How B-DNA Dynamics Decipher Sequence-Selective Protein Recognition

The rules governing sequence-specific DNA–protein recognition are under a long-standing debate regarding the prevalence of base versus shape readout mechanisms to explain sequence specificity and of the conformational selection versus induced fit binding paradigms to explain binding-related conformational changes in DNA. Using a combination of atomistic simulations on a subset of representative sequences and mesoscopic simulations at the protein–DNA interactome level, we demonstrate the prevalence of the shape readout model in determining sequence-specificity and of the conformational selection paradigm in defining the general mechanism for binding-related conformational changes in DNA. Our results suggest that the DNA uses a double mechanism to adapt its structure to the protein: it moves along the easiest deformation modes to approach the bioactive conformation, while final adjustments require localized rearrangements at the base-pair step and backbone level. Our study highlights the large impact of B-DNA dynamics in modulating DNA–protein binding.     

Structure–function correlation for the EcoRV restriction enzyme: from non-specific binding to specific DNA cleavage

The EcoRV restriction endonuclease cleaves DNA at its recognition sequence at least a million times faster than at any other DNA sequence. The only cofactor it requires for activity is Mg²⁺: but in binding to DNA in the absence of Mg²⁺, the EcoRV enzyme shows no specificity for its recognition site. Instead, the reason why EcoRV cuts one DNA sequence faster than any other is that the rate of cleavage is controlled by the binding of Mg²⁺ to EcoRV-DNA complexes: the complex at the recognition site has a high affinity for Mg²⁺, while the complexes at other DNA sequences have low affinities for Mg²⁺. The structures of the EcoRV endonuclease, and of its complexes with either 8pecific or non-specific DNA, have been solved by X-ray crystallography. In the specific complex, the protein interacts with the bases in the recognition sequence and the DNA takes up a highly distorted structure. In the non-specific complex with an unrelated DNA sequence, there are virtually no interactions with the bases and the DNA retains a B-like structure. Since the free energy changes for the formation of specific and non-specific complexes are the same, the energy from the specific interactions balances that required for the distortion of the DNA. The distortion inserts the phosphate at the scissile bond into the active site of the enzyme, where it forms part of the binding site for Mg²⁺. Without this distortion, the EcoRV–DNA complex would be unable to bind Mg²⁺ and thus unable to cleave DNA. The specificity of the EcoRV restriction enzyme is therefore governed, not by DNA binding as such, but by its ability to organize the structure of the DNA to which it is bound.     

Sequence-specific recognition of cancer drug-DNA adducts by HMGB1a repair protein

The efficacy of cancer drugs such as cisplatin (Cp) and oxaliplatin (Ox), which covalently bind to DNA to form drug-DNA adducts, is linked to their recognition by repair proteins such as HMGB1a. Previous experimental studies showed that HMGB1a's binding affinity for Cp- and Ox-DNA varies with the drug used and the local DNA sequence context of the adduct. We link this differential binding affinity to the free energy of deforming (bending and minor groove opening) the drug-DNA molecule during HMGB1a binding. Specifically, the minimal binding affinity of HMGB1a for Ox-DNA in the TGGA context is explained by its larger deformation free energy compared with Cp-DNA or Ox-DNA in other sequence contexts. Methyl groups on neighboring thymine bases in Ox-TGGA crowd the minor groove and sterically hinder the motion of the diaminocyclohexane ring of Ox, leading to this reduced deformability and resultant decrease in HMGB1a's binding affinity.     

Immunoglobulin recognition of synthetic and natural left-handed Z DNA conformations and sequences

