Structure of Three-Way DNA Junctions 1. Non-Planar DNA Geometry

Abstract

Three-way junctions were obtained by annealing two synthetic DNA-oligomers. One of the strands contains a short palindrome sequence, leading to the formation of a hairpin with four base pairs in the stem and four bases in the loop. Another strand is complementary to the linear arms of the first hairpin-containing strand. Both strands were annealed to form a three-way branched structure with sticky ends on the linear arms. The branched molecules were ligated, and the ligation mixture was analysed on a two-dimensional gel in conditions which separated linear and circular molecules. Analysis of 2D-electrophoresis data shows that circular molecules with high mobility are formed. Formation of circular molecules is indicative of bends between linear arms. We estimate the magnitude of the angle between linear arms from the predominant size of the circular molecules formed. When the junction-to-junction distance is 20–21 bp, trimers and tetramers are formed predominately, giving an angle between linear arms as small as 60–90°. Rotation of the hairpin position in the three- way junction allowed us to measure angles between other arms, yielding similar values. These results led us to conclude that the three-way DNA junction possesses a non-planar pyramidal geometry with 60–90° between the arms. Computer modeling of the three-way junction with 60° pyramidal geometry showed a predominantly B-form structure with local distortions at the junction points that diminish towards the ends of the helices. The size distributions of circular molecules are rather broad indicating a dynamic flexibility of three-way DNA junctions.     

Stability and structure of three-way DNA junctions containing unpaired nucleotides.

Non-paired nucleotides stabilize the formation of three-way helical DNA junctions. Two or more unpaired nucleotides located in the junction region enable oligomers ten to fifteen nucleotides long to assemble, forming conformationally homogeneous junctions, as judged by native gel electrophoresis. The unpaired bases can be present on the same strand or on two different strands. Up to five extra bases on one strand have been tested and found to produce stable junctions. The formation of stable structures is favored by the presence of a divalent cation such as magnesium and by high monovalent salt concentration. The order-disorder transition of representative three-way junctions was monitored optically in the ultraviolet and analyzed to quantify thermodynamically the stabilization provided by unpaired bases in the junction region. We report the first measurements of the thermodynamics of adding an unpaired nucleotide to a nucleic acid three-way junction. We find that delta delta G degrees (37 degrees C) = +0.5 kcal/mol for increasing the number of unpaired adenosines from two to three. Three-way junctions having reporter arms 40 base-pairs long were also prepared. Each of the three reporter arms contained a unique restriction site 15 base-pairs from the junction. Asymmetric complexes produced by selectively cleaving each arm were analyzed on native gels. Cleavage of the double helical arm opposite the strand having the two extra adenosines resulted in a complex that migrated more slowly than complexes produced by cleavage at either of the other two arms. It is likely that the strand containing the unpaired adenosines is kinked at an acute angle, forming a Y-shaped, rather than a T-shaped junction.     

Long-distance radical cation reactions in DNA three-way junctions: inter-arm interaction and migration through the junction

DNA three-way junctions (TWJ) are branched molecules having three ‘arms’. We studied long-distance radical cation migration in these assemblies by incorporating anthraquinone (AQ) groups linked by a covalent tether to one strand of one arm of the TWJ. Excitation of the AQ at 350 nm results in one-electron oxidation of the DNA, which generates a base radical cation. This leads to relatively inefficient (compared with duplex DNA) strand cleavage at guanines following piperidine treatment of the irradiated samples. When the AQ is linked to the 5′-terminus of arm III by a flexible tether, gel electrophoretic analysis shows that strand cleavage occurs at the guanines in all three arms. We also investigated a TWJ in which the anthraquinone is specifically intercalated in arm III. In this case, a different pattern of strand cleavage is detected. We conclude that there are at least two mechanisms for long-distance radical cation migration in TWJs: (i) by inefficient charge hopping through the junction; (ii) by a through-space, cross-arm interaction when the AQ is on a flexible tether.     

Structures of bulged three-way DNA junctions.

We have studied a series of three-way DNA junctions containing unpaired bases on one strand at the branch-point of the junctions. The global conformation of the arms of the junctions has been analysed by means of polyacrylamide gel electrophoresis, as a function of conditions. We find that in the absence of added metal ions, all the results for all the junctions can be accounted for by extended structures, with the largest angle being that between the arms defined by the strand containing the extra bases. Upon addition of magnesium (II) or hexamine cobalt (III) ions, the electrophoretic patterns change markedly, indicative of ion-dependent folding transitions for some of the junctions. For the junction lacking the unpaired bases, the three inter-arm angles appear to be quite similar, suggesting an extended structure. However, the addition of unpaired bases permits the three-way junction to adopt a significantly different structure, in which one angle becomes smaller than the other two. These species also exhibit marked protection against osmium addition to thymine bases at the point of strand exchange. These results are consistent with a model in which two of the helical arms undergo coaxial stacking in the presence of magnesium ions, with the third arm defining an angle that depends upon the number of unpaired bases.     

