J. Biochem. 118, 474-479 (1995)

Crystal Structure of the Unliganded Alkaline from Pseudomonas aeruginosa IF03O8O and Its Conformational Changes on Ligand Binding 1

Hideyuki Miyatake,* Yasuo Hata,*,2 Tomomi Fujii,* Kensaku Hamada,•õ

Kazuyuki Morihara,•ö and Yukiteru Katsube

* Institute for Chemical Research , Kyoto University, Uji, Kyoto 611; •õFaculty of Science, Shimane University, Matsue, Shimane 690; •öInstitute for Applied Life Science, University of East Asia, 2-1 Ichinomiya-Gakuen-cho, Shimonoseki, Yamaguchi 751; and Institute for Protein Research, Osaka University, Suita, Osaka 565

Received for publication, May 31, 1995

The crystal structure of the unliganded alkaline protease from Pseudomonas aeruginosa IF03080 has been determined at 2.0 A resolution by the X-ray method. The consists of N-terminal catalytic and C-terminal ƒÀ-helix domains. On structural comparison between the present unliganded enzyme and structurally-known liganded enzyme, some structural changes were observed around the . In the unliganded enzyme, Y216 serves as the fifth ligand for the active site zinc ion. On ligand binding, Y216 may move to form a hydrogen-bond with the carbonyl oxygen of the P1 residue of a ligand peptide. D191 in the flexible loop, Y190 to D196, over the active site cleft forms hydrogen-bonds with the backbone atoms of the P1 and P2 residues of the ligand to close the entrance of the cleft. The water molecule which is the fourth ligand for the zinc ion is replaced by the carbonyl oxygen of the P1 residue. These structural changes around the active site may reflect the substrate-binding mode during the enzymatic reaction.

Key words: alkaline protease, ƒÀ-helix, crystal structure, metalloprotease, Pseudomonas aeruginosa.

Pseudomonas aeruginosa is well known to secrete two alkaline protease. Neutral metalloproteases such as ther metalloproteases: a 50 kDa alkaline protease and a 33 kDa molysin and P. aeruginosa elastase, which are more com elastase, which belong to the serralysin and mon than alkaline metalloproteases, are typical "N(NH2) families, respectively. These were originally type" enzymes which hydrolyze the peptide bonds of isolated from P. aeruginosa IF03080 and IF03455, and certain amino acid residues on the amino group side through characterized as zinc-dependent metalloproteases by specific recognition. On the contrary, a P. aeruginosa Morihara et al. (1-3). In particular, the alkaline protease alkaline protease is a "C (COOH) -type" endoprotease which from P. aeruginosa has been paid much attention in the digests its substrate on the carboxyl side of certain amino medical field for the development of anti-infection agents acid residues (13), and has a hydrophobic (S2') against the bacterium (4). The enzyme is composed of a in addition to the major recognition site (S1) for Arg or polypeptide chain and one catalytic zinc atom, and requires probably Lys (14). Thus, the alkaline protease is an Arg (or several calcium ions for stabilization of its folding (2). The Lys)-specific C-type metalloprotease, although its sub amino acid sequences of 470 residues for P. aeruginosa strate specificity is relatively broad (12, 15). The patho alkaline have been deduced from the DNA logical aspects of the alkaline protease have been extensive sequences of its strains, IF03455 (5) and PA01 (6). The ly investigated already. The enzyme exhibits the potential crystal structure of the alkaline protease from P. aeru anti-coagulant capacity to hydrolyze natural substrates of ginosa was determined using a complex of the enzyme with plasmin, such as fibrin and fibrinogen, with similar specific a mixture of tetrapeptide products (7). The structure was activities to plasmin (13). From these properties of the used to determine the crystal structure of the 50 kDa alkaline protease from P. aeruginosa , it is inferred that the metalloprotease from Serratia marcescens (8), a member enzyme may play a key role in infection of host cells by the of the serralysin family (9-11). From biochemical studies bacterium through the inactivation of various physiological on the catalysis by a P. aeruginosa alkaline protease, it is activators such as some complement components , im well known that the potential catalytic capability of the munoglobulins A and G, and many protease inhibitors (16). enzyme is optimum in the alkaline pH range of 8 to 10 (1, In order to elucidate the characteristic mechanism of its 12). The alkaline optimum pH is characteristic of the catalysis, and to determine the reasons for its alkaline optimum pH and the C(COOH)-type recognition on sub 1 This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas (No. 06235204) from the Ministry of strate cleavage, we have determined the crystal structure Education, Science and Culture of Japan. of the alkaline protease from P. aeruginosa IF03080 by the 2 To whom correspondence should be addressed . Phone: +81-774-32 X-ray diffraction method . Here we report the three-dimen 3111 (Ext. 2159), Fax: +81-774-33-1247 sional structure of the unliganded alkaline protease from P.

