Jaap J. Beintema

Crystal structures of hevamine, a plant defence protein with chitinase and lysozyme activity, and its complex with an inhibitor

BACKGROUND Hevamine is a member of one of several families of plant chitinases and lysozymes that are important for plant defence against pathogenic bacteria and fungi. The enzyme can hydrolyze the linear polysaccharide chains of chitin and peptidoglycan. A full understanding of the structure/function relationships of chitinases might facilitate the production of transgenic plants with increased resistance towards a wide range of pathogens. RESULTS The crystal structure of hevamine has been determined to a resolution of 2.2 A, and refined to an R-factor of 0.169. The enzyme possesses a (beta alpha)8-barrel fold. An inhibitor binding study shows that the substrate-binding cleft is located at the carboxy-terminal end of the beta-barrel, near the conserved Glu127. Glu127 is in a position to act as the catalytic proton donor, but no residue that might stabilize a positively charged oxocarbonium ion intermediate was found. A likely mechanism of substrate hydrolysis is by direct attack of a water molecule on the C1 atom of the scissile bond, resulting in inversion of the configuration at C1. CONCLUSIONS The structure of hevamine shows a completely new lysozyme/chitinase fold and represents a new class of polysaccharide-hydrolyzing (beta alpha)8-barrel enzymes. Because the residues conserved in the family to which hevamine belongs are important for maintaining the structure of the (beta alpha)8-barrel, all members of the family, including fungal, bacterial and insect chitinases, are likely to share this architecture. The crystal structure obtained provides a basis for protein engineering studies in this family of chitinases.

Structural features of plant chitinases and chitin-binding proteins.

Structural features of plant chitinases and chitin‐binding proteins are discussed. Many of these proteins consist of multiple domains, of which the chitin‐binding hevein domain is a predominant one. X‐ray and NMR structures of representatives of the major classes of these proteins are available now, and are used to describe the structures of the other ones. Conserved positions of Cys residues can be taken as evidence for identically located disulfide bridges or cysteine residues. The current classification of chitinases is unsatisfactory and needs to be replaced by an evolutionarily more correct one. As the currently known three‐dimensional structures of chitinases are those from barley and the rubber tree, Hevea brasiliensis, it is proposed to adopt the designation b‐type (classes I, II and IV) and h‐type (classes III and V) chitinases, respectively.

Primary structures of two ribonucleases from ginseng calluses - New members of the PR-10 family of intracellular pathogenesis-related plant proteins

The amino acid sequences of two ribonucleases from a callus cell culture of Panax ginseng were determined. The two sequences differ at 26% of the amino acid positions. Homology was found with a large family of intracellular pathogenesis‐related proteins, food allergens and tree pollen allergens from both dicotyledonous and monocotyledonous plant species. There is about 30% sequence difference with proteins from species belonging to the same plant order (Apiales: parsley and celery), 60% with those from four other dicotyledonous plant orders and about 70% from that of the monocotyledonous asparagus. More thorough evolutionary analyses of sequences lead to the conclusion that the general biological function of members of this protein family may be closely related to the ability to cleave intracellular RNA and that they have an important role in cell metabolism. As the three‐dimensional structure of one of the members of this protein family has been determined recently [Gajhede et al., Nature Struct Biol 3 (1996) 1040–1045], it may be possible to assign active‐site residues in the enzyme molecule and make hypotheses about its mode of action. Structural features in addition to the cellular site of biosynthesis indicate that this family of ribonucleases is very different from previously investigated ones.

Use of endoproteinase Lys-C from Lysobacter enzymogenes in protein sequence analysis

Endoproteinase Lys-C from Lysobacter enzymogenes, which is commercially available, proved to be useful in the determination of primary structures of proteins. The enzyme preferentially cleaves at the carboxyl side of lysine residues.

Subunit composition, X-ray diffraction, amino acid analysis and oxygen binding behaviour of Panulirus interruptus hemocyanin

The subunit composition of Panulirus interruptus hemocyanin has been studied. Equilibrium sedimentation, gel chromatography and sodium dodecyl sulphate-polyacrylamide electrophoresis reveal a hexameric structure for this hemocyanin, with a molecular weight of 450,000 for the undissociated protein and 75,000 for the subunits. Amino acid analysis data suggest homogeneity of the polypeptide chain. X-ray diffraction gives the cell parameters a = 119·8, b = 193·1 and c = 122·2 , β = 118·1 o . The asymmetric unit contains one molecule. Oxygen binding data show the undissociated protein to be co-operative while the dissociated protein binds oxygen non-co-operatively with a low oxygen affinity compared to that of whole molecules.

The amino acid sequence of human pancreatic ribonuclease

The primary structure of human (Homo sapiens) pancreatic ribonuclease has been determined by automatic sequencing of the native protein and by analysis of peptides obtained by cleavage with proteolytic enzymes, cyanogen bromide, and hydroxylamine. The following sequence was deduced: (sequence in text). Human pancreatic ribonuclease differs at 37 positions from bovine pancreatic ribonuclease. In addition the human enzyme has three more residues at the C-terminus. About half of the enzyme molecules contain carbohydrate attached to the sequence Asn-Met-Thr (34-36). Two other Asn-X-Ser/Thr sequences are carbohydrate free. Human pancreatic ribonuclease contains many positively charged residues, especially near the N-terminus, while negatively charged residues are more concentrated near the C-terminus.

