Construction of new ligand binding sites in proteins of known structure. I. Computer-aided modeling of sites with pre-defined geometry

Computational Design of a Biologically Active Enzyme

Rational design of enzymes is a stringent test of our understanding of protein chemistry and has numerous potential applications. Here, we present and experimentally validate the computational design of enzyme activity in proteins of known structure. We have predicted mutations that introduce triose phosphate isomerase activity into ribose-binding protein, a receptor that normally lacks enzyme activity. The resulting designs contain 18 to 22 mutations, exhibit 10(5)- to 10(6)-fold rate enhancements over the uncatalyzed reaction, and are biologically active, in that they support the growth of Escherichia coli under gluconeogenic conditions. The inherent generality of the design method suggests that many enzymes can be designed by this approach.

Computational design of a Zn2+ receptor that controls bacterial gene expression

The control of cellular physiology and gene expression in response to extracellular signals is a basic property of living systems. We have constructed a synthetic bacterial signal transduction pathway in which gene expression is controlled by extracellular Zn2+. In this system a computationally designed Zn2+-binding periplasmic receptor senses the extracellular solute and triggers a two-component signal transduction pathway via a chimeric transmembrane protein, resulting in transcriptional up-regulation of a beta-galactosidase reporter gene. The Zn2+-binding site in the designed receptor is based on a four-coordinate, tetrahedral primary coordination sphere consisting of histidines and glutamates. In addition, mutations were introduced in a secondary coordination sphere to satisfy the residual hydrogen-bonding potential of the histidines coordinated to the metal. The importance of the secondary shell interactions is demonstrated by their effect on metal affinity and selectivity, as well as protein stability. Three designed protein sequences, comprising two distinct metal-binding positions, were all shown to bind Zn2+ and to function in the cell-based assay, indicating the generality of the design methodology. These experiments demonstrate that biological systems can be manipulated with computationally designed proteins that have drastically altered ligand-binding specificities, thereby extending the repertoire of genetic control by extracellular signals.

A consensus-binding structure for adenine at the atomic level permits searching for the ligand site in a wide spectrum of adenine-containing complexes

Attempts to derive structural features of ligand-binding sites have traditionally involved seeking commonalities at the residue level. Recently, structural studies have turned to atomic interactions of small molecular fragments to extract common binding-site properties. Here, we explore the use of larger ligand elements to derive a consensus binding structure for the ligand as a whole. We superimposed multiple molecular structures from a nonredundant set of adenosine-5'-triphosphate (ATP) protein complexes, using the adenine moiety as template. Clustered binding-site atoms of compatible atomic classes forming attractive contacts with the adenine probe were extracted. A set of atomic clusters characterizing the adenine binding pocket was then derived. Among the clusters are three vertices representing the interactions of adenine atom N6 with its protein-binding niche. These vertices, together with atom C6 of the purine ring system, complete the set of four vertices for the pyramid-like structure of the N6 anchor atom. Also, the sequence relationship for the adenine-binding loop interacting with the C2-N6 end of the conjugated ring system is expanded to include a third hydrophilic cluster interacting with atom N1. A search procedure involving interatomic distances between cluster centers was formulated and applied to seek putative binding sites in test cases. The results show that a consensus network of clusters, based on an adenine probe and an ATP-complexed training set of proteins, is sufficient to recognize the experimental cavity for adenine in a wide spectrum of ligand-protein complexes.

Designing redox metalloproteins from bottom-up and top-down perspectives

The design of redox-active metalloproteins has been approached from two different directions. The de novo design approach has recently reached an important stage, at which structural information on several different designed metalloproteins has been obtained. This new information highlights the real challenge of this approach. The alternative approach involving re-engineering of evolved proteins has also made significant advances recently.

Geometry of interaction of the histidine ring with other planar and basic residues

Among the aromatic residues in protein structures, histidine (His) is unique, as it can exist in the neutral or positively charged form at the physiological pH. As such, it can interact with other aromatic residues as well as form hydrogen bonds with polar and charged (both negative and positive) residues. We have analyzed the geometry of interaction of His residues with nine other planar side chains containing aromatic (residues Phe, Tyr, Trp, and His), carboxylate (Asp and Glu), carboxamide (Asn and Gln) and guanidinium (Arg) groups in 432 polypeptide chains. With the exception of the aspartic (Asp) and glutamic (Glu) acid side-chains, all other residues prefer to interact in a face-to-face or offset-face-stacked orientation with the His ring. Such a geometry is different from the edge-to-face relative orientation normally associated with the aromatic-aromatic interaction. His-His pair prefers to interact in a face-to-face orientation; however, when both the residues bind the same metal ion, the interplanar angle is close to 90 degrees. The occurrence of different interactions (including the nonconventional N-H...pi and C-H...pi hydrogen bonds) have been correlated with the relative orientations between the interacting residues. Several structural motifs, mostly involved in binding metal ions, have been identified by considering the cases where His residues are in contact with four other planar moieties. About 10% of His residues used here are also found in sequence patterns in PROSITE database. There are examples of the amino end of the Lys side chain interacting with His residues in such a way that it is located on an arc around a ring nitrogen atom.