The relative immunogenicities of the poly[d(G-C)] and poly[d(A-C) · d(G-T)] families of helices have been determined. The specificities of the resultant immunoglobulins have been characterized for recognition of different synthetic and natural left-handed sequences and conformations. Certain modifications of poly[d(G-C)] in the sugar-phosphate bacbone and cytosine C-5 potentiate the right(R)-to-left(L) (B → Z) transition under physiological conditions. The resulting polynucleotides, poly[d(G^S-C)], poly[d(G-io⁵C)], poly[d(G-br⁵C)] and poly[d(G-m⁵C)], are also highly immunogenic. In contrast, DNAs incapable of assuming the left-handed conformation under physiological salt concentrations are weakly or non-immunogenic. These include unmodified poly[d(G-C)] as well as members of the poly[d(A-C) · d(G-T)] family of sequences bearing pyrimidine C-5 substitutions (methyl, bromo, iodo). These polynucleotides undergo the R → L isomerization under more stringent ionic and thermal conditions.The specificities of purified polyclonal and monoclonal anti-Z DNA immunoglobulins (IgG) were measured by binding to radiolabeled polynucleotides, by electrophoretic analysis of IgG bound to covalent closed circular DNAs, and by immunofluorescent staining of polytene chromosomes. The salt-induced left-handed forms of poly[d(G-C)] and its derivatives (including the cytidine C-5 methyl, bromo, iodo, and N-5 aza substituted polynucleotides) and of the modified poly[d(A-C) · d(G-T)] polymers are bound to varying degrees by different antibodies. The patterns of substrate recognition demonstrate the existence of several antigenic domains in left-handed DNAs, including the helix convex surface and the sugar-phosphate backbone. Substitutions in these regions can produce enhancing (required substitutions), neutral, or inhibitory effects on subsequent IgG binding. Additionally, certain modifications of either the convex surface of Z DNA at the C-5 position of cytidine (i.e. a methyl group) or of the backbone (i.e. phosphorothioate substitution) can lead to polymorphic lefthanded conformations that are compatible with antibody binding when present individually but not in combination. The recognition patterns exhibited with DNA substrates from the two DNA families indicate that some, but not all, IgGs show specificity for different nucleotide sequences.The anti-Z DNA IgGs were used to probe for specific left-handed Z DNA determinants on plasmid (e.g. pBR322) or viral (e.g. simian virus 40 (SV40)) DNAs and on the acid-fixed polytene chromosomes of dipteran larvae. At their extracted superhelical density, the negatively supercoiled form I, but not the relaxed, nicked, or linear forms of all tested plasmid and viral DNAs specifically bind sequence-independent anti-Z IgGs. Dimers, trimers and higher oligomers of form I DNA cross-linked by bivalent anti-Z IgGs are formed with numerous (e.g. φX174, SV40, pBR322) genomes. Their occurrence depends upon IgG concentration and specificity, the conditions of ionic strength and temperatures and the DNA genome. The IgG cross-linked DNA multimers are converted to monomers by dithiothreitol reduction. Sequence-independent monovalent anti-Z Fab fragments bind form I DNA but do not generate oligomeric species. Multimers of order >2 indicate the existence of at least two anti-Z Ig binding sites per molecule, as in the case of SV40. IgGs differ in their ability to form stable complexes with some sites on natural DNAs, presumably due to their sequence and conformation binding specificities. A differential binding of these antibodies is also observed in certain bands of polytene chromosomes, such as the telomeric regions that are involved in chromosome associations.     

Role of sequence evolution and conformational dynamics in the substrate specificity and oligomerization mode of thymidylate kinases

Thymidylate kinase (TMK) is a key enzyme for the synthesis of DNA, making it an important target for the development of anticancer, antibacterial, and antiparasitic drugs. TMK homologs exhibit significant variations in sequence, residue conformation, substrate specificity, and oligomerization mode. However, the influence of sequence evolution and conformational dynamics on its quaternary structure and function has not been studied before. Based on extensive sequence and structure analyses, our study detected several non-conserved residues which are linked by co-evolution and are implicated in the observed variations in flexibility, oligomeric assembly, and substrate specificity among the homologs. These lead to differences in the pattern of interactions at the active site in TMKs of different specificity. The method was further tested on TMK from Sulfolobus tokodaii (StTMK) which has substantial differences in sequence and structure compared to other TMKs. Our analyses pointed to a more flexible dTMP-binding site in StTMK compared to the other homologs. Binding assays proved that the protein can accommodate both purine and pyrimidine nucleotides at the dTMP binding site with comparable affinity. Additionally, the residues responsible for the narrow specificity of Brugia malayi TMK, whose three-dimensional structure is unavailable, were detected. Our study provides a residue-level understanding of the differences observed among TMK homologs in previous experiments. It also illustrates the correlation among sequence evolution, conformational dynamics, oligomerization mode, and substrate recognition in TMKs and detects co-evolving residues that affect binding, which should be taken into account while designing novel inhibitors.     

