In vitro evolution of the binding specificity of neocarzinostatin,an enediyne-binding chromoprotein

Neocarzinostatin is the most studied member of the enediyne-chromoprotein family, and is clinically used as an antitumoral agent. Neocarzinostatin could be a promising drug delivery vehicle if new binding specificities could be conferred to its protein scaffold. We used in vitro evolution methods to demonstrate that this approach is feasible. We created large libraries containing between 1.7 x 10(8) and 1.4 x 10(9) independent clones, where up to 13 side chains pointing toward the binding crevice were randomly substituted. We then used phage display to select variants that bind to a model ligand (testosterone) which is unrelated to the natural ligand of neocarzinostatin. Several different binders were selected from each library. The corresponding proteins were expressed in Escherichia coli and their affinities and specificities were characterized in detail. K(D) values of about 20 nM were obtained for streptavidin-bound testosterone. The K(D) of selected proteins for free soluble testosterone are between 7 and 55 microM and therefore higher than the K(D) for streptavidin-bound testosterone. The spacer and streptavidin used during selection contributed to the high affinity of the selected binders for the target. Binding studies of 15 different steroids related to testosterone allowed us to determine that C3, 4, 5, 6, and 7 on cycles A and B and the conjugated 3 oxo group of the steroid molecule were essential for molecular recognition. Other testosterone analogues substituted on C1, 2, 9, 11, 15, and 17 were not discriminated from testosterone. These results demonstrate that the binding specificity of this protein family can be extended to compounds that are completely unrelated to the natural enediyne chromophore family. This type of highly expressed, stable proteins with tailored binding properties have a wide potential range of applications.     

Two distinct and independent sites on IL-6 trigger gp 130 dimer formation and signalling.

The helical cytokine interleukin (IL) 6 and its specific binding subunit IL-6R alpha form a 1:1 complex which, by promoting homodimerization of the signalling subunit gp130 on the surface of target cells, triggers intracellular responses. We expressed differently tagged forms of gp130 and used them in solution-phase binding assays to show that the soluble extracellular domains of gp130 undergo dimerization in the absence of membranes. In vitro receptor assembly reactions were also performed in the presence of two sets of IL-6 variants carrying amino acid substitutions in two distinct areas of the cytokine surface (site 2, comprising exposed residues in the A and C helices, and site 3, in the terminal part of the CD loop). The binding affinity to IL-6R alpha of these variants is normal but their biological activity is poor or absent. We demonstrate here that both the site 2 and site 3 IL-6 variants complexed with IL-6R alpha bind a single gp130 molecule but are unable to dimerize it, whereas the combined site 2/3 variants lose the ability to interact with gp130. The binding properties of these variants in vitro, and the result of using a neutralizing monoclonal antibody directed against site 3, lead to the conclusion that gp130 dimer is formed through direct binding at two independent and differently oriented sites on IL-6. Immunoprecipitation experiments further reveal that the fully assembled receptor complex is composed of two IL-6, two IL-6R alpha and two gp130 molecules. We propose here a model representing the IL-6 receptor complex as hexameric, which might be common to other helical cytokines.     

A single amino acid mutation in zebrafish (Danio rerio) liver bile acid-binding protein can change the stoichiometry of ligand binding

In all of the liver bile acid-binding proteins (L-BABPs) studied so far, it has been found that the stoichiometry of binding is of two cholate molecules per internal binding site. In this paper, we describe the expression, purification, crystallization, and three-dimensional structure determination of zebrafish (Danio rerio) L-BABP to 1.5A resolution, which is currently the highest available for a protein of this family. Since we have found that in zebrafish, the stoichiometry of binding in the protein cavity is of only one cholate molecule per wild type L-BABP, we examined the role of two crucial amino acids present in the binding site. Using site-directed mutagenesis, we have prepared, crystallized, and determined the three-dimensional structure of co-crystals of two mutants. The mutant G55R has the same stoichiometry of binding as the wild type protein, whereas the C91T mutant changes the stoichiometry of binding from one to two ligand molecules in the cavity and therefore appears to be more similar to the other members of the L-BABP family. Based on the presence or absence of a single disulfide bridge, it can be postulated that fish should bind a single cholate molecule, whereas amphibians and higher vertebrates should bind two. Isothermal titration calorimetry has also revealed the presence in the wild type protein and the G55R mutant of an additional binding site, different from the first and probably located on the surface of the molecule.     

