Mutations in the zinc fingers of ADR1 that change the specificity of DNA binding and transactivation

Prediction of DNA-binding proteins from relational features

Background

The process of protein-DNA binding has an essential role in the biological processing of genetic information. We use relational machine learning to predict DNA-binding propensity of proteins from their structures. Automatically discovered structural features are able to capture some characteristic spatial configurations of amino acids in proteins.

Results

Prediction based only on structural relational features already achieves competitive results to existing methods based on physicochemical properties on several protein datasets. Predictive performance is further improved when structural features are combined with physicochemical features. Moreover, the structural features provide some insights not revealed by physicochemical features. Our method is able to detect common spatial substructures. We demonstrate this in experiments with zinc finger proteins.

Conclusions

We introduced a novel approach for DNA-binding propensity prediction using relational machine learning which could potentially be used also for protein function prediction in general.

Expression, purification and characterization of rat zinc finger protein Mipu1 in Escherichia coli

The novel gene Mipu1 was recently identified in rat due to its up-regulation in response to myocardial ischemia preconditioning. We previously demonstrated that Mipu1 was a nuclear protein and a transcriptional repressor. In this study, Mipu1 was expressed in E. coli and purified using a recombinant expression system and a purification protocol. Milligram quantities of highly purified Mipu1 were obtained. The purified protein was characterized using western blotting, size exclusion chromatography and EMSA. The Mipu1 protein was also used to generate antiserum in rabbits, which was used to detect the expression of Mipu1 protein under normal and stress conditions, by western blotting.

NMR chemical shift perturbation mapping of DNA binding by a zinc-finger domain from the yeast transcription factor ADR1

Mutagenesis studies have revealed that the minimal DNA-binding domain of the yeast transcription factor ADR1 consists of two Cys2-His2 zinc fingers plus an additional 20 residues proximal and N-terminal to the fingers. We have assigned NMR 1H, 15N, and 13C chemical shifts for the entire minimal DNA-binding domain of ADR1 both free and bound to specific DNA. 1H chemical shift values suggest little structural difference between the zinc fingers in this construct and in single-finger constructs, and 13C alpha chemical shift index analysis indicates little change in finger structure upon DNA binding. 1H chemical shift perturbations upon DNA binding are observed, however, and these are mapped to define the protein-DNA interface. The two zinc fingers appear to bind DNA with different orientations, as the entire helix of finger 1 is perturbed, while only the extreme N-terminus of the finger 2 helix is affected. Furthermore, residues N-terminal to the first finger undergo large chemical shift changes upon DNA binding suggesting a role at the protein-DNA interface. A striking correspondence is observed between the protein-DNA interface mapped by chemical shift changes and that previously mapped by mutagenesis.

Recognition of diverse sequences by class I zinc fingers: asymmetries and indirect effects on specificity in the interaction between CF2II and A+T-rich elements

The Drosophila CF2II protein, which contains zinc fingers of the Cys2His2 type and recognizes an A+T-rich sequence, behaves in cell culture as an activator of a reporter chloramphenicol acetyltransferase gene. This activity depends on C-terminal but not N-terminal zinc fingers, as does in vitro DNA binding. By site-specific mutagenesis and binding site selection, we define the critical amino acid-base interactions. Mutations of single amino acid residues at the leading edge of the recognition helix are rarely neutral: many result in a slight change in affinity for the ideal DNA target site; some cause major loss of affinity; and others change specificity for as many as two bases in the target site. Compared to zinc fingers that recognize G+C-rich DNA, CF2II fingers appear to bind to A+T-rich DNA in a generally similar manner, but with additional flexibility and amino acid-base interactions. The results illustrate how zinc fingers may be evolving to recognize an unusually diverse set of DNA sequences.