Lanthanide-binding helix-turn-helix peptides: solution structure of a designed metallonuclease

A designed lanthanide-binding chimeric peptide based on the strikingly similar geometries of the EF-hand and helix-turn-helix (HTH) motifs was investigated by NMR and CD spectroscopy and found to retain the same overall solution structure of the parental motifs. CD spectroscopy showed that the 33-mer peptide P3W folds on binding lanthanides, with an increase in alpha-helicity from 20% in the absence of metal to 38% and 35% in the presence of excess Eu(III) and La(III) ions, respectively. The conditional binding affinities of P3W for La(III) (5.9 +/- 0.3 microM) and for Eu(III) (6.2 +/- 0.3 microM) (pH 7.8, 5 mM Tris) were determined by tryptophan fluorescence titration. The La(III) complex of peptide P3, which differs from P3W by only one Trp-to-His substitution, has much less signal dispersion in the proton NMR spectra than LaP3W, indicating that the Trp residue is a critical hydrophobic anchor for maintaining a well-folded helix-turn-helix structure. A chemical-shift index analysis indicates the metallopeptide has a helix-loop-helix secondary structure. A structure calculated by using nuclear Overhauser effect and other NMR constraints reveals that P3W not only has a tightly folded metal-binding loop but also retains the alpha-alpha corner supersecondary structure of the parental motifs. Although the solution structure is undefined at both the N and C termini, the NMR structure confirms the successful incorporation of a metal-binding loop into a HTH sequence.

Introduction and characterization of a functionally linked metal ion binding site at the exposed heme edge of myoglobin

A binding site for metal ions has been created on the surface of horse heart myoglobin (Mb) near the heme 6-propionate group by replacing K45 and K63 with glutamyl residues. One-dimensional (1)H NMR spectroscopy indicates that Mn(2+) binds in the vicinity of the heme 6-propionate as anticipated, and potentiometric titrations establish that the affinity of the new site for Mn(2+) is 1.28(4) x 10(4) M(-1) (pH 6.96, ionic strength I = 17.2 microM, 25 degrees C). In addition, these substitutions lower the reduction potential of the protein and increase the pK(a) for the water molecule coordinated to the heme iron of metmyoglobin. The peroxidase [2,2'-azinobis(3-ethylbenzthiazoline-6-sulfonic acid), ABTS, as substrate] and the Mn(2+)-peroxidase activity of the variant are both increased approximately 3-fold. In contrast to wild-type Mb, both the affinity for azide and the midpoint potential of the variant are significantly influenced by the addition of Mn(2+). The structure of the variant has been determined by x-ray crystallography to define the coordination environment of bound Mn(2+) and Cd(2+). Although slight differences are observed between the geometry of the binding of the two metal ions, both are hexacoordinate, and neither involves coordination by E63.

Construction of a fluorescent biosensor family

Bacterial periplasmic binding proteins (bPBPs) are specific for a wide variety of small molecule ligands. bPBPs undergo a large, ligand-mediated conformational change that can be linked to reporter functions to monitor ligand concentrations. This mechanism provides the basis of a general system for engineering families of reagentless biosensors that share a common physical signal transduction functionality and detect many different analytes. We demonstrate the facility of designing optical biosensors based on fluorophore conjugates using 8 environmentally sensitive fluorophores and 11 bPBPs specific for diverse ligands, including sugars, amino acids, anions, cations, and dipeptides. Construction of reagentless fluorescent biosensors relies on identification of sites that undergo a local conformational change in concert with the global, ligand-mediated hinge-bending motion. Construction of cysteine mutations at these locations then permits site-specific coupling of environmentally sensitive fluorophores that report ligand binding as changes in fluorescence intensity. For 10 of the bPBPs presented in this study, the three-dimensional receptor structure was used to predict the location of reporter sites. In one case, a bPBP sensor specific for glutamic and aspartic acid was designed starting from genome sequence information and illustrates the potential for discovering novel binding functions in the microbial genosphere using bioinformatics.

Control of a biomolecular motor-powered nanodevice with an engineered chemical switch

The biophysical and biochemical properties of motor proteins have been well-studied, but these motors also show promise as mechanical components in hybrid nano-engineered systems. The cytoplasmic F(1) fragment of the adenosine triphosphate synthase (F1-ATPase) can function as an ATP-fuelled rotary motor and has been integrated into self-assembled nanomechanical systems as a mechanical actuator. Here we present the rational design, construction and analysis of a mutant F1-ATPase motor containing a metal-binding site that functions as a zinc-dependent, reversible on/off switch. Repeated cycles of zinc addition and removal by chelation result in inhibition and restoration, respectively, of both ATP hydrolysis and motor rotation of the mutant, but not of the wild-type F1 fragment. These results demonstrate the ability to engineer chemical regulation into a biomolecular motor and represent a critical step towards controlling integrated nanomechanical devices at the single-molecule level.