Structural basis of the highly efficient trapping of the HIV Tat protein by an RNA aptamer

An RNA aptamer containing two binding sites exhibits extremely high affinity to the HIV Tat protein. We have determined the structure of the aptamer complexed with two argininamide molecules. Two adjacent U:A:U base triples were formed, which widens the major groove to make space for the two argininamide molecules. The argininamide molecules bind to the G bases through hydrogen bonds. The binding is stabilized through stacking interactions. The structure of the aptamer complexed with a Tat-derived arginine-rich peptide was also characterized. It was suggested that the aptamer structure is similar for both complexes and that the aptamer interacts with two different arginine residues of the peptide simultaneously at the two binding sites, which could explain the high affinity to Tat.     

Crystal structure of the human leukocyte Fc receptor, Fc gammaRIIa.

Fc gamma receptors bind IgG to initiate cellular responses against pathogens and soluble antigens. We have determined the three-dimensional structure of the extracellular portion of human Fc gammaRIIa to 2.0 A resolution providing a structural basis for the unique functions of the leukocyte FcR family. The receptor is composed of two immunoglobulin domains and arranged to expose the ligand-binding site at one end of domain 2. Using alanine mutants we find that the binding sites for IgG1 and 2 are similar but the relative importance of specific regions on the receptor varies. In crystals, Fc gammaRIIa molecules associate to resemble V(L)V(H) dimers, suggesting that two Fc gammaRIIa molecules could cooperate to bind IgG in an asymmetric manner.     

Tropomodulin binds two tropomyosins: a novel model for actin filament capping

Tropomodulin, a tropomyosin-binding protein, caps the slow-growing (pointed) end of the actin filament regulating its dynamics. Tropomodulin, therefore, is important for determining cell morphology, cell movement, and muscle contraction. For the first time we show that one tropomodulin molecule simultaneously binds two tropomyosin molecules in a cooperative manner. On the basis of the tropomodulin solution structure and predicted secondary structure, we introduced a series of point mutations in regions important for tropomyosin binding and actin capping. Capping activity of these mutants was assayed by measuring actin polymerization using pyrene fluorescence. Using direct methods (circular dichroism and native gel electrophoresis) for detecting tropomodulin/tropomyosin binding, we localized the second tropomyosin-binding site to residues 109-144. Despite previous reports that the second binding site is for erythrocyte tropomyosin only, we found that both short nonmuscle and long muscle alpha-tropomyosins bind there as well, though with different affinities. We propose a model for actin capping where one tropomodulin molecule can bind to two tropomyosin molecules at the pointed end.     

Binding modes of the initiator and inhibitor forms of the replication protein pi to the gamma ori iteron of plasmid R6K

Discerning the interactions between initiator protein and the origin of replication should provide insights into the mechanism of DNA replication initiation. In the gamma origin of plasmid R6K, the Rep protein, pi, is distinctive in that it can bind the seven 22-bp iterons in two forms; pi monomers activate replication, whereas pi dimers act as inhibitors. In this work, we used wild type and variants of the pi protein with altered monomer/dimer ratios to study iteron/pi interactions. High resolution contact mapping was conducted using multiple techniques (missing base contact probing, methylation protection, base modification, and hydroxyl radical footprinting), and the electrophoretic separation of nucleoprotein complexes allowed us to discriminate between contact patterns produced by pi monomers and dimers. We also isolated iteron mutants that affected the binding of pi monomers (only) or both monomers and dimers. The mutational studies and footprinting analyses revealed that, when binding DNA, pi monomers interact with nucleotides spanning the entire length of the iteron. In contrast, pi dimers interact with only the left half of the iteron; however, the retained interactions are strikingly similar to those seen with monomers. These results support a model in which Rep protein dimerization disturbs one of two DNA binding domains important for monomer/iteron interaction; the dimer/iteron interaction utilizes only one DNA binding domain.     

