar Sletten and Nils ge Frystein

1.1 Introduction

The biological significance of the interaction between metal ions and nucleic acids has become a rather well-established fact. One may mention the observed necessity for the presence of metal ions in many natural processes where nucleic acids play the dominant role. The effect of platinum-based chemotherapeutic drugs probably originates from their attack on DNA. Another aspect of metals in biological systems is the increased flux of metals in the environment during the last decades. An assessment of the toxic effect of an unnatural metal ion concentration must include information on the processes in which the metal can participate. In a comparison of metal carcinogenicity in humans based on several experimental factors, Cr and Ni turned out to be the most potent carcinogens.

The nucleic acid monomers, guanine (G), adenine (A), thymine (T) and cytosine (C) have different metal ion affinities. The order of stability of 3d transition metal ion-nucleobase complexes are: G > A, C > T. At physiological pH the preferred binding sites on the nucleobases are: guanine N7, adenine N1 and/or N7, cytosine N3, thymine O4. For nucleotides the relationship between phosphate and base binding is dependent on the type of metal ion. Eichhorn and Shin studied the effect of various metal ions on the melting temperature of DNA (Figure 1.1). The authors suggest that magnesium ions increase [T.sub.m] by binding to phosphate and stabilizing the double helix, whereas copper ions decrease [T.sub.m] by binding to the bases and destabilizing the double helix. Based on the metal-induced variation in [T.sub.m] they suggested that the relative metal affinity to the phosphate backbone of DNA follows the order [Mg.sup.2+] > [Co.sup.2+] > [Ni.sup.2+] > [Mn.sup.2+] > [Zn.sup.2+] > [Cd.sup.2+] > [Cu.sup.2+].

This implies that the binding of an individual metal ion may involve phosphate and base on the same molecule or form a linkage between two different nucleotides. An example of the latter situation is the mononucleotide-metal ion binding pattern observed for the Cu-(GMP) complex, where [Cu.sup.2+] ions are bridging the GMP ligands through alternating N7-Cu-phosphate bonds (Figure 1.2).

When nucleobases are incorporated into a duplex DNA matrix, the affinities towards metal ions are modified. It has been shown that several divalent metal ions, like [Mn.sup.2+], [Cu.sup.2+] and [Pt.sup.2+] prefer GC-rich regions, while [Hg.sup.2+], for example prefer AT-rich regions. A more detailed picture indicates that metal binding to base residues is sequence-dependent, i.e. not all guanines in a particular sequence show identical affinity towards a specific type of metal ion. As a consequence, one may envisage designing metal complexes that can bind selectively to chosen sequences of DNA. Such complexes may be used as drugs that block specific gene expression associated with a certain disease.

Metal ions can interact with nucleic acids in two distinct modes of binding: diffuse binding and site binding, both of which are important for the structure and function of nucleic acids. In the diffuse binding mode the metal and the nucleic acid retain their hydration layer and the interaction is through water molecules. This is a long-range Coulombic interaction, in which positive metal ions accumulate around the nucleic acid in a delocalized manner; for example, the counterion atmosphere that all nucleic acids possess is made up of diffusely bound positive ions. In the site-binding mode the metal is coordinated to specific ligands on the nucleic acid; the coordination can either be direct (termed inner-sphere) or through a water molecule (termed outer-sphere). In the outer-sphere binding mode only the innermost hydration layer of the metal is kept intact, and the metal and the nucleic acid ligand(s) to which the metal is coordinated share solvation shells. In inner-shell binding there is direct contact between the metal and the nucleic acid. Dehydration of the metal ion and the nucleic acid binding site therefore has to occur before an inner-shell bond is formed.

The mechanism of inner-sphere binding is likely to be initiated by a diffuse binding mode, in which the metal and the nucleic acid are separated by no more than two layers of solvent molecules. This step is diffusion controlled. The next step is that the metal ion and the nucleic acid form an outer-sphere complex, separated only by one layer of solvent molecules. This step primarily depends on electrostatic attractions and hydrogen bonding between the metal and the nucleic acid. In the final step the metal and the nucleic acid come into direct contact (inner-sphere binding). Here the nucleophilicity of the coordination site plays a crucial role. In the last two steps steric effects are also important. Several attempts have been made to quantify the importance of accessibility and molecular electrostatic potential (MEP) at the site where the inner-sphere complex/covalent bond is formed. In these studies a reasonable correlation between these two important factors exists, and has been used to predict which DNA site is the most reactive to metalation or methylation.

In this chapter we present data on sequence-selective interactions between metal complexes and nucleic acids. In the outline we will distinguish between (i) site-selective inner-sphere metal coordination of nucleobases, and (ii) the selectivity of fully hydrated species located in the minor or major groove through hydrogen bonding and electrostatic interaction. In the former case a further distinction will be made between labile and nonlabile metals.

