Structure refined to 2A of a nicked DNA octanucleotide complex with DNase I

The cutting rates of bovine pancreatic deoxyribonuclease I (DNase I) vary along a given DNA sequence, indicating that the enzyme recognizes sequence-dependent structural variations of the DNA double-helix. In an attempt to define the helical parameters determining this sequence-dependence, we have co-crystallized a complex of DNase I with a self-complementary octanucleotide and refined the crystal structure at 2 A resolution. This structure confirms the basic features of an early model, namely that an exposed loop of DNase I binds in the minor groove of B-type DNA and that interactions do occur with the backbone of both strands. Nicked octamer duplexes that have lost a dinucleotide from the 3'-end of one strand are hydrogen-bonded across a two-fold axis in the crystal to form a quasi-continuous double helix of 14 base pairs. The DNA 14-mer has a B-type conformation and shows substantial distortion of both local and overall helix parameters, induced mainly by the tight interaction of Y73 and R38 in the unusually wide minor groove. Directly coupled to the widening of the groove by approximately 3A is a 21.5 degree bend of the DNA away from the bound enzyme towards the major groove, suggesting that both DNA stiffness and groove width are important in determining the sequence-dependence of the enzyme cutting rate. A second cut of the DNA which is induced by diffusion of Mn2+ into the co-crystals suggests that there are two active sites in DNase I separated by more than 15A.     

Importance of DNA stiffness in protein-DNA binding specificity

From the first high-resolution structure of a repressor bound specifically to its DNA recognition sequence it has been shown that the phage 434 repressor protein binds as a dimer to the helix. Tight, local interactions are made at the ends of the binding site, causing the central four base pairs (bp) to become bent and overtwisted. The centre of the operator is not in contact with protein but repressor binding affinity can be reduced at least 50-fold in response to a sequence change there. This observation might be explained should the structure of the intervening DNA segment vary with its sequence, or if DNA at the centre of the operator resists the torsional and bending deformation necessary for complex formation in a sequence dependent fashion. We have considered the second hypothesis by demonstrating that DNA stiffness is sequence dependent. A method is formulated for calculating the stiffness of any particular DNA sequence, and we show that this predicted relationship between sequence and stiffness can explain the repressor binding data in a quantitative manner. We propose that the elastic properties of DNA may be of general importance to an understanding of protein-DNA binding specificity.     

Structure of DNase I at 2.0 A resolution suggests a mechanism for binding to and cutting DNA

Bovine pancreatic deoxyribonuclease I (DNase I), an endonuclease that degrades double-stranded DNA in a nonspecific but sequence-dependent manner, has been used as a biochemical tool in various reactions, in particular as a probe for the structure of chromatin and for the helical periodicity of DNA on the nucleosome and in solution. Limited digestion by DNase I, termed DNase I 'footprinting', is routinely used to detect protected regions in DNA-protein complexes. Recently, we have solved the three-dimensional structure of this glycoprotein (relative molecular mass 30,400) by X-ray structure analysis at 2.5 A resolution and have subsequently refined it crystallographically at 2.0 A. Based on the refined structure and the binding of Ca2+-thymidine 3',5'-diphosphate (Ca-pTp) at the active site, we propose a mechanism of action and present a model for the interaction of DNase I with double-stranded DNA that involves the binding of an exposed loop region in the minor groove of B-DNA and electrostatic interactions of phosphates from both strands with arginine and lysine residues on either side of this loop. We explain DNase I cleavage patterns in terms of this model and discuss the consequences of the extended DNase I-DNA contact region for the interpretation of DNase I footprinting results.     

