Search

The influenza virus neuraminidase (NA)-specific inhibitor zanamivir (4-guanidino-Neu5Ac2en) is effective in humans when administered topically within the respiratory tract. The search for compounds with altered pharmacological properties has led to the identification of a novel series of influenza virus NA inhibitors in which the triol group of zanamivir has been replaced by a hydrophobic group linked by a carboxamide at the 6 position (6-carboxamide). NWS/G70C variants generated in vitro, with decreased sensitivity to 6-carboxamide, contained hemagglutinin (HA) and/or NA mutations. HA mutants bound with a decreased efficiency to the cellular receptor and were cross-resistant to all the NA inhibitors tested. The NA mutation, an Arg-to-Lys mutation, was in a previously conserved site, Arg292, which forms part of a triarginyl cluster in the catalytic site. In enzyme assays, the NA was equally resistant to zanamivir and 4-amino-Neu5Ac2en but showed greater resistance to 6-carboxamide and was most resistant to a new carbocyclic NA inhibitor, GS4071, which also has a hydrophobic side chain at the 6 position. Consistent with enzyme assays, the lowest resistance in cell culture was seen to zanamivir, more resistance was seen to 6-carboxamide, and the greatest resistance was seen to GS4071. Substrate binding and enzyme activity were also decreased in the mutant, and consequently, virus replication in both plaque assays and liquid culture was compromised. Altered binding of the hydrophobic side chain at the 6 position or the triol group could account for the decreased binding of both the NA inhibitors and substrate.

Variation in the proteins produced by animal viruses allows the virus to reinfect the same host, but is constrained by the requirement to maintain critical viral functions, in particular engagement with cellular receptors. The fundamental characteristics of proteins and their interactions with each other suggest that this may not be so much of a constraint at all.

Summary The antigen binding site of antibodies is a chemically and structurally variable surface of amino acids located at the extremities of the arms of the -Y- shaped molecule. Six hypervariable loops (Complementarily Determining Regions) contribute to this surface and determine the specificity of the antibody. Three of these CDRs are on the V H domain and three are on the V L domain, and their relative positions in space are determined by the interactions at the V L -V H dimer interface. That interface has unusual architectural features when compared with the database of protein structures. Antibodies behave like other protein molecules when they associate with a ligand. They may change their structure, at the level either of the peptide backbone or side chain conformation, to achieve improved fit with the antigen. In addition, the V L and V H domains themselves may undergo some rearrangement across the V L -V H interface upon binding an antigen. The V L -V H interface plays the role of an adaptor, permitting movement of V L CDRS relative to V H CDRS SO that shape complementarity of the entire CDR surface to the antigen can be optimized. The diversity of CDR sequences, coupled with the ability to adopt a number of related conformations, results in a formidable armada of antibody specificities.

The three-dimensional structures of (1-->3)-beta-glucanase (EC 3.2.1.39) isoenzyme GII and (1-->3,1-->4)-beta-glucanase (EC 3.2.1.73) isoenzyme EII from barley have been determined by x-ray crystallography at 2.2- to 2.3-A resolution. The two classes of polysaccharide endohydrolase differ in their substrate specificity and function. Thus, the (1-->3)-beta-glucanases, which are classified amongst the plant "pathogenesis-related proteins," can hydrolyze (1-->3)- and (1-->3,1-->6)-beta-glucans of fungal cell walls and may therefore contribute to plant defense strategies, while the (1-->3,1-->4)-beta-glucanases function in plant cell wall hydrolysis during mobilization of the endosperm in germinating grain or during the growth of vegetative tissues. Both enzymes are alpha/beta-barrel structures. The catalytic amino acid residues are located within deep grooves which extend across the enzymes and which probably bind the substrates. Because the polypeptide backbones of the two enzymes are structurally very similar, the differences in their substrate specificities, and hence their widely divergent functions, have been acquired primarily by amino acid substitutions within the groove.

A new statistic Sc, which has a number of advantages over other measures of packing, is used to examine the shape complementarity of protein/protein interfaces selected from the Brookhaven Protein Data Bank. It is shown using Sc that antibody/antigen interfaces as a whole exhibit poorer shape complementarity than is observed in other systems involving protein/protein interactions. This result can be understood in terms of the fundamentally different evolutionary history of particular antibody/antigen associations compared to other systems considered, and in terms of the differing chemical natures of the interfaces.