1.2 Ab initio Calculations and Photo-Cleavage Studies

The highest occupied molecular orbital (HOMO) of DNA nucleobases plays a crucial role in metal coordination by interacting with the lowest unoccupied molecular orbital (LUMO) of metal ions. The calculations of HOMOs of macromolecules such as duplex DNA are extremely difficult. Consequently, there has been little focus on the role of the HOMO/LUMO in studies of DNA-metal ion interactions. Theoretical calculations of DNA bases have mostly focused on ionization potentials (IP) of monomeric nucleobases and the stability of the nucleobase pair in neutral and radical cation states. About ten years ago the Saito group published the first extensive studies on the variation of nucleobase IPs and localization of HOMOs as a function of base stacking, using high-level ab initio calculations. The IPs of four base monomers and 16 sets of nearest-neighbour stacked nucleobases in the B form were calculated. It was found that the GG/CC system has the lowest IP among ten possible stacked nucleobase pairs and that approximately 70% of the HOMO is localized on the 5′-G of 5′-GG-3′. These calculations indicate that the 5′-G of 5′-GG-3′ is the most electron-donating site in B-DNA. The origin of IP lowering as a result of base stacking was further investigated by calculations of HOMO energy distribution as a function of the twist angle of a GG dimer. Within a normal range of twist angles for B-DNA (-25 to -45), IP values of GG are between 7.2 and 7.4eV and the HOMO is predominantly localized on 5-G. This implies that in B-form DNA the 5′-side G of the 5′-GG-3′ sequence is the most strongly interacting site with electrophiles. This principle may be very important in governing sequence-selective metal binding to DNA.

According to the theoretical calculations, HOMOs of GG sites in duplex DNA should serve as the most reactive one-electron oxidation sites. In order to verify this assumption experimentally, Saito et al. performed laser flash photolysis of duplex DNA oligonucleotides with added photosensitizer (photocleaving aminoacid, PCA) and subsequent hydrolysis dephosphorylation with alkaline phosphatase (Scheme 1.1).

The [G.sub.3]:[G.sub.4] ratio (84:16) determined by HPLC implies that the major degradation pathway of the hexamer involves the [G.sub.3] site – the most readily oxidizable site according to theoretical calculations. It should be noted that photo-irradiation of the single-strand alone results in a nonselective cleavage at [G.sub.3] and [G.sub.4] in the ratio 1:1.

Further work by the Saito group involved ab initio calculations of HOMOs for a wide variety of double-stranded G-containing 5-mer sequences with B-form geometry using GUSSIAN [9.sup.x1] at the HF/6-[31G.sup.*] level. For the quantum mechanical studies, all of the sugar backbones of the 5-mers were removed from the coordinate file and replaced by methyl groups. A few examples of the distributions of HOMOs in the duplex 5-mers are shown in Figure 1.3.

The general trend for HOMO distribution is that the HOMO of stacked GG doublets is localized overwhelmingly on the 5′-G, regardless of the 3′- and 5′-flanking residues (A, C or T). Bearing in mind that the model used is rather crude (all of the sugar backbone replaced by methyl groups), further discussion of more subtle sequence-specific differences is not warranted.

Further experimental support for the theoretical results was obtained by studying the oxidation of oligodeoxynucleotides (ODN) with [Co.sup.2+] ions and benzoyl peroxide using PAGE analysis of the reaction mixture after hot piperidine treatment. Sequence-dependent G-cleavage was observed for double-stranded ODN, whereas nonselective equal G cleavage was observed for single-stranded ODN. The relative rates of sequence oxidation were determined by densitometric assay of the ODN cleavage bands. Experimentally observed relative rates of G oxidation matches well with the calculated HOMOs of the G-containing sequences, implying that the [Co.sup.2+] ion is coordinated more strongly to the G having the larger HOMO.

Comparable theoretical calculations by Senthikumar et al. on stacked XGY triplets with B-form geometries, including sugar and phosphate groups, show that the site energy is strongly influenced by the type of nucleobase at the 3′ position. When C or T is present at the 3′ position, the site energy at the guanine was found to be up to 0.44eV higher than for A or G at this position. The influence of the base at the 5′ position was much smaller, the variation in site energy being less than 0.1eV. The amount of charge on a certain G was calculated from the coefficients of the guanine fragment orbitals. It was concluded that the neighbouring base at the 3’position to a large extent determines the charge distribution and therefore the oxidative damage on a sequence of guanine bases.

Further work by the Saito group on experimental mapping of G-rich hot spots included photo-induced cleavage of double-stranded [sup.32]P-end-labelled oligodeoxy-nucleotides (ODNs) 30-mers possessing two different G-containing sequences (5′-TXGYT-3′) and a 5′-TTGGT-3′ step as a standard on the same strand. Under low photo-irradiation conditions, only the cleavage bands of 5′-Gs of the two GG steps and the middle step of the GGG triplet were observed by hot piperidine treatment. Quantitative densitometric assay was used to determine the relative amounts of cleavage products. The experimental results were compared to IP values calculated for 16 sets of base-paired G- and GG-containing 5-mers. A plot of the log of the relative reactivity ([k.sup.rel]) toward photo-induced one-electron oxidation versus calculated IP is shown in Figure 1.4.

