Metallointercalators as Probes of DNA Recognition and Reactions
Katherine E. Augustyn, Valerie C. Pierre and Jacqueline K. Barton, California Institute of Technology, Pasadena, California
Here we describe studies of metallointercalators bound to DNA. These octahedral transition metal complexes primarily bind noncovalently by stacking within the DNA helix. Given the rich photophysics and photochemistry of the ruthenium and rhodium complexes we employ, we have used a variety of biophysical studies to characterize their interactions with DNA. X-ray crystallography has also provided atomic resolution detail as to their binding to the duplex. Complexes have been designed that target DNA with high specificity. We have, for example, designed metal complexes that bind specifically to mismatched sites in the DNA duplex, and these have found application in the detection of single nucleotide polymorphisms and studies of mismatch repair deficiency. The photophysical properties of the metal complexes along with their intercalative stacking have been useful in particular as tools to characterize long-range charge transport in DNA. Using metallointercalators tethered to the duplex, oxidative damage to DNA from a distance has been demonstrated. The metallointercalators may serve as models for DNA-binding proteins, not only in binding DNA sites with high specificity, but also in carrying out electron transfer chemistry mediated by the DNA base-pair stack. Certainly these metallointercalators have proven to be powerful probes of this chemistry.
Our laboratory has focused on studies of metallointercalators that bind to duplex DNA through an ensemble of noncovalent interactions. Here we describe some of our studies with these complexes, including experimental design and applications. We focus attention in particular on the utility of these complexes in probing DNA-mediated charge transport chemistry. Our review here is not intended to be exhaustive but instead is focused on some examples of work from our laboratory to illustrate the effectiveness of these complexes in probing recognition and reactions with DNA (see Reference 1 for additional references). Indeed these complexes in several respects can be regarded as small mimics of DNA-binding proteins, but ones where their photophysics and photochemistry, as well as their inherent stability, allow us to sensitively probe their chemistry with DNA.
Investigations of the interactions of metallointercalators with DNA must start with a range of photophysical, nuclear magnetic resonance (NMR), and crystallographic studies. The complexes that we use are all substitutionally inert so that no direct coordination with the DNA bases occurs. The primary interaction involves intercalation of one ligand into the DNA base stack from the major groove. There is a concomitant doubling of the base-pair rise, from 3.4 A to 6.8 A, and the base pairs separate to accept the intercalating ligand. In the stack, the intercalator seems like a new base pair (2). A bulky metallointercalator that is not so easily accommodated in this stacked structure binds through an alternative means, where the sterically demanding ligand inserts instead from the minor groove side at a thermodynamically destabilized mismatch site, with ejection of mispaired bases. In this case, the stacked ligand replaces a base pair. Both with intercalation at B-DNA sites and insertion at mismatched DNA sites, chiral discrimination in binding of the octahedral complexes is evident; that is, there is a necessity sterically to match the chirality of the metal complex to that of the double helix (Fig. 1). Indeed, the A enantiomer favors interaction with right-handed B-DNA, whereas the A enantiomer preferentially binds left-handed or Z-DNA 3. These enantiomeric preferences control further the reactions of complexes on the DNA helix. DNA-mediated charge transport involving ∆-isomers similarly is more efficient than with left-handed complexes that are not as well coupled electronically into the right-handed double helix.
Figure 1. Examples of chiral metallointercalators that bind DNA with little site selectivity (above) and with high specificity for the targeted sites shown (below). In the center is shown schematically the basis for enantiomeric discrimination in stacking in the right-handed DNA helix. For the ∆-isomer, the ancillary ligands have a right-handed orientation in the DNA groove, whereas for the left-handed ∆-isomer, steric clashes between the ancillary ligands and phosphate backbone can develop.
Metallointercalator/DNA Interactions and Site-Specific Targeting
Early photophysical and photochemical studies
Major groove intercalators bind DNA with high affinity (Ka > 106 M-1) and, in some cases, high sequence specificity. Indeed, an extended aromatic system on the ligand outward from the metal center, as in the case of the phi (9,10-phenanthrenequinone diimine) or dppz (dipyrido[3,2-a:2',3'-c]phenazine) ligands, favors its intercalative stacking between the base pairs of the double helix. The intercalating ligand of these complexes thus behaves as a stable anchor in the major groove, oriented parallel to the base pairs, and directing the orientation of functionalized ancillary ligands with respect to the DNA duplex. Photophysical studies first provided support for intercalation (4). Extensive NMR studies and a crystal structure detailed the nature of the intercalation for the metal complexes via the major groove of the DNA (2). Interestingly, although not all of these complexes are sequence specific, they still demonstrate chiral discrimination: The ∆ enantiomer interacts preferentially with right-handed B-DNA through stacking within the DNA duplex.
Importantly, these complexes possess rich photochemical and photophysical characteristics that have been exploited advantageously both to probe interactions with DNA and to understand further aspects of DNA chemistry. A well-studied example is [Ru(bpy)2(dppz)]2+, which has found many uses as a molecular light switch (5). The Ru complex shows solvatochromic luminescence in organic solutions. In aqueous solution, however, it does not luminesce because of the ability of water to deactivate the excited state through hydrogen bonding with the phenazine nitrogen atoms of the intercalating ligands. Upon intercalation in DNA in aqueous solution, it is brightly luminescent, reflecting the shielding of the intercalating ligand from bulk solvent. This is akin to introducing the complex into a local organic solvent that shields the ring nitrogens on the intercalating ligand from protonation.
If ruthenium complexes have shown uses as molecular light switches, rhodium analogs have been proven to be efficient agents for photoactivated DNA strand cleavage (6). This reactivity enables us to mark directly the site where the metal complex intercalates in the double helix and thus characterizes the recognition properties of the complex. The observed reactivity upon excitation at short wavelengths (313-325 nm) leads to radical formation on the intercalating ligand with subsequent hydrogen atom abstraction from the adjacent deoxyribose ring. Degradation of the sugar radical then leads to direct DNA strand cleavage. In the absence of oxygen, photolysis of [Rh(phen)2phi]3+ or [Rh(phi)2bpy]3+ bound to DNA results in the formation of 3' phosphate and 5' phosphate termini, as well as free bases. In the presence of oxygen, different products result: Direct strand cleavage is observed, but products include, instead, the 5’ phosphate terminus, base propenoic acid, and a 3’ phosphoglycaldehyde end. These results are consistent with the previously described radical chemistry at the C3' position. However, because both the crystal and the NMR structure of the major groove intercalator revealed that the C2'H of the sugar is closer to the intercalating ligand than the C3’H, we propose that initial reaction of the photo-excited intercalator occurs with the C2’H abstraction followed by H-migration to form the C3’ radical and subsequent degradation of the sugar ring (2).
