Sequence-dependent dynamics of duplex DNA: the applicability of a dinucleotide model.

The short-time (submicrosecond) bending dynamics of duplex DNA were measured to determine the effect of sequence on dynamics. All measurements were obtained from a single site on duplex DNA, using a single, site-specific modified base containing a rigidly tethered, electron paramagnetic resonance active spin probe. The observed dynamics are interpreted in terms of single-step sequence-dependent bending force constants, determined from the mean squared amplitude of bending relative to the end-to-end vector using the modified weakly bending rod model. The bending dynamics at a single site are a function of the sequence of the nucleotides constituting the duplex DNA. We developed and examined several dinucleotide-based models for flexibility. The models indicate that the dominant feature of the dynamics is best explained in terms of purine- and pyrimidine-type steps, although distinction is made among all 10 unique steps: It was found that purine-purine steps (which are the same as pyrimidine-pyrimidine steps) were near average in flexibility, but the pyrimidine-purine steps (5' to 3') were nearly twice as flexible, whereas purine-pyrimidine steps were more than half as flexible as average DNA. Therefore, the range of stepwise flexibility is approximately fourfold and is characterized by both the type of base pair step (pyrimidine/purine combination) and the identity of the bases within the pair (G, A, T, or C). All of the four models considered here underscore the complexity of the dependence of dynamics on DNA sequence with certain sequences not satisfactorily explainable in terms of any dinucleotide model. These findings provide a quantitative basis for interpreting the dynamics and kinetics of DNA-sequence-dependent biological processes, including protein recognition and chromatin packaging.

Monoalkylation of DNA by reductively activated FR66979.

The antitumor antibiotic FR66979 has previously been shown to form interstrand cross-links in duplex DNA at the sequence [5'-d(CG)]2, linking the exocyclic amino groups (N2) of deoxyguanosine (dG) residues. During the reaction of reductively activated FR66979 with DNA. products are formed which have electrophoretic mobility in denaturing polyacrylamide gels which is intermediate between that of unmodified and interstrand cross-linked DNA. We show here that these products are monoadducts between FR66979 and DNA and provide strong evidence for the site of alkylation being N2 of dG. Moreover, the sequence selectivity of monoalkylation reactions between FR66979 and DNA containing either 5'-d(CG).5'-d(CI) or [5'-d(CG)]2 was observed to be ca. 5-fold less than for the related antitumor antibiotic mitomycin C (MC). The mechanistic implications of this result are discussed. Furthermore, it was demonstrated that contrary to a previous report, FR66979 requires DNA to be in duplex form for efficient monoadduct formation.

Chemical synthesis of cross-link lesions found in nitrous acid treated DNA: a general method for the preparation of N2-substituted 2'-deoxyguanosines.

Treatment of DNA with nitrous acid results in the formation of DNA-DNA cross-links. Two cross-link lesions have previously been isolated and their structures assigned based on spectroscopic data. The major lesion has been proposed to consist of two deoxyguanosine (dG) nucleosides sharing a common N2 atom (1), while the structure of the minor lesion has been proposed to consist of a common nitrogen atom linking C2 of a dG nucleoside to C6 of deoxyadenosine (2). The chemical synthesis of 1 and 2, utilizing a palladium-catalyzed coupling, is described herein. It is demonstrated that the spectroscopic properties of synthetic 1 are identical to that of lesion 1 obtained from nitrous acid cross-linked DNA, thus providing a proof of its structure. Comparison of the limited spectroscopic data available for lesion 2 originating from nitrous acid cross-linked DNA to synthetic 2 supports its structural assignment. The synthetic approach used for synthesis of 1 and 2 is shown to be a general method for the preparation of a variety of N2-substituted dG nucleosides in good yields.

Sequence-dependent dynamics in duplex DNA.

The submicrosecond bending dynamics of duplex DNA were measured at a single site, using a site-specific electron paramagnetic resonance active spin probe. The observed dynamics are interpreted in terms of the mean squared amplitude of bending relative to the end-to-end vector defined by the weakly bending rod model. The bending dynamics monitored at the single site varied when the length and position of a repeated AT sequence, distant from the spin probe, were changed. As the distance between the probe and the AT sequence was increased, the mean squared amplitude of bending seen by the probe due to that sequence decreased. A model for the sequence-dependent internal flexural motion of duplex DNA, which casts the mean squared bending amplitudes in terms of sequence-dependent bending parameters, has been developed. The best fit of the data to the model occurs when the (AT)(n) basepairs are assumed to be 20% more flexible than the average of the basepairs within the control sequence. These findings provide a quantitative basis for interpreting the kinetics of biological processes that depend on duplex DNA flexibility, such as protein recognition and chromatin packaging.

