Preliminary joint neutron time-of-flight and X-ray crystallographic study of human ABO(H) blood group A glycosyltransferase.

The biosyntheses of oligosaccharides and glycoconjugates are conducted by glycosyltransferases. These extraordinarily diverse and widespread enzymes catalyze the formation of glycosidic bonds through the transfer of a monosaccharide from a donor molecule to an acceptor molecule, with the stereochemistry about the anomeric carbon being either inverted or retained. Human ABO(H) blood group A α-1,3-N-acetylgalactosaminyltransferase (GTA) generates the corresponding antigen by the transfer of N-acetylgalactosamine from UDP-GalNAc to the blood group H antigen. To understand better how specific active-site-residue protons and hydrogen-bonding patterns affect substrate recognition and catalysis, neutron diffraction studies were initiated at the Protein Crystallography Station (PCS) at Los Alamos Neutron Science Center (LANSCE). A large single crystal was subjected to H/D exchange prior to data collection and time-of-flight neutron diffraction data were collected to 2.5 Šresolution at the PCS to ∼85% overall completeness, with complementary X-ray diffraction data collected from a crystal from the same drop and extending to 1.85 Šresolution. Here, the first successful neutron data collection from a glycosyltransferase is reported.

Seasonal variation in cytochrome P450 immunopositive protein levels, lipid peroxidation and genetic toxicity in digestive gland of the mussel Mytilus edulis.

The relationship between cytochrome P450 1A- and 2E-immunopositive proteins, lipid peroxidation and DNA strand breaks (SBs) was studied in Mytilus edulis digestive gland at different seasons and at different sites around the UK coast. Cytochrome P4501A (CYP1A)-immunopositive protein and DNA strand breaks were generally lowest in December but there was no correlation between PAH exposure (indicated by chemical measurement and CYP1A-immunopositive protein expression) and DNA strand breaks which was highest at the relatively non-polluted site (Port Quin). As with CYP1A, CYP2E1-immunopositive protein was maximal at most sites in May. Lipid peroxidation, in contrast, did not alter markedly throughout the year. In conclusion, DNA strand breakage was not correlated with any of the above parameters although it did correlate with "scope for growth" as did the inverse of PAH levels. The study highlights the need to establish the relative contribution of DNA damage and DNA repair processes to the production of DNA strand breaks and emphasises the need to consider seasonal variation in interpretation of biomarkers.

Crystallization and preliminary X-ray diffraction analysis of two homologous antigen-binding fragments in complex with different carbohydrate antigens.

The antigen-binding fragments (Fab) of two murine monoclonal antibodies (mAb) S25-2 and S45-18, specific for carbohydrate epitopes in the lipopolysacchaide of the bacterial family Chlamydiaceae, have been crystallized in the presence and absence of synthetic oligosaccharides corresponding to their respective haptens. Crystals of both Fabs show different morphology depending on the presence of antigens. The sequence of mAb S45-18 was determined and shows a remarkable homology to that reported for mAb S25-2. These crystals offer an unparalleled opportunity to compare the structure and modes of binding of two homologous antibodies to similar but distinct carbohydrate epitopes.

Antibody recognition of a conformational epitope in a peptide antigen: Fv-peptide complex of an antibody fragment specific for the mutant EGF receptor, EGFRvIII.

Epitope mapping studies and the determination of the structure to 1.8 A resolution have been carried out for the antigen-binding fragment MR1 in complex with peptide antigen. MR1 is specific for the novel fusion junction of the mutant epidermal growth factor receptor EGFRvIII and has been reported to have a high degree of specificity for the mutant EGFRvIII over the wild-type EGF receptor. The structure of the complex shows that the peptide antigen residue side-chains found by epitope mapping studies to be critical for recognition are accommodated in pockets on the surface of the Fv. However, the most distinctive portion of the peptide antigen, the novel fusion glycine residue, makes no contact to the Fv and does not contribute directly to the epitope. The specificity of MR1 lies in the ability of this glycine residue to assume the restricted conformation needed to form a type II' beta-hairpin turn more easily, and demonstrates that a peptide antigen can be used to generate a conformational epitope.

Characterization of high affinity monoclonal antibodies specific for chlamydial lipopolysaccharide.

