Beyond the "recognition code": structures of two Cys2His2 zinc finger/TATA box complexes.

BACKGROUND: Several methods have been developed for creating Cys2His2 zinc finger proteins that recognize novel DNA sequences, and these proteins may have important applications in biological research and gene therapy. In spite of this progress with design/selection methodology, fundamental questions remain about the principles that govern DNA recognition. One hypothesis suggests that recognition can be described by a simple set of rules--essentially a "recognition code"--but careful assessment of this proposal has been difficult because there have been few structural studies of selected zinc finger proteins. RESULTS: We report the high-resolution cocrystal structures of two zinc finger proteins that had been selected (as variants of Zif268) to recognize a eukaryotic TATA box sequence. The overall docking arrangement of the fingers within the major groove of the DNA is similar to that observed in the Zif268 complex. Nevertheless, comparison of Zif268 and the selected variants reveal significant differences in the pattern of side chain-base interactions. The new structures also reveal side chain-side chain interactions (both within and between fingers) that are important in stabilizing the protein-DNA interface and appear to play substantial roles in recognition. CONCLUSIONS: These new structures highlight the surprising complexity of zinc finger-DNA interactions. The diversity of interactions observed at the protein-DNA interface, which is especially striking for proteins that were all derived from Zif268, challenges fundamental concepts about zinc finger-DNA recognition and underscores the difficulty in developing any meaningful recognition code.

Selected peptide extension contacts hydrophobic patch on neighboring zinc finger and mediates dimerization on DNA.

Protein-protein interactions often play a crucial role in stabilizing protein-DNA complexes and thus facilitate site-specific DNA recognition. We have worked to incorporate such protein-protein contacts into our design and selection strategies for short peptide extensions that promote cooperative binding of zinc finger proteins to DNA. We have determined the crystal structure of one of these fusion protein-DNA complexes. The selected peptide extension was found to mediate dimerization by reaching across the dyad axis and contacting a hydrophobic patch on the surface of the zinc finger bound to the adjacent DNA site. The peptide-zinc finger protein interactions observed in this structure are similar to those of some homeodomain heterodimers. We also find that the region of the zinc finger surface contacted by the selected peptide extension corresponds to surfaces that also make key interactions in the zinc finger proteins GLI and SWI5.

Design and selection of novel Cys2His2 zinc finger proteins.

Cys2His2 zinc finger proteins offer a stable and versatile framework for the design of proteins that recognize desired target sites on double-stranded DNA. Individual fingers from these proteins have a simple beta beta alpha structure that folds around a central zinc ion, and tandem sets of fingers can contact neighboring subsites of 3-4 base pairs along the major groove of the DNA. Although there is no simple, general code for zinc finger-DNA recognition, selection strategies have been developed that allow these proteins to be targeted to almost any desired site on double-stranded DNA. The affinity and specificity of these new proteins can also be improved by linking more fingers together or by designing proteins that bind as dimers and thus recognize an extended site. These new proteins can then be modified by adding other domains--for activation or repression of transcription, for DNA cleavage, or for other activities. Such designer transcription factors and other new proteins will have important applications in biomedical research and in gene therapy.

Rearrangement of side-chains in a Zif268 mutant highlights the complexities of zinc finger-DNA recognition.

Structural and biochemical studies of Cys(2)His(2) zinc finger proteins initially led several groups to propose a "recognition code" involving a simple set of rules relating key amino acid residues in the zinc finger protein to bases in its DNA site. One recent study from our group, involving geometric analysis of protein-DNA interactions, has discussed limitations of this idea and has shown how the spatial relationship between the polypeptide backbone and the DNA helps to determine what contacts are possible at any given position in a protein-DNA complex. Here we report a study of a zinc finger variant that highlights yet another source of complexity inherent in protein-DNA recognition. In particular, we find that mutations can cause key side-chains to rearrange at the protein-DNA interface without fundamental changes in the spatial relationship between the polypeptide backbone and the DNA. This is clear from a simple analysis of the binding site preferences and co-crystal structures for the Asp20-->Ala point mutant of Zif268. This point mutation in finger one changes the specificity of the protein from GCG TGG GCG to GCG TGG GC(G/T), and we have solved crystal structures of the D20A mutant bound to both types of sites. The structure of the D20A mutant bound to the GCG site reveals that contacts from key residues in the recognition helix are coupled in complex ways. The structure of the complex with the GCT site also shows an important new water molecule at the protein-DNA interface. These side-chain/side-chain interactions, and resultant changes in hydration at the interface, affect binding specificity in ways that cannot be predicted either from a simple recognition code or from analysis of spatial relationships at the protein-DNA interface. Accurate computer modeling of protein-DNA interfaces remains a challenging problem and will require systematic strategies for modeling side-chain rearrangements and change in hydration.

