Visualization of drug-nucleic acid interactions at atomic resolution. III. Unifying structural concepts in understanding drug-DNA interactions and their broader implications in understanding protein-DNA interactions.

Structural information afforded by the X-ray crystallographic studies of ethidium-dinucleoside monophosphate crystalline complexes described in the preceding two papers has led to a detailed model for ethidium-DNA binding. Features of ethidium-DNA binding, in turn, have led to unifying structural concepts in understanding a wide range of drug-DNA interactions. It is possible that these concepts have still broader implications in understanding the nature of protein-DNA interactions.This paper begins by summarizing the stereochemical aspects of ethidium-DNA, actinomycin-DNA and irehdiamine-DNA binding, molecules that use intercalative and kinked-type geometries in binding to DNA. It then describes superhelical DNA structures formed by kinking DNA periodically varying numbers of base-pairs apart. κ-kinked B DNA, a structure formed by kinking DNA every ten base-pairs, is a left-handed superhelical structure that may be utilized in the organization of DNA within the nucleosome in chromatin. β-kinked B DNA is a right-handed superhelical structure formed by kinking DNA every two base-pairs. It is possible that premelting conformational changes occur in DNA which utilize elements of this structure. This would expose base-pairs to solvent denaturation, and could lower the activation energy necessary for strand separation during DNA denaturation. RNA polymerase and other DNA melting proteins could capitalize on this type of premelting conformational change when binding to DNA.The concept that conformational flexibility exists in DNA structure (and that drug intercalation is a phenomenon that reflects this flexibility) can, in addition, explain a wide variety of physicochemical data about DNA. In this paper we discuss the nature of these data in detail.     

Introduction of interrupted secondary structure in supercoiled DNA as a function of superhelix density: consideration of hairpin structures in superhelical DNA.

PM2 DNA was prepared with different superhelical densities (sigma) in order to examine the relationship betweenn supercoiling and the occurrence of a region(s) of unpaired bases in this DNA. A previous study showed that CH3HgOH reacts with native superhelical PM2 DNA more rapidly than the nicked form II. This evaluation of binding, monitored through the change of sedimentation velocity, was repeated on PM2 DNA I with different superhelical densities. Early binding is detected by an increase in sedimentation velocity and occurs with molecules with sigma' values betwee -0.025 and -0.037. The conversion of form I to form II with the single-strand-specific endonuclease from Neurospora crassa also occurs above a sigma value of -0.025. This data strongly supports the view that supercoiling produces interrupted secondary structure. The question whether the interrupted regions remain single stranded in character or form small intrastrand hairpin regions is considered by examining which model best fits the CH3HgOH- induced sedimentation velocity changes and the standard sedimentation velocity versus the superhelical density curve for the in vitro made DNAs. The hairpin model offers the most satisfactory explanations for all the results of this and previous studies.     

Ligand binding isotherm for DNA in the presence of supercoil-induced non-B form: a theoretical analysis

A binding isotherm in the form of a modified McGhee-Von Hippel equation is proposed, on the basis of thermodynamical considerations, to include the non-cooperative binding of extended ligands to supercoiled DNA, where a stretch of non-B form may be present under superhelical stress. It is then studied, on the basis of a non-linear Scatchard plot, how the presence of an intercalating ligand can relax the supercoiled molecule and thus destabilise the non-B stretch, which may be recognised by the existence of a significant kink in the Scatchard plot.     

Further analysis of the altered secondary structure of superhelical DNA. Sensitivity to methylmercuric hydroxide a chemical probe for unpaired bases

A previous study in our laboratory of the reaction of formaldehyde with super-helical DNAs (φX replicative form and PM2) has led to a model for superhelical DNA in which there is a region or regions of altered secondary structure containing unpaired bases. Similar experiments using the nicked circular DNA gave no evidence of interruptions of base pairing. In this study we present additional data, which support the above model as well as extending our analysis of the secondary structure of superhelical DNA and the dynamics of the early denaturation process. In a series of experiments involving the binding of methyl-mercury as a chemical probe of unpaired bases, we obtained the following results. (1) Initially, both s⁰_20w and the buoyant density of the superhelical form of phage PM2 DNA increased as a function of methylmercuric hydroxide concentration, whereas the nicked form did not. (2) This initial binding is accompanied by an increase in superhelical content τ from ?41 to ?46 turns. (3) The binding analysis allows us to estimate that 3.7% of the bases contain methylmercury in this phase of the transition. This is in excellent agreement with the extent of formylation. (4) Such a preformylated molecule shows a shift in the transition to lower mercurial concentrations. These results are interpreted as follows. The initial increase in ?τ excludes the possibility that binding occurs to normal base-paired structures, since this would produce a coupled unwinding of duplex and superhelical turns. The additive effects of formylation and methylmercury binding support the concept that both chemical probes attack the same sites and induce similar structural changes. Thus the evidence clearly supports the view that superhelical DNA contains localized region(s) of interrupted base pairing. Recent studies from other laboratories using single strand-specific endonucleases are in complete agreement with this model.     

