DNA sequence-specific reading by echinomycin: role of hydrogen bonding and stacking interactions

Insight into the sequence-specific elements leading to increased DNA bending and ligase-mediated circularization propensity by antitumor trabectedin

DNA curvature is the result of a combination of both intrinsic features of the double helix and external distortions introduced by the environment and the binding of proteins or drugs. The propensity of certain double-stranded DNA (dsDNA) sequences to bend is essential in crucial biological processes, such as replication and transcription, in which proteins are known to either recognize noncanonical DNA conformations or promote their formation upon DNA binding. Trabectedin (Yondelis®) is a clinically used antitumor drug which, following covalent bond formation with the 2-amino group of guanine, induces DNA curvature and enhances the circularization ratio, upon DNA ligation, of several dsDNA constructs but not others. By means of unrestrained molecular dynamics simulations using explicitly solvated all-atom models, we rationalize these experimental findings in structural terms and shed light on the crucial, albeit possibly underappreciated, role played by T4 DNA ligase in stabilizing a bent DNA conformation prior to cyclization. Taken together, our results expand our current understanding on how DNA shape modification by trabectedin may affect both the sequence-specific recognition by transcription factors to promoter sites and RNA polymerase II binding.

Synthesis, G-quadruplex binding properties and cytotoxicity of naphthalimide–thiourea conjugates

The hydrogen bonding between a thiourea moiety and a G-quadruplex plays a crucial role in the sequence-specific DNA binding of naphthalimide–thiourea conjugates.

A historical account of Hoogsteen base-pairs in duplex DNA

In 1957, a unique pattern of hydrogen bonding between N3 and O4 on uracil and N7 and N6 on adenine was proposed to explain how poly(rU) strands can associate with poly(rA)-poly(rU) duplexes to form triplexes. Two years later, Karst Hoogsteen visualized such a noncanonical A-T base-pair through X-ray analysis of co-crystals containing 9-methyladenine and 1-methylthymine. Subsequent X-ray analyses of guanine and cytosine derivatives yielded the expected Watson-Crick base-pairing, but those of adenine and thymine (or uridine) did not yield Watson-Crick base-pairs, instead favoring "Hoogsteen" base-pairing. More than two decades ensued without experimental "proof" for A-T Watson-Crick base-pairs, while Hoogsteen base-pairs continued to surface in AT-rich sequences, closing base-pairs of apical loops, in structures of DNA bound to antibiotics and proteins, damaged and chemically modified DNA, and in polymerases that replicate DNA via Hoogsteen pairing. Recently, NMR studies have shown that base-pairs in duplex DNA exist as a dynamic equilibrium between Watson-Crick and Hoogsteen forms. There is now little doubt that Hoogsteen base-pairs exist in significant abundance in genomic DNA, where they can expand the structural and functional versatility of duplex DNA beyond that which can be achieved based only on Watson-Crick base-pairing. Here, we provide a historical account of the discovery and characterization of Hoogsteen base-pairs, hoping that this will inform future studies exploring the occurrence and functional importance of these alternative base-pairs.

DNA structure and function

The proposal of a double-helical structure for DNA over 60 years ago provided an eminently satisfying explanation for the heritability of genetic information. But why is DNA, and not RNA, now the dominant biological information store? We argue that, in addition to its coding function, the ability of DNA, unlike RNA, to adopt a B-DNA structure confers advantages both for information accessibility and for packaging. The information encoded by DNA is both digital - the precise base specifying, for example, amino acid sequences - and analogue. The latter determines the sequence-dependent physicochemical properties of DNA, for example, its stiffness and susceptibility to strand separation. Most importantly, DNA chirality enables the formation of supercoiling under torsional stress. We review recent evidence suggesting that DNA supercoiling, particularly that generated by DNA translocases, is a major driver of gene regulation and patterns of chromosomal gene organization, and in its guise as a promoter of DNA packaging enables DNA to act as an energy store to facilitate the passage of translocating enzymes such as RNA polymerase.