Flexibility of duplex DNA on the submicrosecond timescale.

Using a site-specific, Electron Paramagnetic Resonance (EPR)-active spin probe that is more rigidly locked to the DNA than any previously reported, the internal dynamics of duplex DNAs in solution were studied. EPR spectra of linear duplex DNAs containing 14-100 base pairs were acquired and simulated by the stochastic Liouville equation for anisotropic rotational diffusion using the diffusion tensor for a right circular cylinder. Internal motions have previously been assumed to be on a rapid enough time scale that they caused an averaging of the spin interactions. This assumption, however, was found to be inconsistent with the experimental data. The weakly bending rod model is modified to take into account the finite relaxation times of the internal modes and applied to analyze the EPR spectra. With this modification, the dependence of the oscillation amplitude of the probe on position along the DNA was in good agreement with the predictions of the weakly bending rod theory. From the length and position dependence of the internal flexibility of the DNA, a submicrosecond dynamic bending persistence length of around 1500 to 1700 A was found. Schellman and Harvey (Biophys. Chem. 55:95-114, 1995) have estimated that, out of the total persistence length of duplex DNA, believed to be about 500 A, approximately 1500 A is accounted for by static bends and 750 A by fluctuating bends. A measured dynamic persistence length of around 1500 A leads to the suggestion that there are additional conformations of the DNA that relax on a longer time scale than that accessible by linear CW-EPR. These measurements are the first direct determination of the dynamic flexibility of duplex DNA in 0.1 M salt.

Direct demonstration in synthetic oligonucleotides that N,N'-bis(2-chloroethyl)-nitrosourea cross links N1 of deoxyguanosine to N3 of deoxycytidine on opposite strands of duplex DNA.

The sequence specificity and covalent structure of the lesion caused by the DNA interstrand cross-linking reaction of N,N'-bis(2-chloroethyl)-nitrosourea (BCNU) were investigated using synthetic oligonucleotides. The efficiency of interstrand cross-linking was found to parallel the efficiency of monoadduct formation, preferring deoxyguanosine-deoxycytidine-rich duplexes and, particularly, runs of deoxyguanosine. No explicit sequence specificity was observed. Enzymatic digestion of purified, interstrand cross-linked DNA returned primarily the unmodified deoxynucleosides, along with 1-[N3-deoxycytidyl]-2-[N1-deoxyguanosyl]ethane. This substance was characterized by comparison of its mass spectrum, high-pressure liquid chromatography retention time, and UV spectrum to an authentic standard prepared by chemical synthesis. These studies provide the first direct evidence that BCNU has no strong sequence preference for interstrand cross-linking and that substance 4, which has been previously isolated from BCNU-treated DNA, derives from alkylation on opposite strands of DNA. The lack of sequence preference and lesion structure together suggest that one source of BCNU interstrand cross-links is linkage of deoxyguanosine and deoxycytidine partners from a single bp.

Quantification of formaldehyde-mediated covalent adducts of adriamycin with DNA.

Duplex DNA incubated with adriamycin, dithiothreitol (DTT), and Fe3+ under aerobic, aqueous conditions yields double-stranded (DS) DNA bands by denaturing polyacrylamide gel electrophoresis (DPAGE) analysis, characteristic of DNAs which are interstrand cross-linked. Another laboratory has provided evidence that formaldehyde produced under these conditions promotes the covalent linkage of adriamycin to one strand of DNA and suggested that this complex results in the anomalous DPAGE behavior. We provide herein strong support for this interpretation. We show: (a) that mixtures of DNA and adriamycin incubated with DTT/Fe3+, H2O2, or formaldehyde all show DS DNA bands on DPAGE, (b) that the DS DNA bands and the formaldehyde-mediated lesion (detected by an indirect, GC-MS analysis) form with similar time courses, and in similar amounts, and (c) that the DNA in the DS DNA bands contains approximately one such lesion per DNA, whereas the single-stranded DNA is devoid of it. These results further support the interpretation that adriamycin does not create interstrand cross-links in DNA, and that the DS DNA observed in DPAGE experiments derives from the formaldehyde-mediated monoadduct.

Effect of nucleosome structure on DNA interstrand cross-linking reactions.