Mutational alterations in the homotetrameric chaperone SecB that implicate the structure as dimer of dimers.

Variant forms of SecB with substitutions of aminoacyl residues in the region from 74 to 80 were analyzed with respect to their ability to bind a physiological ligand, precursor galactose-binding protein, and to their oligomeric states. SecBL75Q and SecBE77K are tetramers with affinity for ligand indistinguishable from that of the wild-type SecB, and thus the export defect exhibited by strains producing these variants must result from an effect on interactions between SecB and other components. SecBF74I is tetrameric but binds ligand with a lower affinity. Substitutions at positions 76, 78, and 80 cause a shift in the equilibrium so that the SecB tetramer dissociates into dimers. We conclude that the tetramer is a dimer of dimers and that the residues Cys76, Val78, and Gln80 must be involved either directly or indirectly in forming the interface between dimers. These variant species are defective in binding ligand; however, because their oligomeric state is altered no conclusion can be drawn concerning the direct role of these residues in ligand binding.     

Analysis of the functional modules of the tRNA 3' endonuclease (tRNase Z)

tRNA 3' processing is one of the essential steps during tRNA maturation. The tRNA 3'-processing endonuclease tRNase Z was only recently isolated, and its functional domains have not been identified so far. We performed an extensive mutational study to identify amino acids and regions involved in dimerization, tRNA binding, and catalytic activity. 29 deletion and point variants of the tRNase Z enzyme were generated. According to the results obtained, variants can be sorted into five different classes. The first class still had wild type activity in all three respects. Members of the second and third class still formed dimers and bound tRNAs but had reduced catalytic activity (class two) or no catalytic activity (class three). The fourth class still formed dimers but did not bind the tRNA and did not process precursors. Since this class still formed dimers, it seems that the amino acids mutated in these variants are important for RNA binding. The fifth class did not have any activity anymore. Several conserved amino acids could be mutated without or with little loss of activity.     

First structural evidence of a specific inhibition of phospholipase A2 by alpha-tocopherol (vitamin E) and its implications in inflammation: crystal structure of the complex formed between phospholipase A2 and alpha-tocopherol at 1.8 A resolution

This is the first structural evidence of alpha-tocopherol (alpha-TP) as a possible candidate against inflammation, as it inhibits phospholipase A2 specifically and effectively. The crystal structure of the complex formed between Vipera russelli phospholipase A2 and alpha-tocopherol has been determined and refined to a resolution of 1.8 A. The structure contains two molecules, A and B, of phospholipase A2 in the asymmetric unit, together with one alpha-tocopherol molecule, which is bound specifically to one of them. The phospholipase A2 molecules interact extensively with each other in the crystalline state. The two molecules were found in a stable association in the solution state as well, thus indicating their inherent tendency to remain together as a structural unit, leading to significant functional implications. In the crystal structure, the most important difference between the conformations of two molecules as a result of their association pertains to the orientation of Trp31. It may be noted that Trp31 is located at the mouth of the hydrophobic channel that forms the binding domain of the enzyme. The values of torsion angles (phi, psi, chi(1) and chi(2)) for both the backbone as well as for the side-chain of Trp31 in molecules A and B are -94 degrees, -30 degrees, -66 degrees, 116 degrees and -128 degrees, 170 degrees, -63 degrees, -81 degrees, respectively. The conformation of Trp31 in molecule A is suitable for binding, while that in B hinders the passage of the ligand to the binding site. Consequently, alpha-tocopherol is able to bind to molecule A only, while the binding site of molecule B contains three water molecules. In the complex, the aromatic moiety of alpha-tocopherol is placed in the large space at the active site of the enzyme, while the long hydrophobic channel in the enzyme is filled by hydrocarbon chain of alpha-tocopherol. The critical interactions between the enzyme and alpha-tocopherol are generated between the hydroxyl group of the six-membered ring of alpha-tocopherol and His48 N(delta1) and Asp49 O(delta1) as characteristic hydrogen bonds. The remaining part of alpha-tocopherol interacts extensively with the residues of the hydrophobic channel of the enzyme, giving rise to a number of hydrophobic interactions, resulting in the formation of a stable complex.     