Pathogens belonging to the genus Chlamydia contain lipopolysaccharide with a 3-deoxy-D- manno- oct-2-ulosonic acid (Kdo) trisaccharide of the sequence alpha-Kdo-(2-->8)-alpha-Kdo-(2-->4)-alpha-Kdo. This lipopolysaccharide is recognized in a genus-specific pattern by murine monoclonal antibodies (mAbs), S25-23 and S25-2 (both IgG1kappa), which bind as the minimal structures the trisaccharide and the terminal Kdo-disaccharide, respectively. The variable domains of these mAbs were reverse transcribed from mRNA which was isolated from hybridomas and cloned as single-chain variable fragments (scFvs) in E.coli TG1. The kinetics of binding of whole antibodies, Fab fragments and scFvs to natural and synthetically modified ligands were determined by surface plasmon resonance (SPR) using synthetic neoglycoconjugates. As examples of an antibody-carbohydrate interaction involving anionic carboxyl groups on the ligand, we report that the affinities of these antibodies are higher than usually observed in carbo-hydrate-protein interactions (K(D)of 10(-3)to 10(-5)M). SPR analy-ses of monovalent Fab and scFv binding to the natural trisaccharide epitope gave dissociation constants of 770 nM for S25-2 and 350 nM for S25-23, as determined by global fitting (simultaneous fitting of several measurements at different antibody concentrations) of sensorgram data to a one-to-one interaction model. Local fitting (separate fitting of individual sensorgram data at different antibody concentrations) and Scatchard analysis of the data gave kinetic and affinity constants that were in good agreement with those obtained by global fitting. The SPR data also showed that while S25-2 bound well to several Kdo disaccharides and carboxyl-reduced Kdo ligands, S25-23 did not. Identification of amino acids in the complementarity determining regions revealed the presence of a large number of positively charged amino acids which were located towards the center of the combining site, thus suggesting a different recognition mechanism than that observed for neutral ligands. The latter mainly involves aromatic amino acids for hydrophobic stacking inter-actions and hydrogen bonds.

Characterization of protein-glycolipid recognition at the membrane bilayer.

A growing number of important molecular recognition events are being shown to involve the interactions between proteins and glycolipids. Glycolipids are molecules in which one or more monosaccharides are glycosidically linked to a lipid moiety. The lipid moiety is generally buried in the cell membrane or other bilayer, leaving the oligosaccharide moiety exposed but in close proximity to the bilayer surface. This presents a unique environment for protein-carbohydrate interactions, and studies to determine the influence of the bilayer on these phenomena are in their infancy. One important property of the bilayer is the ability to orient and cluster glycolipid species, as strong interactions in biological systems are often achieved through multivalency arising from the simultaneous association of two or more proteins and receptors. This is especially true of protein-carbohydrate binding because of the unusually low affinities that characterize the monovalent interactions. More recent studies have also shown that the composition of the lipid bilayer is a critical parameter in protein-glycolipid recognition. The fluidity of the bilayer allows for correct geometric positioning of the oligosaccharide head group relative to the binding sites on the protein. In addition, there are activity-based and structural data demonstrating the impact of the bilayer microenvironment on the modulation of oligosaccharide presentation. The use of model membranes in biosensor-based methods has supplied decisive evidence of the importance of the membrane in receptor presentation. These data can be correlated with three-dimensional structural information from X-ray crystallography, NMR, and molecular mechanics to provide insight into specific protein-carbohydrate inter--actions at the bilayer.

Donor substrate specificity of recombinant human blood group A, B and hybrid A/B glycosyltransferases expressed in Escherichia coli.

The human blood group A and B glycosyltransferases catalyze the transfer of GalNAc and Gal, to the (O)H-precursor structure Fuc alpha (1-2)Gal beta-OR to form the blood group A and B antigens, respectively. Changing four amino acids (176, 235, 266 and 268) alters the specificity from an A to a B glycosyltransferase. A series of hybrid blood group A/B glycosyltransferases were produced by interchanging these four amino acids in synthetic genes coding for soluble forms of the enzymes and expressed in Escherichia coli. The purified hybrid glycosyltransferases were characterized by two-substrate enzyme kinetic analysis using both UDP-GalNAc and UDP-Gal donor substrates. The A and B glycosyltransferases were screened with other donor substrates and found to also utilize the unnatural donors UDP-GlcNAc and UDP-Glc, respectively. The kinetic data demonstrate the importance of a single amino acid (266) in determining the A vs. B donor specificity.

The role of homophilic binding in anti-tumor antibody R24 recognition of molecular surfaces. Demonstration of an intermolecular beta-sheet interaction between vh domains.