DNA binding sites for the transcriptional activator/repressor YY1

YY1 is ubiquitously expressed zinc finger DNA binding protein. It can act as a transcriptional repressor or activator and, when binding at the initiator element, as a component of the basal transcription complex. Binding sites for YY1 have been reported in a wide variety of promoters and they exhibit substantial diversity in their sequence. To better understand how YY1 interacts with DNA and to be able to predict the presence of YY1 sites in a more comprehensive fashion, we have selected YY1 binding sites from a random pool of oligonucleotides. The sites display considerable heterogeneity, but contain a conserved 5'-CAT-3' core flanked by variable regions, generating the consensus 5'-(C/g/a)(G/t)(C/t/a)CATN(T/a)(T/g/c)-3', where the upper case letters represent the preferred base. This high degree of flexibility in DNA recognition can be predicted by modeling the interaction of the four YY1 zinc fingers with DNA and a detailed model for this interaction is presented and discussed.

DNA-binding specificity of NGFI-A and related zinc finger transcription factors

NGFI-A is the prototypic member of a family of immediate-early gene-encoded transcription factors which includes NGFI-C, Egr3, and Krox20. These proteins possess highly homologous DNA-binding domains, composed of three Cys2-His2 zinc fingers, and all bind to and activate transcription from the sequence GCGGGGGCG. We used a PCR-mediated random site selection protocol to determine whether other sites could be bound by these proteins and the extent to which their binding site preferences are similar or different. The high-affinity consensus sites generated from the selection data are similar, and the combined consensus sequence is T-G-C-G-T/g-G/A-G-G-C/a/t-G-G/T (lowercase letters indicate bases selected less frequently). Using gel shift assays, we found that sequences that diverge from the consensus were bound by NGFI-A, confirming that there is greater variability in binding sites than has generally been acknowledged. We also provide evidence that protein-DNA interactions not noted, or whose importance was not apparent from the X-ray cocrystal structure of the NGFI-A zinc fingers complexed with DNA, contribute significantly to the binding energy of these proteins and confirm that an optimal site is at least 10 instead of 9 nucleotides in length. In contrast to the similarities in binding specificity among these proteins we found that while NGFI-A, Egr3, and Krox20 have comparable DNA binding affinities and kinetics of dissociation, the affinity of NGFI-C is more than threefold lower. This could result in differential regulation of target genes in cells where NGFI-C and the other proteins are coexpressed. Furthermore, we show that this affinity difference is a property not of the zinc fingers themselves but rather of the protein context of the DNA-binding domain.

Distinct residues of human p53 implicated in binding to DNA, simian virus 40 large T antigen, 53BP1, and 53BP2

We identified a minimal domain of human p53 required for the transactivation of a p53 response element in Saccharomyces cerevisiae. This domain contains the central region of p53 sufficient for specific DNA binding, which colocalizes with the region responsible for binding simian virus 40 large T antigen, 53BP1, and 53BP2. Thirty amino acid positions, including natural mutational hot spots (R175, R213, R248, R249, and R273), in the minimal DNA-binding domain were mutated by alanine substitution. Alanine substitutions at positions R213, R248, R249, D281, R282, R283, E286, and N288 affected transactivation but allowed binding to at least one of the three interacting proteins; these amino acids may be involved in amino acid-base pair contacts. Surprisingly, alanine substitution at the mutational hot spot R175 did not affect DNA binding, transactivation, or T-antigen binding, although it nearly eliminated binding to 53BP1 and 53BP2. Mutation of H168 significantly affected only T-antigen binding, and mutation of E285 affected only 53BP1 binding. Thus, we implicate specific residues of p53 in different DNA and protein interactions.