A different explanation for the enhanced reactivity of the 5′-G of a GG step compared to the 3′-G is presented by the Schuster group. Theoretical calculations of base-paired quartets, d(5′-XGGX-3′)/d(5′-YCCY-3′), suggested that electronic factors may not be the primary determinant of the reaction selectivity for GG steps. Instead the authors propose, based on molecular dynamic (MD) simulations on B-DNA oligomers, d(5′-GXXGGXXG-3′)/d(5′-GYYGGYYG-3′), where X = A,T,U and Y is the complementary base, that ‘there is an important steric contribution to the preference for reaction at the 5′-G in the GG doublets’. Photo-cleavage experiments carried out on the series of B-DNA oligomers showed that for GG steps in the context AGGA, the ratio of 5′ to 3′ reactivity was 1.8 0.1, and for GG in the context TGGT, the ratio was 6.1 0.3. The authors propose that the accessibility of [H.sub.2]O to the reaction site determined by the steric blocking by the methyl group plays the dominant role for the observed sequence-selectivity, rather than electronic effects.

1.3 NMR Spectroscopic Studies of Metal Binding to DNA Oligonucleotides

1.3.1 NMR Methodology

Most early nuclear magnetic resonance (NMR) studies on DNA involved complementary homopolymers and self-complementary, alternating copolymers, e.g. poly(dA)/poly/dT). The development of efficient and rapid methods of large-scale oligonucleotide syntheses has made it possible to design heteropolymeric sequences of high purity. Dodecamer (12 base pair) sequences adopting a normal B-DNA double-helical conformation, are assumed to complete a full turn of a right-handed helix. The structure of such a mini-helix is probably sufficiently close to that of real DNA to serve as a realistic model for determining preferred metal binding sites.

The effects of adding paramagnetic metal ions to an aqueous solution of DNA fragments may be monitored by observing the decrease in spin-lattice ([T.sub.1]) and spin-spin ([T.sub.2]) relaxation times (related to line-broadening) for protons close to the metal centres. Paramagnetic metal ions may be classified according to their electronic correlation times, i.e. as relaxation probes producing broad lines or as paramagnetic shift probes producing narrow lines. Divalent manganese is a typical relaxation probe with an estimated electronic relaxation time ([t.sub.s] = [T.sub.1c] = [T.sub.c]) of [10.sup.-8] – [10.sup.-9]s, while nickel, which has a shorter [t.sub.s] in the range [10.sup.-10] – [10.sup.-12]s, is a typical chemical shift probe. Cobalt(II) in a low-spin coordination environment has an estimated [t.sub.s] between that of [Mn.sup.2+] and [Ni.sup.2+] in kinetically labile metal complexes. At low metal to nucleotide ratios paramagnetic shift effects of [Ni.sup.2+] are difficult to detect. In this case geometric information about metal binding sites is most effectively obtained by measuring proton spin-lattice relaxation times ([T.sub.1]).

Paramagnetic relaxation arises in NMR spectroscopy when an unpaired electron spin interacts with a nuclear spin. The large magnetogyric ratio of the electron compared to that of the proton makes the dipolar coupling to the electron spin a very effective means of relaxation for the nuclear spin. Scalar interactions between the electron and nuclear spins have similar effects. In the simplest possible case, a ligand molecule exchanges between a paramagnetic environment (e.g. bound to Mn(II), S = 5/2) and a ‘free’ state, when the ligand is present in solution in vast excess to the paramagnetic centre (e.g. [10.sup.2] – [10.sup.4]). The effect of paramagnetic metal ions located at specific binding sites on DNA is observed as differential line-broadening of proton signals close to the binding site. Often, in 1D spectra of oligonucleotide molecules containing ten base pairs or more, key proton resonances may be severely overlapped, preventing an accurate assessment of the influence of the added metal ions. In these cases, 2D NOESY experiments may be used to obtain sufficient resolution.

For diamagnetic metal ions (no unpaired spin) the formation of a chemical bond is usually found to cause changes in the chemical shifts of proton resonances of hydrogen atoms in the proximity of the metal binding site. However, the coordination of [Hg.sup.2+] ions to single nucleobases induces only rather insignificant [sup.1]H shift changes, as shown for thymidine and guanosine. This could be explained by a down-field chemical shift change, induced by metal binding, being cancelled by an up-field chemical shift caused by changes in ring current effects due to altered nucleobase stacking. The heteronuclei [sup.13]C, [sup.15]N and [sup.31]P may experience large shift-effects when a metal ion binds to nucleobase or phosphate groups.

(Continues…)

Excerpted from Metal Complex – DNA Interactions Copyright © 2009 by Blackwell Publishing Ltd. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.