The murine antibody R24 and mouse-human Fv-IgG1(kappa) chimeric antibody chR24 are specific for the cell-surface tumor antigen disialoganglioside GD3. X-ray diffraction and surface plasmon resonance experiments have been employed to study the mechanism of "homophilic binding," in which molecules of R24 recognize and bind to other molecules of R24 though their heavy chain variable domains. R24 exhibits strong binding to liposomes containing disialoganglioside GD3; however, the kinetics are unusual in that saturation of binding is not observed. The binding of chR24 to GD3-bearing liposomes is significantly weaker, suggesting that cooperative interactions involving antibody constant regions contribute to R24 binding of membrane-bound GD3. The crystal structures of the Fabs from R24 and chR24 reveal the mechanism for homophilic binding and confirm that the homophilic and antigen-binding idiotopes are distinct. The homophilic binding idiotope is formed largely by an anti-parallel beta-sheet dimerization between the H2 complementarity determining region (CDR) loops of two Fabs, while the antigen-binding idiotope is a pocket formed by the three CDR loops on the heavy chain. The formation of homophilic dimers requires the presence of a canonical conformation for the H2 CDR in conjunction with participation of side chains. The relative positions of the homophilic and antigen-binding sites allows for a lattice of GD3-specific antibodies to be constructed, which is stabilized by the presence of the cell membrane. This model provides for the selective recognition by R24 of cells that overexpress GD3 on the cell surface.

Crystallization and preliminary X-ray diffraction analysis of antigen-binding fragments which are specific for antigenic conformations of sialic acid homopolymers.

Meningococcal meningitis is a severe childhood disease which often results in significant disability or death. Two major etiological agents of meningitis are the group B meningococci and capsular type K1 E. coli. The virulence of these organisms is attributable to structural mimicry between their common alpha(2-8)-polysialic acid capsular polysaccharide and human tissue antigens, which allows the bacteria to evade immune surveillance. There is currently no effective vaccine to protect against this infection. It has been demonstrated that the capsular polysaccharide of the bacteria can adopt a unique 'antigenic conformation'. This antigenic conformation has formed the basis for the development of an N-propionylated polysialic acid vaccine. Immunization trials in mice with this vaccine show the production of two groups of antibodies, of which only N-propionylated polysialic acid-specific were protective. Knowledge of the structure of the antigen-binding site which recognizes the protective epitope is essential to determining the antigenic conformation of the polysaccharides, and is a critical aspect in understanding and improving the action of potential vaccines. The antigen-binding fragments (Fab) of one protective (13D9) and one non-protective (6B9) monoclonal antibody specific for the capsular polysaccharides of group B meningococci have been crystallized and have undergone preliminary X-ray diffraction analysis. Both crystals are observed to scatter X-rays to approximately 1.7 A resolution at the A1 station at the Cornell High-Energy Synchrotron Source. 13D9 has an orthorhombic unit cell with a = 41.8, b = 102.3, c = 134.7 A, with space group P212121. Fab 6B9 has an orthorhombic unit cell with a = 89.6, b = 132.0 and c = 36.9 A, with space group P21212.

Production, crystallization and diffraction to atomic resolution of an antibody Fv specific for the blood-group A oligosaccharide antigen.

The histoblood-group ABO carbohydrate antigens are well known as important factors in blood transfusions, but they can also act as receptors for infectious agents and have been implicated in susceptibility to certain carcinomas. A single-chain variable-domain antigen-binding fragment (scFv) gene based on the known sequence of an anti-blood-group-A monoclonal antibody (AC1001) has been synthesized and expressed in Escherichia coli. The purified scFv preparation existed primarily in the monomeric form but also contained large amounts of dimeric and higher oligomeric forms. The corresponding variable-domain antigen-binding fragment (Fv) was generated by cleaving the VL-VH linker with subtilisin, and its activity was demonstrated by surface plasmon resonance with an immobilized bovine serum albumin-A-trisaccharide conjugate (KD = 290 microM). AC1001 Fv crystals grown in the presence of N-acetylgalactosamine diffracted to 0.93 A resolution. This is the first reported example of a crystal of an antibody antigen-binding fragment diffracting to atomic resolution.