474 J. Biochem. Unliganded Alkaline Protease Structure 475

aeruginosa determined at 2,0 A resolution, and the struc technique (19). The solvent-flattened map was good enough tural differences between the unliganded and liganded to follow the entire main-chain folding and most of the forms of the enzyme. side-chain orientations of the enzyme molecule. An initial A lyophilized sample of the alkaline protease from P. model of the enzyme was built with the program package, aeruginosa IFO3080, which was a gift from Nagase Bio TURBO-FRODO (20), on a computer graphics system, chemical, was subjected to crystallization without further IRIS Indigo Elan. The initial crystallographic R value was purification. Crystallization trials were conducted using the 43.5% for 16,182 independent reflections [I>3ƒÐ(I)] hanging drop vapor diffusion method. Small crystals were within the resolution of 10.0 to 2.8A. The model was obtained at 25•‹C using a precipitant solution of 6% (w/v) refined by means of simulated annealing techniques using polyethylene glycol 6000 (50mM acetate buffer including 1 the molecular dynamics program, X-PLOR (21, 22), in mM NaN3, pH 5.6). For the crystallization, the droplets stalled on a CRAY Y-MP2E/264 supercomputer. After were prepared by adding 2.5,ƒÊl of the precipitant solution several cycles of refinement, manual rebuilding of the to an equal volume of a 2% protein solution (pH 7.0) model was carried out using 2Fo -Fc and Fo -Fc maps. The including 5mM CaCl2 and 1mM NaN3. The macroseeding resolution was gradually increased to 2.0 A during iteration technique was effective for growing the small crystals to a of refinement and rebuilding. The total numbers of 335 size (1.0 x 0.6 x 0.3mm) large enough for X-ray experi water molecules and 8 calcium ions were included for the ments. The prismatic crystals have the orthorhombic space further refinement below 2.5 A resolution. The final model group, P2,212,, with cell dimensions of a=77.16, b= has an R factor of 19.8% for 33,698 independent reflections 176.69, and c=51.12 A. They contain four enzyme mole [I > 2ƒÐ (I), 80.0% complete] within the resolution of 8.0 to cules in the unit cell, and exhibit a Vm value of 3.56 A3/Da, 2.0 A. The root-mean-square (r.m.s) deviations from ideal which corresponds to the solvent content, 66%, in the bond distances and angles were 0.009 A and 1.58•‹, respec crystals. Three kinds of isomorphous heavy-atom deriva tively. The model exhibits a reasonable conformation with tive crystals were prepared by soaking the native crystals excellent geometry. The model was inspected with the in heavy-atom solutions of 15mM CH3HgC1, for 3 days, 2.5 program, PROCHECK (23). All residues were observed in mM HgCl2, for 1 day, and 1mM UO2(CH3COO)2, for 1 day, the allowed region of the Ramachandran plot, and 87.6% of respectively. the total residues lay in the most favored region. The All diffraction intensity data were collected using the coordinates of the final model will be deposited in the synchrotron radiation X-ray source (wavelength = 1.00 Brookhaven Protein Data Bank. A) at the Photon Factory of the National Laboratory for Stereoscopic drawings and a folding diagram of the High Energy Physics, Tsukuba, Japan. X-ray diffraction alkaline protease are shown in Figs. 1 and 2, respectively. patterns were recorded on Fuji-imaging plates with a The enzyme molecule has an elongated ellipsoidal shape screenless Weissenberg camera at the BL6A2 station. Two with approximate dimensions of 90 X 42 x 35 A. Its struc native crystals were mounted with the crystallographic b* ture comprises two structurally distinct domains: N and and c* axes parallel to the spindle axis, respectively, and C-terminal domains. the derivative crystals with the c* axis parallel to the Apart from the a -helix comprising the first 17 residues, spindle. The intensity data were processed with the pro which is associated with the C-terminal domain, the N gram, WEIS (17), and calculations of phases and electron terminal domain, residues 18-250, is a catalytic domain density maps were carried out with the program, PHASES with one zinc ion in the active center and shares a basically (18). The statistics for data collection and phase calculation common folding topology with other zinc metalloendopro are summarized in Table I. teases, as suggested by Blundell (24). The catalytic domain A 2.8 A resolution electron density map of the alkaline has an open-sandwich topology, as shown in Fig. 2, in which