Geometric analysis and comparison of protein-DNA interfaces: why is there no simple code for recognition?

Structural studies of protein-DNA complexes have shown that there are many distinct families of DNA-binding proteins, and have shown that there is no simple "code" describing side-chain/base interactions. However, systematic analysis and comparison of protein-DNA complexes has been complicated by the diversity of observed contacts, the sheer number of complexes currently available and the absence of any consistent method of comparison that retains detailed structural information about the protein-DNA interface. To address these problems, we have developed geometric methods for characterizing the local structural environment in which particular side-chain/base interactions are observed. In particular, we develop methods for analyzing and comparing spatial relationships at the protein-DNA interface. Our method involves attaching local coordinate systems to the DNA bases and to the C(alpha) atoms of the peptide backbone (these are relatively rigid structural units). We use these tools to consider how the position and orientation of the polypeptide backbone (with respect to the DNA) helps to determine what contacts are possible at any given position in a protein-DNA complex. Here, we focus on base contacts that are made in the major groove, and we use spatial relationships in analyzing: (i) the observed patterns of side-chain/base interactions; (ii) observed helix docking orientations; (iii) family/subfamily relationships among DNA-binding proteins; and (iv) broader questions about evolution, altered specificity mutants and the limits for the design of new DNA-binding proteins. Our analysis, which highlights differences in spatial relationships in different complexes and at different positions in a complex, helps explain why there is no simple, general code for protein-DNA recognition.

DNA recognition by Cys2His2 zinc finger proteins.

Cys2His2 zinc fingers are one of the most common DNA-binding motifs found in eukaryotic transcription factors. These proteins typically contain several fingers that make tandem contacts along the DNA. Each finger has a conserved beta beta alpha structure, and amino acids on the surface of the alpha-helix contact bases in the major groove. This simple, modular structure of zinc finger proteins, and the wide variety of DNA sequences they can recognize, make them an attractive framework for attempts to design novel DNA-binding proteins. Several studies have selected fingers with new specificities, and there clearly are recurring patterns in the observed side chain-base interactions. However, the structural details of recognition are intricate enough that there are no general rules (a "recognition code") that would allow the design of an optimal protein for any desired target site. Construction of multifinger proteins is also complicated by interactions between neighboring fingers and the effect of the intervening linker. This review analyzes DNA recognition by Cys2His2 zinc fingers and summarizes progress in generating proteins with novel specificities from fingers selected by phage display.

Combining structure-based design with phage display to create new Cys(2)His(2) zinc finger dimers.

BACKGROUND: Several strategies have been reported for the design and selection of novel DNA-binding proteins. Most of these studies have used Cys(2)His(2) zinc finger proteins as a framework, and have focused on constructs that bind DNA in a manner similar to Zif268, with neighboring fingers connected by a canonical (Krüppel-type) linker. This linker does not seem ideal for larger constructs because only modest improvements in affinity are observed when more than three fingers are connected in this manner. Two strategies have been described that allow the productive assembly of more than three canonically linked fingers on a DNA site: connecting sets of fingers using linkers (covalent), or assembling sets of fingers using dimerization domains (non-covalent). RESULTS: Using a combination of structure-based design and phage display, we have developed a new dimerization system for Cys(2)His(2) zinc fingers that allows the assembly of more than three fingers on a desired target site. Zinc finger constructs employing this new dimerization system have high affinity and good specificity for their target sites both in vitro and in vivo. Constructs that recognize an asymmetric binding site as heterodimers can be obtained through substitutions in the zinc finger and dimerization regions. CONCLUSIONS: Our modular zinc finger dimerization system allows more than three Cys(2)His(2) zinc fingers to be productively assembled on a DNA-binding site. Dimerization may offer certain advantages over covalent linkage for the recognition of large DNA sequences. Our results also illustrate the power of combining structure-based design with phage display in a strategy that assimilates the best features of each method.

Exploring the role of glutamine 50 in the homeodomain-DNA interface: crystal structure of engrailed (Gln50 --> ala) complex at 2.0 A.

