Structure-based secondary structure-independent approach to design protein ligands: Application to the design of Kv1.2 potassium channel blockers.

We have developed a structure-based approach to the design of protein ligands. This approach is based on the transfer of a functional binding motif of amino acids, often referred as to the "hot spot", on a host protein able to reproduce the functional topology of these residues. The scaffolds were identified by a systematic in silico search in the Protein Data Bank for proteins possessing a group of residues in a topology similar to that adopted by the functional motif in a reference ligand of known 3D structure. In contrast to previously reported studies, this search is independent of the particular secondary structure supporting the functional motif. To take into account the global properties of the host protein, two additional criteria were taken into account in the selection process: (1) Only those scaffolds sterically compatible with the positioning of the functional motif as observed in a reference complex model were retained. (2) Host proteins displaying electrostatic potentials, in the region of the transferred functional motif, similar to that of the reference ligand were selected. This approach was applied to the development of protein ligands of the Kv1.2 channel using BgK, a small protein isolated from the sea anemone Bunodosoma granulifera, as the reference ligand. Four proteins obtained by this approach were produced for experimental evaluation. The X-ray structure of one of these proteins was determined to check for similarity of the transferred functional motif with the structure it adopts in the reference ligand. Three of these protein ligands bind the Kv1.2 channel with inhibition constants of 0.5, 1.5, and 1.6 microM. Several mutants of these designed protein ligands gave binding results consistent with the presumed binding mode. These results show that protein ligands can be designed by transferring a binding motif on a protein host selected to reproduce the functional topology of this motif, irrespective to the secondary structure supporting the functional motif, if the host protein possesses steric and electrostatic properties compatible with the binding to the target. This result opens the way to the design of protein ligands by taking advantage of the considerable structural repertoire of the Protein Data Bank.

Artificial evolution of an enzyme active site: structural studies of three highly active mutants of Escherichia coli alkaline phosphatase.

The crystal structure of three mutants of Escherichia coli alkaline phosphatase with catalytic activity (k(cat)) enhancement as compare to the wild-type enzyme is described in different states. The biological aspects of this study have been reported elsewhere. The structure of the first mutant, D330N, which is threefold more active than the wild-type enzyme, was determined with phosphate in the active site, or with aluminium fluoride, which mimics the transition state. These structures reveal, in particular, that this first mutation does not alter the active site. The second mutant, D153H-D330N, is 17-fold more active than the wild-type enzyme and activated by magnesium, but its activity drops after few days. The structure of this mutant was solved under four different conditions. The phosphate-free enzyme was studied in an inactivated form with zinc at site M3, or after activation by magnesium. The comparison of these two forms free of phosphate illustrates the mechanism of the magnesium activation of the catalytic serine residue. In the presence of magnesium, the structure was determined with phosphate, or aluminium fluoride. The drop in activity of the mutant D153H-D330N could be explained by the instability of the metal ion at M3. The analysis of this mutant helped in the design of the third mutant, D153G-D330N. This mutant is up to 40-fold more active than the wild-type enzyme, with a restored robustness of the enzyme stability. The structure is presented here with covalently bound phosphate in the active site, representing the first phosphoseryl intermediate of a highly active alkaline phosphatase. This study shows how structural analysis may help to progress in the improvement of an enzyme catalytic activity (k(cat)), and explains the structural events associated with this artificial evolution.

Highly specific anti-estradiol antibodies: structural characterisation and binding diversity.

Subtle modulation of antibody-binding properties by protein engineering often lies with an accurate structural and energetic description of how an antigen is recognised. Thus, with the intent to increase the affinity and add a bias in favour of natural estradiol compared with its chemically modified immunogen, we have determined the crystal structure of two anti-estradiol monoclonal antibodies, 10G6D6 and 17E12E5. Although generated against the same estradiol derivative, these antibodies share little sequence identity, which is reflected in dissimilar binding pockets and in different positioning of the steroid. In both antibodies the characteristic 17-hydroxyl group is buried deeply at the bottom of hydrophobic pockets and stabilised by hydrogen bonds. Apart from this similarity, the steroid is oriented differently in the respective binding pockets. The high specificity of both antibodies has been mapped out, and even closely related steroids show low cross-reactivity. The structural studies of the complex formed between 10G6D6 and 6-CMO-estradiol have identified contacts between the 6-CMO coupling linker and an arginine residue from the heavy chain CDR2 segment. This segment is now being targeted by random mutagenesis to select mutants with a preference for natural estradiol compared to the branched hapten.

Structural evidence for a functional role of human tissue nonspecific alkaline phosphatase in bone mineralization.