Although rhodium intercalators efficiently cleave DNA upon photoactivation, DNA cleaving agents that do not require photoactivation have the advantage of being more convenient to use in many research laboratories. Bifunctional agents have thus been developed in which metal coordinating peptides were covalently tethered to a metallointercalator (7). In these bifunctional conjugates, [Rh(phi)2bpy’]3+ behaves as a targeting vector that delivers the metal ions to the sugar phosphate backbone. The Zn(II) and Cu(II) centers of the metallopeptide, once delivered, promote DNA strand cleavage.
In a similar approach, luminescent DNA cross-linking probes were designed by conjugating short peptides to ruthenium intercalators (8). In this case, [Ru(phen)(bpy’)(dppz)]2+ delivers the peptide to the oligonucleotide and oxidizes it upon irradiation in the presence of an oxidative quencher. This then enables the nearby tethered peptide to cross-link with the oxidized sites of the DNA. Although delivery of the peptide by the metallointercalator is not essential for cross-linking, this technique advantageously yields cross-linking adducts that are luminescent and are thus easily detectable. Furthermore, these crosslinks may resemble those found in vivo under conditions of oxidative stress.
DNA recognition based on shape and functionalities
One of the earlier and important findings with the metalloin- tercalators is the importance of matching the chirality of the metal complex with that of the double helix. The discrimination basically depends on the size of the ancillary ligands relative to that of the DNA groove. Although some selectivity is observed for intercalation into B-DNA with metal complexes containing phenanthroline or bipyridine as ancillary ligands, the most striking stereospecificity is observed with metal complexes with bulky ancillary ligands such as [Rh(DPB)2phi]3+ (9). The ∆ enantiomer of the rhodium complex selectively binds and cleaves the sequence 5'-CTCTAGAG-3' upon photoactivation. However, no intercalation and cleavage is observed with the A enantiomer even with 1000-fold excess of metal complex.
The remarkable specificity of ∆-[Rh(DPB)2phi]3+ enables the efficient inhibition of the restriction endonuclease XbaI. Notably, no comparable inhibition of XbaI is achieved with other metallointercalators and ∆-[Rh(DPB)2phi]3+ also does not inhibit restriction enzymes that bind to alternative sequences. Thus, this coordination complex effectively mimics a DNA-binding protein.
Sequence-selective metallointercalators were also designed de novo by matching the functionality of the ancillary ligands positioned in the major groove with that of the targeted base pairs. Targeting of 5'-CG-3', for instance, is achieved with the complexes [Rh(NH3)4phi]3+, [RhaneN4phi]3+, and ∆-[Rh(en)2phi]3+ through hydrogen bonding between the axial amines of the metalallointercalators and O6 of guanine (1). The predictive design of sequence-specific metallointercalators was expanded with ∆ — α-[Rh[(R,R)- Me2trien]phi]3+, which is a complex that directly reads out the sequence 5’-TGCA-3’ (Fig. 2) (10). The targeting of this site was based on predicted hydrogen bonding contacts between the axial amines and the O6 of guanine, as well as van der Waals contacts between the pendant methyl groups on the metal complex and the methyl groups on the flanking thymines. A high-resolution NMR solution structure followed by the first high-resolution crystal structure of a metallointercalator bound to DNA later revealed atomic resolution details of the interaction 3. As predicted, the DNA unwinds to enable complete and deep intercalation of the phi ligand of the metal complex, which intercalates through the major groove, thereby causing a doubling of the rise at the intercalation site. The metallointercalator thus behaves as a newly inserted base pair, which the DNA accommodates with minimal structural perturbation.
Metallointercalators have also been designed to target specific sequences based both on their shape and on their functionalities. A derivative of [Rh(phen)2phi]3+, called 1-[Rh(MGP)2phi]5+, contains pendant guanidinium groups on the ancillary phenanthroline ligands, and it was designed to bind a subset of sequences recognized by the former. As predicted, the ∆ enantiomer recognizes the sequence 5'-CATCTG-3' specifically. Surprisingly, however, the ∆ enantiomer does bind DNA and recognizes the sequence 5'-CATATG-3' despite the large steric size of the ancillary ligands (11). Plasmid unwinding assays and NMR studies established that the ∆ enantiomer of the metallointercalator binds DNA by unwinding it up to 70°. It is in this conformation that the complex can span the entire six base-pair binding site and contact the N7 position of the flanking guanines with the pendant guanidinium groups. Replacing these flanking guanines with deazaguanine demonstrated that the absence of the N7 nitrogen removed selectivity for the site. It is therefore the guanidinium functionalities of the ancillary ligands that are responsible for the recognition of the guanine, whereas it is the shape of the metallointercalator that enables the recognition of the “twistable” central 5'-ATAT-3' sequence.
This peculiar binding of a bulky ∆ enantiomer and unwinding of the DNA has found biological application in inhibiting transcription factor binding to DNA. ∆-1-[Rh(MGP)2phi]5+ has been used to site specifically inhibit a transcription factor from binding to a modified activator recognition region (12). In competition experiments with yeast Activator Protein 1 (yAP-1), the metal complex was able to compete with the protein at concentrations as low as 120 nM. This work illustrates the potential applicability of the complexes as therapeutic agents, in inhibiting transcription factors sequence-specifically.
Figure 2. Binding of ∆ - α-[Rh[(R,R)Me2trien]phi]3+ to 5'-TGCA-3'. Schematic representation of the sequence-specific interactions of the complex (left) and crystal structure (right) of the complex intercalated into DNA from the major groove (2).
Detection of mismatched DNA sites
In contrast to the sequence specificity of major groove metallointercalators that is achieved through interactions of the ancillary ligands, the site-specificity for mismatched DNA is conferred by the intercalating ligand. DNA mismatch detection requires an extended or bulky intercalating ligand too wide to intercalate readily in well-matched B-DNA. The chrysene quinone diimine (chrysi) ligand of [Rh(bpy)2chrysi]3+ is 0.5 A wider than the span of matched DNA and 2.1 zA wider than the intercalating phi ligand (Fig. 3) (13). The resulting rhodium complex recognizes and cleaves upon photoactivation over 80% of mismatch sites in all possible single base-pair sequence contexts around the mispaired bases (15). The extremely high selectivity of [Rh(bpy)2chrysi]3+ for thermodynamically destabilized sites was demonstrated through the ability of the metallointercalator to recognize and photocleave a single mismatch within a 2725 base-pair plasmid (16).