Characterization of an Ig VH idiotope that results in specific homophilic binding and increased avidity for antigen.

mAb against GD3 ganglioside demonstrate homophilic binding in which soluble anti-GD3 mAb bind, through the GD3 binding site, to a VH idiotope (designated IdHOM) on solid phase anti-GD3 mAb. In this way, homophilic binding provides a mechanism for amplifying the binding of mAb to cell surface GD3. We show that serine 52a, within CDR2, is required for IdHOM expression, homophilic binding, and high avidity binding to cell surface GD3. Computer modeling based on the crystal structure of anti-GD3 mAb R24 showed serine 52a situated at the mouth of the GD3 binding pocket, but not directly involved with GD3 binding. Substitutions at position 52a predicted to maintain the GD3 binding pocket (e.g., threonine) resulted in the loss of IdHOM expression and homophilic binding and markedly decreased binding to cell surface GD3, but maintained low avidity GD3 binding as measured by ELISA. All other substitutions at position 52a were predicted to significantly distort the GD3 binding pocket and resulted in the loss of both homophilic binding and any detectable avidity for GD3. We have structurally defined IdHOM and conclude that this idiotope is not required for the GD3 binding pocket, but that the idiotope is necessary for homophilic binding, which is required for high avidity binding to cell surface GD3. We speculate that selection of certain VH genes may result in the expression of idiotopes that allow homophilic binding, and this may represent a general mechanism for increasing the avidity of Abs against T cell-independent Ags.

Experiments in microgravity: a comparison of crystals of a carbohydrate-binding fab grown on the ground, on space shuttle Discovery and on space station Mir.

The Fab fragment of the hybridoma antibody (YsT9.1) specific to Brucella abortus has been crystallized on earth using both Linbro plates and ground-based models of the flight hardware, as well as in microgravity on board the space shuttle Discovery and the space station Mir. Large-scale experiments using Linbro plates gave two different crystal morphologies, pyramidal and rhomboid, depending on conditions. The pyramidal crystals proved to scatter X-rays to higher resolution, and conditions within the ground-based flight hardware for both Discovery and Mir were adjusted to produce crystals with this morphology. The experiment on Discovery produced large crystals in each of ten chambers. The experiment on Mir produced crystals in only one of the five assigned chambers, despite the fact that the simultaneous ground-based experiment produced large crystals in every corresponding chamber. Data collection was attempted for crystals from both space and ground-based experiments. Higher resolution data was obtained from crystals grown on Discovery than from either Mir or ground-based crystals, even though the crystals obtained from Discovery were smaller and forced to grow over a much shorter period of time because of the shorter length of the shuttle mission.

Bifunctional fusion proteins consisting of a single-chain antibody and an engineered lanthanide-binding protein.

The combination of an antibody fragment with a lanthanide chelating protein has desirable characteristics for fluorescence-based immunoassays and tumor radioimmunotherapy. As a model for this design, a fusion protein consisting of a single-chain antibody linked to an engineered version of oncomodulin, a protein with two Ca(2+)-binding motifs (the CD and EF loops), was produced by secretion from Escherichia coli in good yield. The single-chain antibody was specific for a Salmonella O-polysaccharide. The CD loop of oncomodulin had been redesigned to bind lanthanide ions with high affinity. The fusion protein was shown to have antigen-binding activity that was comparable to that of the unfused single-chain antibody, to bind Tb3+ with very high affinity and to give strong, sensitized Tb3+ luminescence via excitation of the tryptophan residue in the CD loop. A second fusion protein containing a 30-residue helix-loop-helix motif as the lanthanide-binding component was also prepared, but showed considerably lower solubility. Competition for Tb3+ binding by a series of metal chelators indicated that the affinities of the oncomodulin and 30 residue fusions for Tb3+ were approximately 10(11) M-1 and 10(7) M-1, respectively. Time-resolved lanthanide luminescence photography of electrophoresis gels demonstrated that the helix-loop-helix Ca(2+)-binding could be used to specifically visualize the scFv fragment.

Evidence for the extended helical nature of polysaccharide epitopes. The 2.8 A resolution structure and thermodynamics of ligand binding of an antigen binding fragment specific for alpha-(2-->8)-polysialic acid.