Antitumor Activity, X-ray Crystal Structure, and DNA Binding Properties of Thiocoraline A, a Natural Bisintercalating Thiodepsipeptide

The marine natural product thiocoraline A displayed approximately equal cytotoxic activity at nanomolar concentrations in a panel of 12 human cancer cell lines. X-ray diffraction analyses of orthorhombic crystals of this DNA-binding drug revealed arrays of docked pairs of staple-shaped molecules in which one pendent hydroxyquinoline chromophore from each cysteine-rich molecule appears intercalated between the two chromophores of a facing molecule. This arrangement is in contrast to the proposed mode of binding to DNA that shows the two drug chromophores clamping two stacked base pairs, in agreement with the nearest-neighbor exclusion principle. Proof of DNA sequence recognition was obtained from both classical DNase I footprinting experiments and determination of the melting temperatures of several custom-designed fluorescently labeled oligonucleotides. A rationale for the DNA-binding behavior was gained when models of thiocoraline clamping a central step embedded in several octanucleotides were built and studied by means of unrestrained molecular dynamics simulations in aqueous solution.

Role of stacking interactions in the binding sequence preferences of DNA bis-intercalators: insight from thermodynamic integration free energy simulations

The major structural determinant of the preference to bind to CpG binding sites on DNA exhibited by the natural quinoxaline bis-intercalators echinomycin and triostin A, or the quinoline echinomycin derivative, 2QN, is the 2-amino group of guanine (G). However, relocation of this group by means of introduction into the DNA molecule of the 2-aminoadenine (=2,6-diaminopurine, D) base in place of adenine (A) has been shown to lead to a drastic redistribution of binding sites, together with ultratight binding of 2QN to the sequence DTDT. Also, the demethylated triostin analogs, TANDEM and CysMeTANDEM, which bind with high affinity to TpA steps in natural DNA, bind much less tightly to CpI steps, despite the fact that both adenosine and the hypoxanthine-containing nucleoside, inosine (I), provide the same hydrogen bonding possibilities in the minor groove. To study both the increased binding affinity of 2QN for DTDT relative to GCGC sites and the remarkable loss of binding energy between CysMeTANDEM and ICIC compared with ATAT, a series of thermodynamic integration free energy simulations involving conversions between DNA base pairs have been performed. Our results demonstrate that the electrostatic component of the stacking interactions between the heteroaromatic rings of these compounds and the bases that make up the intercalation sites plays a very important role in the modulation of their binding affinities.

Interaction of echinomycin with guanine: electrochemistry and spectroscopy studies

The interaction of antitumor antibiotic, echinomycin (Echi) with guanine (Gua) was thoroughly investigated by adsorptive transfer stripping cyclic voltammetry, ultraviolet and visible adsorption spectra (UV/Vis) and Fourier-transform infrared spectroscopy (FTIR). Electrochemistry provided a simple tool for verifying the occurrence of interaction between Echi and Gua. Echi could be accumulated from the solution and give well-defined electrochemical signals in 0.1 M phosphate buffer solution (pH 7.0) only when Gua was present on the surface of the electrochemically pretreated glass carbon electrode (GCE), suggesting a strong binding of Echi to Gua. All the acquired spectral data showed that a new adduct between Echi and Gua was formed, and two pairs of adjacent intermolecular hydrogen bonds between the Ala backbone atoms in Echi and Gua (Ala-NH to Gua-N3 and Gua-NH2 to Ala-CO) played a dominating role in the interaction. Electrochemistry coupled with spectroscopy techniques could provide a relatively easy way to obtain useful insights into the molecular mechanism of drug-DNA interactions, which should be important in the development of new anticancer drugs with specific base recognition.

Energetics of echinomycin binding to DNA

Differential scanning calorimetry and UV thermal denaturation have been used to determine a complete thermodynamic profile for the bis-intercalative interaction of the peptide antibiotic echinomycin with DNA. The new calorimetric data are consistent with all previously published binding data, and afford the most rigorous and direct determination of the binding enthalpy possible. For the association of echinomycin with DNA, we found DeltaG degrees = -7.6 kcal mol(-1), DeltaH = +3.8 kcal mol(-1) and DeltaS = +38.9 cal mol(-1) K(-1) at 20 degrees C. The binding reaction is clearly entropically driven, a hallmark of a process that is predominantly stabilized by hydrophobic interactions, though a deeper analysis of the free energy contributions suggests that direct molecular recognition between echinomycin and DNA, mediated by hydrogen bonding and van der Waals contacts, also plays an important role in stabilizing the complex.