protease calculated by the isomorphous replacement two ƒ¿-helices (ƒ¿5 and ƒ¿6) lie against a twisted mixed method was modified by means of the solvent-flattening ƒÀ-sheet of five strands (ƒÀ3-ƒÀ7). The ƒÀ-strands are ar

TABLE I. Statistics for data collection and phase calculation.

Statistics for phase calculation at 2.8 A resolution

R merge (%)=100ƒ°i|-I]/ƒ°I 6 is the average of I; over all observed symmetry equivalents. 6 Phasing power=/E; the mean amplitude of the heavy atom structure factor divided by the r.m.s. lack-of-closure error.

Vol. 118, No. 3, 1995.

a 476 H. Miyatake et al. ranged on the sheet in the order of ƒÀ4, ƒÀ3, ƒÀ5, ƒÀ7, and ƒÀ6, This helix contains a part of the consensus sequence motif, with strand ƒÀ6 antiparallel to the others. The large am H1766EXXHXXGXXH186,for zinc binding, in which three phipathic ƒ¿-helix, ƒ¿5, which is inserted between parallel histidine residues, H176, H180, and H186, represent three ƒÀ-strands ƒÀ3 and ƒÀ4 extends over the mixed ƒÀ-sheet . of the five zinc ligands. In the present unliganded enzyme, ƒ¿-helix ƒ¿6 comprising residues 169-182 , which is located Y216 following the conserved Met-turn represents a fifth near the active site zinc ion, is called the active site helix. zinc ligand (11), while in the liganded enzyme this residue

Fig. 1. (A) The stereoview of an a-carbon trace of the unliganded alkaline protease from Pseudo monas aeruginosa IFO3080. (B) A ribbon drawing of the alkaline

protease depicted with the pro gram, MOLSCRIPT (37). In the figure, ƒÀ-strands are denoted by arrows, the active site Zn2+ ion by a black ball, and Ca" ions by white balls.

J. Biochem. Unliganded Alkaline Protease Structure 477

moves away from the zinc ion to form a hydrogen bond with the carbonyl oxygen of the P1 residue in the ligand peptide (7, 8). Similar movement of a tyrosine residue near the active site zinc ion is observed in the case of Y248 of carboxypeptidase A on the binding of a peptide in the active site (25). In carboxypeptidase A, Y248 included in a flexible loop on the molecular surface is not a zinc ligand, unlike Y216 of the alkaline protease, but the phenolic oxygen of Y248 is in the vicinity of the active site, being about 7 A from the zinc ion. The crystal structures of various carboxypeptidase A-ligand complexes show that when each ligand binds to the enzyme, Y248 undergoes a substrate-induced conformational change resulting in movement of the phenolic hydroxyl group from the surface of the enzyme to within hydrogen-bonding distance of the nitrogen of the peptide bond to be cleaved. In arsanilazo tyrosine-248 carboxypeptidase A, it has been shown that Y248 interacts with the zinc ion in the native enzyme and moves away from the zinc ion to interact with the scissile bond on ligand binding (26). There is some controversy regarding the role of Y248 in the mechanism of action of carboxypeptidase A. However, it has been confirmed by a mutagenesis study that the phenolic hydroxyl group of Y248 in carboxypeptidase A does not act as a proton donor for the amino nitrogen in the cleaved peptide bond but participates in ligand binding (25). In the case of the alkaline protease, when a substrate is bound to the enzyme, the phenolic side chain of Y216, the fifth zinc ligand, moves by rotating around the Cƒ¿-CƒÀ bond to form a hydrogen bond with the bound substrate in a similar manner to that in the case of carboxypeptidase A. Thus, the conformation al change of Y216 in the alkaline protease may be important to stabilize the bound substrate, like that of Y248 in carboxypeptidase A, although the former movement is small in comparison with that of the latter. The electron density map around the active site zinc ion is shown in Fig. 3, and the geometry of the zinc ligation is shown in Table II. The four ligand residues and one ligand water are responsible for trigonal bipyramidal ligation with the zinc ion. In the zinc ligation, the imidazole nitrogens of H176 and H186, and the water oxygen are located on the