Two potent inhibitors based on the crystal structure of influenza virus sialidase have been designed. These compounds are effective inhibitors not only of the enzyme, but also of the virus in cell culture and in animal models. The results provide an example of the power of rational, computer-assisted drug design, as well as indicating significant progress in the development of a new therapeutic or prophylactic treatment for influenza infection.

The variable heavy (VH) and variable light (VL) genes of NC10, a monoclonal antibody with specificity toward N9 neuraminidase (NA), were cloned and sequenced. A single chain Fv (scFv) fragment of NC10, consisting of VH and VL domains joined by a peptide linker, was designed, constructed and expressed in the E. coli expression vector pPOW. The N-terminal secretion signal PelB directed the synthesized protein into the periplasm where it was associated with the insoluble membrane fraction. An octapeptide (FLAG) tail was fused to the C-terminus of the single chain Fv to aid in its detection and remained intact throughout the protein purification process. NC10 scFv was purified by solubilization of the E. coli membrane fraction with guanidinium hydrochloride followed by column chromatography. The purified NC10 scFv showed binding affinity for its antigen, NA, 2-fold lower than that of the parent Fab. The complex between NA and the scFv has been crystallized by the vapor diffusion method. The crystals are tetragonal, space group P42(1)2, with unit cell dimensions a = b = 141 A, c = 218 A.

The site on influenza virus N9 neuraminidase recognized by NC41 monoclonal antibody comprises 19 amino acid residues that are in direct contact with 17 residues on the antibody. Single sequence changes in some of the neuraminidase residues in the site markedly reduce antibody binding. However, two mutants have been found within the site, Ile368 to Arg and Asn329 to Asp selected by antibodies other than NC41, and these mutants bind NC41 antibody with only slightly reduced affinity. The three-dimensional structures of the two mutant N9-NC41 antibody complexes as derived from the wild-type complex are presented. Both structures show that some amino acid substitutions can be accommodated within an antigen-antibody interface by local structural rearrangements around the mutation site. In the Ile368 to Arg mutant complex, the side-chain of Arg368 is shifted by 2.9 A from its position in the uncomplexed mutant and a shift of 1.3 A in the position of the light chain residue HisL55 with respect to the wild-type complex is also observed. In the other mutant, the side-chain of Asp329 appears rotated by 150 degrees around C alpha-C beta with respect to the uncomplexed mutant, so that the carboxylate group is moved to the periphery of the antigen-antibody interface. The results provide a basis for understanding some of the potential structural effects of somatic hypermutation on antigen-antibody binding in those cases where the mutation in the antibody occurs at antigen-contacting residues, and demonstrate again the importance of structural context in evaluating the effect of amino acid substitutions on protein structure and function.

The polypeptides of the trimeric seed storage protein phaseolin comprise two structurally similar units each made up of a beta-barrel and an alpha-helical domain. The beta-barrel has the 'jelly-roll' folding topology of the viral coat proteins and the alpha-helical domain shows structural similarity to the helix-turn-helix motif found in certain DNA-binding proteins.

We previously determined, by X-ray crystallography, the three-dimensional structure of a complex between influenza virus N9 neuraminidase (NA) and the Fab fragments of monoclonal antibody NC-41 [P. M. Colman, W. G. Laver, J. N. Varghese, A. T. Baker, P. A. Tulloch, G. M. Air, and R. G. Webster, Nature (London) 326:358-363, 1987]. This antibody binds to an epitope on the upper surface of the NA which is made up of four polypeptide loops over an area of approximately 600 A2 (60 nm2). We now describe properties of NC-41 and other monoclonal antibodies to N9 NA and the properties of variants selected with these antibodies (escape mutants). All except one of the escape mutants had single amino acid sequence changes which affected the binding of NC-41 and which therefore are located within the NC-41 epitope. The other one had a change outside the epitope which did not affect the binding of any of the other antibodies. All the antibodies which selected variants inhibited enzyme activity with fetuin (molecular weight, 50,000) as the substrate, but only five, including NC-41, also inhibited enzyme activity with the small substrate N-acetylneuramin-lactose (molecular weight, 600). These five probably inhibited enzyme activity by distorting the catalytic site of the NA. Isolated, intact N9 NA molecules form rosettes in the absence of detergent, and these possess high levels of hemagglutinin activity (W.G. Laver, P.M. Colman, R.G. Webster, V.S. Hinshaw, and G.M. Air, Virology 137:314-323, 1984). The enzyme activity of N9 NA was inhibited efficiently by 2-deoxy-2,3-dehydro-N-acetylneuraminic acid, whereas hemagglutinin activity was unaffected. The NAs of several variants with sequence changes in the NC-41 epitope lost hemagglutinin activity without any loss of enzyme activity, suggesting that the two activities are associated with separate sites on the N9 NA head.