Influence of DNA stiffness in protein-DNA recognition

Protein-DNA recognition plays an essential role in the regulation of gene expression. The protein-DNA binding specificity is based on direct atomic contacts between protein and DNA and/or the conformational properties of DNA. In this work, we have analyzed the influence of DNA stiffness (E) to the specificity of protein-DNA complexes. The average DNA stiffness parameters for several protein-DNA complexes have been computed using the structure based sequence dependent stiffness scale. The relationship between DNA stiffness and experimental protein-DNA binding specificity has been brought out. We have investigated the importance of DNA stiffness with the aid of experimental free energy changes (DeltaDeltaG) due to binding in several protein-DNA complexes, such as, ETS proteins, 434, lambda, Mnt and trp repressors, 434 cro protein, EcoRV endonuclease V and zinc fingers. We found a correlation in the range 0.65-0.97 between DeltaDeltaG and E in these examples. Further, we have qualitatively analyzed the effect of mutations in the target sequence of lambda repressor and we observed that the DNA stiffness could correctly identify 70% of the correct bases among the considered nine positions.     

Computer simulation of the interaction of α3hclix of λ Cro protein with DNA and origin of sequence specific recognition

The conformation of the α3 helix of Cro protein (residues 27–36) of bacteriophageλ is optimised by the damped least square minimization technique, with the steric constraint that Cα atom positions should match the crystallographic data available to date. On the basis of minimization of total interaction and conformation energy, models for complexes of this peptide sequence with heptanucleotide duplexes from native and altered O_R3 operator are obtained in the major groove of B DNA. Analysis of the energetics for 3 sequences of the DNA show that binding strength is derived mainly from the interaction of side chains of the peptide with DNA. Sequence specificity (maximum difference in binding energy for different DNA sequences) is due to hydrogen bonding interaction. A small amount of sequence specificity is derived from non-bonded interaction also. Stereochemical aspects of peptide DNA interaction and their role in DNA recognition are discussed in this paper.     

Recognition of ATGA sequences by the unfused aromatic dication DB293 forming stacked dimers in the DNA minor groove

Furamidine and related diamidines represent a promising series of drugs active against widespread parasites, in particular the Pneumocystic carinii pathogen. In this series, the phenylfuranbenzimidazole diamidine derivative DB293 was recently identified as the first unfused aromatic dication capable of forming stacked dimers in the DNA minor groove of GC-containing sequences. Here we present a detailed biochemical and biophysical characterization of the DNA sequence recognition properties of DB293. Three complementary footprinting techniques using DNase I, Fe(II)-EDTA, and an anthraquinone photonuclease were employed to locate binding sites for DB293 in different DNA restriction fragments. Two categories of sites were identified by DNase I footprinting: (i) 4/5 bp sequences containing contiguous A.T pairs, such as 5'-AAAA and 5'-ATTA; and (ii) sequences including the motif 5'-ATGA.5'-TCAT. In particular, a 13-bp sequence including two contiguous ATGA motifs provided a highly preferential recognition site for DB293. Quantitative footprinting analysis revealed better occupancy of the 5'-ATGA site compared to the AT-rich sites. Preferential binding of DB293 to ATGA sites was also observed with other DNA fragments and was confirmed independently by means of hydroxyl radical footprinting generated by the Fe(II)-EDTA system, as well as by a photofootprinting approach using the probe anthraquinone-2-sulfonate (AQS). In addition, this photosensitive reagent revealed the presence of sites of enhanced cutting specific to DB293. This molecule, but not other minor groove binders such as netropsin, induces specific local structural changes in DNA near certain binding sites, as independently shown by DNase I and the AQS probe. Recognition of the ATGA sequence by DB293 was investigated further using melting temperature experiments and surface plasmon resonance (SPR). The use of different hairpin oligonucleotides showed that DB293 can interact with AT sites via the formation of 1:1 drug-DNA complexes but binds much more strongly, and cooperatively, to ATGA-containing sequences to form 2:1 drug-DNA complexes. DB293 binds strongly to ATGA sequences with no significant context dependence but is highly sensitive to the orientation of the target sequence. The formation of 2:1 DB293/DNA complexes is abolished by reversing the sequence 5'-ATGA-->3'-ATGA, indicating that directionality plays an important role in the drug-DNA recognition process. Similarly, a single mutation in the A[T-->G]GA sequence is very detrimental to the dimer interactions of DB293. From the complementary footprinting and SPR data, the 5'-ATGA sequence is identified as being a highly favored dimer binding site for DB293. The data provide clues for delineating a recognition code for diamidine-type minor groove binding agents, and ultimately to guide the rational design of gene regulatory molecules targeted to specific sites of the genetic material.     