Chromophore-modified bisnaphthalimides: DNA recognition, topoisomerase inhibition, and cytotoxic properties of two mono- and bisfuronaphthalimides

Bisnaphthalimides represent a promising group of DNA-targeted anticancer agents. In this series, the lead compounds elinafide and bisnafide have reached clinical trials, and the search for more potent analogues remains a priority. In the course of a medicinal chemistry program aimed at discovering novel antitumor drugs based on the naphthalimide skeleton, different dimeric molecules containing two tetracyclic neutral DNA intercalating chromophores were synthesized. The naphthalimide unit has been fused to a benzene ring (azonafide derivatives), an imidazole, a pyrazine, or, as reported here, a furan ring which increases the planar surface of the chromophore and enhances its stacking properties. We report a detailed investigation of the DNA binding capacity of the dimeric molecule MCI3335 composed of two furonaphthalimide units connected by a 12 A long amino alkyl linker [(CH(2))(2)-NH-(CH(2))(3)-NH-(CH(2))(2)] identical to that of elinafide. Qualitative and quantitative binding studies, in particular using surface plasmon resonance, establish that the dimer binds considerably more tightly to DNA (up to 1000 times) than the corresponding monomer and exhibits a higher sequence selectivity for GC-rich sequences. DNase I footprinting experiments attest that the dimer, and to a lesser extent the monomer, preferentially intercalate at GC sites. The strong binding interaction between the drugs and DNA perturbs the relaxation of supercoiled DNA by topoisomerases, but the test compounds do not promote DNA cleavage by topoisomerase I or II. Despite the lack of poisoning effect toward topoisomerase II, MCI3335 displays a very high cytotoxicity toward CEM human leukemia cells, with an IC(50) in the low nanomolar range, approximately 4 times inferior to that of the reference drug elinafide. Confocal microscopy observations indicate that the monomer shows a stronger tendency to accumulate in the cell nuclei than the dimer. The extremely high cytotoxic potential of MCI3335 is attributed to its enhanced capacity to bind to DNA and to inhibit DNA synthesis, as evidenced by flow cytometry experiments using the BrdU assay. The results provide novel mechanistic information that furthers the understanding of the structure-activity relationships in the bisnaphthalimide series and identify MCI3335 as a novel lead compound for further preclinical investigations.

Critical effect of the N2 amino group on structure, dynamics, and elasticity of DNA polypurine tracts

Unrestrained 5-20-ns explicit-solvent molecular dynamics simulations using the Cornell et al. force field have been carried out for d[GCG(N)11GCG]2 (N, purine base) considering guanine*cytosine (G*C), adenine*thymine (A*T), inosine*5-methyl-cytosine (I*mC), and 2-amino-adenine*thymine (D*T) basepairs. The simulations unambiguously show that the structure and elasticity of N-tracts is primarily determined by the presence of the amino group in the minor groove. Simulated A-, I-, and AI-tracts show almost identical structures, with high propeller twist and minor groove narrowing. G- and D-tracts have small propeller twisting and are partly shifted toward the A-form. The elastic properties also differ between the two groups. The sequence-dependent electrostatic component of base stacking seems to play a minor role. Our conclusions are entirely consistent with available experimental data. Nevertheless, the propeller twist and helical twist in the simulated A-tract appear to be underestimated compared to crystallographic studies. To obtain further insight into the possible force field deficiencies, additional multiple simulations have been made for d(A)10, systematically comparing four major force fields currently used in DNA simulations and utilizing B and A-DNA forms as the starting structure. This comparison shows that the conclusions of the present work are not influenced by the force field choice.

DNA recognition and nucleosome organization

The affinity of a DNA sequence for the histone octamer in a core nucleosome depends on the intrinsic flexibility of the DNA. This parameter can be affected both by the sequence-dependent conformational preferences of individual base steps and by the nature and location of the exocyclic groups of the DNA bases. By adopting highly preferred conformations particular types of base step can influence the rotational positioning of the DNA on the surface of the histone octamer. The asymmetry of the next higher order of chromatin structure is determined in part by the asymmetric binding of the globular domain of histone H5 to the core nucleosome.