The three-dimensional structure of antibody and its fragments has been the subject of several recent crystallographic studies. The determination of 6 A resolution of the structure of a myeloma protein [l] was followed by high resolution studies of a BenceJones dimer [2] and Fab fragments [3,4]. These latter studies [2-41 have demonstrated the independent folding of domains along the polypeptide chains [S] at least for the Fab fragment. We have recently reported the structure of a crystalline variable domain from a Bence-Jones protein REI [6] for which the complete amino acid sequence is known [ 71. In section 2 of this communication we outline some of the features of this structure. It seems likely that the observed structural homology between variable (V) and constant (C) domains in the Fab fragment [2,3] is extended to the Fc fragment. Studies of such homology can be made using the Rotation Function [B]. The results given in section 3 indicate a molecular symmetry for

Complexes of influenza virus neuraminidase both with antigen-binding (Fab) fragments and with whole monoclonal antibody molecules have been crystallized. Uniformly thin platelet microcrystals suitable for structure analysis by electron diffraction, yielding reflections to approximately 4.3 A resolution, have been grown from one neuraminidase-Fab complex, that of N9 neuraminidase with 32/3 Fab, and thicker crystals of a second neuraminidase-Fab complex (N9 neuraminidase-NC35 Fab) diffract X-rays to approximately 4.0 A resolution. Electron microscope lattice images of microcrystals both of Fab and of immunoglobulin G complexed with neuraminidase have been interpreted in terms of negatively stained images of the respective individual complex protomers. The sites of binding of the antibodies to the antigen are consistent with the notion that single amino acid changes observed in monoclonal variants of neuraminidase occur in binding epitopes for the antibody used for their selection.

The crystal structure of a human Fc fragment was analysed at 4 A resolution. A partial interpretation of the electron-density map in terms of domain structure was possible. The molecule has the shape of a mickey mouse. The spherical domain was interpreted visually and by domain Patterson function interpretation as the CH3 dimer. This dimer resembles closely the CH1-CL dimer found in Fab structures. The ellipsoidal "ears" of the molecule represent the CH2 domains. They are widely separated from each other, but closely connected to CH3. Their tertiary structure must be different from CH1, as Patterson domain interpretations were unsuccessful. A chain tracing in CH2 was not yet possible.

An influenza virus neuraminidase (NA) of the N9 subtype also has hemagglutinin (HA) activity (W. G. Laver, P. M. Colman, R. G. Webster, V. S. Hinshaw, and G. M. Air (1984), Virology 137, 314-323). To determine sequence relationships between this NA and other known NA and HA subtype sequences, and as a necessary step toward a complete structure determination, we have cloned a full-length copy of the coding sequence of the N9 NA of influenza virus A/tern/Australia/G70C/75 into the plasmid pUC9 using SalI linkers. The gene was sequenced by directed subcloning into the single-stranded phage vectors M13mp19 and M13mp18 and use of the dideoxy procedure. Most of the NA sequence was also obtained by direct protein sequencing of tryptic peptides. The N9 NA has 43 and 44% homology when compared to N1 or N2 sequences, respectively. There is no significant homology to any known HA sequence, or to the HN protein of the paramyxovirus SV5. Like the other NA molecules, the N9 NA is anchored in the membrane by an N-terminal hydrophobic region, from which biologically active heads can be released by pronase.