Computational de novo design, and characterization of an A(2)B(2) diiron protein

Diiron proteins are found throughout nature and have a diverse range of functions; proteins in this class include methane monooxygenase, ribonucleotide reductase, Delta(9)-acyl carrier protein desaturase, rubrerythrin, hemerythrin, and the ferritins. Although each of these proteins has a very different overall fold, in every case the diiron active site is situated within a four-helix bundle. Additionally, nearly all of these proteins have a conserved Glu-Xxx-Xxx-His motif on two of the four helices with the Glu and His residues ligating the iron atoms. Intriguingly, subtle differences in the active site can result in a wide variety of functions. To probe the structural basis for this diversity, we designed an A(2)B(2) heterotetrameric four-helix bundle with an active site similar to those found in the naturally occurring diiron proteins. A novel computational approach was developed for the design, which considers the energy of not only the desired fold but also alternatively folded structures. Circular dichroism spectroscopy, analytical ultracentrifugation, and thermal unfolding studies indicate that the A and B peptides specifically associate to form an A(2)B(2) heterotetramer. Further, the protein binds Zn(II) and Co(II) in the expected manner and shows ferroxidase activity under single turnover conditions.

Metal-binding studies for a de novo designed calcium-binding protein

To understand the key determinants in calcium-binding affinity, a calcium-binding site with pentagonal bipyramid geometry was designed into a non-calcium-binding protein, domain 1 of CD2. This metal-binding protein has five mutations with a net charge in the coordination sphere of -5 and is termed DEEEE. Fluorescence resonance energy transfer was used to determine the metal-binding affinity of DEEEE to the calcium analog terbium. The addition of protein concentration to Tb(III) solution results in a large enhancement of Tb(III) fluorescence due to energy transfer between terbium ions and aromatic residues in CD2-D1. In addition, both calcium and lanthanum compete with terbium for the same desired metal binding pocket. Our designed protein exhibits a stronger affinity for Tb(III), with a K(d) of 21 microM, than natural calcium-binding proteins with a similar Greek key scaffold.

Structural analysis, identification, and design of calcium-binding sites in proteins

Assigning proteins with functions based on the 3-D structure requires high-speed techniques to make a systematic survey of protein structures. Calcium regulates many biological systems by binding numerous proteins in different biological environments. Despite the great diversity in the composition of ligand residues and bond angles and lengths of calcium-binding sites, our structural analysis of 11 calcium-binding sites in different classes of proteins has shown that common local structural parameters can be used to identify and design calcium-binding proteins. Natural calcium-binding sites in both EF-hand proteins and non-EF-hand proteins can be described with the smallest deviation from the geometry of an ideal pentagonal bipyramid. Further, two different magnesium-binding sites in parvalbumin and calbindin(D9K) can also be identified using an octahedral geometry. Using the established method, we have designed de novo calcium-binding sites into the scaffold of non-calcium-binding proteins CD2 and Rop. Our results suggest that it is possible to identify calcium- and magnesium-binding sites in proteins and design de novo metal-binding sites.

De novo design of biocatalysts

The challenging field of de novo enzyme design is beginning to produce exciting results. The application of powerful computational methods to functional protein design has recently succeeded at engineering target activities. In addition, efforts in directed evolution continue to expand the transformations that can be accomplished by existing enzymes. The engineering of completely novel catalytic activity requires traversing inactive sequence space in a fitness landscape, a feat that is better suited to computational design. Optimizing activity, which can include subtle alterations in backbone conformation and protein motion, is better suited to directed evolution, which is highly effective at scaling fitness landscapes towards maxima. Improved rational design efforts coupled with directed evolution should dramatically improve the scope of de novo enzyme design.

Enzyme-like proteins by computational design

We report the development and initial experimental validation of a computational design procedure aimed at generating enzyme-like protein catalysts called "protozymes." Our design approach utilizes a "compute and build" strategy that is based on the physical/chemical principles governing protein stability and catalytic mechanism. By using the catalytically inert 108-residue Escherichia coli thioredoxin as a scaffold, the histidine-mediated nucleophilic hydrolysis of p-nitrophenyl acetate as a model reaction, and the ORBIT protein design software to compute sequences, an active site scan identified two promising catalytic positions and surrounding active-site mutations required for substrate binding. Experimentally, both candidate protozymes demonstrated catalytic activity significantly above background. One of the proteins, PZD2, displayed "burst" phase kinetics at high substrate concentrations, consistent with the formation of a stable enzyme intermediate. The kinetic parameters of PZD2 are comparable to early catalytic Abs. But, unlike catalytic Ab design, our design procedure is independent of fold, suggesting a possible mechanism for examining the relationships between protein fold and the evolvability of protein function.

The crystal structure of human protein farnesyltransferase reveals the basis for inhibition by CaaX tetrapeptides and their mimetics