vertices of the triangular base plane which accommodates Fig. 2. A schematic representation of the secondary structure the active site zinc ion. The imidazole nitrogen of H180 and elements of the Pseudomonas aeruginosa alkaline protease. ƒ¿-Helices are depicted as black cylinders the phenolic oxygen of Y216 lie perpendicular to this , ƒÀ-strands as arrows, the active site zinc ion as an open circle, and calcium ions as filled circles. triangle and occupy the vertices of the bipyramid. A similar The conserved Met-turn is located around M214. The flexible loop ligation system is observed in the native conformations of region (Y190-D196) is represented by a striped coil. (27) and Serratia protease (8), but it differs from that of the liganded alkaline protease (7, 8). The water molecule which serves as a fourth zinc ligand forms a hydrogen bond with E177. The conserved E177 in the form of the enzyme, on the other hand, this loop closes the alkaline protease is the catalytic residue juxtaposed to the entrance to the active site so that N191 forms hydrogen

potential equivalent of E143 in thermolysin. Taking into bonds with the backbone atoms of the P1 and P2 residues of account the knowledge of the reaction mechanism of ther the ligand peptide (7, 8). These structural differences molysin (28, 29), this water molecule in the alkaline between the unliganded and liganded forms of the alkaline protease may serve as a nucleophile in cooperation with the protease may reflect conformational changes of the enzyme general base residue, E177. which occur during its enzymatic reaction, and partly reveal The loop region of Y190 to D196 (Y190-N-A-G-E-G the mode of binding between the enzyme and its substrate. D196) in the present unliganded alkaline protease exhibit The structures of metzincins (11) complexed with in ed unclear electron density. This loop located at the hibitors are now available at atomic resolution. The crystal entrance to the active site is entirely formed by atoms structures of the catalytic domains of human fibroblast which have high temperature factors. Therefore, the loop (30), neutrophil collagenase (31), and strome seems to have a highly flexible conformation to allow lysin-1 (32) in complexes with their substrate-analogous opening of the entrance to the active site. In the liganded inhibitors all reveal that their inhibitors bind to the active

Vol. 118, No. 3, 1995 478 H. Miyatake et al.

Fig. 3. A stereoview of the 2Fo-Fc map around the active site zinc ion with the superimposed skeletal model. The contours are drawn at 1.5ƒÐ level.