Figure 3. Monofunctional (top) and bifunctional (bottom) bulky metallointercalators that target single base mismatches in DNA. In the center is shown a view of the crystal structure of the complex inserted into the DNA from the minor groove at the mismatched DNA site, with ejection of the mismatched bases (14).
The bulky rhodium complex binds single base mismatches with binding affinities of 0.3-20 x 106 M-1 as determined by quantitative photocleavage titrations. Importantly, the mismatch- specific binding affinities directly correlate with independent measurements of thermodynamic destabilization of the single base mismatch, thereby supporting the hypothesis that helix destabilization is a crucial factor in determining the binding affinity of the metal complex for the mismatched site (15). This hypothesis was recently confirmed by NMR studies and the crystal structure of the bulky metallointercalator bound to its target single base mismatch (14). Significantly, in contrast to phi complexes, ∆-[Rh(bpy)2chrysi]3+ inserts into the DNA stack via the minor groove and ejects both mismatched bases out of the double helix; the bulky chrysi ligand thus replaces a destabilized base pair. Nonetheless, the metallointercalator only minimally distorts the DNA, which accommodates insertion of the extended intercalating ligand simply by opening its phosphate backbone. Furthermore, as opposed to major groove intercalation, where DNA strand cleavage involves abstraction of the C2'H of the deoxyribose ring, ∆-[Rh(bpy)2chrysi]3+ preferentially abstracts the closer C1’H of the sugar adjacent to the mismatched site, resulting in different cleavage products (17).
Similar selectivity for thermodynamically destabilized sites in DNA is achieved with the use of other extended intercalating ligands. An analog of the chrysi complex, [Rh(bpy)2phzi]3+, also targets with high selectivity single base mismatches and promotes direct DNA strand scission upon photoactivation (18). The phzi complex binds its target with higher affinity than does the chrysi analog; site-specific photocleavage is evident even at nanomolar concentrations. This increase in affinity is attributed to greater stability in the stacking of the heterocyclic intercalating ligand with the flanking base pairs upon insertion. Notably, the increased affinity is not detrimental to the high selectivity of the metal complex, which binds mispaired versus well-paired sites in the same ratio as [Rh(bpy)2chrysi]3+.
Metallointercalators that selectively and efficiently target single base mismatches have found several applications both as biologic probes and as potential chemotherapeutic agents. For instance, [Rh(bpy)2phzi]3+ was used to probe the relative frequency of mismatched sites in cell lines deficient versus proficient in their mismatch repair machinery (18). The relative cleavage observed with the phzi complex in healthy cell lines was low compared with that in cancer cell lines that carried mutations in essential repair proteins. These results support previous studies on the association of mismatch repair deficiency and cancer.
We have also investigated the design of fluorescent probes for mismatched DNA. The luminescent ruthenium complex, [Ru(bpy)2(tactp)]2+, containing a bulky intercalating ligand that is analogous to the dppz ligand of the popular molecular light switch was prepared and showed luminescence enhancement with mismatched DNA (19). A more efficient fluorescent probe for mismatched DNA was later accomplished by tethering a charged fluorophore to the bulky metallointercalator (20). In the Rhodium-Oregon Green conjugate, ion pairing between the cationic rhodium and the anionic fluorophore moieties dramatically quenches the fluorescence of the conjugate in aqueous solution and in the presence of matched DNA. However, with mismatched DNA, the bulky rhodium complex binds the DNA polyanion, and the resulting electrostatic repulsion with the anionic Oregon Green fluorophore drives the latter away from the rhodium center so as to reduce intramolecular quenching. The fluorescence of the conjugate is thus increased over 300% upon binding to a mismatch site.
Mismatch targeting metallointercalators have also been applied to the discovery of single nucleotide polymorphisms (SNPs). SNPs are the largest source of genetic variation in humans; yet their detection remains difficult as current methods have poor signal-to-noise ratio and yield many false positives. In this regard, mismatch selective metallointercalators have proven to be valuable new tools (21). When pooled genomic samples containing low-frequency SNPs are amplified, denatured, and annealed, mismatches are statistically generated at the polymorphic DNA sites. With photoactivation, these DNA mismatches are cleaved selectively by [Rh(bpy)2chrysi]3+ or [Rh(bpy)2phzi]3+. Fluorescent labeling of the cleaved products and separation by capillary electrophoresis thus permits rapid identification with single-base resolution of the SNP site. This method is remarkably sensitive, and minor allele frequencies as low as 5% can be readily detected.
Can these mismatch recognition agents be targeted inside cells? Intracellular delivery was first achieved by tethering a cell-penetrating peptide such as D-octaarginine to the rhodium complex (22). The resulting conjugate binds and with photoactivation it selectively cleaves DNA neighboring single-base mismatches, although the presence of the oligoarginines is found to increase nonspecific binding of the conjugates for both matched and mismatched DNA. Noticeably, the peptide does not affect the selectivity of the rhodium-induced photocleavage of the mismatched site. Similarly, the rhodium complex does not interfere with the delivery properties of the cell-penetrating peptide and the conjugates rapidly localize in the nucleus of HeLa cells.
A different strategy for the cellular uptake of metallointercalators consists of increasing the “greasiness” of the ancillary ligands (23). In agreement with studies on cis-platin analogs, increasing the lipophilicity of dppz complexes of ruthenium favors their passive uptake by HeLa cells. Importantly, the metal complexes are stable to the intracellular environment. Indeed no degradation in luminescence is evident, as would be expected based on changes in complex coordination.
As mentioned, deficiencies in the mismatch repair machinery (MMR) of cells are associated with an increased susceptibility to cancerous transformation. We have developed bifunctional metallointercalators as potential chemotherapeutic agents in which the rhodium complex serves as a targeting vector toward mismatches. For instance, the bulky metallointercalator was tethered to an aniline mustard known to form covalent adducts to 5'-GNC-3' sites (24). The bifunctional agent demonstrates preferential alkylation of mismatched over fully matched DNA at concentrations where untethered organic mustards show little reaction. Notably, the tethered alkylator does not inhibit binding of the intercalator at the mismatch site, and similarly, the metallointercalator does not hinder alkylation of the DNA. The site-selective alkylation at mismatched DNA thus renders these conjugates useful tools not only for the covalent tagging of DNA base-pair mismatches but also as new chemotherapeutic agents.