The antigen binding fragment from an IgG2a kappa murine monoclonal antibody with specificity for alpha-(2-->8)-linked sialic acid polymers has been prepared and crystallized in the absence of hapten. Crystals were grown by vapor diffusion equilibrium with 16-18% polyethylene glycol 4000 solutions. The structure was solved by molecular replacement methods and refined to a conventional R factor of 0.164 for data to 2.8 A. The binding site is observed to display a shape and distribution of charges that is complementary to that of the predicted conformation of the oligosaccharide epitope. A thermodynamic description of ligand binding has been compiled for oligosaccharides ranging in length from 9 to 41 residues, and the data for the largest ligand has been used in a novel way to estimate the size of the antigen binding site. A model of antigen binding is presented that satisfies this thermodynamic data, as well as a previously reported requirement of conformational specificity of the oligosaccharide. X-ray crystallographic and thermodynamic evidence are consistent with a binding site that accommodates at least eight sialic acid residues.

Location of the active site of the bean alpha-amylase inhibitor and involvement of a Trp, Arg, Tyr triad.

Seeds of the common bean contain three homologous proteins: phytohaemagglutinin E, phytohaemagglutinin L and the lectin-like protein alpha-amylase inhibitor (alpha AI). Whereas the active site of lectins has been studied in great detail, there is no information on the active site of the related protein alpha AI, which exerts its biological activity by making a 1:1 complex with alpha-amylase. alpha-amylase inhibitor is synthesized as a 30 kDa precursor glycoprotein that needs to be processed at Asn77 to form an active molecule. Comparison of the amino acid sequence of the bean alpha AI with that of the bacterial amylase inhibitor, tendamistat, suggested that a region around Trp188 might be involved in the inhibitory site. When a three-dimensional model of the bean alpha AI was constructed based on its homology to the legume lectins, this Trp region was alongside Asn77. To test this site hypothesis, mutants of alpha AI were created by site-directed mutagenesis of the cDNA and expressed in transgenic tobacco. The mutant proteins R74N and WSY188-190GNV, as well as the double mutant, were inactive as inhibitors. These findings suggest that the active site of alpha AI consists of W188, R74 and Y190, in analogy to the Trp-Arg-Tyr motif of tendamistat, and that the processing of the polypeptide at Asn77 may be necessary to bring these residues in close proximity.

Exploring the mimicry of polysaccharide antigens by anti-idiotypic antibodies. The crystallization, molecular replacement, and refinement to 2.8 A resolution of an idiotope-anti-idiotope Fab complex and of the unliganded anti-idiotope Fab.

The monoclonal antibody YsT9.1 is specific for the lipopolysaccharide A antigen of Brucella abortus. A complex formed between the Fab of YsT9.1 (Ab1) and the Fab of its antidiotopic monoclonal antibody T91AJ5 (Ab2) has been crystallized, and data have been collected to 2.8 A resolution. The space group is monoclinic P2(1), with one molecule per asymmetric unit. The structure was solved using a limited Patterson-correlation search over the asymmetric-unit of rotation space, and has been refined to an R-factor of 0.174. The complex is a head-to-head dimer, with the contact between the two Fabs almost completely restricted to their hypervariable loops. The interactions between the two Fabs in the complex are dominated by tyrosine residues, not only in the formation of hydrogen bonds, but in their participation in an aromatic ring network that spans the two Fv domains. The anti-idiotope was found to be unable to carry an internal image of the antigen and induce polysaccharide-specific "anti-anti-idiotopes" (Ab3) because the polysaccharide binding cleft on the Ab1 is too narrow and deep to allow comprehensive contact with the binding site of Ab2. The antibody T91AJ5 therefore is a class gamma anti-idiotope. The contact surfaces of the two Fabs are highly complementary; however, they are distinctly different in character. Each Fab has two separate binding surfaces of approximately equal size, but while the two binding surfaces of Ab1 are partitioned between the light and heavy chains, the two binding surfaces of Ab2 are shared between the chains, with the heavy chain responsible for about 60% of the total binding area. The structure of the unliganded Fab of Ab2 has also been solved by molecular replacement, and refined to an R-factor of 0.152 at 2.8 A resolution. This Fab crystallizes in the orthorhombic space group P2(1)2(1)2(1), with one molecule per asymmetric unit. The second hypervariable loop of the heavy chain of the Ab2 Fab is observed to undergo a significant and necessary conformational rearrangement in going from the unliganded to complexed form, and thus complex formation is an example of "induced fit" of antigen to antibody. The unliganded Ab1 Fab packs in the standard head-to-tail fashion observed for other Fabs. This mode of packing is precluded in the complex, by its head-to-head nature, and it is found to pack in tilted layers with most intermolecular contacts made between adjacent Ab2 Fab molecules.(ABSTRACT TRUNCATED AT 400 WORDS)