Antitumor agents of the nitrogen mustard family and mitomycin C form interstrand cross-links in duplex DNA. To provide information about the cellular mechanism by which these compounds exert their cytotoxic effects, we examined cross-linking of a nucleosomal core particle formed on a fragment of the 5S RNA gene of Xenopus borealis. For the mustards mechlorethamine, chlorambucil, and melphalan, both sites of monoalkylation and interstrand cross-linking were similar in nucleosomal and free DNA. Some small (two- to three- fold) differences in intensity of cross-linking at some sites were apparent. However, these differences did not appear to correlate with rotational or translational positioning. For mitomycin C, cross-linking was inhibited five- to ten-fold at the nucleosomal dyad and showed attenuation of inhibition toward the ends. Furthermore, rotational positioning also appeared to be a factor, with sites facing inward in the nucleosome less accessible for mitomycin cross-linking. None of these agents demonstrated the 10-base pair periodicity exhibited by hydroxyl radical cleavage of nucleosomal DNA.

An NMR study of [d(CGCGAATTCGCG)]2 containing an interstrand cross-link derived from a distamycin-pyrrole conjugate.

Minor groove binding compounds related to distamycin A bind DNA with high sequence selectivity, recognizing sites which contain various combinations of A.T and G.C base pairs. These molecules have the potential to deliver cross-linking agents to the minor groove of a target DNA sequence. We have studied the covalent DNA-DNA cross-linked complex of 2,3- bis(hydroxymethyl)pyrrole-distamycin and [d(CGCGAATTCGCG)]2. The alkylating pyrrole design is based on the pharmacophore of mitomycin C and is similar in substructure to another important class of natural products, the oxidatively activated pyrrolizidine alkaloids. Ligand-DNA NOEs confirm that the tri(pyrrole-carboxamide) unit of the ligand is bound in the minor groove of the central A+T tract. Unexpectedly, it is shifted by 1 bp with respect to the distamycin A binding site on this DNA sequence. The cross-link bridges the 2-amino position of two guanine residues, G4 and G22. The C3.G22 and G4.C21 base pairs exhibit Watson-Crick base pairing, with some local distortion, as evidenced by unusual intensities observed for DNA-DNA NOE cross-peaks. The model is compared with a related structure of a cross-linked mitomycin C:DNA complex.

Bending of DNA by the mitomycin C-induced, GpG intrastrand cross-link.

Mitomycin C (MC) forms interstrand and intrastrand cross-link adducts and monoalkylation products (monoadducts) with DNA. Each of the three types of adducts was incorporated site-specifically into both a 15-mer and a 21-mer oligodeoxyribonucleotide duplex. The adduct-containing duplexes were 32P-phosphorylated and ligated to form multimers, which were then analyzed for anomalous electrophoretic mobility by nondenaturing polyacrylamide gel electrophoresis, using the method of Koo and Crothers [(1988) Proc. Natl. Acad. Sci. U.S.A. 85, 1763-1767] in order to detect DNA curvature caused by the adducts. The intrastrand cross-link adduct was found to induce a 14.6 +/- 2.0 degrees DNA bend per lesion (minimum value) while no DNA bending was detected for either the interstrand cross-link or the monoadduct. Molecular mechanics modeling indicated that the possible origin of the bend lies in a considerable deviation from parallel of the normals to the best planes of the intrastrand cross-linked guanines, due to a shorter than normal distance between their N2 atoms forced upon them by the cross-link. The observed bending by the MC intrastrand lesion may be the cause of the increased flexibility of MC-modified DNA, localized to distinct regions, as observed in earlier work by hydrodynamic methods and electron microscopy. The MC adduct-caused DNA bend may serve as a recognition site for certain DNA-binding proteins.

Solution structure of a cisplatin-induced DNA interstrand cross-link.

The widely used antitumor drug cis-diamminedichloroplatinum(II) (cisplatin or cis-DDP) reacts with DNA, cross-linking two purine residues through the N7 atoms, which reside in the major groove in B-form DNA. The solution structure of the short duplex [d(CAT-AGCTATG)]2 cross-linked at the GC:GC site was determined by nuclear magnetic resonance (NMR). The deoxyguanosine-bridging cis-diammineplatinum(II) lies in the minor groove, and the complementary deoxycytidines are extrahelical. The double helix is locally reversed to a left-handed form, and the helix is unwound and bent toward the minor groove. These findings were independently confirmed by results from a phase-sensitive gel electrophoresis bending assay. The NMR structure differs markedly from previously proposed models but accounts for the chemical reactivity, the unwinding, and the bending of cis-DDP interstrand cross-linked DNA and may be important in the formation and repair of these cross-links in chromatin.

DNA-DNA interstrand cross-linking by cis-diamminedichloroplatinum(II): N7(dG)-to-N7(dG) cross-linking at 5'-d(GC) in synthetic oligonucleotides.