Similarly, a potential chemotherapeutic drug was designed via a bimetallic conjugate that combines a metallointercalator specific for DNA mismatches tethered to a reactive cis -platinum analog that coordinates DNA and inhibits transcription and replication (25). The recognition of a DNA mismatch by the bulky rhodium intercalator directs the reactivity of the platinum unit to a site close to the mismatch that may or may not be the preferred site for platinum coordination. Indeed, in the latter case, the rhodium targeting dominates over the platinum reactivity. This ability to tune the reactivity of the cis-platinum analog could lead to therapeutic agents for MMR-deficient cell lines.
Interestingly, the parent monofunctional metallointercalators have also shown promise as potential chemotherapeutic agents targeted to MMR-deficient tumor cell lines. Indeed, both [Rh(bpy)2chrysi]3+ and [Rh(bpy)2phzi]3+ inhibit cellular proliferation differentially in MMR-deficient cells compared with cells that are MMR-proficient (18). Significantly, the inhibition of cellular proliferation depends strictly on the mismatch repair deficiency of the cell and thus correlates with the ability of the bulky metallointercalators to target DNA mismatches. For instance, it is the A enantiomer of [Rh(bpy)2chrysi]3+ that is active both in targeting the mismatches and in inhibiting DNA synthesis; neither mismatch binding nor inhibition of cellular proliferation is observed with the ∆ enantiomer. Additionally, the cellular response is enhanced with photoactivation, which is an effect that correlates with the ability of the rhodium intercalators to promote strand cleavage at the mismatch site upon photoactivation. Targeting DNA mismatches may thus provide a cell-selective strategy for chemotherapeutic design.
Metallointercalators as Probes of DNA-Mediated Charge Transport
Early photophysical studies
Rhodium and ruthenium intercalators have served as powerful probes of DNA-mediated charge transport. Since DNA-mediated charge transport depends so sensitively on n-stacking, it is reasonable that a probe that intercalates into DNA, with optimum n-stacking, might also serve as a powerful probe of this chemistry. Our earliest studies of DNA-mediated charge transport employed a ruthenium complex containing dppz as the photooxidant and a rhodium complex containing phi as the electron acceptor. Assemblies containing 5'-tethered [Ru(phen’)2(dppz)]2+ (phen’ = 5-amido-glutarate-1,10-phenanthroline) with and without 5'-tethered [Rh(phi)2(phen’)]3+ were designed where fluorescence quenching of the photooxidant was observed only in the presence of the electron acceptor (26) (Fig. 4). Given a 0.75-eV driving force, little spectral overlap between the excited state of ruthenium and the ground state of rhodium, and the fact that the tethered complexes are well separated on a 15-mer duplex, the observed results were consistent with DNA-mediated charge transfer. This work set the stage for many varied experiments using metallointercalators to characterize this interesting chemistry.
Additional experiments spectrally identifying the ruthenium (III) intermediate confirmed that the quenching mechanism was caused by charge transfer (27). When bound to DNA, [Ru(DMP)2(dppz)]2+ decays with two lifetimes corresponding to the two orientations of the intercalating dppz ligand. After excitation at 480 nm of intercalating [Ru(DMP)2(dppz)]2+ (DMP = 4,7-dimethylphenanthroline) in the presence of [Rh(phi)2(bpy)]3+, a negative transient was observed on the microsecond timescale at 440 nm after the initial bleach corresponding to the decay of the ruthenium excited state. As the rhodium concentration was increased, the observed decrease in luminescence intensity but not in lifetime of the ruthenium excited state indicated that the quenching and hence the rate of charge transport was fast relative to the measurement. The same transient was also observed when [Ru(NH3)6]3+ was used as the quencher, albeit with a slower rate of formation. As expected, the decay kinetics of the transient was similar for both quenchers. Differences in quenching kinetics between [Rh(phi)2(bpy)]3+ and [Ru(NH3)6]3+ are attributed to intercalation: The intercalative [Rh(phi)2(bpy)]3+ exhibits static quenching, whereas the diffusional [Ru(NH3)6]3+ shows dynamic quenching.
Figure 4. Schematic representation of a doubly metallated DNA duplex used to probe photoinduced electron transfer in DNA (26).
Long-range oxidative damage from a distance
Oxidative conditions within the cell can lead to damage of the DNA bases. Guanine, which has been experimentally determined to have the lowest oxidation potential of the naturally occurring bases, is the most easily damaged (28). Upon oxidation, the neutral guanine radical can react with water or oxygen to form permanently damaged products such as 8-oxo-G, oxazolone, or imidazalone (29). Many organic photooxidants such as anthraquinone, riboflavin, and napthalimide have been shown to specifically damage the 5' guanine of a guanine doublet (30).
The rhodium intercalator, which is tethered to the terminus of an oligonucleotide, was first employed to demonstrate oxidative damage to DNA from a distance through DNA charge transport (30, 31) (Fig. 5). With an excited state potential greater than 2.0 eV versus NHE, the rhodium intercalator serves as a potent photooxidant. Irradiation at 365 nm of 5’ radioactively labeled DNA duplexes containing the tethered rhodium intercalator results in oxidative damage through long-range hole transport that can be revealed by gel electrophoresis after treatment in hot piperidine; piperidine promotes strand breaks neighboring the base lesion 29. If these same duplexes are irradiated at shorter wavelength (313 nm), hydrogen abstraction leads to direct scission of the DNA backbone, indicating the exact position of intercalation. Comparison of the irradiation products of 15-meroligonucleotides containing two sets of guanine doublets with a 5'-tethered rhodium intercalator versus one where the photooxidant is intercalated noncovalently reveals that the damage patterns show little distinction in oxidation of the proximal and distal guanine doublets 31. However, the 5’ guanine in both doublets was more susceptible to damage than the 3’ guanine. This finding is consistent with ab initio molecular orbital calculations that have indicated that the HOMO is localized on the 5' guanine of a guanine doublet 32. This preferential reaction at the 5’-G of guanine doublets has become a signature for one-electron DNA oxidation 30. Irradiation at 313 nm reveals that the covalently tethered rhodium intercalates three bases from the tethered end, whereas the noncovalent complex intercalates throughout the duplex. As the covalently bound rhodium intercalates far away from the observed damaged guanines, oxidation must therefore occur through long-range DNA-mediated charge transport. Additional analysis of the damaged products by high performance liquid chromatography (HPLC) after enzymatic digestion showed that the primary damage product was 8-oxo-G.