Linker histone interaction shows divalent character with both supercoiled and linear DNA

The interaction of linker histone H1 with both linear and superhelical double-stranded DNA has been investigated at low ionic strengths. Gel mobility retardation experiments demonstrate strikingly different behavior for the two forms of DNA. First, the experiments strongly suggest that linker histone binds to superhelical DNA in a negatively cooperative mode. In contrast, binding of linker histone to linear DNA under the conditions employed here shows no cooperativity. Second, binding of linker histone to linear DNA results in aggregation of histone-DNA complexes, even at very low levels of input histone H1. Because H1 has been shown to interact as a monomer, this aggregation is evidence of the divalent character of the linker histone, for without H1's ability to bind to two duplex strands of DNA, aggregation could not occur. Although aggregation can be made to occur with superhelical DNA, it can do so only at near-saturation levels of input histone H1. Finally, in direct competition, linker histone binds to superhelical DNA to the complete exclusion of linear DNA, indicating that the linker histone's function is related to the crossover structures that differentiate superhelical DNA from linear DNA. We develop a model that explains the observed behavior of binding of linker histone to superhelical DNA that is consistent with both the divalent character of the linker histone and the negative cooperativity by which linker histone and superhelical DNA interact.     

Specific binding of chartreusin, an antitumor antibiotic, to DNA

Chartreusin, an antitumor and antibacterial antibiotic, was found to inhibit negatively superhelical DNA-relaxation catalyzed by prokaryotic topoisomerase I and conversion of the superhelical DNA into unit length linear form catalyzed by single-strand-specific S1 nuclease. The inhibitory effect of the agent was due to the binding to DNA causing the alteration of tertiary structure. To characterize the binding specificity, we investigated the protection of DNA against cleavages by various restriction endonucleases. It was evidenced that the binding of the agent is not at random and correlates to the sequence 5'CGC 3' 3'GCG 5' on DNA stretch.     

Interaction between gene II protein and the DNA replication origin of bacteriophage f1

The origin of DNA replication of the filamentous bacteriophage f1 binds its initiator protein (gene II protein) in vitro to form a complex that can be trapped on nitrocellulose filters. The binding occurs with both superhelical form DNA and linear DNA fragments. A number of defective mutants of the origin were tested for the ability to bind gene II protein. The region of DNA required for the binding is around a second palindrome downstream from the palindrome that contains the DNA replication initiation site. It overlaps, but is not identical to, the region required for the nicking reaction by the protein. The nicking site itself was dispensable for the binding. In vivo, a number of defective deletion mutants of the origin, when in a plasmid, inhibited growth of superinfecting phage if the intracellular level of gene II protein was low. In addition, these defective origins inhibited the activity of the functional phage origin located on the same replicon. The domain of the DNA sequence required for inhibition in vivo was consistent with that for the binding in vitro.     

Catenation and supercoiling in the products of bacteriophage λ integrative recombination in vitro

Catenanes (interlocked circular DNA molecules) are the exclusive products of the bacteriophage λ integrative recombination reaction in vitro when the substrate is a supercoiled DNA molecule containing both the attP and attB sites. It is proposed that the catenation results from the superhelical form of the substrate DNA. We also show that both circular DNA products of a single recombination event can be recovered as superhelical molecules with a superhelical density approximately that of the substrate DNA. The recombination reaction must therefore occur as a coupled process which does not permit free rotation around single-strand breaks at any stage.     