Endonuclease VII resolves Y-junctions in branched DNA in vitro.

Endonuclease VII (gp 49 of phage T4) resolves four-way junctions in branched DNAs. We have extended our investigations of the specificity of endo VII and tested its activity with three-way junctions (Y-structures) constructed in vitro. Both 'closed' and 'open' Y-structures were made, absolutely identical in sequence but differing from each other by a single nick in one of the three arms. Pure Y-structures were obtained on a preparative scale by annealing plus and minus strands from two M13mp strains. One strain has an inverted repeat of 2 X 31 nucleotides cloned into the single EcoRI site while in the other strain this repeat is absent. The structures were used in reactions with endo VII, which recognizes the branch point of both structures and introduces a characteristic number of nicks, 3' to the junction in each arm of the structure. Strong and weak sites could be distinguished and the cleavage pattern differed significantly between the two structures. The observed resolution of Y-junctions by endo VII in vitro is compatible with a model for the resolution of recombinant Y-branches in DNA.     

Structure of branched DNA molecules: gel retardation and atomic force microscopy studies.

DNA heteroduplexes as models for slipped strand DNA have been analyzed by polyacrylamide gel migration and atomic force microscopy (AFM). All heteroduplexes containing one hairpin or loop have reduced electrophoretic mobilities compared with that expected for their molecular weights. The retarded gel mobility correlates with the formation of a sharp kink detected by AFM. Increasing the hairpin length from 7 bp to 50 bp results in a monotonous decrease in gel mobility of heteroduplexes. This secondary retardation effect appears to depend only on the hairpin size since the AFM data show no dependence of the kink angle on the hairpin length. Heteroduplex isomers with a loop or hairpin in opposite strands migrate with distinct mobilities. Analysis of gel migration of heteroduplexes with altered hairpin orientations as well as of truncated heteroduplexes indicates that the difference in mobility is due to an inherent curvature in one of the long arms. This is confirmed by the end-to-end distance measurements from AFM images. In addition, significant variation of the end-to-end distances is consistent with a dynamic structure of heteroduplexes at the three-way junction. Double heteroduplexes containing one hairpin in each of the complementary strands also separate in a gel as two isomers. Their appearance in AFM showed a complicated pattern of flat representations of the three-dimensional structure and may indicate a certain degree of interaction between complementary parts of the hairpins that are several helical turns apart.     

The three-way DNA junction is a Y-shaped molecule in which there is no helix-helix stacking.

We have studied the structure of a number of three-way DNA junctions that were closely related in sequence to four-way junctions studied previously. We observe that the electrophoretic mobility of the species derived by selective shortening of one arm of a junction are very similar whichever arm is shortened, and that this remains so whether or not magnesium is present in the buffer. This suggests that the angles subtended between the arms of the three-way junctions are similar. All thymine bases located immediately at the junction are reactive to osmium tetroxide, indicating that out-of-plane attack is not prevented by helix-helix stacking, and this is also independent of the presence or absence of metal cations. The results suggest that the three-way junction cannot undergo an ion-induced conformational folding involving helical stacking, but remains fixed in a Y-shaped extended conformation. Thus the three- and four-way junctions are quite different in character in the presence of cations.     

The structure of intramolecular triplex DNA: atomic force microscopy study

We applied atomic force microscopy (AFM) for direct imaging of intramolecular triplexes (H-DNA) formed by mirror-repeated purine-pyrimidine repeats and stabilized by negative DNA supercoiling. H-DNA appears in atomic force microscopy images as a clear protrusion with a different thickness than DNA duplex. Consistent with the existing models, H-DNA formation results in a kink in the double helix path. The kink forms an acute angle so that the flanking DNA regions are brought in close proximity. The mobility of flanking DNA arms is limited compared with that for cruciforms and three-way junctions. Structural properties of H-DNA may be important for promoter-enhancer interactions and other DNA transactions.     