Protein farnesyltransferase (FTase) catalyzes the attachment of a farnesyl lipid group to the cysteine residue located in the C-terminal tetrapeptide of many essential signal transduction proteins, including members of the Ras superfamily. Farnesylation is essential both for normal functioning of these proteins, and for the transforming activity of oncogenic mutants. Consequently FTase is an important target for anti-cancer therapeutics. Several FTase inhibitors are currently undergoing clinical trials for cancer treatment. Here, we present the crystal structure of human FTase, as well as ternary complexes with the TKCVFM hexapeptide substrate, CVFM non-substrate tetrapeptide, and L-739,750 peptidomimetic with either farnesyl diphosphate (FPP), or a nonreactive analogue. These structures reveal the structural mechanism of FTase inhibition. Some CaaX tetrapeptide inhibitors are not farnesylated, and are more effective inhibitors than farnesylated CaaX tetrapeptides. CVFM and L-739,750 are not farnesylated, because these inhibitors bind in a conformation that is distinct from the TKCVFM hexapeptide substrate. This non-substrate binding mode is stabilized by an ion pair between the peptide N terminus and the alpha-phosphate of the FPP substrate. Conformational mapping calculations reveal the basis for the sequence specificity in the third position of the CaaX motif that determines whether a tetrapeptide is a substrate or non-substrate. The presence of beta-branched amino acids in this position prevents formation of the non-substrate conformation; all other aliphatic amino acids in this position are predicted to form the non-substrate conformation, provided their N terminus is available to bind to the FPP alpha-phosphate. These results may facilitate further development of FTase inhibitors.

Review: protein design--where we were, where we are, where we're going

Protein design has become a powerful approach for understanding the relationship between amino acid sequence and 3-dimensional structure. In the past 5 years, there have been many breakthroughs in the development of computational methods that allow the selection of novel sequences given the structure of a protein backbone. Successful design of protein scaffolds has now paved the way for new endeavors to design function. The ability to design sequences compatible with a fold may also be useful in structural and functional genomics by expanding the range of proteins used for fold recognition and for the identification of functionally important domains from multiple sequence alignments.

Conversion of a maltose receptor into a zinc biosensor by computational design

We have demonstrated that it is possible to radically change the specificity of maltose binding protein by converting it into a zinc sensor using a rational design approach. In this new molecular sensor, zinc binding is transduced into a readily detected fluorescence signal by use of an engineered conformational coupling mechanism linking ligand binding to reporter group response. An iterative progressive design strategy led to the construction of variants with increased zinc affinity by combining binding sites, optimizing the primary coordination sphere, and exploiting conformational equilibria. Intermediates in the design series show that the adaptive process involves both introduction and optimization of new functions and removal of adverse vestigial interactions. The latter demonstrates the importance of the rational design approach in uncovering cryptic phenomena in protein function, which cannot be revealed by the study of naturally evolved systems.

Construction of protein binding sites in scaffold structures

We have developed a strategy for grafting a protein-protein interface based on the known crystal structure of a native ligand and receptor proteins in a complex. The key interaction residues at the ligand protein binding interface are transferred onto a scaffold protein so that the mutated scaffold protein will bind the receptor protein in the same manner as the ligand protein. First, our method identifies key residues and atoms in the ligand protein, which strongly interact with the receptor protein. Second, this method searches the scaffold protein for combinations of candidate residues, among which the distance between any two candidate residues is similar to that between relevant key interaction residues in the ligand protein. These candidate residues are mutated to key interaction residues in the ligand protein respectively. The scaffold protein is superposed onto the ligand protein based upon the coordinates of corresponding atoms, which are assumed to strongly interact with the receptor protein. Complementarity between scaffold and receptor proteins is evaluated. Scaffold proteins with a low superposing rms difference and high complementary score are accepted for further analysis. Then, the relative position of the scaffold protein is adjusted so that the interfaces between the scaffold and receptor proteins have a reasonable packing density. Other mutations are also considered to reduce the desolvation energy or bad steric contacts. Finally, the scaffold protein is cominimized with the receptor protein and evaluated. To test the method, the binding interface of barstar, the inhibitor of barnase, was grafted onto small proteins. Four scaffold proteins with high complementary scores are accepted.

Tailoring new enzyme functions by rational redesign

Site-directed mutagenesis is still a very efficient strategy to elaborate improved enzymes. Recently, advances have been made in developing rational strategies aimed at reshaping enzyme specificities and mechanisms, and at engineering biocatalysts through molecular assembling. These knowledge-based studies greatly benefit from the most recent computational analyses of enzyme structures and functions. The combination of rational and combinatorial methods opens up new vistas in the design of stable and efficient enzymes.

De novo design and structural characterization of proteins and metalloproteins

De novo protein design has recently emerged as an attractive approach for studying the structure and function of proteins. This approach critically tests our understanding of the principles of protein folding; only in de novo design must one truly confront the issue of how to specify a protein's fold and function. If we truly understand proteins, it should be possible to design receptors, enzymes, and ion channels from scratch. Further, as this understanding evolves and is further refined, it should be possible to design proteins and biomimetic polymers with properties unprecedented in nature.

De novo design of heterotrimeric coiled coils

The three-helix bundle is a common structural motif among natural proteins. It has been observed in numerous important proteins, such as fibrinogen, laminin, spectrin, dystrofin, hemagglutinin, and mannose binding proteins. The three-helix bundle is a simple structure in which three alpha-helices pack against each other, with a slight left-handed twist. Because of its simplicity relative to other structural motifs, the three-helix bundle can be conveniently used both to clarify the forces responsible for the protein folding and stability, and for the design of novel proteins. In this paper we describe the design, synthesis, and characterization of three peptides that self-assemble into antiparallel, heterotrimeric coiled coils. The experimental results, obtained from CD spectroscopy and ultracentrifugation equilibrium sedimentation, indicate that the mixture of the three peptides preferentially forms heterotrimers; moreover, these aggregates represent attractive systems for combinatorial design of libraries of pseudo C3 symmetric ligands or binding sites.