The influence of the exocyclic pyrimidine 5-methyl group on DNAse I cleavage and sequence recognition by drugs

Incorporation of modified nucleotides into DNA, using the PCR, has allowed us to probe the influence that the exocyclic 5-methyl group of pyrimidines has on DNAse I cleavage and sequence recognition by drugs. The results show that removal of the methyl group from the major groove, made possible by substituting uridine for thymidine, allows DNAse I to cleave more readily at AT-rich regions compared to normal DNA. By contrast, addition of an extra methyl group, contrived by substituting 5-methylcytidine for normal cytidine, allows DNAse I to cleave more readily at GC-rich regions compared to normal DNA. In the cutting pattern of DNA containing both uridine and 5-methyl cytosine, we find the cleavage characteristics of both the single-substituted DNA species combined. Thus, the presence or absence of the exocyclic 5-methyl group in the major groove has a strong influence on the relative intensity of cleavage of phosphodiester bonds by DNAse I. These nucleotide substitutions can also influence the sequence-selective binding of drugs to DNA. Whereas removal of the methyl group (replacement of T with U) generally has little effect on sequence recognition by a variety of drugs, addition of a methyl group (replacement of C with M) generates new binding sites for some intercalators, namely daunomycin, DACA and SN16713.

The exocyclic groups of DNA modulate the affinity and positioning of the histone octamer

To investigate the nature of the chemical determinants in DNA required for nonspecific binding and bending by proteins we have created a novel DNA in which inosine-5-methylcytosine and 2, 6-diaminopurine-uracil base pairs are substituted for normal base pairs in a defined DNA sequence. This procedure completely switches the patterns of the base pair H bonding and attachment of exocyclic groups. We show that this DNA binds a histone octamer more tightly than normal DNA but, surprisingly, does not alter the orientation of the sequence on the surface of the protein. However, in general, the addition or removal of DNA exocyclic groups reduces or increases, respectively, the affinity for the histone octamer. The average incremental change in binding energy for a single exocyclic group is approximately 40 J/mol. The orientation of the DNA in core nucleosomes also is sensitive to the number and nature of the exocyclic groups present. Notably, substitution with the naturally occurring cytosine analogue, 5-methylcytosine, shifts the preferred rotational position by 3 bp, whereas incorporating 2,6-diaminopurine shifts it 2 bp in the opposite direction. These manipulations potentially would alter the accessibility of a protein recognition sequence on the surface of the histone octamer. We propose that exocyclic groups impose steric constraints on protein-induced DNA wrapping and are also important in determining the orientation of DNA on a protein surface. In addition, we consider the implications of the selection of A-T and G-C base pairs in natural DNA.

Stacking interactions and intercalative DNA binding

The DNA-binding properties of many ligands can be rationalized on the basis of their structural and electronic complementarity with the functional groups present in the minor and major grooves of particular DNA sequences. Specific hydrogen bonding patterns are particularly useful for the purpose of sequence recognition. Less obvious, however, is the influence of base composition on the conformational preferences of individual base steps and on the binding of intercalating moieties which become sandwiched between contiguous base pairs. Improved knowledge of stacking interactions may lead to a better understanding of the architecture and inherent flexibility of particular DNA sequences and may provide insight into the principles that dictate the structural changes and specificity patterns observed in the binding of some intercalating ligands to DNA.

Binding of daunomycin to diaminopurine- and/or inosine-substituted DNA

The binding of the anticancer drug daunomycin to double-helical DNA has been investigated by DNase I footprinting and fluorescence titration, using a series of polymerase chain reaction (PCR) synthesized DNA fragments that contained systematic base substitutions to alter the disposition of functional groups within the minor groove. The 160 bp tyrT DNA fragment constituted the starting material. Fragments in which (i) inosine was substituted for guanosine, (ii) diaminopurine was substituted for adenine, and (iii) both inosine and diaminopurine were substituted for guanosine and adenine, respectively, were studied. These fragments permit the role of the 2-amino group in the minor groove to be systematically explored. The results of DNase I footprinting experiments confirmed that daunomycin binds preferentially to 5'(A/T)GC and 5'(A/T)CG triplets in the normal fragment. Substitution of inosine for guanosine, with the concomitant loss of the N-2 in the minor groove, weakened binding affinity but did not dramatically alter the sequence preference associated with daunomycin binding. Complete reversal of the location of the N-2 group by the double substitution, however, completely altered the sequence preference of daunomycin and shifted its binding from the canonical triplets to ones with a 5'IDD motif. These results have critically tested and confirmed the proposed key roles of the daunosamine moiety and the 9-OH group of daunomycin in dictating binding to preferred sites. In a parallel study, both macroscopic and microscopic binding to the normal tyrT fragment were investigated, experiments made possible by using PCR to prepare large quantities of the long, defined DNA sequence. The results of these experiments underscored the complexity of the interaction of the drug with the DNA lattice and revealed unequivocal heterogeneity in its affinity for different binding sites. A class of high-affinity sites, most probably corresponding to the 5'(A/T)GC and 5'(A/T)CG triplets, was identified and characterized in macroscopic binding isotherms.