Homologous pairing in genetic recombination: recA protein makes joint molecules of gapped circular DNA and closed circular DNA

The recA protein, which is essential for genetic recombination in E. coli, promotes the homologous pairing of double-stranded DNA and linear single-stranded DNA, thereby forming a three-stranded joint molecule called a D loop. Single-stranded DNA stimulates recA protein to unwind double-stranded DNA. By a presumably related mechanism, recA protein promoted the homologous pairing of two circular double-stranded molecules when one of them had a gap in one strand. The two molecules were joined at homologous sites by noncovalent bonds. The covalently closed molecule remained intact and was not topologically linked to the intact circular strand of the gapped substrate. Electron microscopy showed that molecules were usually linked at two or more nearby points. The junctions in most molecules were shorter than 300 nucleotides. Sometimes the region between two extreme points was separated into two arms, producing an ellipsoidal loop (called an eye loop). The junctions in these biparental joint molecules were frequently remote from the site of the gap. We infer that a free end of the interrupted strand crossed over to form a structure like a D loop which moved away from the gap by branch migration.     

Sequence-dependent folding of DNA three-way junctions

Three-way DNA junctions can adopt several different conformers, which differ in the coaxial stacking of the arms. These structural variants are often dominated by one conformer, which is determined by the DNA sequence. In this study we have compared several three-way DNA junctions in order to assess how the arrangement of bases around the branch point affects the conformer distribution. The results show that rearranging the different arms, while retaining their base sequences, can affect the conformer distribution. In some instances this generates a structure that appears to contain parallel coaxially stacked helices rather than the usual anti-parallel arrangement. Although the conformer equilibrium can be affected by the order of purines and pyrimidines around the branch point, this is not sufficient to predict the conformer distribution. We find that the folding of three-way junctions can be separated into two groups of dinucleotide steps. These two groups show distinctive stacking properties in B-DNA, suggesting there is a correlation between B-DNA stacking and coaxial stacking in DNA junctions.     

Resolution of Holliday junction analogs by T4 endonuclease VII can be directed by substrate structure

Endonuclease VII is an enzyme from bacteriophage T4 capable of resolving four-arm Holliday junction intermediates in recombination. Since natural Holliday junctions have homologous (2-fold) sequence symmetry, they can branch migrate, creating a population of substrates that have the branch point at different sites. We have explored the substrate requirements of endonuclease VII by using immobile analogs of Holliday junctions that lack this homology, thereby situating the branch point at a fixed site in the molecule. We have found that immobile junctions whose double-helical arms contain fewer than nine nucleotide pairs do not serve as substrates for resolution by endonuclease VII. Scission of substrates with 2-fold symmetrically elongated arms produces resolution products that are a function of the particular arms that are lengthened. We have confirmed that the scission products are those of resolution, rather than nicking of individual strands, by using shamrock junction molecules formed from a single oligonucleotide strand. A combination of end-labeled and internally labeled shamrock molecules has been used to demonstrate that all of the scission is due to coordinated cleavage of DNA on opposite sides of the junction, 3' to the branch point. Endonuclease VII is known to cleave the crossover strands of Holliday junctions in this fashion. The relationship of the long arms to the cleavage direction suggests that the portion of the enzyme which requires the minimum arm length interacts with the pair of arms containing the 3' portion of the crossover strands on the bound surface of the antiparallel junction.     

Assembly and characterization of five-arm and six-arm DNA branched junctions.

DNA branched junctions have been constructed that contain either five arms or six arms surrounding a branch point. These junctions are not as stable as junctions containing three or four arms; unlike the smaller junctions, they cannot be shown to migrate as a single band on native gels when each of their arms contains eight nucleotide pairs. However, they can be stabilized if their arms contain 16 nucleotide pairs. Ferguson analysis of these junctions in combination with three-arm and four-arm junctions indicates a linear increase in friction constant as the number of arms increases, with the four-arm junction migrating anomalously. The five-arm junction does not appear to have any unusual stacking structure, and all strands show similar responses to hydroxyl radical autofootprinting analysis. By contrast, one strand of the six-arm junction shows virtually no protection from hydroxyl radicals, suggesting that it is the helical strand of a preferred stacking domain. Both junctions are susceptible to digestion by T4 endonuclease VII, which resolves Holliday junctions. However, the putative helical strand of the six-arm junction shows markedly reduced cleavage, supporting the notion that its structure is largely found in a helical conformation. Branched DNA molecules can be assembled into structures whose helix axes form multiply connected objects and networks. The ability to construct five-arm and six-arm junctions vastly increases the number of structures and networks that can be built from branched DNA components. Icosahedral deltahedra and 11 networks with 432 symmetry, constructed from Platonic and Archimedean solids, are among the structures whose construction is feasible, now that these junctions can be made.     

Helical stacking in DNA three-way junctions containing two unpaired pyrimidines: proton NMR studies.