The human tissue nonspecific alkaline phosphatase (TNAP) is found in liver, kidney, and bone. Mutations in the TNAP gene can lead to Hypophosphatasia, a rare inborn disease that is characterized by defective bone mineralization. TNAP is 74% homologous to human placental alkaline phosphatase (PLAP) whose crystal structure has been recently determined at atomic resolution (Le Du, M. H., Stigbrand, T., Taussig, M. J., Ménez, A., and Stura, E. A. (2001) J. Biol. Chem, 276, 9158-9165). The degree of homology allowed us to build a reliable TNAP model to investigate the relationship between mutations associated with hypophosphatasia and their probable consequences on the activity or the structure of the enzyme. The mutations are clustered within five crucial regions, namely the active site and its vicinity, the active site valley, the homodimer interface, the crown domain, and the metal-binding site. The crown domain and the metal-binding domain are mammalian-specific and were observed for the first time in the PLAP structure. The crown domain contains a collagen binding loop. A synchrotron radiation x-ray fluorescence study confirms that the metal in the metal-binding site is a calcium ion. Several severe mutations in TNAP occur around this calcium site, suggesting that calcium may be of critical importance for the TNAP function. The presence of this extra metal-binding site gives new insights on the controversial role observed for calcium.

Crystal structure of alkaline phosphatase from human placenta at 1.8 A resolution. Implication for a substrate specificity.

Human placental alkaline phosphatase (PLAP) is one of three tissue-specific human APs extensively studied because of its ectopic expression in tumors. The crystal structure, determined at 1.8-A resolution, reveals that during evolution, only the overall features of the enzyme have been conserved with respect to Escherichia coli. The surface is deeply mutated with 8% residues in common, and in the active site, only residues strictly necessary to perform the catalysis have been preserved. Additional structural elements aid an understanding of the allosteric property that is specific for the mammalian enzyme (Hoylaerts, M. F., Manes, T., and Millán, J. L. (1997) J. Biol. Chem. 272, 22781-22787). Allostery is probably favored by the quality of the dimer interface, by a long N-terminal alpha-helix from one monomer that embraces the other one, and similarly by the exchange of a residue from one monomer in the active site of the other. In the neighborhood of the catalytic serine, the orientation of Glu-429, a residue unique to PLAP, and the presence of a hydrophobic pocket close to the phosphate product, account for the specific uncompetitive inhibition of PLAP by l-amino acids, consistent with the acquisition of substrate specificity. The location of the active site at the bottom of a large valley flanked by an interfacial crown-shaped domain and a domain containing an extra metal ion on the other side suggest that the substrate of PLAP could be a specific phosphorylated protein.

Improving Escherichia coli alkaline phosphatase efficacy by additional mutations inside and outside the catalytic pocket.

We describe a strategy that allowed us to confer on a bacterial (E. coli) alkaline phosphatase (AP) the high catalytic activity of the mammalian enzyme while maintaining its high thermostability. First, we identified mutations, at positions other than those occupied by essential catalytic residues, which inactivate the bacterial enzyme without destroying its overall conformation. We transferred concomitantly into the bacterial enzyme four residues of the mammalian enzyme, two being in the catalytic pocket and two being outside. Second, the gene encoding the inactive mutant was submitted to random mutagenesis. Enzyme activity was restored upon the single mutation D330N, at a position that is 12 A away from the center of the catalytic pocket. Third, this mutation was combined with other mutations previously reported to increase AP activity slightly in the presence of magnesium. As a result, at pH 10.0 the phosphatase activity of both mutants D330N/D153H and D330N/D153G was 17-fold higher than that of the wild-type AP. Strikingly, although the two individual mutations D153H and D153G destabilize the enzyme, the double mutant D330N/D153G remained highly stable (T(m)=87 degrees C). Moreover, when combining the phosphatase and transferase activities, the catalytic activity of the mutant D330N/D153G increased 40-fold (k(cat)=3200 s-1) relative to that of the wild-type enzyme (k(cat)=80 s-1). Due to the simultaneous increase in K(m), the resulting k(cat)/K(m) value was only increased by a factor of two. Therefore, a single mutation occurring outside a catalytic pocket can dramatically control not only the activity of an enzyme, but also its thermostability. Preliminary crystallographic data of a covalent D330N/D153G enzyme-phosphate complex show that the phosphate group has significantly moved away from the catalytic pocket, relative to its position in the structure of another mutant previously reported.

Observation of an unexpected third receptor molecule in the crystal structure of human interferon-gamma receptor complex.