Retrostructural analysis of metalloproteins: application to the design of a minimal model for diiron proteins

De novo protein design provides an attractive approach for the construction of models to probe the features required for function of complex metalloproteins. The metal-binding sites of many metalloproteins lie between multiple elements of secondary structure, inviting a retrostructural approach to constructing minimal models of their active sites. The backbone geometries comprising the metal-binding sites of zinc fingers, diiron proteins, and rubredoxins may be described to within approximately 1 A rms deviation by using a simple geometric model with only six adjustable parameters. These geometric models provide excellent starting points for the design of metalloproteins, as illustrated in the construction of Due Ferro 1 (DF1), a minimal model for the Glu-Xxx-Xxx-His class of dinuclear metalloproteins. This protein was synthesized and structurally characterized as the di-Zn(II) complex by x-ray crystallography, by using data that extend to 2.5 A. This four-helix bundle protein is comprised of two noncovalently associated helix-loop-helix motifs. The dinuclear center is formed by two bridging Glu and two chelating Glu side chains, as well as two monodentate His ligands. The primary ligands are mostly buried in the protein interior, and their geometries are stabilized by a network of hydrogen bonds to second-shell ligands. In particular, a Tyr residue forms a hydrogen bond to a chelating Glu ligand, similar to a motif found in the diiron-containing R2 subunit of Escherichia coli ribonucleotide reductase and the ferritins. DF1 also binds cobalt and iron ions and should provide an attractive model for a variety of diiron proteins that use oxygen for processes including iron storage, radical formation, and hydrocarbon oxidation.

Rational design of nascent metalloenzymes

Understanding the early genesis of new enzymatic functions is one of the challenges in protein design, mechanistic enzymology, and molecular evolution. We have experimentally mimicked starting points in this process by introducing primitive iron and oxygen binding sites at various locations in thioredoxin, a small protein lacking metal centers, by using computational design. These rudimentary active sites show emerging enzymatic activities that select to varying degrees between different oxygen chemistries. Even within these nascent enzymes, mechanisms by which different reactions are controlled can be discerned. These involve both stabilizing and destabilizing interactions imposed on the metal center by the surrounding protein matrix.

Grafting of protein-protein interaction epitope

Transferring the biological function of one protein to another is a key issue in understanding the structure and function relationship of proteins. We have developed a strategy for grafting protein-protein interaction epitopes. As a first step, residues at the interface of the ligand protein which strongly interact with the receptor protein were identified. Then protein scaffolds were docked onto receptor protein based on geometric complementarity. Only high docking score matches were saved. For each saved match, the scaffold protein was accepted if it had suitable positions for grafting key interaction residues of the ligand protein. These candidate residues were mutated to corresponding residues in the ligand protein at each relevant position and the mutated scaffold protein was co-minimized with receptor protein. Finally, the minimized complexes were evaluated by a scoring function deduced from statistical analysis of rigid binding data sets. As a test case, the binding epitope of barstar, the inhibitor of barnase, was grafted onto smaller proteins. Pheromone Er-1 (PDB entry 1erc) has been found to be a good scaffold. The calculated binding free energy for mutated Pheromone Er-1 is equivalent to that of barstar.

From genes to protein structure and function: novel applications of computational approaches in the genomic era

The genome-sequencing projects are providing a detailed 'parts list' of life. A key to comprehending this list is understanding the function of each gene and each protein at various levels. Sequence-based methods for function prediction are inadequate because of the multifunctional nature of proteins. However, just knowing the structure of the protein is also insufficient for prediction of multiple functional sites. Structural descriptors for protein functional sites are crucial for unlocking the secrets in both the sequence and structural-genomics projects.

Use of a quantitative structure-property relationship to design larger model proteins that fold rapidly

A quantitative structure-property relationship (QSPR) was used to design model protein sequences that fold repeatedly and relatively rapidly to stable target structures. The specific model was a 125-residue heteropolymer chain subject to Monte Carlo dynamics on a simple cubic lattice. The QSPR was derived from an analysis of a database of 200 sequences by a statistical method that uses a genetic algorithm to select the sequence attributes that are most important for folding and a neural network to determine the corresponding functional dependence of folding ability on the chosen attributes. The QSPR depends on the number of anti-parallel sheet contacts, the energy gap between the native state and quasi-continuous part of the spectrum and the total energy of the contacts between surface residues. Two Monte Carlo procedures were used in series to optimize both the target structures and the sequences. We generated 20 fully optimized sequences and 60 partially optimized control sequences and tested each for its ability to fold in dynamic MC simulations. Although sequences in which either the number of anti-parallel sheet contacts or the energy of the surface residues is non-optimal are capable of folding almost as well as fully optimized ones, sequences in which only the energy gap is optimized fold markedly more slowly. Implications of the results for the design of proteins are discussed.

Branch-and-terminate: a combinatorial optimization algorithm for protein design

BACKGROUND

Several deterministic and stochastic combinatorial optimization algorithms have been applied to computational protein design and homology modeling. As structural targets increase in size, however, it has become necessary to find more powerful methods to address the increased combinatorial complexity.