Oxidative DNA damage has also been studied using the ruthenium intercalator, [Ru(phen)(dppz)(bpy’)]2+. When excited by visible light and quenched by non-intercalating quenchers such as [Ru(NH3)6]3+, [Co(NH3)5Cl]2+, or methyl viologen, a powerful ground state oxidant, Ru(III), is generated. With a 1.5 V potential versus NHE, Ru(III) can oxidize DNA bases in a similar manner to photoexcited [Rh(phi)2(bpy’)]3+ (30). This flash quench technique, originally developed by Gray and coworkers to study electron transfer reactions in proteins, can also be applied to study DNA charge transport (33). DNA duplexes containing 5' tethered [Ru(phen)(dppz)(bpy’)]2+ were irradiated at 442 nm in the presence of either methyl viologen or [Ru(NH3)6]3+, and preferential damage was observed at the 5' guanine of a guanine doublet, which is consistent with one-electron oxidation chemistry. HPLC analysis showed 8-oxo guanine also as a primary oxidation product. To rule out guanine damage caused by singlet oxygen sensitization, duplexes containing tethered ruthenium were irradiated in the absence of quencher. Damage was only observed at guanines near the ruthenium intercalation site, and was not 5' specific. Furthermore, this damage without quencher increased when the experiments were performed in D2O, a characteristic of singlet oxygen chemistry.
Figure 5. Schematic representation of a metallated duplex designed to probe long-range oxidative damage in DNA. Oxidative damage is found at the 5'-G of the guanine doublets.
Spectroscopy of charge transfer intermediates
Spectroscopic studies provide a means to characterize DNA- mediated charge transport in more detail and to provide a link to biochemical observations. The flash quench technique was first used in experiments involving the synthetic oligonucleotide poly(dG-dC) (30). In the presence of poly(dG-dC), the negative absorbance at 440 nm, caused by excited state quenching of intercalated [Ru(phen)2dppz]2+ by non-intercalating [Ru(NH3)6]3+, disappeared concomitantly with a rise in a positive signal at 390 nm, which is consistent with the formation of the guanine radical. A difference spectrum of this species was obtained with strong positive features at 390 and 550 nm, indicative of the deprotonated neutral guanine radical. Importantly, this signal was not observed in the presence of the synthetic oligomer poly(dA-dT) nor in the absence of quencher in poly(dG-dC). Formation of the radical occurs in less than 10-7 s or within the time scale of quenching of the ruthenium excited state.
Spectroscopic studies of long-range DNA charge transport were then carried out on assemblies containing tethered [Ru(phen)(dppz)(bpy’)]2+ as the oxidant and the artificial base 4-methylindole as a guanine analog (34). The oxidation potential of 4-methylindole is lower than that of guanine, and the higher extinction coefficient of its radical at 600 nm renders it particularly amenable to spectroscopic studies of DNA-mediated charge transport. Excitation of ruthenium with oxidative quenching induces charge injection into the duplex. Hole migration to the methylindole base gives rise to a positive absorbance at 600 nm. However, if a mismatch is introduced into the intervening π stack of the duplex, DNA-mediated charge transport is disrupted, which leads to a complete attenuation of methylindole radical formation.
The effect of sequence on charge transfer rate was assessed with assemblies containing a tethered ruthenium photooxidant separated from the methylindole hole trap by a series of A-T base pairs of increasing length. Over distances of 17-37 A, methyl-indole radical formation is found to occur concomitantly with quenching of the Ru(II) excited state. Thus the rate of radical formation over this distance through AT tracts is greater than 107 s-1, and over these distances, charge transport through the DNA is not rate-limiting. Furthermore, guanine radical formation can compete with that of the methylindole radical as charge equilibrates across the duplex (35).
Distance dependence of charge transport
Our earliest results had indicated that DNA charge transport might be significant over long molecular distances and certainly over longer distances than had been demonstrated in studies of protein electron transfer. The tethered rhodium intercalator was employed in the design of assemblies to examine relative oxidative damage at two guanine doublets within a 28-mer duplex (36). The proximal guanine doublet was kept at a constant distance from the rhodium intercalator, whereas the distal doublet was placed in two base-pair increments at increasing distances from the photooxidant. The damage ratio between the proximal and distal double guanine sites then served as an indicator of charge transport efficiency. Remarkably the damage ratio did not show significant diminution as a function of distance over 75 A. DNA charge transport thus shows a very shallow distance dependence that is not a result of helical phasing.
Similar studies were also carried out using the ruthenium intercalator, [Ru(phen)(dppz)(bpy’)]2+. A comparison of oxidative damage with the rhodium and ruthenium photooxidants was carried out in a 63 base-pair duplex containing six sets of guanine doublets arranged at 10 base-pair increments. Both complexes were able to oxidize all six sets of double guanines, indicating that DNA charge transport chemistry could be observed at distances up to 197 A away from the intercalation site! Oxidative damage to DNA in fact can develop over biologically significant distances.
Cyclopropyl amine-substituted bases have provided very fast traps for DNA charge transport using an irreversible ring opening reaction associated with oxidation (37). Guanine is in fact a poor oxidative trap, because the guanine radical reacts only on the microsecond-to-millisecond timescale with water and oxygen to form irreversible products; model studies suggest that irreversible ring opening of the N2-cyclopropylguanine radical occurs on the picosecond time scale. Recently our laboratory has designed a series of rhodium-tethered duplexes, in which the oxidative trap is N2-cyclopropylguanine (38). With intervening adenine tracts as the bridge between the intercalator and the trap, a shallow distance dependence is observed, now with prominent periodic features. Interestingly, if the same experiments are repeated monitoring guanine damage, the periodicities are not apparent. These periodicities were also absent in assemblies containing AT base-pair bridges, and they were less pronounced when the intervening bridge consisted of AI, ATIC, or AITC (I = inosine) repeats.
An analogous hole trap, cyclopropylcytosine, has also been used to monitor charge transport through the higher energy pyrimidine bases (39). Rhodium-tethered duplexes containing a distally placed cyclopropylcytosine were irradiated and ring opening was observed. This striking result indicated that rhodium-induced charge transport can oxidize not only guanine bases but also cytosines. These findings indicate that charge transport through DNA must involve all the DNA bases, not only the low energy guanines we observed using gel electrophoresis.