We describe here, for the first time, crystals of antibodies bound to a viral antigen which diffract X-rays to beyond 3 A. Crystals have been grown of Fab fragments of monoclonal antibody NC41, complexed with influenza virus neuraminidase (NA) of the N9 subtype and with a variant of N9 NA having a sequence change of Asn to Asp at position 331. This reduces, but does not abolish, the binding of NC41 antibody (in the case of another variant, Ser 371 to Leu, binding of NC41 antibody appears to be abolished). We are presenting data on the three-dimensional structure of these two complexes which indicates that NC41 antibody binds isosterically to the wild type and variant neuraminidase molecules.

The crystal structure of the Fab‡ fragment of the IgG protein Kol was analysed and an electron density map calculated at 3 A resolution based on phases obtained from multiple isomorphous replacement. An atomic model was constructed but the lack of amino acid sequence data causes some uncertainty, in particular concerning the amino acid side-chains. Several cycles of constrained crystallographic refinement (Deisenhofer & Steigemann, 1975) produced an R value of 0.36. The resulting model was compared with the intact Kol IgG crystal structure (Colman et al., 1976), which had also been subjected to constrained crystallographic refinement, and two Fab fragments (Davies et al., 1975). The elbow angle was found open and only 8 ° more bent than in intact Kol, in contrast to the other Fab fragments, which show a strongly bent elbow angle. The intramolecular longitudinal V-C contacts are very similar in Kol Fab and Kol IgG and considerably fewer in number than in the other Fab fragments. The lateral V-V and C-C association is virtually identical in Kol Fab and Kol IgG. Although Kol Fab and Kol IgG crystallize in different lattices, they exhibit a nearly identical head to tail packing of the Fab parts, with the C modules touching the hypervariable segments of the V modules. This strongly conserved particular mode of aggregation might be reflected in the property of cold precipitation of the Kol cryoglobulin.

Isolated intact influenza virus neuraminidase (NA) molecules of the N9 subtype have been found to possess hemagglutinin (HA) activity which, at equivalent protein concentration, was fourfold higher than that of isolated hemagglutinin molecules of the H3 subtype. The amino-terminal sequence of the N9 NA is the same as in neuraminidases of the eight other influenza A virus NA subtypes previously reported. Viruses possessing N9 NA therefore have two different HA activities and antibody to either HA or NA alone was incapable of inhibiting hemagglutination by the virus. However, antibody to the HA of an H1N9 virus neutralized its infectivity as effectively as it neutralized H1N1 or H1N2 viruses whose neuraminidases have no HA activity. (Antibodies to N9 NA did not neutralize the infectivity of viruses with N9 neuraminidase). 2-deoxy-2,3-dehydro-N-acetyl-neuraminic acid inhibited N9 NA activity but had no effect on the HA activity of the isolated N9 NA. One interpretation of this result would be that the HA and NA activities are located in separate sites. Pronase-released N9 NA heads form crystals suitable for X-ray diffraction studies and preliminary data to 2.9 A establish the space group as cubic, I432 with cell dimension a = 184 A. Data extend to beyond 1.9 A resolution, and these will be collected in the future.

The three-dimensional structure of plastocyanin, a ‘blue’ or ‘Type 1’ copper-protein, has been determined at a resolution of 2.7 A. The copper atom has a highly distorted tetrahedral coordination geometry. It is coordinated by a cysteine thiol group, a methionine thioether group, and two histidine imidazole groups.

An Fab fragment from a monoclonal antibody (NTS10/1) has been crystallized. The antigen, influenza virus A/Tokyo/3/67 (N2) neuraminidase, is also crystalline and its structure analysis is in progress. The Fab crystals are trigonal, space group P3121 with cell dimensions a = 132·3 A , c = 73·8 A . This crystalline material forms a complex with the neuraminidase with an equilibrium binding constant in excess of 1011 m −1. The molecular weight of the complex is 406,000 ± 20,000 indicating that four Fab fragments are attached to each neuraminidase tetramer. The separate crystallization of antigen and Fab fragment opens the way to map, for the first time, the complementary surfaces of an antigen-antibody complex.