BACKGROUND: Molecular interactions among cytokines and cytokine receptors form the basis of many cell-signaling pathways relevant to immune function. Interferon-gamma (IFN-gamma) signals through a multimeric receptor complex consisting of two different but structurally related transmembrane chains: the high-affinity receptor-binding subunit (IFN-gammaRalpha) and a species-specific accessory factor (AF-1 or IFN-gammaRbeta). In the signaling complex, the two receptors probably interact with one another through their extracellular domains. Understanding the atomic interactions of signaling complexes enhances the ability to control and alter cell signaling and also provides a greater understanding of basic biochemical processes. RESULTS: The crystal structure of the complex of human IFN-gamma with the soluble, glycosylated extracellular part of IFN-gammaRalpha has been determined at 2.9 A resolution using multiwavelength anomalous diffraction methods. In addition to the expected 2:1 complex, the crystal structure reveals the presence of a third receptor molecule not directly associated with the IFN-gamma dimer. Two distinct intermolecular contacts, involving the edge strands of the C-terminal domains, are observed between this extra receptor and the 2:1 receptor-ligand complex thereby forming a 3:1 complex. CONCLUSIONS: The observed interactions in the 2:1 complex of the high-affinity cell-surface receptor with the IFN-gamma cytokine are similar to those seen in a previously reported structure where the receptor chains were not glycosylated. The formation of beta-sheet packing interactions between pairs of IFN-gammaRalpha receptors in these crystals suggests a possible model for receptor oligomerization of Ralpha and the structurally homologous Rbeta receptors in the fully active IFN-gamma signaling complex.

Do structural deviations between toxins adopting the same fold reflect functional differences?

Three-finger proteins form a structurally related family of compounds that exhibit a great variety of biological properties. To address the question of the prediction of functional areas on their surfaces, we tentatively conferred the acetylcholinesterase inhibitory activity of fasciculins on a short-chain curaremimetic toxin. For this purpose, we assimilated the three-dimensional structure of fasciculin 2 with the one of toxin alpha. This comparison revealed that the tips of the first and second loops, together with the C terminus residue, deviated most. A first recombinant fasciculin/toxin alpha chimera was designed by transferring loop 1 in its entirety together with the tip of loop 2 of fasciculin 2 into the toxin alpha scaffold. A second chimera (rChII) was obtained by adding the point Asn-61 --> Tyr substitution. Comparison of functional and structural properties of both chimeras show that rChII can accommodate the imposed modifications and displays nearly all the acetylcholinesterase-blocking activities of fasciculins. The three-dimensional structure of rChII demonstrates that rChII adopts a typical three-fingered fold with structural features of both parent toxins. Taken together, these results emphasize the great structural flexibility and functional adaptability of that fold and confirm that structural deviations between fasciculins and short-chain neurotoxins do indeed reflect functional diversity.

Stability of a structural scaffold upon activity transfer: X-ray structure of a three fingers chimeric protein.

Fasciculin 2 and toxin alpha proteins belong to the same structural family of three-fingered snake toxins. They act on different targets, but in each case the binding region involves residues from loops I and II. The superimposition of the two structures suggests that these functional regions correspond to structurally distinct zones. Loop I, half of loop II and the C-terminal residue of fasciculin 2 were therefore transferred into the toxin alpha. The inhibition constant of the resulting chimera is only 15-fold lower than that of fasciculin 2, and as expected the potency of binding to the toxin alpha target has been lost. In order to understand the structure-function relationship between the chimera and its "parent" molecules, we solved its structure by X-ray crystallography. The protein crystallized in space group P3(1)21 with a=b=58.5 A, and c=62.3 A. The crystal structure was solved by molecular replacement and refined to 2.1 A resolution. The structure belongs to the three-fingered snake toxin family with a core of four disulphide bridges from which emerge the three loops I, II and III. Superimposition of the chimera on fasciculin 2 or toxin alpha revealed an overall fold intermediate between those of the two parent molecules. The regions corresponding to toxin alpha and to fasciculin 2 retained their respective geometries. In addition, the chimera protein displayed a structural behaviour similar to that of fasciculin 2, i.e. dimerization in the crystal structure of fasciculin 2, and the geometry of the region that binds to acetylcholinesterase. In conclusion, this structure shows that the chimera retains the general structural characteristics of three-fingered toxins, and the structural specificity of the transferred function.

Molecular evolution of snake toxins: is the functional diversity of snake toxins associated with a mechanism of accelerated evolution?

Recent studies revealed that animal toxins with unrelated biological functions often possess a similar architecture. To tentatively understand the evolutionary mechanisms that may govern this principle of functional prodigality associated with a structural economy, two complementary approaches were considered. One of them consisted of investigating the rates of mutations that occur in cDNAs and/or genes that encode a variety of toxins with the same fold. This approach was largely adopted with phospholipases A2 from Viperidae and to a lesser extent with three-fingered toxins from Elapidae and Hydrophiidae. Another approach consisted of investigating how a given fold can accommodate distinct functional topographies. Thus, a number of topologies by which three-fingered toxins exert distinct functions were investigated either by making chemical modifications and/or mutational analyses or by studying the three-dimensional structure of toxin-target complexes. This review shows that, although the two approaches are different, they commonly indicate that most if not all the surface of a snake toxin fold undergoes natural engineering, which may be associated with an accelerated rate of evolution. The biochemical process by which this phenomenon occurs remains unknown.