Footprinting of echinomycin and actinomycin D on DNA molecules asymmetrically substituted with inosine and/or 2,6-diaminopurine

In order to clarify the role of the purine 2-amino group in the recognition of DNA by small molecules we have examined the binding of actinomycin D and echinomycin to artificial DNA molecules asymmetrically substituted with inosine and/or 2,6-diaminopurine (DAP) in one of the complementary strands. These DNAs, prepared by a method based upon PCR, present various potential sites for antibiotic binding, including several containing only a single purine 2-amino group in different configurations. The results show unambiguously that the presence of two 2-amino groups is mandatory for binding of actinomycin D to double-stranded DNA. In the case of echinomycin only one purine 2-amino group is required for remarkably strong binding to the asymmetric TpDAP.TpA dinucleotide step, but the CpDAP.TpI step (which also contains only a single purine-2 amino group) does not afford a binding site. Evidently, removing a 2-amino group (G-->I substitution) is dominant over adding one (A-->DAP substitution). No sequences containing just a single guanine residue are acceptable. The possibility is raised that replacing guanosine with inosine may do more than remove a group endowed with hydrogen bonding capability and interfere with ligand binding in other ways. The new methodology developed to construct asymmetrically substituted DNA substrates for this work provides a novel strategy that should be generally applicable for studying ligand-DNA interactions, beyond the specific interest in drug binding to DNA, and may help to elucidate how proteins and oligonucleotides recognize their target sites.

Structure-affinity relationships for the binding of actinomycin D to DNA

Molecular models of the complexes between actinomycin D and 14 different DNA hexamers were built based on the X-ray crystal structure of the actinomycin-d(GAAGCTTC)2 complex. The DNA sequences included the canonical GpC binding step flanked by different base pairs, nonclassical binding sites such as GpG and GpT, and sites containing 2,6-diamino-purine. A good correlation was found between the intermolecular interaction energies calculated for the refined complexes and the relative preferences of actinomycin binding to standard and modified DNA. A detailed energy decomposition into van der Waals and electrostatic components for the interactions between the DNA base pairs and either the chromophore or the peptidic part of the antibiotic was performed for each complex. The resulting energy matrix was then subjected to principal component analysis, which showed that actinomycin D discriminates among different DNA sequences by an interplay of hydrogen bonding and stacking interactions. The structure-affinity relationships for this important antitumor drug are thus rationalized and may be used to advantage in design of novel sequence-specific DNA-binding agents.

Molecular dynamics simulations of the bis-intercalated complexes of ditercalinium and Flexi-Di with the hexanucleotide d(GCGCGC)2: theoretical analysis of the interaction and rationale for the sequence binding specificity

The X-ray crystal structures of the complexes of ditercalinium and Flexi-Di with d(CGCG)2 have been studied by computational chemistry methods in an attempt to rationalize their distinct structural features. In addition, the complexes of these two bisintercalating drugs with d(GCGCGC)2 have been modeled and subjected to 0.5 ns of molecular dynamics simulations in explicit solvent with the aim of evaluating the relative importance of hydrogen bonding and stacking interactions in the sequence binding specificity of these compounds. According to our calculations, the electrostatic term is attractive for the stacking interactions between the pyridocarbazole chromophores of these drugs and the base pairs that make up the sandwiched GpC step. On the contrary, this energy term is repulsive for the base pairs that make up the boundaries of the bisintercalation site. This differential electrostatic binding energy component, which is shown to have a strong orientational dependence, could lie at the origin of the observed binding preferences of these drugs. In addition, both the Lennard-Jones and the electrostatic energy terms contribute to stabilizing the underwound central GpC step. The attractive electrostatic interactions between the linkers and the major groove are in concert with the stacking specificities for the sandwiched GpC step, which is thus very effectively stapled by the drugs. The hydrogen-bonding potential of the linkers, however, appears to be reduced in an aqueous medium due to competing interactions with water. Binding of either ditercalinium or Flexi-Di to d(GCGCGC)2 appears to favor the A-type conformation that this DNA molecule most likely adopts in the free state. The possible relevance of these findings to the process of bis-intercalation and to the pharmacological action of these compounds is discussed.