Backbone dynamics of sequence specific recognition and binding by the yeast Pho4 bHLH domain probed by NMR

Backbone dynamics of the basic/helix-loop-helix domain of Pho4 from Saccharomyces cerevisae have been probed by NMR techniques, in the absence of DNA, nonspecifically bound to DNA and bound to cognate DNA. Alpha proton chemical shift indices and nuclear Overhauser effect patterns were used to elucidate the secondary structure in these states. These secondary structures are compared to the co-crystal complex of Pho4 bound to a cognate DNA sequence (Shimizu T. Toumoto A, Ihara K, Shimizu M, Kyogou Y, Ogawa N, Oshima Y, Hakoshima T, 1997, EMBO J 15: 4689-4697). The dynamic information provides insight into the nature of this DNA binding domain as it progresses from free in solution to a specifically bound DNA complex. Relative to the unbound form, we show that formation of either the nonspecific and cognate DNA bound complexes involves a large change in conformation and backbone dynamics of the basic region. The nonspecific and cognate complexes, however, have nearly identical secondary structure and backbone dynamics. We also present evidence for conformational flexibility at a highly conserved glutamate basic region residue. These results are discussed in relation to the mechanism of sequence specific recognition and binding.     

A Structural Portrait of the PDZ Domain Family

PDZ (PSD-95/Discs-large/ZO1) domains are interaction modules that typically bind to specific C-terminal sequences of partner proteins and assemble signaling complexes in multicellular organisms. We have analyzed the existing database of PDZ domain structures in the context of a specificity tree based on binding specificities defined by peptide-phage binding selections. We have identified 16 structures of PDZ domains in complex with high-affinity ligands and have elucidated four additional structures to assemble a structural database that covers most of the branches of the PDZ specificity tree. A detailed comparison of the structures reveals features that are responsible for the diverse specificities across the PDZ domain family. Specificity differences can be explained by differences in PDZ residues that are in contact with the peptide ligands, but these contacts involve both side-chain and main-chain interactions. Most PDZ domains bind peptides in a canonical conformation in which the ligand main chain adopts an extended β-strand conformation by interacting in an antiparallel fashion with a PDZ β-strand. However, a subset of PDZ domains bind peptides with a bent main-chain conformation and the specificities of these non-canonical domains could not be explained based on canonical structures. Our analysis provides a structural portrait of the PDZ domain family, which serves as a guide in understanding the structural basis for the diverse specificities across the family.     

A new method for evaluating the specificity of indirect readout in protein-DNA recognition

Proteins recognize a specific DNA sequence not only through direct contact (direct readout) with base pairs but also through sequence-dependent conformation and/or flexibility of DNA (indirect readout). However, it is difficult to assess the contribution of indirect readout to the sequence specificity. What is needed is a straightforward method for quantifying its contributions to specificity. Using Bayesian statistics, we derived the probability of a particular sequence for a given DNA structure from the trajectories of molecular dynamics (MD) simulations of DNAs containing all possible tetramer sequences. Then, we quantified the specificity of indirect readout based on the information entropy associated with the probability. We tested this method with known structures of protein-DNA complexes. This method enabled us to correctly predict those regions where experiments suggested the involvement of indirect readout. The results also indicated new regions where the indirect readout mechanism makes major contributions to the recognition. The present method can be used to estimate the contribution of indirect readout without approximations to the distributions in the conformational ensembles of DNA, and would serve as a powerful tool to study the mechanism of protein-DNA recognition.