TABLE II. Geometries around the active site zinc ion. The sheets of the central ƒÀ-helix have only a slight twist, which is in striking contrast to parallel ƒÀ-sheets found in the parallel ƒ¿/ƒÀ domains of proteins. This flatness of the sheets in the ƒÀ-helix appears to be a consequence of the characteristic Ca2+ binding discussed below. This also is in contrast to the nature of the ƒÀ-sheets in the other regions of the ƒÀ-sandwich structure, which have more individual twists. Parallel ƒÀ-helix structures similar to the central ƒÀ-helix of the alkaline protease are also found in pectate C and E from Erwinia chrysanthemi (33, 34), and the tailspike protein of Salmonella typhimurium phage P22 (35). The parallel f3-helix structures in the pectate lyases sites in parallel with the ƒÀ-strand ƒÀ6 proximate to the and tailspike protein are all composed of three parallel active site ƒ¿-helix ƒ¿6 and form hydrogen bonds with the ƒÀ-sheets. In the pectate lyases, two ƒÀ-sheets form a ƒÀ-strand . On the other hand, in the case of the liganded ƒÀ-sandwich, the other one being positioned perpendicular to alkaline protease, a peptide ligand seems to bind to the the ƒÀ-sandwich. In the tailspike protein, two ƒÀ-sheets, A active site nearly perpendicular to the equivalent ƒÀ-strand. and C, form a ƒÀ-sandwich, and ƒÀ-sheets A and B enclose an These findings suggest that the substrate binding mode of angle of approximately 120•‹. The core of the ƒÀ-helices enzymes of the serratia family, possibly like the astacin contains stacks of polar side-chains in the pectate lyases, family, might be different from those of the other families, and hydrophobic residues in the tailspike protein. The i. e. replolysins and adamalysins, in the metzincin super ƒÀ-helix in the alkaline protease does not have the third family (9-11). More detailed information on conformation parallel ƒÀ-sheet that is found in the pectate lyases from E. al changes, obtained through crystallographic studies of chrysanthemi and the tailspike protein of S. typhimurium metzincins and/or their complexes with substrate-ana phage P22, but does have five internal Ca2+ ions (Ca3-Ca7) logues, is required for elucidation of the reaction mecha which may stabilize short turns between the successive nisms of these enzymes. ƒÀ-strands through a special Ca2+ -mediated coordination The C-terminal domain comprising residues 251-470 network (7, 8). with eight Ca2+ ions (Cal-Ca8) is a ƒÀ-strand-rich domain The functional significance of the C-terminal f3-helix formed by eighteen ƒÀ-strands, ƒÀ8 to ƒÀ25, which are domain in the alkaline protease of P. aeruginosa remains incorporated in a two-layer ƒÀ-sandwich structure. The unclear. It has been reported that proteases B and C from domain can be subdivided into three separate regions on the E. chrysanthemi, which are homologous with the alkaline basis of structural features. The amino-proximal region, protease, are synthesized as inactive higher molecular which includes five strands, ƒÀ8 to ƒÀ12, adopts a mixed weight precursors with N-terminal extensions which are

parallel/antiparallel ƒÀ-sheet structure with irregularly secreted into the external medium, where divalent cation wound loops. The region is connected to the catalytic mediated activation occurs (36). The alkaline protease of P. domain by a loop linking ƒ¿9 to ƒÀ8, and is in contact with the aeruginosa is known to be secreted as a zymogen which has catalytic domain through one external face of this region. nine N-terminal residues extended from the mature pro The C-terminal region consisting of strands ƒÀ20 to ƒÀ25 also tein. These proteins, however, do not require N-terminal has a mixed ƒÀ-sheet topology with rather irregular loop signal sequences for membrane translocation in secretion, connections between the successive strands. The central and therefore may share an independent secretion pathway region consisting of six strands, ƒÀ14 to ƒÀ19, has an unusual in which the premature polypeptide-foldings in their C parallel ƒÀ-helix structure formed by all the strands ar terminal domains are involved. The f3-helix of the alkaline ranged regularly in a right-handed spiral. The N-terminal protease is stabilized through Ca2+ binding to the special a -helix, a 1, is packed predominantly on one external face ized sites with the Ca2+ binding sequence motif, GGXGXD of the ƒÀ-sheet (ƒÀ15, ƒÀ17, ƒÀ19) of this region. In the ƒÀ-helix XBX (B: bulky hydrophobic residue, ideally leucine). structure, two parallelƒÀ-sheets, (ƒÀ15, ƒÀ17,ƒÀ19) and (ƒÀ14, These findings imply that the ƒÀ-helix structure in the C ƒÀ16, ƒÀ18), are packed together in an antiparallel manner. terminal domain of the P. aeruginosa alkaline protease