Taken together, our results from gel electrophoresis and cyclopropylamine ring opening can be rationalized by considering a novel mechanism for DNA-mediated charge transport: conformationally gated domain hopping (40). Two general mechanistic extremes are used to describe DNA-mediated charge transport, including superexchange, in which the donor and acceptor are lower in energy than the orbitals of the intervening DNA base-pair bridge, and hopping, in which the donor and acceptor have similar energies to the DNA base-pair bridge. In the superexchange mechanism, the charge tunnels from the donor to the acceptor without actually occupying the DNA bridge.
In the hopping mechanism, the charge hops from the donor to the acceptor, transiently occupying discrete sites on the bridge. A superexchange mechanism would exhibit an exponential distance dependence as a function of donor acceptor separation, but a hopping mechanism would result in a more shallow distance dependence so long as the hopping rate exceeds the rate of radical trapping. Although our results cannot be explained by either of these mechanistic extremes alone, we propose a mechanism in which the charge migrates through the DNA by hopping between transiently generated delocalized domains, defined by base sequence and dynamics. We describe a domain as a series of four to five bases acting in concert, over which a charge can delocalize. As the gel electrophoresis experiments measure guanine radical trapping on the millisecond timescale, contributions from base dynamics are not easily revealed. However, when the faster assay of oxidation-induced ring opening is used, additional effects from base dynamics can be discerned. Base motions, occurring on picosecond time scales, contribute to conformational gating of the charge transfer events, both limiting and facilitating the migration of charge between domains (40).
Sensitivity of charge transport to DNA conformation and dynamics
The importance of DNA conformation to DNA-mediated charge transport was evident also in many of our early photophysical studies. In fluorescence quenching experiments using assemblies containing the organic intercalator, ethidium, tethered to one end of the duplex, and the rhodium intercalator tethered to the other end, quenching by photoinduced electron transfer was found with a well-matched duplex, but no significant quenching was observed with an intervening CA mismatch (41). A well n-stacked array of heterocyclic aromatic bases is essential to the efficient transport of charge over a duplex. Perturbations in the intervening n-stack inhibit long-range oxidative damage. Assemblies containing a series of single base mismatches located between proximal and distal guanine doublet sites relative to a tethered ruthenium photooxidant were designed to explore the effect of stacking disruption on charge transport yield (42). These studies showed the dependence of charge transport efficiency on the dynamics of a mismatch; those mismatches that are relatively well stacked, as in purine-purine mismatches, cause only small attenuations in charge transport yield, whereas disruptive mismatches cause significant attenuations. In addition to mismatches, bulges can also disrupt the integrity of charge transport (43). Duplexes containing a tethered rhodium photooxidant and an ATA bulge positioned in between a proximal and a distal double guanine site showed a drastic decrease in damage at the distal guanine doublet site, again underscoring the necessity of a well-stacked duplex. Indeed, the sensitivity in charge transport yield to intervening perturbations in base stacking had two important consequences: 1) the path of charge transport must be through the bases rather than through the sugar-phosphate backbone, and 2) the reaction can report sensitively upon the integrity of the DNA duplex.
An interesting study using the base flipping enzyme Methyltransferase HhaI (M.HhaI) showed that disruption of the π-stack by protein binding with insertion of a nonaromatic amino acid side chain can also significantly attenuate charge transport (44). M.Hhal performs its alkylation reaction on DNA after flipping out the central cytosine in the 5'-GCGC-3' sequence and inserting a glutamine residue in its place. An assembly containing a covalently tethered rhodium photooxidant and proximal and distal 5'-GG-3' doublets separated by the M.HhaI target site was used to investigate charge transport yield in the presence versus absence of the enzyme. Site-specific binding of the enzyme to its target sequence was effective in eliminating oxidation at the distal double guanine site. Moreover, when a mutant enzyme containing tryptophan in place of glutamine in the wild type was used instead, insertion of the aromatic amino acid served to restore the base-pair stack, leading to extensive damage at the distal site. From these studies it seems that the binding of DNA-binding proteins can both inhibit and activate long-range DNA charge transport.
Biological opportunities for DNA charge transport
We have extensively studied the effect of sequence, structure, and distance dependence of DNA-mediated charge transport. However, the ultimate question remains: Is DNA-mediated charge transport an issue, indeed perhaps even a useful reaction, within the cell? We already knew that charge transport chemistry could occur over long enough distances to be biologically relevant, and that DNA-binding proteins could modulate the chemistry, but we needed also to determine whether this transport chemistry could occur within the tightly packed nucleosome structures found in cells. Within these structures, the DNA is highly bent, wrapped around a positively charged histone core. A nucleosome core particle, containing a 5' tethered rhodium was therefore constructed (45, 46) (Fig. 6); upon photoactivation of the tethered rhodium, oxidative damage at a distance to guanines within the core particle was observed. In fact, the efficiency of damage was similar to that observed on the same DNA in the absence of bound histones. We have also shown that rhodium can induce DNA damage in the nucleus of HeLa cells 47. Moreover, in these studies, if we compare sites of rhodium binding with those of strong oxidative damage, we determine that oxidative damage can occur at a distance within the cell nucleus. Long-range charge transport through DNA does develop within the cell.
Figure 6. A nucleosome core particle is shown containing a 146 base-pair DNA duplex wrapped around a histone octamer with a rhodium intercalating photooxidant tethered to the DNA terminus. This particle was constructed to probe DNA charge transport through a nucleosome (45). Seven sets of guanine doublets are located at the red positions along the duplex. Oxidative damage initiated by rhodium photoactivation is observed at the guanine doublets, demonstrating long-range oxidation within the nucleosome.
Additional evidence indicating that DNA charge transport may be biologically relevant comes from studies with DNA repair proteins such as MutY and EndoIII that contain [4Fe-4S] clusters (48). Bound to DNA, the redox potential of the [4Fe-4S] cluster in these proteins is found to be shifted so that the protein is more easily oxidized. By comparing potentials both bound to DNA and free, we estimate that the binding affinity of the oxidized form is at least three orders of magnitude higher than that of the reduced form. Based on these studies, we have proposed a model in which base excision repair enzymes can locate damaged DNA using DNA-mediated charge transport. A repair protein in its reduced state can bind to DNA, becoming more easily oxidized so as to transfer an electron to another DNA repair protein bound at a distal site, reducing the distally bound protein, and promoting its dissociation. But this DNA-mediated reaction can only occur if the intervening DNA base stack is intact and well stacked; if not, the protein remains associated with the DNA and on a slower time scale can progressively migrate to the damaged site. DNA-mediated charge transport thus serves to redistribute the repair proteins in regions of the genome near damage. The redistribution of repair proteins to a damage site using DNA-mediated charge transport essentially provides a way for the proteins to scan large regions of DNA without physically binding to each base. Significantly this scanning for damage is particularly important under conditions of oxidative stress, when guanine radicals are generated. We also used the flash quench method with ruthenium intercalators to show that guanine radicals can provide the first signal for repair, promoting the oxidation also of the DNA repair proteins in a DNA-mediated reaction (49).