Single-molecule analysis of epidermal growth factor binding on the surface of living cells

Global cellular responses induced by epidermal growth factor (EGF) receptor (EGFR) occur immediately with a less than 1% occupancy among tens of thousands of EGFR molecules on single cell surface. Activation of EGFR requires the formation of a signaling dimer of EGFR bound with a single ligand to each molecule. How sufficient numbers of signaling dimers are formed at such low occupancy rate is still not known. Here, we have analyzed the kinetics of EGF binding and the formation of the signaling dimer using single-molecule imaging and mathematical modeling. A small number of EGFR on the cell surface formed dimeric binding sites, which bound EGF two orders of magnitude faster than the monomeric binding sites. There was a positive cooperative binding of EGF to the dimeric binding sites through a newly discovered kinetic intermediate. These two mechanisms facilitate the formation of signaling dimers of EGF/EGFR complexes.     

Phage display selects for amylases with improved low pH starch-binding

Directed evolution of secreted industrial enzymes is hampered by the lack of powerful selection techniques. We have explored surface display to select for enzyme variants with improved binding performance on complex polymeric substrates. By a combination of saturation mutagenesis and phage display we selected alpha-amylase variants, which have the ability to bind starch substrate at industrially preferred low pH conditions. First we displayed active alpha-amylase on the surface of phage fd. Secondly we developed a selection system that is based on the ability of alpha-amylase displaying phages to bind to cross-linked starch. This system was used to probe the involvement of specific beta-strands in substrate interaction. Finally, a saturated library of alpha-amylase mutants with one or more amino acid residues changed in their Cbeta4 starch-binding domain was subjected to phage display selection. Mutant molecules with good starch-binding and hydrolytic capacity could be isolated from the phage library by repeated binding and elution of phage particles at lowered pH value. Apart from the wild type alpha-amylase a specific subset of variants, with only changes in three out of the seven possible positions, was selected. All selected variants could hydrolyse starch and heptamaltose at low pH. Interestingly, variants were found with a starch hydrolysis ratio at pH 4.5/7.5 that is improved relative to the wild type alpha-amylase. These data demonstrate that useful alpha-amylase mutants can be selected via surface display on the basis of their binding properties to starch at lowered pH values.     

Structure-function studies on human macrophage colony-stimulating factor (M-CSF)

M-CSF (CSF-1) can be produced in a variety of structural forms that may affect function in vivo. Truncated, nonglycosylated forms of recombinant M-CSF (rM-CSF) from E. coli have been refolded in vitro in high yield and shown to be functionally equivalent in vitro to glycosylated rM-CSF secreted from mammalian cells. An N-terminal domain of 149 amino acids is produced by all of the known M-CSF mRNA splice variants and is the region responsible for bioactivity observed in vitro. Heterodimeric rM-CSFs from different splice variants containing this domain were produced in pure form by refolding in vitro, and are fully active, but have yet to be observed in vivo. The circulating half-life of truncated M-CSF forms injected intravenously into rats increased with the MW of the M-CSF used. Large increases in half-life in vivo were observed following chemical addition of a single molecule of 10 kD polyethylene glycol to rM-CSF in vitro. The crystal structure of rM-CSF revealed that M-CSF is a member of a family of molecules related by having a distinctive four-helical-bundle structural core. Site-directed mutagenesis showed that residues in or near helix A and helix C are involved in receptor binding, as reflected by decreased bioactivity and receptor binding of certain mutants. A soluble form of the M-CSF receptor, c-fms, was produced in a baculovirus/Sf9 expression system and purified to homogeneity. The MW of rM-CSF saturated with this soluble receptor was determined by molecular sieve chromatography and light scattering. Each dimeric M-CSF molecule appears to bind two soluble receptor molecules in vitro, supporting the observation that M-CSF signaling is linked to receptor dimerization. Mol Reprod Dev 46:31–38, 1997. © 1997 Wiley-Liss, Inc.     