The structure of a χ-type Bence-Jones protein variable fragment Au has been determined by molecular replacement methods using the known structure of an other Bence-Jones variable fragment Rei (Epp et al., Eur. J. Biochem. 45, 513 (1974)). The crystallographic R factor is 0.31 for about 4000 significantly measured reflections between 6.8 to 2.5 a. The Au protein forms a dimer across a crystallographic two fold axis. The spatial relationship of the two monomers, the conformation of the backbones and of the internal residues is extremely similar to that found in Rei.

An electron density map at 5 A resolution of the human myeloma protein Kol (immunoglobulin G1) is reported. The density describing the Fab‡ fragments has been interpreted in terms of the known structures of a V K fragment, Rei (Epp et al. , 1974), and the mouse McPC603 Fab fragment (Segal et al. , 1974). It shows 'a quaternary structure different from previously reported Fab fragments and provides evidence for flexibility in the switch peptides between variable and constant domains on heavy and light chains. No interpretable density is found C-terminal to the inter-heavy chain disulphide bonds, forcing the conclusion that the Fc fragment is disordered in the crystal lattice and implying flexibility in the Fc fragment in the residues immediately preceding the C H2 domain. The only density not described by the Fab fragments is found linking the two Fab arms of one molecule across the 2-fold symmetry axis. This density can be interpreted as extending as far as the Cys-Pro-Pro-Cys peptide.

An electron density map at 2.3 A resolution has been determined for the thermostable protease thermolysin. Crystallographic details are given for the three isomorphous heavy-atom derivatives which were used in the structure determination. The course of the polypeptide chain could be followed through the electron density map without reference to the chemically determined amino-acid sequence and 53% of the amino acids identified correctly. The electron density map, when combined with the amino-acid sequence determined by Titani, Hermodson, Ericsson, Walsh & Neurath (1972) , revealed the conformation of the thermolysin molecule in considerable detail. The molecule is folded into two distinct lobes, with β-structure predominating in one half and helices in the other. The essential zinc atom lies between the two lobes in a deep cleft and has as ligands two histidines and a glutamic acid as in carboxypeptidase A. Some other elements of the active site are similar to that of carboxypeptidase A, but there are also some striking differences. The over-all structures of thermolysin and carboxypeptidase A are quite different. It is shown that thermolysin binds four calcium ions, which may be replaced to varying degrees by ions of strontium, barium and the rare earth metals.

A description is given of a computer controlled rotating drum film scanner suitable for the measurement of X-ray diffraction photographs. The advantages of such a densitometer for protein crystallography are discussed. By measuring integrated densities 'on line' and taking advantage of the high positional accuracy of the scanner, a very satisfactory combination of speed, accuracy and flexibility can be achieved. It is found that commonly used X-ray films are significantly nonlinear even at optical densities less than 1.2 but by using a parabolic modification of the integrated densities the usable density range can be extended to optical densities of at least 2.5.

SummaryThe immune system must distinguish viable cells from cells damaged by physical and infective processes. The damaged cell-recognition molecule Clec9A is expressed on the surface of the mouse and human dendritic cell subsets specialized for the uptake and processing of material from dead cells. Clec9A recognizes a conserved component within nucleated and nonnucleated cells, exposed when cell membranes are damaged. We have identified this Clec9A ligand as a filamentous form of actin in association with particular actin-binding domains of cytoskeletal proteins. We have determined the crystal structure of the human CLEC9A C-type lectin domain and propose a functional dimeric structure with conserved tryptophans in the ligand recognition site. Mutation of these residues ablated CLEC9A binding to damaged cells and to the isolated ligand complexes. We propose that Clec9A provides targeted recruitment of the adaptive immune system during infection and can also be utilized to enhance immune responses generated by vaccines.

We have recently reported the X-ray crystal structure of a fragment of the fusion protein (F) of Newcastle disease virus (NDV). This work describes the methodology involved in the production and crystallization of that protein in recombinant form. The full-length cDNA of NDV-F was cloned and the ectodomain expressed in both CHO-K1 and Lec-3.2.8.1 cells. The recombinant protein, secreted as a single-chain polypeptide F0′, was purified using a c-myc antibody affinity column followed by gel filtration chromatography. Electron microscopic imaging showed the F0′ product to consist of unaggregated club-shaped particles. Trypsin treatment of F0′ could be used to produce disulfide-linked F2 and F1′ chains. However, imaging revealed extensive rosette-like aggregation of the trypsin-treated material, indicative of a conformational change. Only the non-trypsin-treated product was thus suitable for crystallization and two crystal forms were obtained, diffracting to ca. 3.5 and 4.0 Å, respectively. Both crystal forms were used in the structure determination.