Implications and Conclusions
These experiments hopefully serve to illustrate the use of met- allointercalators in probing recognition and reactions on the DNA helix. Starting with relatively simple coordination complexes that contain a wealth of photochemical and photophysical properties, functionalizing second and third generation derivatives, we have designed metallointercalators of high affinity, high specificity, and high usage in targeting and reacting with DNA. These complexes can serve as mimics of DNA binding proteins, competing effectively with them for DNA sites and perhaps even carrying out comparable electron transfer chemistry. These experiments, however, provide only a sampling of what might be considered in the future. Can these metal complexes, for example, serve as the basis for new chemotherapeutic designs targeted selectively to cancer cells? Perhaps the experiments we have described will inspire a new generation of complexes to follow with even more powerful applications to biology and medicine.
We thank the NIH for financial support of this work. We thank also our many coworkers in these studies for their hard work and critical insights.
1. Erkkila KE, Odom DT, Barton JK. Recognition and reaction of metallointercalators with DNA. Chem. Rev. 1999; 99:2777-2795.
2. Kielkopf CL, Erkkila KE, Hudson BP, Barton JK, Rees DC. Structure of a photoactive rhodium complex intercalated into DNA. Nat. Struct. Biol. 2000; 7:n7-12L
3. Barton JK. Metals and DNA: molecular left-handed complements. Science 1986; 233:727-734.
4. Pyle AM, Rehmann JP, Meshoyrer R, Kumar CV, Turro NJ, Barton JK. Mixed ligand complexes of ruthenium(II)—factors governing binding to DNA. J. Am. Chem. Soc. 1989; 111:3051-3058.
5. Friedman AE, Chambron JC, Sauvage J-P, Turro NJ, Barton JK. Molecular light switch for DNA-Ru(bpy)2(dppz)2+. J. Am. Chem. Soc. 1990; 112:4960-4962.
6. Sitlani A, Long EC, Pyle AM, Barton JK. DNA photocleavage by phenanthrenequinone diimine complexes of rhodium(III): shape-selective recognition and reaction. J. Am. Chem. Soc. 1992; 114:2303-2312.
7. Fitzsimons MP, Barton JK. Design of a synthetic nuclease: DNA hydrolysis by a zinc-binding peptide tethered to a rhodium intercalator. J. Am. Chem. Soc. 1997; 119:3379-3380.
8. Copeland KD, Lueras AMK, Stemp EDA, Barton JK. DNA cross-linking with metallointercalator-peptide conjugates. Biochemistry 2002; 41:12785-12797.
9. Sitlani A, Dupureur CM, Barton JK. Enantiospecific palindromic recognition of 5’-d(CTCTAGAG)-3’ by a novel rhodium intercalator: analogies to a DNA-binding protein. J. Am. Chem. Soc. 1993; 115:12589-12590.
10. Krotz AH, Hudson BP, Barton JK. Assembly of DNA recognition elements on an octahedral rhodium intercalator: predictive recognition of 5/-TGCA-3/ by D-[Rh[(R,R)-Me2trien]phi]3+. J. Am. Chem. Soc. 1993; 115:12577-12578.
11. Franklin SJ, Barton JK. Differential DNA recognition by the enantiomers of 1-Rh(MGP)2phi: a combination of shape selection and direct readout. Biochemistry 1998; 37:16093-16105.
12. Odom DT, Parker CS, Barton JK. Site-specific inhibition of transcription factor binding to DNA by a metallointercalator. Biochemistry 1999; 38:5155-5163.
13. Jackson BA, Barton JK. Recognition of DNA base mismatches by arhodiumintercalator. J.Am. Chem. Soc. 1997; 119:12986-12987.
14. Pierre VC, Kaiser JT, Barton JK. Insights into finding a mismatch through the structure of a mispaired DNA bound by a rhodium Intercalator. PNAS 2007; 104:429-434.
15. Jackson BA, Barton JK. Recognition of base mismatches in DNA by 5,6-chrysenequinone diimine complexes of rhodium(III): a proposed mechanism for preferential binding in destabilized regions of the double helix. Biochemistry 2000; 39:6176-6182.
16. Jackson BA, Alekseyev VY, Barton JK. A versatile mismatch recognition agent: specific cleavage of a plasmid DNA at a single base mispair. Biochemistry 1999; 38:4655-4662.
17. Brunner J, Barton JK. Site-specific DNA photocleavage by rhodium intercalators analyzed by MALDI-TOF mass spectrometry. J. Am. Chem. Soc. 2006; 128:6772-6773.
18. Junicke H, Hart JR, Kisko J, Glebov O, Kirsch IR, Barton JK. A rhodium(III) complex for high-affinity DNA base-pair mismatch recognition. Proc. Natl. Acad. Sci. U.S.A. 2003; 100:3737-3742.
19. Ruba E, Hart JR, Barton JK. [Ru(bpy)2(L)]Cl2: Luminescent metal complexes that bind DNA base mismatches. Inorg. Chem. 2004; 43:4570-4578.
20. Zeglis BM, Barton JK. A mismatch-selective bifunctional rhodium-Oregon Green conjugate: a fluorescent probe for mismatched DNA. J. Am. Chem. Soc. 2006; 128:5654-5655.
21. Hart JR, Johnson MD, Barton JK. Single-nucleotide polymorphism discovery by targeted DNA photocleavage. Proc. Natl. Acad. Sci. U.S.A. 2004; 101:14040-14044.
22. Brunner J, Barton JK. Targeting DNA mismatches with rhodium intercalators functionalized with a cell-penetrating peptide. Biochemistry 2006; 45:12295-12302.