We have determined the crystal structure of a complex containing the engrailed homeodomain Gln50 --> Ala variant (QA50) bound to the wild-type optimal DNA site (TAATTA) at 2.0 A resolution. Biochemical and genetic studies by other groups have suggested that residue 50 is an important determinant of differential DNA-binding specificity among homeodomains (distinguishing among various sites of the general form TAATNN). However, biochemical studies of the QA50 variant had revealed that it binds almost as tightly as the wild-type protein and with only modest changes in specificity. We have now determined the crystal structure of the QA50 variant to help understand the role of residue 50 in site-specific recognition. Our cocrystal structure shows some interesting changes in the water structure at the site of the substitution and shows some changes in the conformations of neighboring side chains. However, the structure, like the QA50 biochemical data, suggests that Gln50 plays a relatively modest role in determining the affinity and specificity of the engrailed homeodomain.

A bacterial two-hybrid selection system for studying protein-DNA and protein-protein interactions.

We have developed a bacterial "two-hybrid" system that readily allows selection from libraries larger than 10(8) in size. Our bacterial system may be used to study either protein-DNA or protein-protein interactions, and it offers a number of potentially significant advantages over existing yeast-based one-hybrid and two-hybrid methods. We tested our system by selecting zinc finger variants (from a large randomized library) that bind tightly and specifically to desired DNA target sites. Our method allows sequence-specific zinc fingers to be isolated in a single selection step, and thus it should be more rapid than phage display strategies that typically require multiple enrichment/amplification cycles. Given the large library sizes our bacterial-based selection system can handle, this method should provide a powerful tool for identifying and optimizing protein-DNA and protein-protein interactions.

Crystal structure of an OCA-B peptide bound to an Oct-1 POU domain/octamer DNA complex: specific recognition of a protein-DNA interface.

We have determined the crystal structure, at 3.2 A, of a ternary complex containing an OCA-B peptide, the Oct-1 POU domain, and an octamer DNA site. The OCA-B peptide binds in the major groove near the center of the octamer site, and its polypeptide backbone forms a pair of hydrogen bonds with the adenine base at position 5 of the octamer DNA. Numerous protein-protein contacts between the OCA-B peptide and the POU domain are also involved in the ternary complex. In particular, the hydrophobic surface from a short alpha-helix of OCA-B helps to stabilize the complex by binding to a hydrophobic pocket on the POU-specific domain. The structure of this ternary complex is consistent with previous biochemical studies and shows how peptide-DNA and peptide-protein contacts from OCA-B provide structural and functional specificity in the regulation of immunoglobulin transcription.

Dimerization of zinc fingers mediated by peptides evolved in vitro from random sequences.

Peptides that mediate dimerization of attached zinc finger DNA-binding domains have been evolved in vitro starting from random sequences. We first used phage display to select dimerization elements from libraries of random 15-residue polypeptides that were fused to the N terminus of the zinc finger domains. We then reoptimized these peptides by sequentially randomizing five-residue blocks (proceeding across the peptide in three steps) and selecting variant peptides that further stabilized the protein-DNA complex. Biochemical experiments confirmed that the selected peptides promote dimerization of the zinc fingers on an appropriate DNA target site. These results demonstrate that dimerization units can be obtained readily from random polypeptide libraries of moderate complexity. Our success reemphasizes the utility of searching random peptide libraries in protein design projects, and the sequences presented here may be useful when designing novel transcription factors.

Binding studies with mutants of Zif268. Contribution of individual side chains to binding affinity and specificity in the Zif268 zinc finger-DNA complex.

The Zif268 zinc finger-DNA complex has served as a model system for understanding how Cys2His2 type zinc fingers recognize DNA. Structural studies of the Zif268-DNA complex revealed that residues at four positions in the alpha helix of each zinc finger play key roles in recognition, but there has been no information about the precise contributions of individual residues. Here we report the results of binding studies involving five mutants of Zif268 that have changes in the base-contacting residues of finger one. These studies let us evaluate the contributions that Arg18 (position -1 of the alpha helix), Asp20 (position 2), Glu21 (position 3), and Arg24 (position 6) make to the overall energy of DNA binding. Our results confirm the important role played by these arginines. By comparing the affinities of the wild type and mutant peptides for various sites, we also prove that Asp20 and Glu21 play important roles in determining binding site specificity.

Basis for recognition of cisplatin-modified DNA by high-mobility-group proteins.