A variant of the influenza virus NWS/G70C has been generated which has decreased sensitivityin vitroto the neuraminidase-specific inhibitor, 4-guanidino-Neu5Ac2en. The virus is 1000-fold less sensitive to the 4-guanidino-Neu5Ac2en in a plaque assay, but only 10-fold less sensitive to 4-amino-Neu5Ac2en. In an enzyme inhibition assay 250-fold more drug was needed to achieve inhibition comparable to that observed with the parent virus. In contrast to the plaque assay, the virus was fully sensitive to 4-amino-Neu5Ac2en in the enzyme inhibition assay. Kinetic analysis of 4-guanidino-Neu5Ac2en binding demonstrated that the variant no longer exhibited the slow binding characteristic seen with the parent and other influenza viruses and inhibition by Neu5Ac2en was also decreased. However, binding to 4-amino-Neu5Ac2en remained the same as the parent. Sequence analysis of this virus revealed a mutation at a previously conserved site in the enzyme active site of the neuraminidase, Glu 119 to Gly. Crystallographic analysis of the mutant neuraminidase with and without bound inhibitor confirmed this mutation and suggested that the reduced affinity for the 4-guanidino-Neu5Ac2en derives partly from the loss of a stabilizing interaction between the guanidino moiety and the carboxylate at residue 119, and partly from alterations to the solvent structure of the active site.

The arrangement of the disulphide bonds in the pronase-released neuraminidase heads of the Asian influenza virus A/Tokyo/3/67 have been examined by cyanogen bromide fragmentation, enzymic digestion and diagonal peptide mapping. There are 9 intrachain disulphide bridges and one interchain bridge which links pairs of monomers at the distal end of the stalk region of the neuraminidase tetramer. The disulphide bond arrangements of the remaining 3 half-cystine residues in the membrane-embedded stalk region of the neuraminidase were not examined.

The three-dimensional structure of the membrane glycoprotein neuraminidase of an escape mutant of the influenza virus strain A/Tokyo/3/67 has been determined to 3 A (1 A = 0.1 nm) resolution by X-ray diffraction. The mutant virus, selected by growing the virus in the presence of a monoclonal antibody to the neuraminidase, is shown to have undergone a single amino acid change of lysine to glutamic acid at residue 368. The three-dimensional structure of the neuraminidase is identical with that reported for A/Tokyo/3/67, except for a purely local adjustment of the structure at position 368.

Phaseolin, the main reserve globulin in seeds of the French bean (Phaseolus vulgaris L.), has been purified for sedimentation equilibrium analysis. The major protomer at neutral pH has a molecular weight of 150 000±5 000 which associates to a tetrameric form of molecular weight 596 000±20 000 at pH 4.5. The trimeric nature of the protomer is apparent from protein crosslinking experiments. Initial structural studies were made using optical rotatory dispersion and circular dichroism measurements. Two different crystal forms of the protein have been grown and characterised by X-ray diffraction; one of these forms appears suitable for high resolution X-ray analysis.

Three different types of crystals were grown from phaseolin, the major storage body protein from French bean. Type I crystals are cubes with space group symmetry P432, a = 67 A. Type II crystals are bipyramids with a rounded basal plane and belong to space group P2(1)2(1)2, a = 128 A, b = 136 A, and c = 162 A. Type III crystals are rhombs grown from phytic acid-free protein. The space group symmetry is P2(1)2(1)2(1), a = 113 A, b = 136 A, and c = 89 A. Both Type II and III crystals are suitable for high resolution x-ray study.

The crystal structure analysis of a human Fc fragment was pursued to 3.5 A resolution and a complete model was built and refined into the isomorphous Fourier map. The CH2 and CH3 domains show the immunoglobulin fold, with CH3 being closely similar to CH1, but CH2 intermediate in structure between V and CH3. The carbohydrate is rigidly attached to CH2, covering the C face. CH3 dimerizes as CH1-CL, but CH2 has no contact to the second chain. Residues involved in the lateral CH3-CH3 and the longitudinal CH3-CH2 contact are conserved in Ig classes and sub-classes. In IgM and IgE the two C-terminal domains also show this characteristic distribution of contact residues.