J. Biochem. Unliganded Alkaline Protease Structure 479 might play an important role in stable folding, or mature Bacterial Enzymes and Virulence (Holder, I.A., ed.) pp. 41-79, active folding, of the polypeptide induced by divalent cation CRC Press, Boca Raton 17. Higashi, T. (1989) The processing of diffraction data taken on a binding, most likely Ca2+ binding, after membrane trans screenless Weissenberg camera for macromolecular crystallogra location. phy. J. Appl. Crystallogr. 22, 9-18 18. Furey, W. and Swaminathan, S. (1990) PHASES-a program We are grateful to Nagase Biochemical Co., Ltd. for the kind gift of package for the processing and analysis of diffraction data from the alkaline protease from P. aeruginosa IFO3080. We are also macromolecules. Am. Crystallogr. Assoc. Annu. Mtg. Program deeply indebted to Drs. Noriyoshi Sakabe, Atsushi Nakagawa and Abst. 18, 73 Nobuhisa Watanabe, Photon Factory, National Laboratory of High 19. Wang, B.C. (1985) Resolution of phase ambiguity in macromo Energy Physics, for their kind support in the synchrotron experi lecular crystallography. Methods Enzymol. 115, 90-112 ments. The computation in this work was performed partly in the 20. Jones, T.A. (1978) A graphics model building and refinement Supercomputer Laboratory, Institute for Chemical Research, Kyoto system for macromolecules. J. Appl. Crystallogr. 15, 23-31 University, and partly in the Research Center for Protein Engineer 21. Brunger, A.T., Kuriyan, J., and Karplus, M. (1987).Crystallo ing, Institute for Protein Research, Osaka University. We would also graphic R factor refinement by molecular dynamics. Science 235, like to thank all the staff in the computation centers for their kind help 458-460 22. Brunger, AT. and Krukowski, A. (1990) Slow-cooling protocols in the computation. for crystallographic refinement by simulated annealing. Acta Cryst. A46, 585-593 REFERENCES 23. Laskowski, R.A., MacArthur, M.W., Moss, D.S., and Thornton, J.M. (1993) PROCHECK: A program to check the stereochemical 1. Morihara, K. (1963) Pseudomonas aeruginosa proteinase I. quality of protein structures. J. Appl. Crystallogr. 26, 283-291 Purification and general properties. Biochim. Biophys. Acta 73, 24. Blundell, T.L. (1994) Metalloproteinase superfamilies and drug 113-124 design. Nature Struct. Biol. 1, 73-75 2. Morihara, K. and Tsuzuki, H. (1964) Pseudomonas aeruginosa 25. Gardell, S.J., Craik, C.S., Hilvert, D., Urdea, M.S., and Rutter, peptidohydrolase III. Some characters as a Ca" -metalloenzyme. W.J. (1985) Site-directed mutagenesis shows that tyrosine 248 Biochim. Biophys. Acta 92, 351-360 of carboxypeptidase A does not play a crucial role in catalysis. 3. Morihara, K., Tsuzuki, H., Oka, T., Inoue, H., and Ebata, M. Nature 317, 551-555 (1969) Pseudomonas aeruginosa elastase: Isolation, crystalliza 26. Johansen, J.T. and Vallee, B.L. (1975) Environment and confor tion, and preliminary characterization. J. Biol. Chem. 240, 3295 mation dependent sensitivity of the arsanilazotyrosine-248 - 3304 carboxypeptidase A chromophore. Biochemistry 14, 649-660 4. Homma, J.Y. and Tanimoto, H. (1987) A multicomponent 27. Gomis-Ruth, F.X., Stocker, W., Huber, R., Zwilling, R., and Pseudomonas aeruginosa vaccine constituting of toxoids of Bode, W. (1993) Refined 1.8 A X-ray crystal structure of astacin, protease, elastase, exotoxin A and a common protective antigen a zinc- from the crayfish Astacus astacus L. (OEP): Application in patients with diffuse panbronchiolitis. Structure determination, refinement, molecular structure and Antibiot. Chemother. 39, 215-221 comparison with thermolysin. J. Mol. Biol. 229, 945-968 5. Okuda, K., Morihara, K., Atsumi, Y., Takeuchi, H., Kawamoto, 28. Kester, W.R. and Matthews, B.W. (1977) Crystallographic study S., Kawasaki, H., Suzuki, K., and Fukushima, J. (1990) Com of the binding of dipeptide inhibitors to thermolysin: Implica plete nucleotide sequence of the structural gene for alkaline tions for mechanism of catalysis. Biochemistry 16, 2506-2516 proteinase from Pseudomonas aeruginosa IFO3455. Infect. 