The nucleotide sequence specificity of the DNA-DNA interstrand cross-linking reaction of cis-diamminedichloroplatinum(II) (cis-DDP) was studied in synthetic oligonucleotides. Of six self-complementary DNAs tested, only those containing the central sequence 5'-d(GC) formed appreciable interstrand cross-linked product, as assayed by denaturing polyacrylamide gel electrophoresis (DPAGE). The nucleotide connectivity of the interstrand cross-link was defined by sequence random oxidative fragmentation followed by DPAGE, revealing that dG residues on opposite strands at the sequence 5'-d(GC) were connected to one another. The covalent structure of the cross-link was established following hydrolysis of the phosphodiester backbone of a structurally homogeneous sample of a cis-DDP interstrand cross-linked DNA tetradecamer. HPLC analysis of the hydrolysate returned all of the deoxynucleoside residues from the starting DNA except for two deoxyguanosine residues. Also returned was diammine-bis-[N7-(2'- deoxyguanosyl)]platinum(II)2+, identified by a combination of spectroscopic methods, and comparison to a synthetic authentic sample. This study directly establishes that cis-DDP forms interstrand cross-links at the duplex sequence 5'-d(GC), linking deoxyguanosine residues on opposite strands at N7 through a bridging platinum atom. Computer simulation of the interstrand cross-linked product using molecular mechanics energy minimization and molecular dynamics revealed significant structural reorganization at the site of the cross-link including a ca 40 degrees angle between the platinated guaninyl residues, which propagated to adjacent residues by base stacking to yield duplexes bent by some 30 degrees toward the major groove.

Monitoring DNA dynamics using spin-labels with different independent mobilities.

The electron paramagnetic resonance (EPR) spectra of spin-labeled DNA duplexes, both bound to DEAE-Sephadex and free in solution, have been analyzed. The nitroxide spin-labels are covalently linked to a deoxyuridine residue using either a monoacetylene or diacetylene tether. This difference in tether length produces a dramatic difference in the independent mobility of the nitroxide relative to the DNA. In the case of the monoacetylene tether, the motion of the nitroxide has previously been shown to be tightly coupled to that of the DNA duplex. With the diacetylene tether, there is considerable independent motion of the probe. The diacetylene tether is intended to minimize the possibility of the nitroxide producing a perturbation of the dynamics of DNA. It is demonstrated here that, when coupled via the diacetylene tether, the nitroxide undergoes a rapid uniaxial rotation about the tether. A detailed analysis of the EPR spectrum of duplex DNA in solution, spin-labeled using the diacetylene tether, demonstrates that the motion of the nitroxide can be modeled in terms of this independent uniaxial rotation together with motion of the DNA which is consistent with the global tumbling of the duplex. As was previously found using the monoacetylene tether, there is no evidence of rapid, large-amplitude motions of the base pair in the EPR spectrum of a nitroxide coupled to duplex DNA via the diacetylene tether. This result confirms the small amplitudes of internal motion, local and collective, previously observed in duplex DNA with the monoacetylene-tethered nitroxide.

A mechlorethamine-induced DNA interstrand cross-link bends duplex DNA.

The dG-to-dG, DNA-DNA interstrand cross-link at the duplex sequence 5'-d(GNC) formed by the antitumor drug mechlorethamine (bis(2-chloroethyl)methylamine) was studied both theoretically and experimentally. Computer models of cross-linked DNA were energy minimized using molecular mechanics. The energy minimized structures possessed local distortion of the DNA helix, especially propeller twisting and buckling, caused by the tether length being too small to bridge the spacing of N7 atoms of dG at the sequence 5'-d(GNC) in B DNA. Overwinding of 2-6 degrees was present at each of the two dinucleotide steps spanned by the cross-link. The predicted structural changes were compatible with the possibility that this cross-link would introduce a static bend into the DNA double helix axis. An experimental study provided evidence for this induced bending of the helix axis in interstrand cross-linked samples. DNAs containing multiple mechlorethamine-induced interstrand cross-links exhibited anomalously low electrophoretic mobility in polyacrylamide gels when the lesions were separated by one or two turns. From the degree of gel retardation, the cross-linked DNAs were estimated to be bent by 12.4-16.8 degrees per lesion; estimation of the extent to which this bend was induced by the lesion was complicated by a preexisting bend in the non-cross-linked DNAs used. The data did not allow distinction of a static from an anisotropic dynamic bend; "universal" and "hinge" joints were excluded. Anomalous mobility was maximal when the lesion spacing was 21 bp, suggesting a helical repeat of 10.5 bp per turn.

DNA-DNA interstrand cross-linking by 2,5-bis(1-aziridinyl)-3,6-bis(carbethoxyamino)-1,4-benzoquinone: covalent structure of the dG-to-dG cross-links in calf thymus DNA and a synthetic DNA duplex.