The proton NMR spectra of DNA three-way junction complexes (TWJ) having unpaired pyrimidines, 5'-TT- and 5'-TC- on one strand at the junction site were assigned from 2D NOESY spectra acquired in H2O and D2O solvents and homonuclear 3D NOESY-TOCSY and 3D NOESY-NOESY in D2O solvent. TWJ are the simplest branched structures found in biologically active nucleic acids. Unpaired nucleotides are common features of such structures and have been shown to stabilize junction formation. The NMR data confirm that the component oligonucleotides assemble to form conformationally homogeneous TWJ complexes having three double-helical, B-form arms. Two of the helical arms stack upon each other. The unpaired pyrimidine bases lie in the minor groove of one of the helices and are partly exposed to solvent. The coaxial stacking arrangement deduced is different from that determined by Rosen and Patel (Rosen, M.A., and D.J. Patel. 1993. Biochemistry. 32:6576-6587) for a DNA three-way junction having two unpaired cytosines, but identical to that suggested by Welch et al. (Welch, J. B., D. R. Duckett, D. M. J. Lilley. 1993. Nucleic Acids Res. 21:4548-4555) on the basis of gel electrophoretic studies of DNA three-way junctions containing unpaired adenosines and thymidines.     

Quasi-equivalence in site-specific recombinase structure and function: crystal structure and activity of trimeric Cre recombinase bound to a three-way Lox DNA junction

The crystal structure of a novel Cre-Lox synapse was solved using phases from multiple isomorphous replacement and anomalous scattering, and refined to 2.05 A resolution. In this complex, a symmetric protein trimer is bound to a Y-shaped three-way DNA junction, a marked departure from the pseudo-4-fold symmetrical tetramer associated with Cre-mediated LoxP recombination. The three-way DNA junction was accommodated by a simple kink without significant distortion of the adjoining DNA duplexes. Although the mean angle between DNA arms in the Y and X structures was similar, adjacent Cre trimer subunits rotated 29 degrees relative to those in the tetramers. This rotation was accommodated at the protein-protein and DNA-DNA interfaces by interactions that are "quasi-equivalent" to those in the tetramer, analogous to packing differences of chemically identical viral subunits at non-equivalent positions in icosahedral capsids. This structural quasi-equivalence extends to function as Cre can bind to, cleave and perform strand transfer with a three-way Lox substrate. The structure explains the dual recognition of three and four-way junctions by site-specific recombinases as being due to shared structural features between the differently branched substrates and plasticity of the protein-protein interfaces. To our knowledge, this is the first direct demonstration of quasi-equivalence in both the assembly and function of an oligomeric enzyme.     

Architecture of the replication fork stalled at the 3' end of yeast ribosomal genes

Every unit of the rRNA gene cluster of Saccharomyces cerevisiae contains a unique site, termed the replication fork barrier (RFB), where progressing replication forks are stalled in a polar manner. In this work, we determined the positions of the nascent strands at the RFB at nucleotide resolution. Within an HpaI-HindIII fragment essential for the RFB, a major and two closely spaced minor arrest sites were found. In the majority of molecules, the stalled lagging strand was completely processed and the discontinuously synthesized nascent lagging strand was extended three bases farther than the continuously synthesized leading strand. A model explaining these findings is presented. Our analysis included for the first time the use of T4 endonuclease VII, an enzyme recognizing branched DNA molecules. This enzyme cleaved predominantly in the newly synthesized homologous arms, thereby specifically releasing the leading arm.     

Role of the origin of transfer in termination of strand transfer during bacterial conjugation.

Conjugal transfer of the broad-host-range plasmid R1162 is initiated and terminated at the nic site within the 38-bp origin of transfer (oriT). Termination involves ligation of the transferred single strand by the plasmid-encoded MobA protein. Several different assays were used to identify the oriT DNA required for termination. For plasmids containing two oriTs, with transfer initiated at one and terminated at the other, the inverted repeat within oriT is important for termination. Deletion of the outer arm reduces the termination frequency; those terminations that do occur probably depend upon nicking at this oriT prior to transfer. The locations of second-site suppressor mutations indicate that base pairing between the arms of the inverted repeat is important for termination. In vitro, the inverted repeat is not required for specific cleavage of single-stranded DNA at nic, but competition experiments indicate that oriTs with the inverted repeat are preferentially cleaved. We propose that the function of the oriT inverted repeat is to trap the plasmid-encoded MobA protein at the end of a round of strand transfer, thus ensuring that the protein is available for the ligation step.     