29. Holmes, M.A. and Matthews, B.W. (1981) Binding of hydrox Immun. 58, 4083-4088 amic acid inhibitors to crystalline thermolysin suggests a penta 6. Doung, F., Lazdunski, A., Cami, B., and Murgier, M. (1992) coordinate zinc intermediate in catalysis. Biochemistry 20, 6912 Sequence of a cluster of genes controlling synthesis and secretion - 6920 of alkaline protease in Pseudomonas aeruginosa. Relationship to 30. Borkakoti, N., Winkler, F.K., Williams, D.H., D'Arcy, A., other secretory pathways. Gene 121, 47-54 Broadhurst, M.J., Brown, P.A., Johnson, W.H., and Murray, 7. Baumann, U., Wu, S., Flaherty, K.M., and McKay, D.B. (1993) E.J. (1994) Structure of the catalytic domain of human fibroblast Three-dimensional structure of the alkaline protease of Pseudo collagenase complexed with an inhibitor. Nature Struct. Biol. 1, monas aeruginosa: A two-domain protein with a calcium binding 106-110 parallel beta roll motif. EMBO J. 12,,,,,,,,,3357-33648 31. Status, T., Spurlino, J.C., Smith, D.L., Wahl, R.C., Ho, T.F., . Baumann, U. (1994) Crystal structure of the 50 kDa metallo Qoronfleh, M.W., Banks, T.M., and Rubin, B. (1994) Structure of protease from Serratia marcescens. J. Mol. Biol. 242, 244-251 human neutrophil collagenase reveals large Sl' specificity pocket. 9. Jiang, W. and Bond, J.S. (1992) Families of metallendopep Nature Struct. Biol. 1, 119-123 tidases and their relationships. FEBS Lett. 312, 110-114 32. Gooley, P.R., O'Connell, J.F., Marcy, A.I., Cuca, G.C., Salowe, 10. Hooper, N.M. (1994) Families of zinc metalloproteases. FEBS S.P., Bush, B.L., Hermes, J.D., Esser, C.K., Hagmann, W.K., Lett. 354, 1-6 Springer, J.P., and Johnson, B.A. (1994) The NMR structure of 11. Bode, W., Gomis-Ruth, F.-X., and Stocker, W. (1993) , the inhibited catalytic domain of human stromelysin-1. Nature serralysins, snake venom and matrix metalloproteinases exhibit Struct. Biol. 1, 111-118 identical zinc-binding environments (HEXXHXXGXXH and 33. Yoder, M., Keen, N.T., and Jurnak, F. (1993) New domain Met-turn) and topologies and should be grouped into a common motif: The structure of pectate C, a secreted plant virulent family, the 'metzincins.' FEBS Lett. 331, 134-140 factor. Science 260, 1503-1507 12. Morihara, K., Tsuzuki, H., and Oka, T. (1973) On the specificity 34. Yoder, M., Lietzke, S.E., and Jurnak, F. (1993) Unusual struc of Pseudomonas aeruginosa alkaline proteinase with synthetic tural features in the parallel f3-helix in pectate lyase. Structure 1, peptides. Biochim. Biophys. Acta 309, 414-429 241-251 13. Shibuya, Y., Yamamoto, T., Morimoto, T., Nishino, N., Kam 35. Steinbacher, S., Seckler, R., Miller, S., Steipe, B., Huber, R., and bara, T., and Okabe, H. (1991) Pseudomonas aeruginosa alkaline Reinemer, P. (1994) Crystal structure of P22 tailspike protein: proteinase might share a biological function with plasmin. Interdigitated subunits in a thermostable trimer. Science 265, Biochim. Biophys. Acta 1077, 316-324 383-386 14. Morimoto, T., Nishino, N., Fujimoto, T., and Yamamoto, T. 36. Delepelaire, P. and Wandersman, C. (1989) Protease secretion (1990) Substrate specificity and inhibition of Pseudomonas by Erwinia chrysanthemi. J. Biol. Chem. 264, 9083-9089 aeruginosa alkaline protease. Pept. Chem. 1989, 387-390 37. Kraulis, P.J. (1991) MOLSCRIPT: A program to produce both 15. Morihara, K. and Tsuzuki, H. (1966) Substrate specificity of detailed and schematic plots ofoprotein structures. J. Appl, elastolytic and nonelastolytic proteinases from Pseudomonas Crystallogr. 24, 946-950 aeruginosa. Arch. Biochem. Biophys. 114, 158-165 16. Morihara, K. and Homma, J.Y. (1985) Pseudomonas proteases in

Vol. 118, No. 3, 1995