High resolution crystal structures of unliganded and liganded human liver ACBP reveal a new mode of binding for the acyl-CoA ligand

The acyl-CoA binding protein (ACBP) is essential for the fatty acid metabolism, membrane structure, membrane fusion, and ceramide synthesis. Here high resolution crystal structures of human cytosolic liver ACBP, unliganded and liganded with a physiological ligand, myristoyl-CoA are described. The binding of the acyl-CoA molecule induces only few structural differences near the binding pocket. The crystal form of the liganded ACBP, which has two ACBP molecules in the asymmetric unit, shows that in human ACBP the same acyl-CoA binding pocket is present as previously described for the bovine and Plasmodium falciparum ACBP and the mode of binding of the 3'-phosphate-AMP moiety is conserved. Unexpectedly, in one of the acyl-CoA binding pockets the acyl moiety is bound in a reversed mode as compared with the bovine and P. falciparum structures. In this binding mode, the myristoyl-CoA molecule is fully ordered and bound across the two ACBP molecules of the crystallographic asymmetric unit: the 3'-phosphate-AMP moiety is bound in the binding pocket of one ACBP molecule and the acyl chain is bound in the pocket of the other ACBP molecule. The remaining binding pocket cavities of these two ACBP molecules are filled by other ligand fragments. This novel binding mode shows that the acyl moiety can flip out of its classical binding pocket and bind elsewhere, suggesting a mechanism for the acyl-CoA transfer between ACBP and the active site of a target enzyme. This mechanism is of possible relevance for the in vivo function of ACBP.     

Differences in the binding of H1 variants to DNA. Cooperativity and linker-length related distribution

A study of the complexes formed between short linear DNA and three H1 variants, a typical somatic H1, and the extreme variants H5, from chicken erythrocytes, and spH1 from sea urchin sperm, has revealed differences between H1, H5 and spH1 that have implications for chromatin structure and folding. 1. All three histones bind cooperatively to DNA in 35 mM NaCl forming similar, but not identical, rod-like complexes. With sufficiently long DNA the complexes may be circular, circles forming more easily with H5 and spH1 than with H1. 2. The binding of H5 and spH1 to DNA is cooperative even in 5 mM NaCl, resulting in well-defined thin filaments that appear to contain two DNA molecules bridged by histone molecules. In contrast, H1 binds distributively over all the DNA molecules in 5 mM NaCl, but forms short stretches similar in appearance to the thin filaments formed with H5 and spH1. Rods appear to arise from the intertwining of regular thin filaments containing cooperatively bound histone molecules on raising the NaCl concentration to 35 mM. 3. The compositions of the rods correspond to one histone molecule for about every 47 bp (H1), 81 bp (H5) and 112 bp (spH1), suggesting average spacings of 24 bp (H1), 41 bp (H5) and 56 bp (spH1) in the component thin (double) filaments. Strikingly, these values are proportional to the linker lengths of the chromatins in which the particular H1 variant is the main or sole H1.     

Targeting human telomeric G-quadruplex DNA with oxazole-containing macrocyclic compounds

Oxazole-containing macrocycles, which include the natural product telomestatin, represent a promising class of anticancer agents that target G-quadruplex DNA. Two synthetic hexaoxazole-containing macrocyclic compounds (HXDV and HXLV-AC) have been characterized with regard to their cytotoxic activities versus human cancer cells, as well as the mode, thermodynamics, and specificity with which they bind to the intramolecular (3+1) G-quadruplex structural motif formed in the presence of K(+) ions by human telomeric DNA. Both compounds exhibit cytotoxic activities versus human lymphoblast (RPMI 8402) and oral carcinoma (KB3-1) cells, with associated IC(50) values ranging from 0.4 to 0.9muM. The compounds bind solely to the quadruplex nucleic acid form, but not to the duplex or triplex form. Binding to the quadruplex is associated with a stoichiometry of two ligand molecules per DNA molecule, with one ligand molecule binding to each end of the host quadruplex via a nonintercalative "terminal capping" mode of interaction. For both compounds, quadruplex binding is primarily entropy driven, while also being associated with a negative change in heat capacity. These thermodynamic properties reflect contributions from favorable ligand-induced alterations in the loop configurational entropies of the quadruplex, but not from changes in net hydration. The stoichiometry and mode of binding revealed by our studies have profound implications with regard to the number of ligand molecules that can potentially bind the 3-overhang region of human telomeric DNA.     