23. Puckett CA, Barton JK. Methods to explore cellular uptake of ruthenium complexes. J. Am. Chem. Soc. 2007; 129:46-47.
24. Schatzschneider U, Barton JK. Bifunctional rhodium intercalator conjugates as mismatch-directing DNA alkylating agents. J. Am. Chem. Soc. 2004; 126:8630-8631.
25. Petitjean A, Barton JK. Tuning the DNA reactivity of cis-platinum: conjugation to a mismatch-specific metallointercalator. J. Am. Chem. Soc. 2004; 126:14728-14729.
26. Murphy CJ, Arkin MR, Jenkins Y, Ghatlia ND, Bossmann SH, Turro NJ, Barton JK. Long-range photoinduced electron-transfer through a DNA helix. Science 1993; 262:1025-1029.
27. Stemp EDA, Arkin MR, Barton JK. Electron-transfer between metallointercalators bound to DNA-spectral identification of the transient intermediate. J. Am. Chem. Soc. 1995; 117:2375-2376.
28. Steenken S, Jovanovic SV. How easily oxidizable is DNA? One-electron reduction potentials of adenosine and guanosine radicals in aqueous solution. J. Am. Chem. Soc. 1997; 119:617-618.
29. Burrows CJ, Muller JG. Oxidative nucleobase modifications leading to strand scission. Chem. Rev. 1998; 98:1109-1151.
30. Delaney S, Barton JK. Long range DNA charge transport. J. Org. Chem. 2003; 68:6475-6483.
31. Hall DB, Holmlin RE, Barton JK. Oxidative DNA damage through long-range electron transfer. Nature 1996; 382:731-735.
32. Saito I, Takayama M, Sugiyama H, Nakatani K, Tsuchida A, Yamamoto M. Photoinduced DNA cleavage by electron transfer— demonstration that guanine residues located 5’ to guanines are the most electron donating sites. J. Am. Chem. Soc. 1995; 117:6406-6407.
33. Stemp EDA, Arkin MR, Barton JK. Oxidation of guanine in DNA by [Ru(phen)2(dppz)]3+ using the flash-quench technique. J. Am. Chem. Soc. 1997; 119:2921-2925.
34. Pascaly M, Yoo J, Barton JK. DNA mediated charge transport: characterization of a DNA radical localized at an artificial nucleic acid base. J. Am. Chem. Soc. 2002; 124:9083-9092.
35. Yoo J, Delaney S, Stemp EDA, Barton JK. Rapid radical formation by DNA charge transport through sequences lacking intervening guanines. J. Am. Chem. Soc. 2003; 125:6640-6641.
36. Nunez ME, Hall DB, Barton JK. Long-range oxidative damage to DNA: effects of distance and sequence. Chem. Biol. 1999; 6:85-97.
37. Nakatani K, Dohno C, Saito I. Design of a hole-trapping nucleobase: termination of DNA-mediated hole transport at N-2-cyclopropyldeoxyguanosine. J. Am. Chem. Soc. 2001; 123:9681-9682.
38. Augustyn K, Shao FW, Genereux J, Davis M, Barton JK. Unpublished results.
39. Shao FW, O’Neill MA, Barton JK. Long-range oxidative damage to cytosines in duplex DNA. Proc. Natl. Acad. Sci. U.S.A. 2004; 101:17914-17919.
40. O’Neill MA, Barton JK. DNA charge transport: conformationally gated hopping through stacked domains. J. Am. Chem. Soc. 2004; 126:11471-11483.
41. Kelley SO, Holmlin RE, Stemp EDA, Barton JK. Photoinduced electron transfer in ethidium-modified DNA duplexes: dependence on distance and base stacking. J. Am. Chem. Soc. 1997; 119:9861-9870.
42. Bhattacharya PK, Barton JK. Influence of intervening mismatches on long-range guanine oxidation in DNA duplexes. J. Am. Chem. Soc. 2001; 123:8649-8656.
43. Hall DB, Barton JK. Sensitivity of DNA-mediated electron transfer to the intervening pi-stack: a probe for the integrity of the DNA base stack. J. Am. Chem. Soc. 1997; 119:5045-5046.
44. Wagenknecht HA, Rajski SR, Pascaly M, Stemp EDA, Barton JK. Direct observation of radical intermediates in protein-dependent DNA charge transport. J. Am. Chem. Soc. 2001; 123:4400-4407.
45. Nunez ME, Noyes KT, Barton JK. Oxidative charge transport through DNA in nucleosome core particles. Chem. Biol. 2002; 9:403-415.
46. Bjorklund CC, Davis WB. Attenuation of DNA charge transport by compaction into a nucleosome core particle. Nucleic Acids Res. 2006; 34:1836-1846.
47. Nunez ME, Holmquist GP, Barton JK. Evidence for DNA charge transport in the nucleus. Biochemistry 2001; 40:12465-12471.
48. Boal AK, Yavin E, Lukianova OA, O’Shea VL, David SS, Barton JK. DNA-bound redox activity of DNA repair glycosylases containing [4Fe-4S] clusters. Biochemistry 2005; 44:8397-8407.
49. Yavin E, Boal AK, Stemp EDA, Boon EM, Livingston AL, O’Shea VL, David SS, Barton JK. Protein-DNA charge transport: redox activation of a DNA repair protein by guanine radical. Proc. Natl. Acad. Sci. U.S.A. 2005; 102:3546-3551.
Boon EM, Barton JK. Charge transport in DNA. Curr. Opin. Struct. Biol. 2002; 12:320-329.
Erkkila KE, Odom DT, Barton JK. Recognition and reaction of metallointercalators with DNA. Chem. Rev. 1999; 99:2777-2795.
Hannon MJ. Supramolecular DNA recognition. Chem. Soc. Rev. 2006; 35:1-16.
Jackson BA, Barton JK. Probing nucleic acid structure with shape-selective rhodium and ruthenium complexes. In: Current Protocols in Nucleic Acid Chemistry. Beaucage SL, Bergstrom DE, Glick GD, Jones RA, eds. 2000. John Wiley & Sons, Inc., New York.
O’Neill MA, Barton JK. DNA-mediated charge transport chemistry and biology. Top. Curr. Chem. 2004; 236:67-115.
Wagenknecht HA. Charge Transfer in DNA: From Mechanism to Application. 2005. Wiley-VCH, New York.
Inorganic Chemistry in Biology
Oxidative DNA Damage, Chemistry of
Physico-Chemical Properties of Nucleic Acids
Nucleic Acid Recognition by Peptides and Drugs
DNA Damage and Carcinogenesis