RESULTS

We present a new deterministic combinatorial search algorithm called 'Branch-and-Terminate' (B&T), which is derived from the Branch-and-Bound search method. The B&T approach is based on the construction of an efficient but very restrictive bounding expression, which is used for the search of a combinatorial tree representing the protein system. The bounding expression is used both to determine the optimal organization of the tree and to perform a highly effective pruning procedure named 'termination'. For some calculations, the B&T method rivals the current deterministic standard, dead-end elimination (DEE), sometimes finding the solution up to 21 times faster. A more significant feature of the B&T algorithm is that it can provide an efficient way to complete the optimization of problems that have been partially reduced by a DEE algorithm.

CONCLUSIONS

The B&T algorithm is an effective optimization algorithm when used alone. Moreover, it can increase the problem size limit of amino acid sidechain placement calculations, such as protein design, by completing DEE optimizations that reach a point at which the DEE criteria become inefficient. Together the two algorithms make it possible to find solutions to problems that are intractable by either algorithm alone.

Histidine placement in de novo-designed heme proteins

The effects of histidine residue placement in a de novo-designed four-alpha-helix bundle are investigated by placement of histidine residues at coiled coil heptad a positions in two distinct heptads and at each position within a single heptad repeat of our prototype heme protein maquette, [H10H24]2 [[Ac-CGGGELWKL x HEELLKK x FEELLKL x HEERLKK x L-CONH2]2]2 composed of a generic (alpha-SS-alpha)2 peptide architecture. The heme to peptide stoichiometry of variants of [H10H24]2 with either or both histidines on each helix replaced with noncoordinating alanine residues ([H10A24]2, [A10H24]2, and [A10A24]2) demonstrates the obligate requirement of histidine for biologically significant heme affinity. Variants of [A10A24]2, [[Ac-CGGGELWKL x AEELLKK x FEELLKL x AEERLKK x L-CONH2]2]2, containing a single histidine per helix in positions 9 to 15 were evaluated to verify the design based on molecular modeling. The bis-histidine site formed between heptad positions a at 10 and 10' bound ferric hemes with the highest affinity, Kd1 and Kd2 values of 1.5 and 800 nM, respectively. Placement of histidine at position 11 (heptad position b) resulted in a protein that bound a single heme with moderate affinity, Kd1 of 9.5 microM, whereas the other peptides had no measurable apparent affinity for ferric heme with Kd1 values >200 microM. The bis-histidine ligation of heme to [H10A24]2 and [H11A24]2 was confirmed by electron paramagnetic resonance spectroscopy. The protein design rules derived from this study, together with the narrow tolerances revealed, are applicable for improving future heme protein designs, for analyzing the results of randomized heme protein combinatorial libraries, as well as for implementation in automated protein design.

Knowledge-based modeling of the serine protease triad into non-proteases

The Asp-His-Ser triad of serine proteases has been regarded, in the present study, as an independent catalytic motif, because in nature it has been incorporated at the active sites of enzymes as diverse as the serine proteases and the lipases. Incorporating this motif into non-protease scaffolds, by rational design and mutagenesis, might lead to the generation of novel catalysts. As an aid to such experiments, a knowledge-based computer modeling procedure has been developed to model the protease Asp-His-Ser triad into non-proteases. Catalytic triads from a set of trypsin family proteases have been analyzed and criteria that characterize the geometry of the triads have been obtained. Using these criteria, the modeling procedure first identifies sites in non-proteases that are suitable for modeling the protease triad. H-bonded Asp-His-Ser triads, that mimic the protease catalytic triad in geometry, are then modeled in at these sites, provided it is stereochemically possible to do so. Thus non-protease sites at which H-bonded Asp-His-Ser triads are successfully modeled in may be considered for mutagenesis experiments that aim at introducing the protease triad into non-proteases. The triad modeling procedure has been used to identify sites for introducing the protease triad in three binding proteins and an immunoglobulin. A scoring function, depending on inter-residue distances, solvent accessibility and the substitution potential of amino acid residues at the modeling sites in the host proteins, has been used to assess the quality of the model triads.

Hydrophobic core packing and protein design

Over the past few years, we have witnessed exciting advances in protein design. Several groups have reported success in the design of hydrophobic cores, and the principles developed in these studies have been recently applied to the full sequence design of a small protein motif and the design of a catalytically active metal center. These successes suggest that designing large, functional proteins in computero is more feasible than ever before.

Functional analysis of the Escherichia coli genome using the sequence-to-structure-to-function paradigm: identification of proteins exhibiting the glutaredoxin/thioredoxin disulfide oxidoreductase activity

The application of an automated method for the screening of protein activity based on the sequence-to-structure-to-function paradigm is presented for the complete Escherichia coli genome. First, the structure of the protein is identified from its sequence using a threading algorithm, which aligns the sequences to the best matching structure in a structural database and extends sequence analysis well beyond the limits of local sequence identity. Then, the active site is identified in the resulting sequence-to-structure alignment using a "fuzzy functional form" (FFF), a three-dimensional descriptor of the active site of a protein. Here, this sequence-to-structure-to-function concept is applied to analysis of the complete E. coli genome, i.e. all E. coli open reading frames (ORFs) are screened for the thiol-disulfide oxidoreductase activity of the glutaredoxin/thioredoxin protein family. We show that the method can identify the active sites in ten sequences that are known to or proposed to exhibit this activity. Furthermore, oxidoreductase activity is predicted in two other sequences that have not been identified previously. This method distinguishes protein pairs with similar active sites from proteins pairs that are just topological cousins, i.e. those having similar global folds, but not necessarily similar active sites. Thus, this method provides a novel approach for extraction of active site and functional information based on three-dimensional structures, rather than simple sequence analysis. Prediction of protein activity is fully automated and easily extendible to new functions. Finally, it is demonstrated here that the method can be applied to complete genome database analysis.