Detection of template strand switching during initiation and termination of DNA replication of porcine circovirus

Nucleotide substitution mutagenesis was conducted to investigate the importance of the inverted repeats (palindrome) at the origin of DNA replication (Ori) of porcine circovirus type 1 (PCV1). Viral genomes with engineered mutations on either arm or both arms of the palindrome were not impaired in protein synthesis and yielded infectious progeny viruses with restored or new palindromes. Thus, a flanking palindrome at the Ori was not essential for initiation of DNA replication, but one was generated inevitably at termination. Among the 26 viruses recovered, 16 showed evidence of template strand switching, from minus-strand genome DNA to palindromic strand DNA, during biosynthesis of the Ori. Here I propose a novel rolling-circle "melting-pot" model for PCV1 DNA replication. In this model, the replicator Rep protein complex binds, destabilizes, and nicks the Ori sequence to initiate leading-strand DNA synthesis. All four strands of the destabilized inverted repeats exist in a "melted" configuration, and the minus-strand viral genome and a palindromic strand are available as templates, simultaneously, during initiation or termination of DNA replication. Inherent in this model is a "gene correction" or "terminal repeat correction" mechanism that can restore mutilated inverted-repeat sequences to a palindrome at the Ori of circular DNAs or at the termini of circularized linear DNAs. Potentially, the melted state of the inverted repeats increases the rate of noncomplementary or illegitimate nucleotide incorporation into the palindrome. Thus, this melting-pot model provides insight into the mechanisms of DNA replication, gene correction, and illegitimate recombination at the Ori of PCV1, and it may be applicable to the replication of other circular DNA molecules.     

An explanation for observed estrogen receptor binding to single-stranded estrogen-responsive element DNA.

Estrogen-inducible genes contain an enhancer called the estrogen response element (ERE), a double-stranded inverted repeat. The estrogen receptor (ER) is generally thought to bind to the double-stranded ERE. However, some reports provide evidence that an ER homodimer can bind a single strand of the ERE and suggest that single-stranded ERE binding is the preferred binding mode for ER. Since these two models describe quite different mechanisms of receptor action, we have attempted to reconcile the observations. Analyzing DNA structure by nuclease sensitivity, we found that two identical molecules of a single strand of DNA containing the ERE sequence can partially anneal in an antiparallel manner. Bimolecular annealing produces double-stranded inverted repeats, with adjacent unannealed tails. The amount of annealing correlates exactly with the ability of ER to bind bimolecular EREs. Either strand of an ERE could anneal to itself in a way that would bind ER. We conclude that ER binds only the annealed double-stranded ERE both in vitro and in vivo.     

Hemicatenanes form upon inhibition of DNA replication

Plasmid DNA incubated in interphase Xenopus egg extracts is normally assembled into chromatin and then into synthetic nuclei which undergo one round of regulated replication. During a study of restriction endonuclease cut plasmid replication intermediates (RIs) by the Brewer–Fangman 2D gel electrophoresis technique, we have observed the formation of a strong spike of X-shaped DNA molecules in extracts that otherwise yield very little or no RIs. Formation of these joint molecules is also efficiently induced in replication-competent extracts upon inhibition of replication fork progression by aphidicolin. Although their electrophoretic properties are quite similar to those of Holliday junctions, 2D gels of doubly cut plasmids show that these junctions can link two plasmid molecules at any pair of DNA sequences, with no regard for sequence homology at the branch points. Neutral–neutral–alkaline 3D gels show that the junctions only contain single strands of parental size and no recombinant strands. A hemicatenane, in which one strand of a duplex is wound around one strand of another duplex, is the most likely structure to account for these observations. The mechanism of formation of these novel joint DNA molecules and their biological implications are discussed.     

Structural recognition between a four-way DNA junction and a resolving enzyme

Resolving enzymes bind highly selectively to four-way DNA junctions, but the mechanism of this structural specificity is poorly understood. In this study, we have explored the role of interactions between the dimeric enzyme and the helical arms of the junction, using junctions with either shortened arms, or circular permutation of arms. We find that DNA-protein contacts in the arms containing the 5' ends of the continuous strands are very important, conferring a significant level of sequence discrimination upon both the choice of conformer and the order of strand cleavage. We have exploited these properties to obtain hydroxyl radical footprinting data on endonuclease I-junction complexes that are not complicated by the presence of alternative conformers, with results that are in good agreement with the arm permutation and shortening experiments. Substitution of phosphate groups at the center of the junction reveals the importance of electrostatic interactions at the point of strand exchange in the complex. Our data show that the form of the complex between endonuclease I and a DNA junction depends on the core of the junction and on interactions with the first six base-pairs of the arms containing the 5' ends of the continuous strands.