Search

Orthorhombic crystals diffracting beyond 1.7 A resolution, have been grown from the stoichiometric complex formed between human leukocyte elastase (HLE) and the third domain of turkey ovomucoid inhibitor (OMTKY3). The crystal and molecular structure has been determined with the multiple isomorphous replacement technique. The complex has been modeled using the known structure of OMTKY3 and partial sequence information for HLE, and has been refined. The current crystallographic R‐value is 0.21 for reflections from 25 to 1.8 A resolution. HLE shows the characteristic polypeptide fold of trypsin‐like serine proteinases and consists of 218 amino acid residues. However, several loop segments, mainly arranged around the substrate binding site, have unique conformations. The largest deviations from the other vertebrate proteinases of known spatial structure are around Cys168. The specificity pocket is constricted by Val190, Val216 and Asp226 to preferentially accommodate medium sized hydrophobic amino acids at P1. Seven residues of the OMTKY3‐binding segment are in specific contact with HLE. This interaction and geometry around the reactive site are similar as observed in other complexes. It is the first serine proteinase glycoprotein analysed, having two sugar chains attached to Asn159 and to residue 109.

Orthorhombic crystals of the complex formed between bovine alpha‐chymotrypsin and a recombinant human mucous proteinase inhibitor (SLPI) were grown. Data to 2.3 A resolution were collected on the area‐detector diffractometer FAST. The crystal structure of the complex was solved by Patterson search techniques using chymotrypsin as a search model. A cyclic procedure of modeling and crystallographic refinement enabled the determination of the SLPI structure. The current crystallographic R‐value is 0.19. SLPI has a boomerang‐like shape with both wings comprising two well separated domains of similar architecture. In each domain the polypeptide chain is arranged like a stretched spiral. Two internal strands form a regular beta‐hairpin loop which is accompanied by two external strands linked by the proteinase binding segment. The polypeptide segment of each domain is interconnected by four disulfide bridges with a connectivity pattern hitherto unobserved. The reactive site loop of the second domain has elastase and chymotrypsin binding properties. It contains the scissile peptide bond between Leu72I and Met73I and has a similar conformation to that observed in other serine proteinase protein inhibitors. Eight residues of this loop, two of the adjacent hairpin loop, the C‐terminal segment and Trp30I are in direct contact with the cognate enzyme. The binding loop of the first domain (probably with anti‐trypsin activity) is disordered due to proteolytic cleavage occurring in the course of crystallization.

A stoichiometric complex formed between human alpha‐thrombin and D‐Phe‐Pro‐Arg chloromethylketone was crystallized in an orthorhombic crystal form. Orientation and position of a starting model derived from homologous modelling were determined by Patterson search methods. The thrombin model was completed in a cyclic modelling‐crystallographic refinement procedure to a final R‐value of 0.171 for X‐ray data to 1.92 A. The structure is in full agreement with published cDNA sequence data. The A‐chain, ordered only in its central part, is positioned along the molecular surface opposite to the active site. The B‐chain exhibits the characteristic polypeptide fold of trypsin‐like proteinases. Several extended insertions form, however, large protuberances; most important for interaction with macromolecular substrates is the characteristic thrombin loop around Tyr60A‐Pro60B‐Pro60C‐Trp60D (chymotrypsinogen numbering) and the enlarged loop around the unique Trp148. The former considerably restricts the active site cleft and seems likely to be responsible for poor binding of most natural proteinase inhibitors to thrombin. The exceptional specificity of D‐Phe‐Pro‐Arg chloromethylketone can be explained by a hydrophobic cage formed by Ile174, Trp215, Leu99, His57, Tyr60A and Trp60D. The narrow active site cleft, with a more polar base and hydrophobic rims, extends towards the arginine‐rich surface of loop Lys70‐Glu80 that probably represents part of the anionic binding region for hirudin and fibrinogen.

Ornithodorin, isolated from the blood sucking soft tick Ornithodoros moubata, is a potent (Ki = 10(‐12) M) and highly selective thrombin inhibitor. Internal sequence homology indicates a two domain protein. Each domain resembles the Kunitz inhibitor basic pancreatic trypsin inhibitor (BPTI) and also the tick anticoagulant peptide (TAP) isolated from the same organism. The 3.1 A crystal structure of the ornithodorin‐thrombin complex confirms that both domains of ornithodorin exhibit a distorted BPTI‐like fold. The N‐terminal portion and the C‐terminal helix of each domain are structurally very similar to BPTI, whereas the regions corresponding to the binding loop of BPTI adopt different conformations. Neither of the two ‘reactive site loops’ of ornithodorin contacts the protease in the ornithodorin‐thrombin complex. Instead, the N‐terminal residues of ornithodorin bind to the active site of thrombin, reminiscent of the thrombin‐hirudin interaction. The C‐terminal domain binds at the fibrinogen recognition exosite. Molecular recognition of its target protease by this double‐headed Kunitz‐type inhibitor diverges considerably from other members of this intensely studied superfamily. The complex structure provides a model to explain the perplexing results of mutagenesis studies on the TAP‐factor Xa interaction.