The anticancer activity of cis-diamminedichloroplatinum(II) (cisplatin) arises from its ability to damage DNA, with the major adducts formed being intrastrand d(GpG) and d(ApG) crosslinks. These crosslinks bend and unwind the duplex, and the altered structure attracts high-mobility-group domain (HMG) and other proteins. This binding of HMG-domain proteins to cisplatin-modified DNA has been postulated to mediate the antitumour properties of the drug. Many HMG-domain proteins recognize altered DNA structures such as four-way junctions and cisplatin-modified DNA, but until now the molecular basis for this recognition was unknown. Here we describe mutagenesis, hydroxyl-radical footprinting and X-ray studies that elucidate the structure of a 1:1 cisplatin-modified DNA/HMG-domain complex. Domain A of the structure-specific HMG-domain protein HMG1 binds to the widened minor groove of a 16-base-pair DNA duplex containing a site-specific cis-[Pt(NH3)2[d(GpG)-N7(1),-N7(2)]] adduct. The DNA is strongly kinked at a hydrophobic notch created at the platinum-DNA crosslink and protein binding extends exclusively to the 3' side of the platinated strand. A phenylalanine residue at position 37 intercalates into a hydrophobic notch created at the platinum crosslinked d(GpG) site and binding of the domain is dramatically reduced in a mutant in which alanine is substituted for phenylalanine at this position.

Crystal structure of the human Pax6 paired domain-DNA complex reveals specific roles for the linker region and carboxy-terminal subdomain in DNA binding.

Pax6, a transcription factor containing the bipartite paired DNA-binding domain, has critical roles in development of the eye, nose, pancreas, and central nervous system. The 2.5 A structure of the human Pax6 paired domain with its optimal 26-bp site reveals extensive DNA contacts from the amino-terminal subdomain, the linker region, and the carboxy-terminal subdomain. The Pax6 structure not only confirms the docking arrangement of the amino-terminal subdomain as seen in cocrystals of the Drosophila Prd Pax protein, but also reveals some interesting differences in this region and helps explain the sequence specificity of paired domain-DNA recognition. In addition, this structure gives the first detailed information about how the paired linker region and carboxy-terminal subdomain contact DNA. The extended linker makes minor groove contacts over an 8-bp region, and the carboxy-terminal helix-turn-helix unit makes base contacts in the major groove. The structure and docking arrangement of the carboxy-terminal subdomain of Pax6 is remarkably similar to that of the amino-terminal subdomain, and there is an approximate twofold symmetry axis relating the polypeptide backbones of these two helix-turn-helix units. Our structure of the Pax6 paired domain-DNA complex provides a framework for understanding paired domain-DNA interactions, for analyzing mutations that map in the linker and carboxy-terminal regions of the paired domain, and for modeling protein-protein interactions of the Pax family proteins.

Structural basis of DNA recognition by the heterodimeric cell cycle transcription factor E2F-DP.

The E2F and DP protein families form heterodimeric transcription factors that play a central role in the expression of cell cycle-regulated genes. The crystal structure of an E2F4-DP2-DNA complex shows that the DNA-binding domains of the E2F and DP proteins both have a fold related to the winged-helix DNA-binding motif. Recognition of the central c/gGCGCg/c sequence of the consensus DNA-binding site is symmetric, and amino acids that contact these bases are conserved among all known E2F and DP proteins. The asymmetry in the extended binding site TTTc/gGCGCc/g is associated with an amino-terminal extension of E2F4, in which an arginine binds in the minor groove near the TTT stretch. This arginine is invariant among E2Fs but not present in DPs. E2F4 and DP2 interact through an extensive protein-protein interface, and structural features of this interface suggest it contributes to the preference for heterodimers over homodimers in DNA binding.

Analysis of zinc fingers optimized via phage display: evaluating the utility of a recognition code.

Cys2His2 zinc finger proteins are composed of modular DNA-binding domains and provide an excellent framework for the design and selection of proteins with novel site specificity. Crystal structures of zinc finger-DNA complexes have shown that many Cys2His2 zinc fingers use a conserved docking arrangement that juxtaposes residues at key positions in the "recognition helix" with corresponding base positions in the three to four base-pair subsite. Several groups have proposed that specificity can be explained with a zinc finger-DNA recognition code that correlates specific amino acids at these key positions in the alpha-helix with specific bases in each position of the corresponding subsite. Here, we explore the utility of such a code through detailed studies of zinc finger variants selected via phage display. These proteins provide interesting systems for detailed analysis since they have affinities and specificities for their sites similar to those of naturally occurring DNA-binding proteins. Comparisons are facilitated by the fact that only key DNA-binding residues are varied in each finger while leaving all other regions of the structure unchanged. We study these proteins in detail by (1) selecting their optimal binding sites and comparing these binding sites with sites that might have been predicted from a code; (2) by examining the "evolutionary history" of these proteins during the phage display protocol to look for evidence of context-dependent effects; and (3) by reselecting finger 1 in the presence of the optimized finger 2/finger 3 domains to obtain further data on finger modularity. Our data for optimized fingers and binding sites demonstrate a clear correlation with contacts that would be predicted from a code. However, there are enough examples of context-dependent effects (not explained by any existing code) that selection is the most reliable method for maximizing the affinity and specificity of new zinc finger proteins.