Method for prediction of protein function from sequence using the sequence-to-structure-to-function paradigm with application to glutaredoxins/thioredoxins and T1 ribonucleases

The practical exploitation of the vast numbers of sequences in the genome sequence databases is crucially dependent on the ability to identify the function of each sequence. Unfortunately, current methods, including global sequence alignment and local sequence motif identification, are limited by the extent of sequence similarity between sequences of unknown and known function; these methods increasingly fail as the sequence identity diverges into and beyond the twilight zone of sequence identity. To address this problem, a novel method for identification of protein function based directly on the sequence-to-structure-to-function paradigm is described. Descriptors of protein active sites, termed "fuzzy functional forms" or FFFs, are created based on the geometry and conformation of the active site. By way of illustration, the active sites responsible for the disulfide oxidoreductase activity of the glutaredoxin/thioredoxin family and the RNA hydrolytic activity of the T1 ribonuclease family are presented. First, the FFFs are shown to correctly identify their corresponding active sites in a library of exact protein models produced by crystallography or NMR spectroscopy, most of which lack the specified activity. Next, these FFFs are used to screen for active sites in low-to-moderate resolution models produced by ab initio folding or threading prediction algorithms. Again, the FFFs can specifically identify the functional sites of these proteins from their predicted structures. The results demonstrate that low-to-moderate resolution models as produced by state-of-the-art tertiary structure prediction algorithms are sufficient to identify protein active sites. Prediction of a novel function for the gamma subunit of a yeast glycosyl transferase and prediction of the function of two hypothetical yeast proteins whose models were produced via threading are presented. This work suggests a means for the large-scale functional screening of genomic sequence databases based on the prediction of structure from sequence, then on the identification of functional active sites in the predicted structure.

The de novo design of a rubredoxin-like Fe site

A redox center similar to that of rubredoxin was designed into the 56 amino acid immunoglobulin binding B1 domain of Streptococcals protein G. The redox center in rubredoxin contains an iron ion tetrahedrally coordinated by four cysteine residues, [Fe(S-Cys)4](-1),(-2). The design criteria for the target site included taking backbone movements into account, tetrahedral metal-binding, and maintaining the structure and stability of the wild-type protein. The optical absorption spectrum of the Co(II) complex of the metal-binding variant is characteristic of tetrahedral chelation by four cysteine residues. Circular dichroism and nuclear magnetic resonance measurements reveal that the metal-free and Cd(II)-bound forms of the variant are folded correctly and are stable. The Fe(III) complex of the metal-binding mutant reproduces the optical and the electron paramagnetic resonance spectra of oxidized rubredoxin. This demonstrates that the engineered protein chelates Fe(III) in a tetrahedral array, and the resulting center is similar to that of oxidized rubredoxin.

Analysis of zinc binding sites in protein crystal structures

The geometrical properties of zinc binding sites in a dataset of high quality protein crystal structures deposited in the Protein Data Bank have been examined to identify important differences between zinc sites that are directly involved in catalysis and those that play a structural role. Coordination angles in the zinc primary coordination sphere are compared with ideal values for each coordination geometry, and zinc coordination distances are compared with those in small zinc complexes from the Cambridge Structural Database as a guide of expected trends. We find that distances and angles in the primary coordination sphere are in general close to the expected (or ideal) values. Deviations occur primarily for oxygen coordinating atoms and are found to be mainly due to H-bonding of the oxygen coordinating ligand to protein residues, bidentate binding arrangements, and multi-zinc sites. We find that H-bonding of oxygen containing residues (or water) to zinc bound histidines is almost universal in our dataset and defines the elec-His-Zn motif. Analysis of the stereochemistry shows that carboxyl elec-His-Zn motifs are geometrically rigid, while water elec-His-Zn motifs show the most geometrical variation. As catalytic motifs have a higher proportion of carboxyl elec atoms than structural motifs, they provide a more rigid framework for zinc binding. This is understood biologically, as a small distortion in the zinc position in an enzyme can have serious consequences on the enzymatic reaction. We also analyze the sequence pattern of the zinc ligands and residues that provide elecs, and identify conserved hydrophobic residues in the endopeptidases that also appear to contribute to stabilizing the catalytic zinc site. A zinc binding template in protein crystal structures is derived from these observations.