Distinct residues of human p53 implicated in binding to DNA, simian virus 40 large T antigen, 53BP1, and 53BP2

We identified a minimal domain of human p53 required for the transactivation of a p53 response element in Saccharomyces cerevisiae. This domain contains the central region of p53 sufficient for specific DNA binding, which colocalizes with the region responsible for binding simian virus 40 large T antigen, 53BP1, and 53BP2. Thirty amino acid positions, including natural mutational hot spots (R175, R213, R248, R249, and R273), in the minimal DNA-binding domain were mutated by alanine substitution. Alanine substitutions at positions R213, R248, R249, D281, R282, R283, E286, and N288 affected transactivation but allowed binding to at least one of the three interacting proteins; these amino acids may be involved in amino acid-base pair contacts. Surprisingly, alanine substitution at the mutational hot spot R175 did not affect DNA binding, transactivation, or T-antigen binding, although it nearly eliminated binding to 53BP1 and 53BP2. Mutation of H168 significantly affected only T-antigen binding, and mutation of E285 affected only 53BP1 binding. Thus, we implicate specific residues of p53 in different DNA and protein interactions.

Identification of potential target genes for Adr1p through characterization of essential nucleotides in UAS1

Adr1p is a regulatory protein in the yeast Saccharomyces cerevisiae that binds to and activates transcription from two sites in a perfect 22-bp inverted repeat, UAS1, in the ADH2 promoter. Binding requires two C2H2 zinc fingers and a region amino terminal to the fingers. The importance for DNA binding of each position within UAS1 was deduced from two types of assays. Both methods led to an identical consensus sequence containing only four essential base pairs: GG(A/G)G. The preferred sequence, TTGG(A/G)GA, is found in both halves of the inverted repeat. The region of Adr1p amino terminal to the fingers is important for phosphate contacts in the central region of UAS1. However, no base-specific contacts in this portion of UAS1 are important for DNA binding or for ADR1-dependent transcription in vivo. When the central 6 bp were deleted, only a single monomer of Adr1p was able to bind in vitro and activation in vivo was severely reduced. On the basis of these results and previous knowledge about the DNA binding site requirements, including constraints on the spacing and orientation of sites that affect activation in vivo, a consensus binding site for Adr1p was derived. By using this consensus site, potential Adr1p binding sites were located in the promoters of genes known to show ADR1-dependent expression. In addition, this consensus allowed the identification of new potential target genes for Adr1p.

Identification of potential target genes for Adr1p through characterization of essential nucleotides in UAS1

Adr1p is a regulatory protein in the yeast Saccharomyces cerevisiae that binds to and activates transcription from two sites in a perfect 22-bp inverted repeat, UAS1, in the ADH2 promoter. Binding requires two C2H2 zinc fingers and a region amino terminal to the fingers. The importance for DNA binding of each position within UAS1 was deduced from two types of assays. Both methods led to an identical consensus sequence containing only four essential base pairs: GG(A/G)G. The preferred sequence, TTGG(A/G)GA, is found in both halves of the inverted repeat. The region of Adr1p amino terminal to the fingers is important for phosphate contacts in the central region of UAS1. However, no base-specific contacts in this portion of UAS1 are important for DNA binding or for ADR1-dependent transcription in vivo. When the central 6 bp were deleted, only a single monomer of Adr1p was able to bind in vitro and activation in vivo was severely reduced. On the basis of these results and previous knowledge about the DNA binding site requirements, including constraints on the spacing and orientation of sites that affect activation in vivo, a consensus binding site for Adr1p was derived. By using this consensus site, potential Adr1p binding sites were located in the promoters of genes known to show ADR1-dependent expression. In addition, this consensus allowed the identification of new potential target genes for Adr1p.

Importance of a flanking AT-rich region in target site recognition by the GC box-binding zinc finger protein MIG1

MIG1 is a zinc finger protein that mediates glucose repression in the yeast Saccharomyces cerevisiae. MIG1 is related to the mammalian Krox/Egr, Wilms' tumor, and Sp1 finger proteins. It has two fingers and binds to a GCGGGG motif that resembles the GC boxes recognized by these mammalian proteins. We have performed a complete saturation mutagenesis of a natural MIG1 site in order to elucidate its binding specificity. We found that only three mutations within the GC box retain the ability to bind MIG1: G1 to C, C2 to T, and G5 to A. This result is consistent with current models for zinc finger-DNA binding, which assume that the sequence specificity is determined by base triplet recognition within the GC box. Surprisingly, we found that an AT-rich region 5' to the GC box also is important for MIG1 binding. This AT box is present in all natural MIG1 sites, and it is protected by MIG1 in DNase I footprints. However, the AT box differs from the GC box in that no single base within it is essential for binding. Instead, the AT-rich nature of this sequence seems to be crucial. The fact that AT-rich sequences are known to increase DNA flexibility prompted us to test whether MIG1 bends DNA. We found that binding of MIG1 is associated with bending within the AT box. We conclude that DNA binding by a simple zinc finger protein such as MIG1 can involve both recognition of the GC box and flanking sequence preferences that may reflect local DNA bendability.