Dissociation kinetics of echinomycin from CpG binding sites in different sequence environments

We have examined the kinetics of dissociation of echinomycin from CpG sites in several DNA fragments containing synthetic DNA inserts by a variation of the footprinting technique. Complexes of the ligand with radiolabeled DNA fragments were dissociated by adding an excess of unlabeled calf thymus DNA. Samples were removed from this mixture at subsequent time intervals and subjected to DNase I footprinting. The rate of disappearance of the footprints varied considerably between the various CpG sites. At 20 degrees C, echinomycin dissociates more slowly from CpG sites flanked by (AT)n (t1/2 approximately 40 min) and (CA)n.(TG)n (t1/2 approximately 11 min) than by An.Tn (t1/2 < 3 min). In each sequence context the dissociation from ACGT is slower than that from TCGA. (TAA)4CG(TTA)4 also represents a very good binding site (t1/2 approximately 35 min), which is less sensitive to changes in temperature than most other sites. Within sequences (AT)10(G/C)4(AT)10, the dissociation from CGGC is slower than that from CCCG or CCGC.

Site selective bis-intercalation of a homodimeric thiazole orange dye in DNA oligonucleotides

We have used one and two dimensional 1H NMR spectroscopy to characterize the binding of a homodimeric thiazole orange dye, 1,1'-(4,4,8,8-tetramethyl-4,8-diaza-undecamethylene)-bis-4- (3-methyl-2,3-dihydro-(benzo-1,3-thiazole)-2-methylidene)-quinolin ium tetraiodide (TOTO), to various double stranded DNA oligonucleotides. TOTO binds strongly to all the oligonucleotides used, but usually more than one complex is observed and exchange between different binding sites broadens the lines in the NMR spectra. Complete precipitation occurs when TOTO is bound to small oligonucleotides. Binding to larger oligonucleotides occurs by bis-intercalation. The 1:1 complex of TOTO with the oligonucleotide d(CCGACTGATGC):d (GCATCAGTCGG) gave only one complex that was shown to be a bis-intercalation in the CTGA:TCAG binding site. The binding to this site was also characterized by studying the TOTO complex with the d(CCGCTGAGC):d(GCTCAGCGG) oligonucleotide. NOE connectivities and molecular modelling were used to characterize the complex. The 1:1 complex of TOTO with the oligonucleotide d(CCGCTAGCG):d(CGCTAGCGG) containing a CTAG:CTAG binding site was similarly characterized by NMR. It was concluded that the binding of TOTO to larger oligonucleotides is site selective with CTAG:CTAG as the preferred binding site.

Binding of echinomycin to d(GCGC)2 and d(CCGG)2: distinct stacking interactions dictate the sequence-dependent formation of Hoogsteen base pairs

Molecular dynamics simulations have been used to explore the behavior of the complexes of echinomycin with the DNA tetramers d(GCGC)2 and d(CCGG)2 in which the terminal bases have been paired according to either a Hoogsteen or a Watson-Crick hydrogen bonding scheme. The energy of the four resulting complexes has been monitored along the dynamics trajectories and the interaction energy between echinomycin and DNA has been decomposed into contributions arising from the planar aromatic systems and the depsipeptide part of the antibiotic. Our calculations predict a large increase in overall stabilization upon protonation of the terminal cytosines and subsequent Hoogsteen pair formation in the complex of echinomycin with d(GCGC)2 but not with d(CCGG)2, in agreement with the experimental evidence [Gao and Patel, Quart. Rev. Biophys. 22, 93-138 (1989)]. The conformational preferences appear to arise mainly from differential stacking interactions in which the electrostatic component is shown to play a dominant role. Differences in hydrogen bonding patterns are also found among the complexes and these are compared in relation to available crystal structures. The binding of echinomycin to DNA appears as a complex process involving many interrelated variables.