Left-handed Z-DNA regions are present in negatively supercoiled bacteriophage PM2 DNA

Bacteriophage PM2 DNA is a 10-kb covalently closed circular (ccc) molecule with a reported superhelical density of sigma = -0.12. Here we describe the binding of anti-Z-DNA antibodies to PM2 form I DNA under high and low salt conditions. The binding to PM2 DNA has been demonstrated by competitive radioimmunoassay (RIA), retardation of the DNA:antibody complexes in agarose gels and visualization by electron microscopy. The antibody binding is dependent on the degree of negative supercoiling. Thus, PM2 form II and form III did not bind the antibody. The low salt RIA results indicated the presence of 200-400 bp of left-handed DNA per PM2 molecule. This could reduce the effective superhelical density to sigma = -0.04 to -0.08, a range comparable with those found for other ccc DNAs in vivo. Electron microscopy revealed that a maximum of 22 antibody molecules bind to PM2. Single-site restriction with HpaII of the fixed DNA:antibody complex showed a cluster of four to five antibody molecules bound near one end of the linear DNA molecule. The evidence presented indicates that PM2 DNA contains regions of left-handed conformation under physiological conditions (low salt concentration) as well as at high salt concentrations. In addition, electrophoretic analyses of PM2 topoisomers indicate the presence of left-handed regions at superhelical densities less than that of isolated PM2 DNA.     

Cell lines from xeroderma pigmentosum complementation group A lack a single-stranded-DNA-binding activity

Human fibroblasts and HeLa cells contain two major DNA-binding activities for superhelical DNA, which can be separated by phosphocellulose chromatography. The DNA-binding activity which elutes first from the column coelutes with and is probably identical to a single-stranded-DNA-binding activity. The second activity has been characterized previously. It binds preferentially to super-helical DNA containing DNA damage, but does not bind to single-stranded DNA. Five cell lines derived from patients with the repairdeficiency syndrome xeroderma pigmentosum (XP) were analyzed for the presence of these binding activities. Four of the cell lines were from the A-complementation group and one was from the D-complementation group of XP. The binding activity with preference for damaged DNA was present in all cell lines. The single-stranded-DNA-binding activity was present in the XP-D cell line but was absent or reduced in all of the four XP-A cell lines tested.     

The effect of H1 histone on the action of DNA-relaxing enzyme.

The action of DNA-relaxing enzyme on H1-DNA complexes was investigated. Complexes of superhelical and relaxed closed circular duplex DNA with H1 were treated with mammalian relaxing enzyme, deproteinized, and electrophoresed on agarose gels. At relatively low ratios of H1 to superhelical DNA, molecules of superhelical density intermediate between those of the starting material and relaxed DNA, the normal product, were generated. At relatively high H1 histone concentrations (H1:DNA greater than 0.4 w/w), the superhelical DNA was not relaxed. Further, no superhelical turns were introduced into relaxed closed duplex DNA at any concentration of H1 tested. Thus, the binding of H1 histone to DNA prevents the action of the relaxing enzyme. Moreover, H1 histone does not appear to unwind the DNA duplex upon binding. The implications of these observations and the previously demonstrated specificity of H1 histone for superhelical DNA are discussed in relation to the structure of chromatin.     

DNA binding properties of Saccharomyces cerevisiae chromatin associated ras-related protein Yp20

Yp20 is an abundant 20 kDa chromatin associated protein which has been shown to be related antigenically to genuine Hras products. Using Southwestern blots we have demonstrated that Yp20 is a DNA binding protein. It is also shown that protein Yp20 like protein HM (an abundant thermostable 20 kDa DNA binding protein isolated from mitochondria) and like the 21 kDa autonomously replicating sequence binding factor II (ABFII) is able to introduce superhelical turns into circular relaxed DNA in the presence of DNA topoisomerase I activity. We suggest that this protein may be important for chromatin structure and function.     

Purification and characterization of DNA-relaxing enzyme from Haemophilus gallinarium.

A DNA-relaxing enzyme capable of concerted nicking and closing of DNA backbone bonds has been purified from Haemophilus gallinarum by two chromatographic steps and gel filtration. The enzyme efficiently catalyzes the removal of superhelical turns from a negatively twisted DNA and requires Mg2+ for this activity. Slight removal of superhelical turns from a positively twisted DNA generated by binding of ethidium bromide is found, but only at high enzyme concentrations. The DNA-relaxing activity is inhibited markedly with heat-denatured DNA, whereas native DNA and RNA have almost no affect on this activity.     