Rhodniin is a highly specific inhibitor of thrombin isolated from the assassin bug Rhodnius prolixus. The 2.6 Angstrum crystal structure of the non‐covalent complex between recombinant rhodniin and bovine alpha‐thrombin reveals that the two Kazal‐type domains of rhodniin bind to different sites of thrombin. The amino‐terminal domain binds in a substrate‐like manner to the narrow active‐site cleft of thrombin; the imidazole group of the P1 His residue extends into the S1 pocket to form favourable hydrogen/ionic bonds with Asp189 at its bottom, and additionally with Glu192 at its entrance. The carboxy‐terminal domain, whose distorted reactive‐site loop cannot adopt the canonical conformation, docks to the fibrinogen recognition exosite via extensive electrostatic interactions. The rather acidic polypeptide linking the two domains is displaced from the thrombin surface, with none of its residues involved in direct salt bridges with thrombin. The tight (Ki = 2 × 10(‐13) M) binding of rhodniin to thrombin is the result of the sum of steric and charge complementarity of the amino‐terminal domain towards the active‐site cleft, and of the electrostatic interactions between the carboxy‐terminal domain and the exosite.

The crystal structure of human neutrophil cathepsin G, complexed with the peptidyl phosphonate inhibitor Suc‐Val‐Pro‐PheP‐(OPh)2, has been determined to a resolution of 1.8 A using Patterson search techniques. The cathepsin G structure shows the polypeptide fold characteristic of trypsin‐like serine proteinases and is especially similar to rat mast cell proteinase II. Unique to cathepsin G, however, is the presence of Glu226 (chymotrypsinogen numbering), which is situated at the bottom of the S1 specificity pocket, dividing it into two compartments. For this reason, the benzyl side chain of the inhibitor PheP residue does not fully occupy the pocket but is, instead, located at its entrance. Its positively charged equatorial edge is involved in a favourable electrostatic interaction with the negatively charged carboxylate group of Glu226. Arrangement of this Glu226 carboxylate would also allow accommodation of a Lys side chain in this S1 pocket, in agreement with the recently observed cathepsin G preference for Lys and Phe at P1. The cathepsin G complex with the covalently bound phosphonate inhibitor mimics a tetrahedral substrate intermediate. A comparison of the Arg surface distributions of cathepsin G, leukocyte elastase and rat mast cell protease II shows no simple common recognition pattern for a mannose‐6‐phosphate receptor‐independent targeting mechanism for sorting of these granular proteinases.

A stoichiometric complex of human stefin B and carboxymethylated papain has been crystallized in a trigonal crystal form. Data to 2.37 A resolution were collected using the area detector diffractometer FAST. The crystal structure of the complex has been solved by Patterson search techniques using papain as search model. Starting from the structure of chicken cystatin, the stefin structure was elucidated through cycles of model building and crystallographic refinement. The current crystallographic R factor is 0.19. Like cystatin, the stefin molecule consists of a five stranded beta‐sheet wrapped around a five turn alpha‐helix, but with an additional carboxy terminal strand running along the convex side of the sheet. Topological equivalence of stefin and cystatin reveal the previous sequence alignment to be incorrect in part, through deletion of the intermediate helix. The conserved residues form a tripartite wedge, which slots into the papain active site as proposed through consideration of the tertiary structures of the individual components (Bode et al., 1988). The main interactions are provided by the amino terminal ‘trunk’ (occupying the ‘unprimed’ subsites of the enzyme), and by the first hairpin loop, containing the highly conserved QVVAG sequence, with minor contributions from the second hairpin loop. The carboxyl terminus of stefin provides an additional interaction region with respect to cystatin. The interaction is dominated by hydrophobic contacts. Inhibition by the cysteine proteinase inhibitors is fundamentally different to that observed for the serine proteinase inhibitors.