X-ray structure of a DNA hairpin molecule

We have solved the crystal structure of a synthetic DNA hexadecanucleotide of sequence: C-G-C-G-C-G-T-T-T-T-C-G-C-G-C-G, at 2.1 A resolution, and observed that it adopts a monomeric hairpin configuration with a Z-DNA hexamer stem. In the T4 loop the bases stack with one another and with neighbouring molecules of the crystal, and not with base pairs of their own hexamer stem. Two thymine T10 rings from different molecules stack between the C1-G16 ends of a third and a fourth hairpin helix, in a manner that suggests T-T base 'pairing' and simulates a long, 13-base-pair helix. Although such T-T interactions would not be present in solution, they illustrate a remarkable tendency of thymines for self-association. Purine-purine G-A base pairs are known to exist in the anti-anti conformation with an increase in local helix width; it may be that more serious consideration should be given to the possible existence of pyrimidine-pyrimidine C-T base pairs with decreased local helix width, particularly where several such base pairs occur sequentially.     

The structure of DNA in the nucleosome core

The 1.9-A-resolution crystal structure of the nucleosome core particle containing 147 DNA base pairs reveals the conformation of nucleosomal DNA with unprecedented accuracy. The DNA structure is remarkably different from that in oligonucleotides and non-histone protein-DNA complexes. The DNA base-pair-step geometry has, overall, twice the curvature necessary to accommodate the DNA superhelical path in the nucleosome. DNA segments bent into the minor groove are either kinked or alternately shifted. The unusual DNA conformational parameters induced by the binding of histone protein have implications for sequence-dependent protein recognition and nucleosome positioning and mobility. Comparison of the 147-base-pair structure with two 146-base-pair structures reveals alterations in DNA twist that are evidently common in bulk chromatin, and which are of probable importance for chromatin fibre formation and chromatin remodelling.     

Stimulation of protein-directed strand exchange by a DNA helicase

The protein-mediated exchange of strands between a DNA double helix and a homologous DNA single strand involves both synapsis and branch migration, which are two important aspects of any general recombination reaction. Purified DNA-dependent ATPases from Escherichia coli (recA protein), Ustilago (rec 1 protein) and phage T4 (uvsX protein) have been shown to drive both synapsis and branch migration in vitro. The T4 gene 32 protein is a helix-destabilizing protein that greatly stimulates uvsX-protein-catalysed synapsis, and the E. coli SSB (single-strand binding) protein stimulates the analogous recA-protein-mediated reaction to a lesser degree. One suspects that several other proteins also play a role in the strand exchange process. For example, a DNA helicase could in principle accelerate branch migration rates by helping to melt the helix at the branch point. The T4 dda protein is a DNA helicase that is required to move the T4 replication fork past DNA template-bound proteins in vitro. Previously, we have shown that the dda protein binds to a column that contains immobilized T4 uvsX protein. We show here that this helicase specifically stimulates the branch migration reaction that the uvsX protein catalyses as a central part of the genetic recombination process in a T4 bacteriophage-infected cell.     

Structure of a repair enzyme interrogating undamaged DNA elucidates recognition of damaged DNA

How DNA repair proteins distinguish between the rare sites of damage and the vast expanse of normal DNA is poorly understood. Recognizing the mutagenic lesion 8-oxoguanine (oxoG) represents an especially formidable challenge, because this oxidized nucleobase differs by only two atoms from its normal counterpart, guanine (G). Here we report the use of a covalent trapping strategy to capture a human oxoG repair protein, 8-oxoguanine DNA glycosylase I (hOGG1), in the act of interrogating normal DNA. The X-ray structure of the trapped complex features a target G nucleobase extruded from the DNA helix but denied insertion into the lesion recognition pocket of the enzyme. Free energy difference calculations show that both attractive and repulsive interactions have an important role in the preferential binding of oxoG compared with G to the active site. The structure reveals a remarkably effective gate-keeping strategy for lesion discrimination and suggests a mechanism for oxoG insertion into the hOGG1 active site.     