Importance of a flanking AT-rich region in target site recognition by the GC box-binding zinc finger protein MIG1

MIG1 is a zinc finger protein that mediates glucose repression in the yeast Saccharomyces cerevisiae. MIG1 is related to the mammalian Krox/Egr, Wilms' tumor, and Sp1 finger proteins. It has two fingers and binds to a GCGGGG motif that resembles the GC boxes recognized by these mammalian proteins. We have performed a complete saturation mutagenesis of a natural MIG1 site in order to elucidate its binding specificity. We found that only three mutations within the GC box retain the ability to bind MIG1: G1 to C, C2 to T, and G5 to A. This result is consistent with current models for zinc finger-DNA binding, which assume that the sequence specificity is determined by base triplet recognition within the GC box. Surprisingly, we found that an AT-rich region 5' to the GC box also is important for MIG1 binding. This AT box is present in all natural MIG1 sites, and it is protected by MIG1 in DNase I footprints. However, the AT box differs from the GC box in that no single base within it is essential for binding. Instead, the AT-rich nature of this sequence seems to be crucial. The fact that AT-rich sequences are known to increase DNA flexibility prompted us to test whether MIG1 bends DNA. We found that binding of MIG1 is associated with bending within the AT box. We conclude that DNA binding by a simple zinc finger protein such as MIG1 can involve both recognition of the GC box and flanking sequence preferences that may reflect local DNA bendability.

Structures of DNA-binding mutant zinc finger domains: implications for DNA binding

Studies of Cys2-His2 zinc finger domains have revealed that the structures of individual finger domains in solution determined by NMR spectroscopy are strikingly similar to the structure of fingers bound to DNA determined by X-ray diffraction. Therefore, detailed structural analyses of single finger domains that contain amino acid substitutions known to affect DNA binding in the whole protein can yield information concerning the structural ramifications of such mutations. We have used this approach to study two mutants in the N-terminal finger domain of ADR1, a yeast transcription factor that contains two Cys2-His2 zinc finger sequences spanning residues 102-159. Two point mutants at position 118 in the N-terminal zinc finger (ADR1b: 102-130) that adversely affect the DNA-binding activity of ADR1 have previously been identified: H118A and H118Y. The structures of wild-type ADR1b and the two mutant zinc finger domains were determined using two-dimensional nuclear magnetic resonance spectroscopy and distance geometry and were refined using a complete relaxation matrix method approach (REPENT) to improve agreement between the models and the nuclear Overhauser effect spectroscopy data from which they were generated. The molecular architecture of the refined wild-type ADR1b domain is presented in detail. Comparisons of wild-type ADR1b and the two mutants revealed that neither mutation causes a significant structural perturbation. The structures indicate that the DNA binding properties of the His 118 mutants are dependent on the identity of the side chain at position 118, which has been postulated to make a direct DNA contact in the wild-type ADR1 protein. The results suggest that the identity of the side chain at the middle DNA contact position in Cys2-His2 zinc fingers may be changed with impunity regarding the domain structure and can affect the affinity of the protein-DNA interaction.