Conformational studies of A23187 with mono-, di- and trivalent metal ions by circular dichroism spectroscopy

CD studies carried out on A23187 indicate a solvent-dependent conformation for the free acid. Alkali metal ions were found to bind to the ionophore weakly. Divalent metal ions such as Mg2+, Ca2+, Sr2+, Ba2+ and Co2+ and trivalent lanthanide metal ions like La3+ were found to form predominantly 2:1 (ionophore-metal ion) complexes at low concentrations of metal ions, but both 2:1 and 1:1 complexes were formed with increasing salt concentration. Mg2+ and Co2+ exhibit similar CD behaviour that differs from that observed for the other divalent and lanthanide metal ions. The structure of 2:1 complexes involves two ligand molecules coordinated to the metal ion through the carboxylate oxygen, benzoxazole nitrogen and keto-pyrrole oxygen from each ligand molecule along with one or more solvent molecules. Values of the binding constant were determined for 2:1 complexes of the ionophore with divalent and lanthanide metal ions.     

Immobilized active monomers of D-glyceraldehyde-3-phosphate dehydrogenase from rabbit skeletal muscles and their coenzyme-binding properties

Experimental conditions favouring the dissociation of tetrameric rabbit muscle D-glyceraldehyde-3-phosphate dehydrogenase into active monomers were elaborated. The urea-induced dissociation of the tetramer was shown to be a stepwise process (in 2 M urea only dimers are formed; an increase in urea concentration up to 3 M causes the splitting of the dimers into monomers). The specific activity of immobilized monomers in the glyceraldehyde-3-phosphate oxidation reaction does not differ from that of the parent immobilized tetrameric form. The tetrameric enzyme molecule binds the coenzyme with a negative cooperativity (the first two NAD+ molecules bind with KD below 0.1 microM; for the third and fourth molecules the dissociation constant was determined to be equal to 5.5 +/- 1.5 microM (50 mM medinal buffer, 10 mM sodium phosphate, pH 8.2). The cooperativity of NAD+ binding is preserved in the immobilized preparation of tetrameric dehydrogenase. The immobilized monomers bind NAD+ with KD of 1.6 +/- 1.0 microM. The experimental results are consistent with the hypothesis according to which the association of catalytically active subunits into a tetramer changes their coenzyme-binding properties in such a way that the first two NAD+ molecules bind more firmly to a tetramer than to a monomer, whereas the third and the fourth NAD+ molecules bind less firmly.     

Biochemical characterization of complex formation by human erythrocyte spectrin, protein 4.1, and actin

Ternary complex formation between the major human erythrocyte membrane skeletal proteins spectrin, protein 4.1, and actin was quantified by measuring cosedimentation of spectrin and band 4.1 with F-actin. Complex formation was dependent upon the concentration of spectrin and band 4.1, each of which promoted the binding of the other to F-actin. Simultaneous measurement of the concentrations of spectrin and band 4.1 in the sedimentable complex showed that a single molecule of band 4.1 was sufficient to promote the binding of a spectrin dimer to F-actin. However, the molar ratio of band 4.1/spectrin in the complex was not fixed, ranging from approximately 0.6 to 2.2 as the relative concentration of added spectrin to band 4.1 was decreased. A mole ratio of 0.6 band 4.1/spectrin suggests that a single molecule of band 4.1 can promote the binding of more than one spectrin dimer to an actin filament. Saturation binding studies showed that in the presence of band 4.1 every actin monomer in a filament could bind at least one molecule of spectrin, yielding ternary complexes with spectrin/actin mole ratios as high as 1.4. Electron microscopy of such complexes showed them to consist of actin filaments heavily decorated with spectrin dimers. Ternary complex formation was not affected by alteration in Mg2+ or Ca2+ concentration but was markedly inhibited by KCl above 100 mM and nearly abolished by 10 mM 2,3-diphosphoglycerate or 10 mM adenosine 5'-triphosphate. Our data are used to refine the molecular model of the red cell membrane skeleton.