Theoretical model for the equilibrium behavior of DNA superhelices

A statistical-mechanical model for superhelical DNA is presented. The partition function for a DNA superhelix is written by using a combinatorial approach in order to allow for the known relation between the number of superhelical twists and the states of the base pairs in the double helix. While the theory allows any factors which might contribute to the free energy of superhelical twisting to be included in the statistical weights of the superhelical twists, only the reduction in configurational entropy is considered in this paper. Similarities between an imperfectly matched DNA double helix and a DNA superhelix are used in the derivation of expressions for the entropy of superhelical DNA. Although the partition function is presented in a general form, permitting many equilibrium properties of DNA superhelices to be treated, the application considered in this paper is the calculation of helix–coil transition curves. Several experimentally observed features of such transitions are predicted. For example, the curves are bimodal, with an early and a late transition relative to that of a nicked molecule. The results are very sensitive to the volume within which two parts of the double helix must meet when forming a superhelical twist. The free energy of superhelix formation is calculated, and the results are compared with those obtained from the data of Bauer and Vinograd for ethidium bromide intercalation. In the present model, the free energy increases less sharply with an increase in the number of superhelical twists than observed experimentally, indicating that factors other than configurational entropy probably make important contributions to the free energy of superhelix formation.     

A single amino acid substitution reduces the superhelicity requirement of a replication initiator protein.

The origin of rolling circle replication in filamentous coliphage consists of a core origin that is absolutely required and an adjacent replication enhancer sequence that increases in vivo replication 30 to 100-fold. The core origin binds the initiator protein (gpII) which either nicks or relaxes negatively superhelical replicative form DNA (RFI). Nicking at the origin, but not relaxation, leads to initiation of DNA replication. Our results indicate that the ratio of nicking to relaxation (nicking-closing) in vitro depends on the superhelical density of the substrate. We have studied the effect of a single amino acid substitution in gpII, which allows wild-type levels of replication in the absence of the enhancer, on origin nicking and binding. The enhancer-independent mutation yields more nicking and less relaxation of RFI, compared to the wild-type protein. The mutant gpII also shows a reduced requirement for superhelicity of the substrate in the nicking reaction. At the same time, the mutant gpII increases the cooperativity of protein-protein interactions in origin binding. We propose that the relaxation activity of gpII negatively regulates replication initiation, and that both increase in the negative superhelicity of the substrate and action of the replication enhancer may antagonize the relaxation activity.     

The Xenopus laevis mitochondrial protein mtDBP-C cooperatively folds the DNA in vitro.

The binding of the Xenopus laevis mitochondrial protein mtDBP-C to DNA was studied by equilibrium density banding, agarose gel electrophoresis and electron microscopy. The results obtained show that the mtDBP-C binds cooperatively to DNA irrespective of whether the DNA is supercoiled, relaxed or linear and it induces the formation of superhelical turns locally leading to the formation of a highly folded structure. It appears that this protein could be involved in the compaction of DNA in the mitochondrial nucleoid.     

Escherichia coli DNA topoisomerase III: purification and characterization of a new type I enzyme

A new topoisomerase capable of relaxing negatively supercoiled DNA in Escherichia coli has been identified during chromatography on novobiocin-Sepharose. A simple and reproducible purification procedure is described to obtain this enzyme, called topoisomerase III (topo III), in a homogeneous form. The protein is a single polypeptide with a molecular weight of 74 000 +/- 2000 and is a type I topoisomerase, changing the linking number of DNA circles in steps of one. It is present in deletion strains lacking the topA gene and further differs from the well-studied topoisomerase I (omega protein; Eco topo I) in (1) its requirement for K+ in addition to Mg2+ to exhibit optimal activity and (2) its affinity to novobiocin-Sepharose. Positively supercoiled DNA is not relaxed during exposure to the enzyme. Topo III has no ATPase activity, and ATP does not show any discernible effect on the reduction of superhelical turns. The purified topoisomerase has no supercoiling activity and is unaffected by high concentrations of oxolinic acid and novobiocin in the relaxing reaction. Single-stranded DNA and spermidine strongly inhibit the topoisomerase activity.     

Linker histone binding to superhelical DNA: Electrophoretic and filter binding studies

It has been known for years that linker histones bind preferentially to supercoiled DNA. This preference has been demonstrated by a number of different techniques including deoxyfibonucleoprotein electrophoresis, sedimentation, and filter binding under non-competitive conditions. In an attempt to further study this issue, we used one- and two-dimensional electrophoretic gels and filter binding under competitive conditions, with all DNA forms of interest being simultaneously present in the incubation mixture. Comparison between results obtained by the two methods showed that whereas the preference for superhelical molecules was clearly seen in the electrophoretic gels, the filter binding assay failed to reveal this preference. These results reveal limitations to the filter binding technique, which must be borne in mind in studies involving superhelical DNA molecules.