The structure of a complex between influenza virus neuraminidase and an antibody displays features inconsistent with the inflexible 'lock and key' model of antigen‐antibody binding. The structure of the antigen changes on binding, and that of the antibody may also change; the interaction therefore has some of the character of a handshake.

Native crystallographic data of tetrameric phosphorylase a crystals, space group P21, have been collected photographically to 3 A resolution. These data have been used in Patterson search methods in reciprocal and real space. The tetramers were found to exhibit molecular 222 symmetry. The cross vector between the centres of the two symmetry related tetramers in the unit cell was determined by two different translation function methods. On the basis of these rotation and translation function results a model for the arrangement of monomers within the tetramer and of tetramers in the unit cell is proposed; The 222 symmetry of the tetrameric molecule is found only when high resolution diffraction data are included (i.e. higher than 6 A). At lower resolution other symmetries dominate. Calculations with the proposed model have shown that these spurious symmetries result from the nonspecific overlap of protein-protein and solvent-solvent cross vectors. These results emphasize the importance of high resolution data when noncrystallographic symmetry of globular proteins is studied.

International audience; On the basis of comparative studies of known antibody structures and sequences it has been argued that there is a small repertoire of main-chain conformations for at least five of the six hypervariable regions of antibodies, and that the particular conformation adopted is determined by a few key conserved residues. These hypotheses are now supported by reasonably successful predictions of the structures of most hypervariable regions of various antibodies, as revealed by comparison with their subsequently determined structures.

The influenza virus neuraminidase glycoprotein is a tetramer with a box-shaped head, 100 X 100 X 60 A, attached to a slender stalk. The three-dimensional structure of neuraminidase heads shows that each monomer is composed of six topologically identical beta-sheets arranged in a propeller formation. The tetrameric enzyme has circular 4-fold symmetry stabilized in part by metal ions bound on the symmetry axis. Sugar residues are attached to four of the five potential glycosylation sequences, and in one case contribute to the interaction between subunits in the tetramer.

Neuraminidase genes from A/tern/Australia/G70C/75 (H11 N9) and A/whale/Maine/1/84 (H13N9) influenza viruses have been sequenced. Seventy-two nucleotide changes were found, 17 of which result in changes in the amino acid sequence of the neuraminidase; 3 in the stalk region and 14 in the heads. To our surprise, all of the sequence changes in the head region are located on the base of the neuraminidase tetramer, resulting in conservation of antigenic sites on top of the neuraminidase which vary extensively in human influenza virus neuraminidase. Whale N9 neuraminidase, like tern N9 neuraminidase, possesses high levels of hemagglutinating activity but, unlike the tern neuraminidase, failed to form large well-ordered crystals. However, when the neuraminidase was complexed with Fab fragments of monoclonal antibodies, which were made against the tern N9 neuraminidase, large crystals of the complexes were obtained which diffract X-rays to beyond 3A.

Publisher Summary Protein folding is an example of a protein–protein interaction. The free energy of the folded state is a result of the balance between competing thermodynamic effects, conformational entropy on the one hand—which favors the unfolded state—and the free energy of hydrophobic interactions, of hydrogen bonds, and of van der Waals' contacts on the other—favoring the folded state. For protein antigens, the antibody–antigen interaction is but another example of a protein–protein interaction. A substantial data base, drawn from high-resolution X-ray diffraction studies, has enabled many of the fundamental general principles of these interactions to be enumerated. One of the central questions that this chapter addresses is if antibody–antigen interactions demonstrate or extend these same principles. In developing this theme, it has been necessary to review different aspects of protein–protein interactions together with the structures of both antibody and antigen. The chapter focuses on the structure of antibodies and the design principles of the antigen combining site, leading to commentary on the issue of antigen-induced conformational change in antibody and vice versa. An antigen-receptor molecule, endowed with the properties suggested by this hypothesis, might seem contrary to the requirements of specificity of the immune system, but an important aspect of the hypothesis is that the reorganization of the V module interface is limited.