The development of new biotechnologies using metalloprotein design

A variety of methodologies are under development to alter the behavior of existing metal centers or create entirely new sites within a protein framework in order to exploit the intrinsic chemical versatility of metals using the exquisite level of control that a protein matrix can exert to modulate their reactivity. Even at this relatively early stage, engineering of metal centers has led to the development of a number of emerging technologies with a wide variety of applications, including affinity purification of proteins, engineering of metal-mediated protein stability, control of protein activity, imaging and therapy, biosensors, and new catalysts.

Computer search algorithms in protein modification and design

The computer-aided design of protein sequences requires efficient search algorithms to handle the enormous combinatorial complexity involved. A variety of different algorithms have now been applied with some success. The choice of algorithm can influence the representation of the problem in several important ways--the discreteness of the configuration, the types of energy terms that can be used and the ability to find the global minimum energy configuration. The use of dead end elimination to design the complete sequence for a small protein motif and the use of genetic and mean-field algorithms to design hydrophobic cores for proteins represent the major themes of the past year.

Hybrid enzymes: manipulating enzyme design

Hybrid enzymes are engineered to contain elements of two or more enzymes. Hybrid-enzyme approaches, by taking advantage of the vast array of enzymatic properties that nature has evolved, as well as the strategies that nature has used to evolve them, are becoming an increasingly important avenue for obtaining novel enzymes with desired activities and properties.

The construction of metal centers in proteins by rational design

Metalloprotein properties result from the interplay between coordination requirements of the metal center, protein stability, and modulation of the metal center by the surrounding protein matrix. Simple metal centers, which exercise control over the protein by affecting stability or enzyme activity, have been created by rational design. Complex centers, which require control by the protein matrix, have also been constructed.

The rational design and construction of a cuboidal iron-sulfur protein

Rational protein design is an emerging approach for testing general theories of protein chemistry through the creation of new structures and functions. Here we present the first successful introduction by rational design of a [Fe4S4] cuboidal cluster into the hydrophobic core of Escherichia coli thioredoxin, a protein normally devoid of metal centers. Cuboidal [Fe4S4] is one of the stable forms of self-assembled iron-sulfur clusters that are thought to represent some of the earliest evolved biological redox centers. [Fe4S4] clusters have been recruited for use in a variety of proteins whose functions are central to many of the major biochemical processes ranging from simple soluble electron-transfer agents, to membrane-bound components of electron-transfer chains, to electron reservoirs in complex metalloenzymes such as nitrogenase. By situating an [Fe4S4] cluster into a protein environment not previously adapted by evolution we can explore the factors by which their activity is modulated by the protein matrix.

Construction of a catalytically active iron superoxide dismutase by rational protein design

The rational protein design algorithm DEZYMER was used to introduce the active site of nonheme iron superoxide dismutase (SOD) into the hydrophobic interior of the host protein, Escherichia coli thioredoxin (Trx), a protein that does not naturally contain a transition metal-binding site. Reconstitution of the designed protein, Trx-SOD, showed the incorporation of one high-affinity metal-binding site. The electronic spectra of the holoprotein and its N3- and F- adducts are analogous to those previously reported for native {Fe3+}SOD. Activity assays showed that {Fe3+}Trx-SOD is capable of catalyzing the dismutation of the superoxide anion; comparative studies with the unrelated wild-type E. coli iron SOD indicated that {Fe3+}Trx-SOD catalyzes the dismutation reaction at a rate on the order of 10(5) M-1s -1. The ability to design catalytically competent metalloenzymes allows for the systematic investigation of fundamental mechanistic questions concerning catalysis at transition metal centers.

The rational design of allosteric interactions in a monomeric protein and its applications to the construction of biosensors

Rational protein design is an emerging approach for testing general theories of structure and function. The ability to manipulate function rationally also offers the possibility of creating new proteins of biotechnological value. Here we use the design approach to test the current understanding of the structural principles of allosteric interactions in proteins and demonstrate how a simple allosteric system can form the basis for the construction of a generic biosensor molecular engineering system. We have identified regions in Escherichia coli maltose-binding protein that are predicted to be allosterically linked to its maltose-binding site. Environmentally sensitive fluorophores were covalently attached to unique thiols introduced by cysteine mutations at specific sites within these regions. The fluorescence of such conjugates changes cooperatively with respect to maltose binding, as predicted. Spatial separation of the binding site and reporter groups allows the intrinsic properties of each to be manipulated independently. Provided allosteric linkage is maintained, ligand binding can therefore be altered without affecting transduction of the binding event by fluorescence. To demonstrate applicability to biosensor technology, we have introduced a series of point mutations in the maltose-binding site that lower the affinity of the protein for its ligand. These mutant proteins have been combined in a composite biosensor capable of measuring substrate concentration within 5% accuracy over a concentration range spanning five orders of magnitude.

Metalloprotein design

The rational design of novel proteins offers a new method of studying structure and function, and makes possible the construction of new biomaterials. The richness of metal chemistry, the relative ease of creating stable complexes, and the remarkable degree of subtle, highly specific control of reactivity imposed by the protein matrix upon the metal center make metalloprotein design a very fruitful area for the exploration and application of design techniques. So far, most designs have concentrated on the exploration of simple metal-chelation properties. Even so, this has led to the development of new methods for protein stabilization and affinity purification, of metal biosensors, of novel strategies for control of protein activity, and of model systems for the exploration of fundamental principles of molecular recognition.