Fungal catabolic gene regulation: molecular genetic analysis of the amdS gene of Aspergillus nidulans

Aspergillus nidulans is an excellent experimental organism for the study of gene regulation. Genetic and molecular analyses of trans-acting and cis-acting mutations have revealed a complex pattern of regulation involving multiple independent controls. Expression of the amdS gene is regulated by the facB and amdA genes which encode positively acting regulatory proteins mediating a major and a minor form of acetate induction respectively. The product of the amdR gene mediates omega amino acid induction of amdS. The binding sites for each of these proteins have been localised through amdS cis-acting mutations which specifically affect the interaction with the regulatory protein. The global controls of nitrogen metabolite repression and carbon catabolite repression regulate the expression of many catabolic genes, including amdS. Nitrogen control is exerted through the positively acting areA gene product and carbon control is dependent on the creA gene product. Each of the characterized regulatory genes encodes a DNA-binding protein which recognises particular sequences in the amdS promoter to activate or repress gene expression. In addition, there is evidence for other genetically uncharacterized proteins, including a CCAAT-binding complex, which interact with the 5' region of the amdS gene.

Use of a zinc-finger consensus sequence framework and specificity rules to design specific DNA binding proteins

We have designed three zinc-finger proteins with different DNA binding specificities. The design strategy combines a consensus zinc-finger framework sequence with previously characterized recognition regions such that the specificity of each protein is predictable. The first protein consists of three identical zinc fingers, each of which was expected to recognize the subsite GCG. This protein binds specifically to the sequence 5'-GCG-GCG-GCG-3' with a dissociation constant of approximately 11 microM. The second protein has three zinc fingers with different predicted preferred subsites. This protein binds to the predicted recognition site 5'-GGG-GCG-GCT-3' with a dissociation constant of 2 nM. Furthermore, selection experiments indicate that this is the optimal binding site. A permuted version of the second protein was also constructed and shown to preferentially recognize the corresponding permuted site 5'-GGG-GCT-GCG-3' over the non-permuted site. These results indicate that earlier observations on the specificity of zinc fingers can be extended to generalized zinc-finger structures and realize the use of zinc fingers for the design of site-specific DNA binding proteins. This consensus-based design system provides a useful model system with which to study details of zinc-finger-DNA specificity.

A mutation outside the two zinc fingers of ADR1 can suppress defects in either finger

A second-site mutation that restored DNA binding to ADR1 mutants altered at different positions in the two zinc fingers was identified. This mutation (called IS1) was a conservative change of arginine 91 to lysine in a region amino terminal to the two zinc fingers and known from previous experiments to be necessary for DNA binding. IS1 increased binding to the UAS1 sequence two- to sevenfold for various ADR1 mutants and twofold for wild-type ADR1. The change of arginine 91 to glycine decreased binding twofold, suggesting that this arginine is involved in DNA binding in the wild-type protein. The increase in binding by IS1 did not involve protein-protein interactions between the two ADR1 monomers, nor did it require the presence of the sequences flanking UAS1. However, the effect of IS1 was influenced by the sequence of the first finger, suggesting that interactions between the region amino terminal to the fingers and the fingers themselves could exist. A model for the role of the amino-terminal region based on these results and sequence homologies with other DNA-binding motifs is proposed.

A mutation outside the two zinc fingers of ADR1 can suppress defects in either finger

A second-site mutation that restored DNA binding to ADR1 mutants altered at different positions in the two zinc fingers was identified. This mutation (called IS1) was a conservative change of arginine 91 to lysine in a region amino terminal to the two zinc fingers and known from previous experiments to be necessary for DNA binding. IS1 increased binding to the UAS1 sequence two- to sevenfold for various ADR1 mutants and twofold for wild-type ADR1. The change of arginine 91 to glycine decreased binding twofold, suggesting that this arginine is involved in DNA binding in the wild-type protein. The increase in binding by IS1 did not involve protein-protein interactions between the two ADR1 monomers, nor did it require the presence of the sequences flanking UAS1. However, the effect of IS1 was influenced by the sequence of the first finger, suggesting that interactions between the region amino terminal to the fingers and the fingers themselves could exist. A model for the role of the amino-terminal region based on these results and sequence homologies with other DNA-binding motifs is proposed.