Structure of fasciculin 2 from green mamba snake venom: evidence for unusual loop flexibility.

The crystal structure of the snake toxin fasciculin 2, a potent acetylcholinesterase inhibitor from the venom of the green mamba (Dendroaspis angusticeps), has been determined by the molecular-replacement method, using the fasciculin 1 model and refined to 2.0 A resolution. The introduction of an overall anisotropic temperature factor improved significantly the quality of the electron-density map. It suggests, as it was also indicated by the packing, that the thermal motion along the unique axis direction is less pronounced than on the (ab) plane. The final crystallographic R factor is 0.188 for a model having r.m.s. deviations from ideality of 0.016 A for bond lengths and 2.01 degrees for bond angles. As fasciculin 1, fasciculin 2 belongs to the three-finger class of Elapidae toxins, a structural group that also contains the alpha-neurotoxins and the cardiotoxins. Although the two fasciculins have, overall, closely related structures, the conformation of loop I differs appreciably in the two molecules. The presence of detergent in crystallization medium in the case of fasciculin 2 appears to be responsible for the displacement of the loop containing Thr9. This conformational change also results in the formation of a crystallographic dimer that displays extensive intermolecular interactions.

Crystal structure of ribonuclease A.d(ApTpApApG) complex. Direct evidence for extended substrate recognition.

The crystal structure of the complex between ribonuclease A and d(ApTpApApG) has been solved by x-ray crystallography using the molecular replacement method. The model includes, besides the enzyme, the d(ApTpApA) 5'-segment (A1T2A3A4) and 68 solvent molecules. The R-factor for the strongest 87% of the measured data that partially extends to 2.3-A resolution is 0.207. The A1 position is well defined; the 5'-O of the deoxyribose establishes a hydrogen bond with a solvent molecule that is, in turn, bonded to the epsilon-amino group of Lys66. The base (B0 site) is well ordered; it interacts with a symmetry-related enzyme molecule. In the crystal, the phosphate group at the p0 site has no direct charge compensation. However, Lys66 is not too far, and, in solution, it could bind to it. The T2 (R1B1p1) site is occupied as in other complex structures, and it is defined by very good electron density. The A3 site shows that the adenine moiety interacts with Asn71 and Gln69 and that the phosphate at p2 forms a salt bridge with Lys7. The most consistent model for the base of A4 (B3), both in terms of electron density and stereochemistry, shows that it forms a hydrogen bond with Gln69 and a g-g- array with the base at B2. The stacking of B2 and B3 may be a general feature of the binding of polyribonucleotides to ribonuclease A. The side chains of Gln69, Asn71, and Glu111 may thus constitute a malleable binding site capable of establishing various hydrogen bonds depending on the nature of the stacked bases. There is no evidence for the 3' G5 site in the electron density map.

1.9-A resolution structure of fasciculin 1, an anti-acetylcholinesterase toxin from green mamba snake venom.

The crystal structure of fasciculin 1, a potent acetylcholinesterase inhibitor from green mamba snake venom, has been solved by the multiple isomorphous replacement method complemented with anomalous scattering and subsequently refined at 1.9-A resolution. The overall structure of fasciculin is similar to those of the short alpha-neurotoxins and cardiotoxins, with a dense core rich in disulfide bridges and three long loops disposed as the central fingers of a hand. A comparison of these three prototypic toxin types shows that fasciculin 1 has structural features that are intermediate between those of the other two molecules. Its core region, which can be defined as a continuous stretch of conserved residues, is very similar to that of erabutoxin b, whereas the orientation of its long loops resembles that of cardiotoxin VII4. This result introduces a new element in the study of phylogenetic relationships of snake toxins and suggests that, after divergency from an ancestral gene, convergent evolution may have played an important factor in the evolution of these proteins. In fasciculin 1, several arginine and lysine residues are well ordered and relatively exposed to the solvent medium and may play a role in the binding to the peripheral site of acetylcholinesterases.

Crystals of fasciculin 2 from green mamba snake venom. Preparation and preliminary x-ray analysis.

Fasciculin 2 from the venom of the green mamba, Dendroaspis angusticeps, has been crystallized. The crystals are tetragonal, with unit cell dimensions a = 48.9 A and c = 82.0 A, space group P 41 21 2 or P 43212. Density measurements and pseudocentering of the hko zone indicate that there are 16 molecules in the unit cell.