Invariant features of the structure of pancreatic ribonuclease: A test of different predictive models

(1) The primary structures of 24 homologous ribonucleases which differ in up to 34% of their amino acids are compared to determine which features are essential for the three-dimensional structure. (2) The distribution pattern of helix-forming residues, as defined by Chou & Fasman (1974a), was rather constant. However, predictions of the helices, based on such a distribution, were not satisfactory, although the method of Chou & Fasman (1974b) gave more uniform results and a better agreement with the X-ray structure of bovine RNA ase than the method of Burgess et al. (1974). The prediction of the β-sheet and the β-bend using these statistical methods was less accurate than the helix prediction. (3) The prediction of the secondary structure by the method of Lim (1974b) which is based on the relative positions of the hydrophobic residues in the α-helix and β-sheet, gave good results in the case of ribonuclease. It turned out that all residues which, according to Lims theory, are essential for the formation of secondary structure are invariant in all ribonucleases tested. (4) The invariability of the distribution of the hydrophobic residues suggests strongly that this is a very important feature of the primary structure. The role of these residues in forming secondary and tertiary structure is discussed.

Affinity chromatography of porcine pancreatic ribonuclease reinvestigation of the N-terminal amino acid sequence.

In 1969 Wilchek et al. [I] described the affinity chromatography of bovine ribonuclease A using Agarose to which a strong competitive inhibitor of ribonuclease, 5’-(4-aminophenyl)-uridine-(2’,3’) phosphate (APUP) was coupled. On this adsorbent binding occurred at low ionic strength and neutral pH; desorption was achieved with 2 M acetic acid. We were unable to repeat their results, but could obtain desorption of ribonuclease with 4 M NaCl. Also, with crude pancreatic extracts, we could only achieve satisfactory binding of the enzyme at an ionic strength of 0.2, which largely prevents aspecific interactions. In this way we have already isolated pure ribonucleases from more than fifteen mammalian species in sufficient amounts for amino acid sequence studies. In this article we describe the purification of porcine ribonuclease. Jackson and Hirs [2] have determined the amino acid sequence of this enzyme. In the N-terminal sequence they found glutamine at the positions 2 and 9, where, in other pancreatic ribonucleases, normally glutamic acid is found, (except for rat [3] which contains lysine at position 9). The side chain of glutamic acid 9 has no distinct function in the structure of bovine ribonuclease, but glutamic acid 2 may form a salt brdige with the side chain of arginine 10 [4], both “constant” residues in pancreatic ribonucleases. Marchiori et al. [5] and Hofmann et al. [4, 61 have demonstrated in their synthetic S-peptide studies that if this ion pair cannot be formed, the resulting partially synthetic ribonucleases S’ are less active. Therefore, we reinvestigated the state of amidation of glutamic acid residues in the N-terminal part of porcine ribonuclease.

Evolution and recombination of bovine DNA repeats

The history of the abundant repeat elements in the bovine genome has been studied by comparative hybridization and PCR. The Bov-A and Bov-B SINE elements both emerged just after the divergence of the Camelidae and the true ruminants. A 31-bp subrepeat motif in satellites of the Bovidae species cattle, sheep, and goat is also present in Cervidae (deer) and apparently predates the Bovidae. However, the other components of the bovine satellites were amplified after the divergence of the cattle and the Caprinae (sheep and goat). A 23-bp motif, which as subrepeat of two major satellites occupies 5% of the cattle genome, emerged only after the split of the water buffalo and other cattle species. During the evolution of the Bovidae the satellite repeat units were shaped by recombination events involving subrepeats, other satellite components, and SINE elements. Differences in restriction sites of homologous satellites indicate a continuing rapid horizontal spread of new sequence variants.

NMR investigations of protein-carbohydrate interactions : Studies on the relevance of Trp/Tyr variations in lectin binding sites as deduced from titration microcalorimetry and NMR studies on hevein domains. Determination of the NMR structure of the complex between pseudohevein and N,N ',N ''-triacetylchitotriose

Model studies on lectins and their interactions with carbohydrate ligands in solution are essential to gain insights into the driving forces for complex formation and to optimize programs for computer simulations. The specific interaction of pseudohevein with N,N′,N′′‐triacetylchitotriose has been analyzed by 1H‐NMR spectroscopy. Because of its small size, with a chain length of 45 amino acids, this lectin is a prime target to solution‐structure determination by NOESY NMR experiments in water. The NMR‐analysis was extended to assessment of the topology of the complex between pseudohevein and N,N′,N′′‐triacetylchitotriose. NOESY experiments in water solution provided 342 protein proton‐proton distance constraints. Binding of the ligand did not affect the pattern of the protein nuclear Overhauser effect signal noticeably, what would otherwise be indicative of a ligand‐induced conformational change. The average backbone (residues 3‐41) RMSD of the 20 refined structures was 1.14 Å, whereas the heavy atom RMSD was 2.18 Å. Two different orientations of the trisaccharide within the pseudohevein binding site are suggested, furnishing an explanation in structural terms for the lectins capacity to target chitin. In both cases, hydrogen bonds and van der Waals contacts confer stability to the complexes. This conclusion is corroborated by the thermodynamic parameters of binding determined by NMR and isothermal titration calorimetry. The association process was enthalpically driven. In relation to hevein, the Trp/Tyr‐substitution in the binding pocket has only a small effect on the free energy of binding in contrast to engineered galectin‐1 and a mammalian C‐type lectin. A comparison of the three‐dimensional structure of pseudohevein in solution to those reported for wheat germ agglutinin (WGA) in the solid state and for hevein and WGA‐B in solution has been performed, providing a data source about structural variability of the hevein domains. The experimentally derived structures and the values of the solvent accessibilities for several key residues have also been compared with conformations obtained by molecular dynamics simulations, pointing to the necessity to further refine the programs to enhance their predictive reliability and, thus, underscoring the importance of this kind of combined analysis in model systems. Proteins 2000;40:218–236.