1.70 A resolution structure of myoglobin from yellowfin tuna. An example of a myoglobin lacking the D helix.

The crystal structure of metmyoglobin from yellowfin tuna (Thunnus albacares) has been determined by molecular replacement methods and refined to a conventional R factor of 0.177 for all observed reflections in the range of 6.0-1.70 A resolution. Like other myoglobins for which a high-resolution structure is available, the polypeptide chain is organized into several helices that cooperate to form a hydrophobic pocket into which the heme prosthetic group is non-covalently bound; however, the D helix observed in other myoglobins is absent in myoglobin from yellowfin tuna and has been replaced with a random coil. As well, the A helix has a pronounced kink due to the presence of Pro16. The differences in structure between this and sperm whale myoglobin can be correlated with their reported dioxygen affinity and dissociation. The structure is in agreement with reported fluorescence data which show an increased Trp14.heme distance in yellowfin tuna compared to sperm whale myoglobin.

Three-dimensional structure of cyanomet-sulfmyoglobin C.

The atomic structure of horse heart cyanomet-sulfmyoglobin C has been established by x-ray crystallographic techniques to a resolution of 2.0 A with an R value of 0.129. The protoheme IX prosthetic group of this thermodynamically stable sulfmyoglobin derivative has been converted to a chlorin in which the pyrrole ring bearing the 4-vinyl group is saturated and possesses an exocyclic thiolene ring. This study provides the three-dimensional structure of a protein with an iron-chlorin prosthetic group. The overall conformation of the surrounding polypeptide chain of the modified protein is very similar to that of the native protein. However, the addition of the sulfur atom has caused a distortion of the prosthetic group from that in the native protein to result in the repositioning of the side chains of some residues in the heme pocket.

SETOR: hardware-lighted three-dimensional solid model representations of macromolecules.

SETOR is designed to exploit the hardware lighting capabilities of the IRIS-4D series graphics workstations to render high-quality raster images of macromolecules that can undergo rotation and translation interactively. SETOR can render standard all-atom and backbone models of proteins or nucleic acids, but focuses on displaying protein molecules by highlighting elements of secondary structure. The program has a very friendly user interface that minimizes the number of input files by allowing the user to interactively edit parameters, such as colors, lighting coefficients, and descriptions of secondary structure via mouse activated dialogue boxes. The choice of polymer chain representation can be varied from standard vector models and van der Waal models, to a B-spline fit of polymer backbones that yields a smooth ribbon that approximates the polymer chain, to strict Cardinal splines that interpolate the smoothest curve possible that will precisely follow the polymer chain. The program provides a photograph mode, save/restore facilities, and efficient generation of symmetry-related molecules and packing diagrams. Additionally, SETOR is designed to accept commands and model coordinates from the standard input stream, and to control standard output. Ancillary programs provide a method to interactively edit hardcopy plots of all vector and many solid models generated by SETOR, and to produce standard HPGL or PostScript files. Examples of figures rendered by SETOR of a number of macromolecules of various classes are presented.

Refinement of recombinant oncomodulin at 1.30 A resolution.

A refinement of the oncomodulin crystal structure at 1.30 A resolution has been carried out with X-ray data from the recombinant protein. The crystallographic R-factor values are 0.169 for 19,995 reflections in the range 6.0 to 1.30 A, which were used for the restrained least-squares refinement, and 0.176 for 20,186 observed reflections in the range 10.0 to 1.30 A. This high resolution refinement has enabled us to make more definitive statements about the molecular structure than was possible heretofore. The present model includes residues 1 to 108, the two Ca2+ of the CD and EF loops, two intermolecular Ca2+, and 103 water molecules per oncomodulin molecule. The electron density maps indicate disordered orientations for ten residues on the hydrophilic surface of the molecule. The pattern of molecular aggregation via intermolecular Ca2+, which occurs in the native rat oncomodulin structure, is also present in the recombinant oncomodulin structure. The Cys18 side-chain is not in a position that would be easily accessible for molecular dimerization via a disulphide bond. The substitution of Glu59, which is preserved in all the determined species of parvalbumin, by Asp59 in oncomodulin seems to break a stabilizing hydrogen bond in the CD loop and render the main-chain in positions 59 to 60 somewhat unstable. This instability in the CD loop, and the strong tendency of oncomodulin for molecular aggregation via intermolecular Ca2+, appear to be the two outstanding features that may account for oncomodulin's biological peculiarities.