The products of the alkylation reaction of reduced 2,5-bis(1-aziridinyl)-3,6-bis(carbethoxyamino)-1,4-benzoquinone (AZQ, 1a) with duplex DNA were studied using calf thymus DNA and a synthetic oligodeoxynucleotide. Reaction of calf thymus DNA with a mixture of AZQ and ascorbic acid followed by enzymatic digestion of the sugar phosphate backbone afforded numerous AZQ-derived products including substances identified as monoadducts of AZQ with both dG and dA (with the former in greater abundance) and two diadducts, as would be expected for intrastrand or interstrand cross-links, with one containing two dG residues per AZQ and the other one each of dG and dA (with the former adduct in greater abundance). The nucleotide connectivity and covalent structure of the dG-to-dG interstrand cross-links were studied in greater detail using a synthetic DNA duplex containing the nucleotide sequence 5'-d(GGGCCC), where it appeared that the predominant interstrand cross-links bridged dG residues on opposite strands and were separated by two intervening base pairs [5'-d(GNNC)]. The covalent structure of this lesion was tentatively identified as 2b, in which the N7 atoms of two dG residues have been alkylated by the aziridine functions of AZQ, based upon the results of piperidine fragmentation and characterization of the enzymatic and acidic hydrolysates of the cross-linked DNA.

Motions of short DNA duplexes: an analysis of DNA dynamics using an EPR-active probe.

The dynamics of a series of four DNA duplexes of length 12, 24, 48, and 96 base pairs have been studied using an electron paramagnetic resonance (EPR) active nitroxide spin-label covalently attached to a thymidine located near the center of each duplex. The linear EPR spectra were simulated by solving the stochastic Liouville equation for anisotropic rotational diffusion. The diffusion tensor for global rotation of the duplex was predicted from hydrodynamic theory for a right circular cylinder. All internal motions, assumed to be rapid, are modeled by reduced electron Zeeman and hyperfine tensor anisotropies. Best fit simulations to the data were then obtained by adjusting the total amplitude of all internal dynamics. The local, length-independent and the collective, length-dependent contributions to the internal dynamics were separated by determining the total amplitude of internal motion as a function of duplex length. The major axis of the spin tensors was determined to be tilted 20 degrees from the helix axis. As a result, the spin-label is most sensitive to flexural motions of the DNA duplex. It is found that the global tumbling of duplex is accurately modeled by hydrodynamic theory. The length-independent motion is characterized by a root-mean-squared amplitude of oscillation of 10 degrees in two dimensions at 20 degrees C and has a strong temperature dependence, indicating that the local structure of the DNA changes with temperature. The length dependence of the internal dynamics leads to an estimate of the dynamic flexural persistence length of 2500 +/- 340 A. There was no statistically significant difference between models assuming a harmonic or a square-well local bending potential.

Site-specific metal-induced damage of mitomycin C-crosslinked DNA fragments in the presence of sodium dithionite.

Purified singly mitomycin C-crosslinked DNA duplex fragments were treated with Cu(II) in the presence of sodium dithionite. Cleavage products were analyzed through single-nucleotide resolving denaturing polyacrylamide gel electrophoresis. In addition to the previously described non-specific cleavage, which may be attributed to the production of hydroxyl radical, a novel site-specific cleavage in the immediate vicinity of the MC-crosslink was observed. Possible mechanisms of this reaction are discussed.

Sequence preferences of DNA interstrand crosslinking agents: quantitation of interstrand crosslink locations in DNA duplex fragments containing multiple crosslinkable sites.

A general approach to the quantitative study of the sequence specificity of DNA interstrand crosslinking agents in synthetic duplex DNA fragments is described. In the first step, a DNA fragment previously treated with an interstrand crosslinking agent is subjected to denaturing PAGE. Not only does this distinguish crosslinked from native or monoadducted DNA, it is shown herein that isomeric crosslinked DNAs differing in position of the crosslink can in some cases be separated. In the second stage, the now fractionated crosslinked DNAs isolated from denaturing PAGE are subjected to fragmentation using iron(II)/EDTA. For those fractions which are structurally homogeneous, analysis of the resulting fragment distribution has previously been shown to reveal the crosslink position at nucleotide resolution. It is shown herein that in fractions which are structurally heterogeneous due to differences in position of crosslink, this analysis quantifies the relative extent of crosslinking at distinct sites. Using this method it is shown that reductively activated mitomycin C crosslinks the duplex sequences 5'-GCGC and 5'-TCGA with 3 +/- 1:1 relative efficiency.