Matrix metalloproteinases are a family of zinc endopeptidases involved in tissue remodelling. They have been implicated in various disease processes including tumour invasion and joint destruction. These enzymes consist of several domains, which are responsible for latency, catalysis and substrate recognition. Human neutrophil collagenase (PMNL‐CL, MMP‐8) represents one of the two ‘interstitial’ collagenases that cleave triple helical collagens types I, II and III. Its 163 residue catalytic domain (Met80 to Gly242) has been expressed in Escherichia coli and crystallized as a non‐covalent complex with the inhibitor Pro‐Leu‐Gly‐hydroxylamine. The 2.0 A crystal structure reveals a spherical molecule with a shallow active‐site cleft separating a smaller C‐terminal subdomain from a bigger N‐terminal domain, composed of a five‐stranded beta‐sheet, two alpha‐helices, and bridging loops. The inhibitor mimics the unprimed (P1‐P3) residues of a substrate; primed (P1′‐P3′) peptide substrate residues should bind in an extended conformation, with the bulky P1′ side‐chain fitting into the deep hydrophobic S1′ subsite. Modelling experiments with collagen show that the scissile strand of triple‐helical collagen must be freed to fit the subsites. The catalytic zinc ion is situated at the bottom of the active‐site cleft and is penta‐coordinated by three histidines and by both hydroxamic acid oxygens of the inhibitor. In addition to the catalytic zinc, the catalytic domain harbours a second, non‐exchangeable zinc ion and two calcium ions, which are packed against the top of the beta‐sheet and presumably function to stabilize the catalytic domain.(ABSTRACT TRUNCATED AT 250 WORDS)

Adamalysin II, a 24 kDa zinc endopeptidase from the snake venom of Crotalus adamanteus, is a member of a large family of metalloproteinases isolated as small proteinases or proteolytic domains of mosaic haemorrhagic proteins from various snake venoms. Homologous domains have recently been detected in multimodular mammalian reproductive tract proteins. The 2.0 A crystal structure of adamalysin II reveals an ellipsoidal molecule with a shallow active‐site cleft separating a relatively irregularly folded subdomain from the calcium‐binding main molecular body composed of a five‐stranded beta‐sheet and four alpha‐helices. The folding of the peptide fragment containing the zinc‐binding motif HExxHxxGxxH bears only a distant resemblance to thermolysin, but is identical to that found in astacin, with the three histidines and a water molecule (linked to the glutamic acid) likewise constituting the zinc ligand; adamalysin II lacks a fifth (tyrosine) zinc ligand, however, leaving its zinc ion tetrahedrally co‐ordinated. Furthermore, adamalysin II and astacin share an identical active‐site basement formed by a common Metturn. Due to their virtually identical active‐site environment and similar folding topology, the snake venom metalloproteinases (hitherto called adamalysins) and the astacins (and presumably also the matrix metalloproteinases/mammalian collagenases and the Serratia proteinase‐like large bacterial proteinases) might be grouped into a common superfamily with distinct differences from the thermolysin family.

The crystal structure of chicken egg white cystatin has been solved by X‐ray diffraction methods using the multiple isomorphous replacement technique. Its structure has been refined to a crystallographic R value of 0.19 using X‐ray data between 6 and 2.0A. The molecule consists mainly of a straight five‐turn alpha‐helix, a five‐stranded antiparallel beta‐pleated sheet which is twisted and wrapped around the alpha‐helix and an appending segment of partially alpha‐helical geometry. The ‘highly conserved’ region from Gln53I to Gly57I implicated with binding to cysteine proteinases folds into a tight beta‐hairpin loop which on opposite sides is flanked by the amino‐terminal segment and by a second hairpin loop made up of the similarly conserved segment Pro103I ‐ Trp104I. These loops and the amino‐terminal Gly9I ‐ Ala10I form a wedge‐shaped ‘edge’ which is quite complementary to the ‘active site cleft’ of papain. Docking experiments suggest a unique model for the interaction of cystatin and papain: according to it both hairpin loops of cystatin make major binding interactions with the highly conserved residues Gly23, Gln19, Trp177 and Ala136 of papain in the neighbourhood of the reactive site Cys25; the amino‐terminal segment Gly9I ‐ Ala10I of bound cystatin is directed towards the substrate subsite S2, but in an inappropriate conformation and too far away to be attacked by the reactive site Cys25. As a consequence, the mechanism of the interaction between cysteine proteinases and their cystatin‐like inhibitors seems to be fundamentally different from the ‘standard mechanism’ defined for serine proteinases and most of their protein inhibitors.

The metalloexozymogen procarboxypeptidase A is mainly secreted in ruminants as a ternary complex with zymogens of two serine endoproteinases, chymotrypsinogen C and proproteinase E. The bovine complex has been crystallized, and its molecular structure analysed and refined at 2.6 A resolution to an R factor of 0.198. In this heterotrimer, the activation segment of procarboxypeptidase A essentially clamps the other two subunits, which shield the activation sites of the former from tryptic attack. In contrast, the propeptides of both serine proproteinases are freely accessible to trypsin. This arrangement explains the sequential and delayed activation of the constituent zymogens. Procarboxypeptidase A is virtually identical to the homologous monomeric porcine form. Chymotrypsinogen C displays structural features characteristic for chymotrypsins as well as elastases, except for its activation domain; similar to bovine chymotrypsinogen A, its binding site is not properly formed, while its surface located activation segment is disordered. The proproteinase E structure is fully ordered and strikingly similar to active porcine elastase; its specificity pocket is occluded, while the activation segment is fixed to the molecular surface. This first structure of a native zymogen from the proteinase E/elastase family does not fundamentally differ from the serine proproteinases known so far.