Unusual helical packing in crystals of DNA bearing a mutation hot spot

The target sequence of the restriction enzyme NarI (GGCGCC) is a hot spot for the -2 frameshift mutagenesis (GGCGCC----GGCC) induced by the chemical carcinogens such as N-2-acetyl-aminofluorene. Of the guanine residues, all of which show equal reactivity towards the carcinogen, only binding to the 3'-most proximal guanine within the NarI site is able to trigger the frameshift event. We selected the non-palindromic dodecamer d(ACCGGCGCCACA), whose sequence corresponds to the most mutagenic NarI site in pBR322 DNA; for X-ray structure analysis. Its molecular structure determined at 2.8 A resolution reveals significant deviations from the structure of canonical B-form DNA, with partial opening of three G-C base pairs, high propeller twist values and sequence-dependent three-centred hydrogen bonds. This crystal structure shows a novel kind of packing in which helices are locked together by groove-backbone interactions. The partial opening of G-C base pairs is induced by interactions of phosphate anionic oxygen atoms with the amino group of cytosine bases. This provides a model for close approach of DNA molecules during biological processes, such as recombination.     

Dependence of the torsional rigidity of DNA on base composition

The Escherichia coli phage 434 repressor binds as a dimer to the operator of the DNA helix. Although the centre of the operator is not in contact with protein, the repressor binding affinity can be reduced at least 50-fold by changing the sequence there: operators with A.T base pairs near their centre bind the repressor more strongly than do operators with G.C base pairs at the same positions. To explain these observations, it has been proposed that the base composition at the centre of the operator affects the affinity of the operator for repressor by altering the ease with which operator DNA can undergo the torsional deformation necessary for complex formation. In this model, the variation in binding affinity would require the torsion constant to have specific values and to change in a sequence-dependent manner. We have now measured torsion constants for DNAs with widely different base compositions. Our results indicate that the torsion constants depend only slightly on the overall composition, and firmly delimit the range of values for each. Even the upper-limit values are much too small to account for the observed changes in affinity of the 434 repressor. These results rule out simple models that rely on substantial generic differences in torsion constant between A.T-rich sequences and G.C-rich sequences, although they do not rule out the possibility of particular sequences having abnormal torsion constants.     

The three-dimensional structure of a DNA duplex containing looped-out bases

Unpaired bases in DNA have been assigned a possible role in the mechanism of frameshift mutagenesis in sequences with repeated base pairs. They also occur in quasipalindromic DNA sequences, which have been implicated in mutagenesis where there are no repeated base pairs, through the formation of single-stranded hairpin loops. The conformation of unpaired bases in DNA has been the subject of numerous thermodynamic as well as high resolution NMR (nuclear magnetic resonance) studies (reviewed in ref. 4). The NMR studies in solution have shown that the duplex of the tridecamer DNA fragment d(CGCAGAATTCGCG) remains intact, and that the unpaired adenosines are stacked into the duplex. Having crystallized this oligonucleotide and determined its structure, we find its conformation in the crystal is close to that of a B-DNA duplex, with the two additional adenosines looped out from the double helix and causing little disruption of the rest of the structure.     

Nuclear factor III, a novel sequence-specific DNA-binding protein from HeLa cells stimulating adenovirus DNA replication

Dissection and reconstitution of the adenovirus DNA replication machinery has led to the discovery of two HeLa nuclear proteins which are required in conjunction with three viral proteins. One of these, nuclear factor I (NF-I), recognizes an internal region of the origin between nucleotides 25 and 40 and by binding to one side of the helix stimulates the initiation reaction up to 30-fold. NFI-binding sites have been observed upstream of several cellular genes, such as chicken lysozyme, human IgM and human c-myc, and coincide in most cases with DNase I hypersensitive regions. Here we report the identification of a novel DNA-binding protein from HeLa nuclei, designated NF-III, that recognizes a sequence in the adenovirus origin very close to the NFI-binding site, between nucleotides 36 and 54. This sequence includes the partially conserved nucleotides TATGATAATGAG. NF-III stimulates DNA replication four- to sixfold by increasing the initiation efficiency. Potential cellular binding sites include promoter elements of the histone H2B gene, the human interferon beta gene, the human and mouse immunoglobulin VK and VH genes and the mammal/chicken/Xenopus laevis U1 and U2 small nuclear RNA genes. Furthermore, a subset of the herpes simplex virus immediate early promoter specific TAATGARAT elements is homologous with the adenovirus 2 (Ad-2) NFIII-binding site.     