Engrailed homeodomain-DNA complex at 2.2 A resolution: a detailed view of the interface and comparison with other engrailed structures.

We report the 2.2 A resolution structure of the Drosophila engrailed homeodomain bound to its optimal DNA site. The original 2.8 A resolution structure of this complex provided the first detailed three-dimensional view of how homeodomains recognize DNA, and has served as the basis for biochemical studies, structural studies and molecular modeling. Our refined structure confirms the principal conclusions of the original structure, but provides important new details about the recognition interface. Biochemical and NMR studies of other homeodomains had led to the notion that Gln50 was an especially important determinant of specificity. However, our refined structure shows that this side-chain makes no direct hydrogen bonds to the DNA. The structure does reveal an extensive network of ordered water molecules which mediate contacts to several bases and phosphates (including contacts from Gln50), and our model provides a basis for detailed comparison with the structure of an engrailed Q50K altered-specificity variant. Comparing our structure with the crystal structure of the free protein confirms that the N and C termini of the homeodomain become ordered upon DNA-binding. However, we also find that several key DNA contact residues in the recognition helix have the same conformation in the free and bound protein, and that several water molecules also are "preorganized" to contact the DNA. Our structure helps provide a more complete basis for the detailed analysis of homeodomain-DNA interactions.

Comparison of X-ray and NMR structures for the Antennapedia homeodomain-DNA complex.

Homeodomains are one of the key families of eukaryotic DNA-binding motifs and provide an important model system for studying protein-DNA interactions. We have crystallized the Antennapedia homeodomain-DNA complex and solved this structure at 2.4 A resolution. NMR and molecular dynamics studies had implied that this homeodomain achieves specificity through an ensemble of rapidly fluctuating DNA contacts. The crystal structure is in agreement with the underlying NMR data, but our structure reveals a well-defined set of contacts and also reveals the locations and roles of water molecules at the protein-DNA interface. The synthesis of X-ray and NMR studies provides a unified, general model for homeodomain-DNA interactions.

High-resolution structures of variant Zif268-DNA complexes: implications for understanding zinc finger-DNA recognition.

BACKGROUND: Zinc fingers of the Cys2-His2 class comprise one of the largest families of eukaryotic DNA-binding motifs and recognize a diverse set of DNA sequences. These proteins have a relatively simple modular structure and key base contacts are typically made by a few residues from each finger. These features make the zinc finger motif an attractive system for designing novel DNA-binding proteins and for exploring fundamental principles of protein-DNA recognition. RESULTS: Here we report the X-ray crystal structures of zinc finger-DNA complexes involving three variants of Zif268, with multiple changes in the recognition helix of finger one. We describe the structure of each of these three-finger peptides bound to its corresponding target site. To help elucidate the differential basis for site-specific recognition, the structures of four other complexes containing various combinations of these peptides with alternative binding sites have also been determined. CONCLUSIONS: The protein-DNA contacts observed in these complexes reveal the basis for the specificity demonstrated by these Zif268 variants. Many, but not all, of the contacts can be rationalized in terms of a recognition code, but the predictive value of such a code is limited. The structures illustrate how modest changes in the docking arrangement accommodate the new sidechain-base and sidechain-phosphate interactions. Such adaptations help explain the versatility of naturally occurring zinc finger proteins and their utility in design.

Getting a handhold on DNA: design of poly-zinc finger proteins with femtomolar dissociation constants.

Structure-based design was used to link zinc finger peptides and make poly-finger proteins that have dramatically enhanced affinity and specificity. Our studies focused on a fusion in which the three-finger Zif268 peptide was linked to a designed three-finger peptide (designated "NRE") that specifically recognizes a nuclear hormone response element. Gel shift assays indicate that this six-finger peptide, 268//NRE, binds to a composite 18-bp DNA site with a dissociation constant in the femtomolar range. We find that the slightly longer linkers used in this fusion protein provide a dramatic improvement in DNA-binding affinity, working much better than the canonical "TGEKP" linkers that have been used in previous studies. Tissue culture transfection experiments also show that the 268//NRE peptide is an extremely effective repressor, giving 72-fold repression when targeted to a binding site close to the transcription start site. Using this strategy, and linking peptides selected via phage display, should allow the design of novel DNA-binding proteins-with extraordinary affinity and specificity-for use in biological research and gene therapy.