The crystal structure of the complex formed between eglin c, an elastase inhibitor from the medical leech, and subtilisin Carlsberg has been determined at 1.2 A resolution by a combination of Patterson search methods and isomorphous replacement techniques. The structure has been refined to a crystallographic R‐value of 0.18 (8‐1.2 A). Eglin consists of a four‐stranded beta‐sheet with an alpha‐helical segment and the protease‐binding loop fixed on opposite sides. This loop, which contains the reactive site Leu45I‐‐Asp46I, is mainly held in its conformation by unique electrostatic/hydrogen bond interactions of Thr44I and Asp46I with the side chains of Arg53I and Arg51I which protrude from the hydrophobic core of the molecule. The conformation around the reactive site is similar to that found in other proteinase inhibitors. The nine residues of the binding loop Gly40I‐‐Arg48I are involved in direct contacts with subtilisin. In this interaction, eglin segment Pro42I‐‐Thr44I forms a three‐stranded anti‐parallel beta‐sheet with subtilisin segments Gly100‐‐Gly102 and Ser125‐‐Gly127. The reactive site peptide bond of eglin is intact, and Ser221 OG of the enzyme is 2.81 A apart from the carbonyl carbon.

The structure of the Gla‐domainless form of the human anticoagulant enzyme activated protein C has been solved at 2.8 A resolution. The light chain is composed of two domains: an epidermal growth factor (EGF)‐like domain modified by a large insert containing an additional disulfide, followed by a typical EGF‐like domain. The arrangement of the long axis of these domains describes an angle of approximately 80 degrees. Disulfide linked to the light chain is the catalytic domain, which is generally trypsin‐like but contains a large insertion loop at the edge of the active site, a third helical segment, a prominent cationic patch analogous to the anion binding exosite I of thrombin and a trypsin‐like Ca[II] binding site. The arrangement of loops around the active site partially restricts access to the cleft. The S2 and S4 subsites are much more polar than in factor Xa and thrombin, and the S2 site is unrestricted. While quite open and exposed, the active site contains a prominent groove, the surface of which is very polar with evidence for binding sites on the primed side, in addition to those typical of the trypsin class found on the non‐primed side.

The clotting cascade system of the horseshoe crab (Limulus) is involved in both haemostasis and host defence. The cascade results in the conversion of coagulogen, a soluble protein, into an insoluble coagulin gel. The clotting enzyme excises the fragment peptide C from coagulogen, giving rise to aggregation of the monomers. The crystal structure of coagulogen reveals an elongated molecule that embraces the helical peptide C fragment. Cleavage and removal of the peptide C would expose an extended hydrophobic cove, which could interact with the hydrophobic edge of a second molecule, leading to a polymeric fibre. The C‐terminal half of the coagulogen molecule exhibits a striking topological similarity to the neurotrophin nerve growth factor (NGF), providing the first evidence for a neurotrophin fold in invertebrates. Similarities between coagulogen and Spatzle, the Drosophila ligand of the receptor Toll, suggest that the neurotrophin fold might be considered more ancient and widespread than previously realized.

Thrombin is a serine protease that plays a central role in blood coagulation. It is inhibited by hirudin, a polypeptide of 65 amino acids, through the formation of a tight, noncovalent complex. Tetragonal crystals of the complex formed between human alpha‐thrombin and recombinant hirudin (variant 1) have been grown and the crystal structure of this complex has been determined to a resolution of 2.95 A. This structure shows that hirudin inhibits thrombin by a previously unobserved mechanism. In contrast to other inhibitors of serine proteases, the specificity of hirudin is not due to interaction with the primary specificity pocket of thrombin, but rather through binding at sites both close to and distant from the active site. The carboxyl tail of hirudin (residues 48‐65) wraps around thrombin along the putative fibrinogen secondary binding site. This long groove extends from the active site cleft and is flanked by the thrombin loops 35‐39 and 70‐80. Hirudin makes a number of ionic and hydrophobic interactions with thrombin in this area. Furthermore hirudin binds with its N‐terminal three residues Val, Val, Tyr to the thrombin active site cleft. Val1 occupies the position P2 and Tyr3 approximately the position P3 of the synthetic inhibitor D‐Phe‐Pro‐ArgCH2Cl. Thus the hirudin polypeptide chain runs in a direction opposite to that expected for fibrinogen and that observed for the substrate‐like inhibitor D‐Phe‐Pro‐ArgCH2Cl.