The SRA domain of UHRF1 flips 5-methylcytosine out of the DNA helix

Maintenance methylation of hemimethylated CpG dinucleotides at DNA replication forks is the key to faithful mitotic inheritance of genomic methylation patterns. UHRF1 (ubiquitin-like, containing PHD and RING finger domains 1) is required for maintenance methylation by interacting with DNA nucleotide methyltransferase 1 (DNMT1), the maintenance methyltransferase, and with hemimethylated CpG, the substrate for DNMT1 (refs 1 and 2). Here we present the crystal structure of the SET and RING-associated (SRA) domain of mouse UHRF1 in complex with DNA containing a hemimethylated CpG site. The DNA is contacted in both the major and minor grooves by two loops that penetrate into the middle of the DNA helix. The 5-methylcytosine has flipped completely out of the DNA helix and is positioned in a binding pocket with planar stacking contacts, Watson-Crick polar hydrogen bonds and van der Waals interactions specific for 5-methylcytosine. Hence, UHRF1 contains a previously unknown DNA-binding module and is the first example of a non-enzymatic, sequence-specific DNA-binding protein domain to use the base flipping mechanism to interact with DNA.     

Crystal structure of T4 endonuclease VII resolving a Holliday junction

Holliday proposed a four-way DNA junction as an intermediate in homologous recombination, and such Holliday junctions have since been identified as a central component in DNA recombination and repair. Phage T4 endonuclease VII (endo VII) was the first enzyme shown to resolve Holliday junctions into duplex DNAs by introducing symmetrical nicks in equivalent strands. Several Holliday junction resolvases have since been characterized, but an atomic structure of a resolvase complex with a Holliday junction remained elusive. Here we report the crystal structure of an inactive T4 endo VII(N62D) complexed with an immobile four-way junction with alternating arm lengths of 10 and 14 base pairs. The junction is a hybrid of the conventional square-planar and stacked-X conformation. Endo VII protrudes into the junction point from the minor groove side, opening it to a 14 A x 32 A parallelogram. This interaction interrupts the coaxial stacking, yet every base pair surrounding the junction remains intact. Additional interactions involve the positively charged protein and DNA phosphate backbones. Each scissile phosphate that is two base pairs from the crossover interacts with a Mg2+ ion in the active site. The similar overall shape and surface charge potential of the Holliday junction resolvases endo VII, RuvC, Ydc2, Hjc and RecU, despite having different folds, active site composition and DNA sequence preference, suggest a conserved binding mode for Holliday junctions.     

Molecular structure of r(GCG)d(TATACGC): a DNA--RNA hybrid helix joined to double helical DNA

The molecule r(GCG)d(TATACGC) is self-complementary and forms two DNA--RNA hybrid segments surrounding a central region of double helical DNA; its molecular structure has been solved by X-ray analysis. All three parts of the molecule adopt a conformation which is close to that seen in the 11-fold RNA double helix. The conformation of the ribonucleotides is partly determined by water molecules bridging between the ribose O2' hydroxyl group and cytosine O2. The hybrid-DNA duplex junction contains no structural discontinuities. However, the central DNA TATA sequence has some structural irregularities.     

Evidence for kinks in DNA folding in the nucleosome

The nucleosome subunit of chromatin consists of DNA folded around a histone core as a 1.8-turn left-handed solenoid. The crystal structure of the nucleosome core particle revealed that it has a dyad symmetry axis and that the minor helix groove faces outwards from the protein core. Richmond et al. noticed that the path traversed by the helix has severe bends at sites approximately one and four helix turns from the dyad axis. We have developed two photochemical methods to study the structure of DNA, and in particular that wrapped around the nucleosome core. One method depends on the sensitization of singlet oxygen production by an eosin analogue. We have monitored the rate at which excited state oxygen diffuses into contact with DNA base planes, and find that it attacks the nucleosome with high specificity. We have also mapped the DNA binding of the intercalating dye methylene blue, and conclude that it binds to the same sites accessible to oxygen by diffusion. On the basis of these results we suggest that the DNA in the nucleosome is bent or kinked at two sites, 1.5 helix turns from the dyad axis.